WO2020118517A1

WO2020118517A1 - Method for establishing and issuing evaluation indicator system for traffic management and control service indexes

Info

Publication number: WO2020118517A1
Application number: PCT/CN2018/120284
Authority: WO
Inventors: 关金平; 关志超; 须成忠
Original assignee: 深圳先进技术研究院
Priority date: 2018-12-11
Filing date: 2018-12-11
Publication date: 2020-06-18

Abstract

Disclosed is a method for establishing and issuing the evaluation indicator system for traffic management and control service indexes. The method comprises: step a: establishing an indicator system for traffic management and control service indexes of the urban traffic system according to selected traffic management service indexes and traffic control service indexes (100); step b: performing modeling according to the correlation and evaluation of indicators for traffic management and control service indexes (300); step c: matching the established evaluation indicator system for traffic management and control service indexes with data in a traffic big data cloud computing platform (500); step d: matching the evaluation indicators for traffic management and control service indexes with corresponding cities to form a big data evaluation system for traffic management and control service indexes (600). The method can promote the rapid development of urban traffic and social economy, optimize and improve the capacity of urban traffic capacity and the living standard of residents, enables scientific decision-making on the construction and development goal of urban traffic system, and ensures more accurate and efficient development strategies for urban traffic.

Description

交通管理与控制服务指数的评价指标体系建立及发布方法Establishment and release method of evaluation index system of traffic management and control service index

技术领域Technical field

本申请属于智能公共交通技术领域，特别涉及一种交通管理与控制服务指数的评价指标体系建立及发布方法。This application belongs to the field of intelligent public transportation technology, and in particular relates to a method for establishing and publishing an evaluation index system of a traffic management and control service index.

背景技术Background technique

当前，中国已进入新型城镇化建设与快速发展时期，党的“十九大”明确提出“交通强国建设”、“交通运输引领支撑社会经济发展”战略，提出构建现代综合交通运输体系的目标，将推广智能化技术在交通运输行业的应用作为一项重点工作内容。伴随着快速城市化发展进程，实施新型城镇化建设，从供给侧进行结构性改革，需求侧管理，着力降成本、补短板、高品质、强服务，更好地服务保障社会经济平稳健康发展。城市在转型，城市交通也在转型，交通***正处在由过去历史的、静态的小数据分析模式向实时的、动态的大数据研判模式的转变。At present, China has entered a period of new-type urbanization and rapid development. The party's "19th National Congress" clearly put forward the strategies of "building a powerful transportation country" and "leading transportation to support social and economic development", and proposed the goal of building a modern comprehensive transportation system. The promotion of the application of intelligent technology in the transportation industry is a key task. With the rapid urbanization development process, the implementation of new urbanization construction, structural reform from the supply side, demand-side management, focus on reducing costs, complementing shortcomings, high-quality, strong services, better services to ensure the stable and healthy development of social economy . The city is transforming, urban transportation is also transforming, and the transportation system is changing from the historical and static small data analysis mode to the real-time and dynamic big data research and judgment mode.

近年来，我国城市的快速城市化建设进程和城市机动化速度不断加快，对交通***需求迅猛增长，交通***问题也日益突出，成为城市社会经济发展的瓶颈问题。如何解决困扰城市发展的交通***问题，是摆在我国城市发展面前的严峻课题。In recent years, the rapid urbanization process of cities in my country and the speed of urban motorization are accelerating. The demand for the transportation system is growing rapidly, and the problems of the transportation system are becoming increasingly prominent. How to solve the traffic system problem that plagues the development of cities is a stern subject facing the development of cities in my country.

交通***管理与控制服务指数评价体系既是交通***管理与控制的评价体系，也是交通***治理处置引导体系。作为交通***服务指数体系，能够衡量一个城市不同时期交通管理与控制水平的变化，也能够评价同一时期不同城市交通管理与控制水平的差异。此外，以交通***管理与控制服务指数分析城市交通问题的关键症结和严重程度，以提高对症下药的治理方案科学性。作为交通***管理与控制的服务指数引导体系，它帮助交通规划、建设、管理与控制等相关政府部门建立***工程的方法，以及解决城市交通***问题的总体思路，给出解决城市交通问题的总体框架和城市交通管理与控制的发展前景，引导城市交通管理的科学化、现代化、国际化、一体化发展进程，进而引导交通***可持续发展的交通管理与控制服务体系建立。The traffic system management and control service index evaluation system is not only the evaluation system of traffic system management and control, but also the guidance system of traffic system governance and disposal. As a transportation system service index system, it can measure the changes of traffic management and control levels of a city in different periods, and can also evaluate the differences in traffic management and control levels of different cities in the same period. In addition, the transportation system management and control service index is used to analyze the key crux and severity of urban traffic problems, so as to improve the scientificity of the appropriate treatment plan. As a service index guidance system for transportation system management and control, it helps transportation planning, construction, management and control and other relevant government departments to establish system engineering methods, as well as general ideas for solving urban transportation system problems, and gives an overall solution to urban transportation problems. The framework and the development prospects of urban traffic management and control will guide the scientific, modern, international and integrated development of urban traffic management, and then guide the establishment of a sustainable traffic management and control service system for the transportation system.

交通管理与控制学科是随车辆与道路交通而生，由交通管理与交通控制两个部分构成；交通管理是对道路上的行车、停车、行人、道路使用，执行交通法规的“执法管理”，用交通工程技术措施对交通运行状况进行改善的“交通治理”统称；交通控制是依靠交通警察或采用交通信号控制设施，随交通变化特性来指挥车辆、行人的通行。从宏观上来说，交通控制是交通管理的某一表现方式，因此在现代交通管理中，交通管理与交通控制是一个有机结合的整体。交通管理与控制措施，按其是否具有法律意义，在性质上可分为两类：具有法律意义且必须强制执行的管理措施，是指在交通法规中规定的，为维护交通秩序，保障交通安全所必须的基本交通规则；用来改善交通状况的工程技术措施，这些措施本身不具有法律意义，但要使这些措施能得到有效实施，还需依靠具有法律意义的管理措施来强制执行，或依靠经济手段来诱导执行。The discipline of traffic management and control is born with vehicles and road traffic, and consists of two parts: traffic management and traffic control; traffic management is the "enforcement management" of traffic laws, parking, pedestrians, and road use, and the implementation of traffic regulations. "Traffic governance" that uses traffic engineering technical measures to improve traffic operation conditions is collectively referred to; traffic control is to rely on traffic police or use traffic signal control facilities to direct the passage of vehicles and pedestrians with the changing characteristics of traffic. From a macro perspective, traffic control is a certain expression of traffic management. Therefore, in modern traffic management, traffic management and traffic control are an organically integrated whole. Traffic management and control measures, according to whether they have legal significance, can be divided into two categories in nature: management measures that have legal significance and must be enforced refer to those prescribed in traffic regulations to maintain traffic order and ensure traffic safety Necessary basic traffic rules; engineering and technical measures used to improve traffic conditions. These measures themselves have no legal significance, but for these measures to be effectively implemented, they must also be enforced by legally meaningful management measures, or rely on Economic means to induce execution.

随着社会及小汽车工业的发展，交通管理与控制的目的也在不断变化。初期的交通管理是最基本的交通要求，即保障交通安全。随着车辆保有量的增加，道路上出现了车辆拥挤、阻塞现象，在保障交通安全的基础上，还要求交通管理与控制达到疏导交通、保障交通畅通。在采取了疏导交通之后，车辆依然在不断地增长，交通拥挤、交通安全、交通污染现象日趋严重；道路交通工程设施的建设速度总是跟不上车辆增长的速度，现有的道路交通设施的交通效率总是有限。因此，近年来在交通管理与控制领域产生了一种新思路，即通过采用交通需求管理的方法，来控制道路上的汽车交通量需求。现代化交通管理与控制服务指数的目的是除了为保障交通安全、疏导交通、提高现有交通设施的通车效率的传统目的外，着重采取各种交通需求管理措施来减少道路上的车辆交通总量、缓解交通拥堵、保障交通安全与畅通，并降低汽车对交通环境的污染，实现采用数字量化衡量交通管理与控制服务水平以及服务模式的趋势。With the development of society and the car industry, the purpose of traffic management and control is constantly changing. The initial traffic management is the most basic traffic requirement, which is to ensure traffic safety. With the increase in the number of vehicles, vehicle congestion and congestion have appeared on the roads. On the basis of ensuring traffic safety, traffic management and control are also required to facilitate traffic and ensure smooth traffic. After adopting traffic diversion, vehicles are still growing, traffic congestion, traffic safety, and traffic pollution are becoming more and more serious; the construction speed of road traffic engineering facilities is always unable to keep up with the growth rate of vehicles. Traffic efficiency is always limited. Therefore, in recent years, a new idea has emerged in the field of traffic management and control, that is, to adopt the method of traffic demand management to control the demand for automobile traffic on the road. The purpose of the modern traffic management and control service index is to focus on adopting various traffic demand management measures to reduce the total amount of vehicle traffic on the road, in addition to the traditional purposes of ensuring traffic safety, channeling traffic, and improving the efficiency of existing traffic facilities. Alleviate traffic congestion, ensure traffic safety and smoothness, and reduce the pollution of cars to the traffic environment, and realize the trend of using digital quantitative measurement of traffic management and control service levels and service models.

交通管理与控制是一个新型的学科领域，包括交通管理、交通控制、交通诱导、交通事故、交通教育、交通执法、交通信息工程、交通大数据云计算、人工智能、交通指挥机器人、交通云机器人、深度学***等诸多方面得到反映。此外，城市交通结构的合理化，城市土地利用的调整以及管理体制、交通投资、交通规划的制定与措施均是解决交通***问题不可缺少的重要内容。在评价交通管理与控制时，必须采取多目标原则，对影响道路交通管理水平的各个方面进行定量计算和定性分析，确定评价标准和方法，综合评价整个交通管理与控制的总体水平和能力。所以，选取交通管理与控制服务指数指标体系时，既要注重交通管理效果评价，也要注重管理手段、管理过程的科学指导。Traffic management and control is a new discipline, including traffic management, traffic control, traffic guidance, traffic accidents, traffic education, traffic law enforcement, traffic information engineering, traffic big data cloud computing, artificial intelligence, traffic command robots, traffic cloud robots , Deep learning and other interdisciplinary composite comprehensive technologies, the management and control effects can be reflected in many aspects such as traffic order, traffic congestion, traffic accidents, and traffic service levels. In addition, the rationalization of the urban transportation structure, the adjustment of urban land use and the formulation and measures of management systems, transportation investment, and transportation planning are all indispensable important elements for solving the problems of the transportation system. When evaluating traffic management and control, the multi-objective principle must be adopted to quantitatively calculate and qualitatively analyze various aspects that affect the level of road traffic management, determine evaluation criteria and methods, and comprehensively evaluate the overall level and capability of the entire traffic management and control. Therefore, when selecting the index system of traffic management and control service index, it is necessary to pay attention not only to the evaluation of traffic management effects, but also to the scientific guidance of management methods and management processes.

现有的大数据环境交通管理与控制服务指数方法及***主要存在以下缺点：The existing big data environment traffic management and control service index method and system mainly have the following disadvantages:

1、当前，交通服务指数领域中通常是采用向市民发放交通出行问卷、归纳统计分析整理问卷后，来发布城市与区域交通服务指数方式进行的。这种方式存在着发放问卷调查周期长、问卷发放范围与取样比例不足、对问卷的问题回答不十分准确、不能进行连续动态对交通服务数据及时进行评价等问题。1. At present, in the field of transportation service index, the urban and regional transportation service index is released by issuing the transportation questionnaire to the citizens, summarizing the statistical analysis and sorting out the questionnaire. This method has the problems of long questionnaire survey cycle, insufficient questionnaire scope and sampling ratio, inaccurate answers to questionnaires, and continuous and dynamic evaluation of transportation service data.

2、在城市道路交通指数单一方面，虽然有了个别的道路交通指数发明，但是对于城市交通体系整体服务指数上，缺乏全面性、***性的总体评价与总体描述城市交通运输体系服务的一体化、可量化、可视化的交通管理与控制服务指数方法发明提出，明显缺乏全面评价城市交通与区域交通服务指数的方法与***。2. In the single aspect of the urban road traffic index, although there are individual road traffic index inventions, the overall service index of the urban transportation system lacks a comprehensive and systematic overall evaluation and overall description of the integration of urban transportation system services The invention of a method for the management of traffic management and control service index based on quantization, quantification and visualization clearly proposes that there is obviously a lack of a method and system for comprehensive evaluation of urban traffic and regional traffic service index.

3、国外发达国家对城市交通服务水平的评价进行了***和周密的研究，形成了较完整的体系，但是构建基于交通大数据实时发布交通管理与控制服务指数的方法与***模式还属空白。在这些国家中城市交通服务水平的指标体系已经成为编制城市交通规划的重要依据，也是政府部门为交通运输企业制订运营任务和目标的工具，更是公众市民对城市交通***服务进行监督和评价标准的手段。3. The developed countries have carried out systematic and thorough research on the evaluation of urban traffic service levels, forming a relatively complete system, but the method and system model for building real-time traffic management and control service index based on traffic big data is still blank. In these countries, the index system of urban transportation service level has become an important basis for the preparation of urban transportation planning, and it is also a tool for government departments to formulate operational tasks and goals for transportation enterprises. It is also a standard for public citizens to supervise and evaluate urban transportation system services. s method.

4、国内目前依然缺乏***地评价城市交通服务水平的指标体系，《城市道路交通规划设计规范》中有关城市交通的章节，其内容侧重于交通设施规划，涉及的指标也多为设施配置与建设指标，并不是城市交通运行服务指标，特别是当前城市交通转型期就更不适用。也有学者在研究城市交通发展水平方面，提出综合评价指标体系时提出一些的运营服务水平指标，然而这些服务水平指标的提及并没有进行与选择交通方式的行为决策过程结合起来，缺乏交通行为(交通行为的运作、交通行为的通道、交通行为的控制与保障)的分析研判，也没有涉及到城市交通服务水平指标等级的划分与确定，更没有建立交通大数据环境下实时发布交通管理与控制服务指数的方法与***。4. At present, there is still a lack of an index system for systematic evaluation of urban transportation service levels in China. The chapter on urban transportation in the "Code for Design of Urban Road Transportation Planning" focuses on transportation facility planning, and the indicators involved are mostly facility allocation and construction. Indicators are not indicators of urban transportation operation services, especially in the current urban transportation transformation period. Some scholars have put forward some operational service level indicators when they put forward a comprehensive evaluation index system in the study of urban transportation development level. However, the mention of these service level indicators has not been combined with the behavioral decision-making process of selecting transportation methods, and lacks traffic behavior ( The analysis and judgment of the operation of traffic behaviors, the channels of traffic behaviors, the control and guarantee of traffic behaviors, nor the classification and determination of the index level of urban traffic service levels, nor the establishment of real-time traffic management and control under the traffic big data environment. Method and system of service index.

综上所述，国内外在大数据环境交通管理与控制服务指数方法及***上，尚未见到完整的、一体化的、可量化的、可视化的发明记载。我国在城市交通服务水平指标体系研究方面和基于交通大数据交通管理与控制服务指数的方法相对滞后的状况，严重影响了城市交通***建设和健康发展，特别是公众出行对城市交通***服务水平监督的需求满足。因此，迫切需要提供可量化、可视化指标体系的方法对交通***服务进行分析与研判。In summary, at home and abroad, there has not been a complete, integrated, quantifiable, and visual record of inventions in the big data environment traffic management and control service index method and system. The relatively lagging situation in the research on the index system of urban traffic service level and the method of traffic management and control service index based on traffic big data in China has seriously affected the construction and healthy development of the urban traffic system, especially the public travel supervision of the urban traffic system service level The needs are met. Therefore, there is an urgent need to provide a quantifiable and visual indicator system to analyze and judge traffic system services.

发明内容Summary of the invention

本申请提供了一种交通管理与控制服务指数的评价指标体系建立及发布方法，旨在至少在一定程度上解决现有技术中的上述技术问题之一。This application provides a method for establishing and publishing an evaluation index system for a traffic management and control service index, which aims to solve at least a certain degree of one of the above technical problems in the prior art.

为了解决上述问题，本申请提供了如下技术方案：一种交通管理与控制服务指数的评价指标体系建立及发布方法，包括：In order to solve the above problems, this application provides the following technical solutions: A method for establishing and publishing an evaluation index system for the traffic management and control service index, including:

步骤a：根据选取的交通管理服务指数及交通控制服务指数构建城市交通***的交通管理与控制服务指数指标体系；Step a: According to the selected traffic management service index and traffic control service index, construct the traffic management and control service index index system of the urban transportation system;

步骤b：根据交通管理与控制服务指数指标关联与评价进行建模；Step b: Modeling based on the correlation and evaluation of traffic management and control service index indicators;

步骤c：通过已建立的交通管理与控制服务指数评价指标体系，与交通大数据云计算平台中的数据进行匹配；Step c: Through the established traffic management and control service index evaluation index system, match with the data in the traffic big data cloud computing platform;

步骤d：匹配交通管理与控制服务指数评价指标对应城市，形成大数据环境交通管理与控制服务指数评价体系。Step d: Match the corresponding index of the traffic management and control service index evaluation index to form a big data environment traffic management and control service index evaluation system.

本申请实施例采取的技术方案还包括：在所述步骤a中，选取的交通管理服务指数包括：交通行政管理服务指数、交通秩序管理服务指数、交通运行管理服务指数、交通优先管理服务指数、交通***管理服务指数、交通需求管理服务指数、交通事件管理服务指数、交通拥挤管理服务指数、道路交通***运行指数、轨道交通***服务指数、常规公交***服务指数、出租车公交服务指数、自行车公交服务指数、快速公交BRT服务指数、公交专用道服务指数、交通信息发布服务指数、交通设施管养服务指数、停车管理***服务指数。The technical solution adopted in the embodiment of the present application further includes: In the step a, the selected traffic management service index includes: traffic administrative management service index, traffic order management service index, traffic operation management service index, traffic priority management service index, Transportation System Management Service Index, Transportation Demand Management Service Index, Traffic Incident Management Service Index, Traffic Congestion Management Service Index, Road Transportation System Operation Index, Rail Transportation System Service Index, Conventional Bus System Service Index, Taxi Bus Service Index, Bicycle Bus Service index, BRT BRT service index, bus lane service index, traffic information release service index, transportation facility maintenance service index, parking management system service index.

本申请实施例采取的技术方案还包括：在所述步骤a中，选取的交通控制服务指数包括：交叉口信号控制运行指数、交叉口通行能力服务指数、交叉口延误控制服务指数、交叉口排队长度监控服务指数、交叉口潮汐车道诱导服务指数、交叉口渠化设计服务指数、交叉口相位设计服务指数、交叉***通流控制服务指数、关键路段协调控制服务指数、公交优先信号控制服务指数、城市快速路控制服务指数、交叉口智能控制指挥机器人、智能车路协同管控云机器人、区域交通信号控制服务指数、城市交通事件控制服务指数、高速公路交通控制服务指数、交通行为管理控制服务指数、城市安全交通控制服务指数。The technical solution adopted in the embodiment of the present application further includes: In the step a, the selected traffic control service index includes: intersection signal control operation index, intersection capacity service index, intersection delay control service index, intersection queuing Length monitoring service index, intersection tidal lane guidance service index, intersection channelization design service index, intersection phase design service index, intersection traffic flow control service index, key road coordination control service index, bus priority signal control service index, Urban expressway control service index, intersection intelligent control command robot, intelligent vehicle road collaborative management and control cloud robot, regional traffic signal control service index, urban traffic incident control service index, highway traffic control service index, traffic behavior management control service index, Urban safety traffic control service index.

本申请实施例采取的技术方案还包括：所述步骤b还包括：设计区域交通与城市交通大数据云计算平台，所述设计区域交通与城市交通大数据云计算平台具体包括：根据交通大数据云计算引擎，构建分布式计算结构、交通管理与控制服务指数关联的NOSQL数据库；采用人工智能的深度学***台即服务PaaS、软件即服务SaaS、容器即服务CaaS的交通大数据云计算平台。The technical solution adopted in the embodiment of the present application further includes: the step b further includes: designing a regional transportation and urban transportation big data cloud computing platform, and the designing of the regional transportation and urban transportation big data cloud computing platform specifically includes: according to traffic big data Cloud computing engine, build a distributed computing structure, traffic management and control service index NOSQL database; use artificial intelligence deep learning technology, under the virtualization and Docker container model, build traffic management and control service index information release environment; architecture Design a transportation big data cloud computing platform with infrastructure as a service IaaS, platform as a service PaaS, software as a service SaaS, and container as a service CaaS.

本申请实施例采取的技术方案还包括：在所述步骤b中，根据交通管理与控制服务指数指标关联与评价进行建模具体为：基于区域交通与城市交通大数据云计算平台，建立交通管理与控制服务数评价指标与交通大数据的匹配关联性评价建模，所述建模采用：多目标决策方法、层次分析法、单纯矩阵评价法、模糊分析法、广义函数法、加权相对偏差距离最小法、集合分析法、模糊综合评判发、主成分分析法和/或因子分析法。The technical solution adopted in the embodiment of the present application further includes: in the step b, modeling based on the correlation and evaluation of the traffic management and control service index indicators is specifically: establishing a traffic management based on the cloud computing platform of regional traffic and urban traffic big data Correlation evaluation modeling for matching with the control service number evaluation index and traffic big data, the modeling uses: multi-objective decision-making method, analytic hierarchy process, simple matrix evaluation method, fuzzy analysis method, generalized function method, weighted relative deviation distance Minimal method, set analysis method, fuzzy comprehensive evaluation, principal component analysis method and/or factor analysis method.

本申请实施例采取的技术方案还包括：所述步骤b还包括：根据交通管理与控制服务指数大数据进行多任务学习，所述多任务学习是采用归纳迁移机制的深度学习方法，利用隐含在多个交通管理与控制服务指数评价任务提炼数据中的特定交通***领域信息来提高泛化能力。The technical solution adopted in the embodiment of the present application further includes: the step b further includes: performing multi-task learning according to the traffic management and control service index big data, the multi-task learning is a deep learning method using an inductive transfer mechanism, using implicit In multiple traffic management and control service index evaluation tasks, the specific traffic system domain information in the data is refined to improve the generalization ability.

本申请实施例采取的技术方案还包括：所述多任务学习的形式包括：联合学习、自主学习和带有辅助任务的学习。The technical solutions adopted in the embodiments of the present application further include: the forms of multi-task learning include: joint learning, autonomous learning, and learning with auxiliary tasks.

本申请实施例采取的技术方案还包括：所述步骤c具体包括：通过已建的交通管理与控制服务指数评价指标体系，与交通大数据云计算平台中的直接关联数据或者间接关联数据进行匹配，形成服务于交通管理与控制服务指数评价的功能要素，通过人工智能的深度学***台环境。The technical solution adopted in the embodiments of the present application further includes: the step c specifically includes: matching the directly related data or the indirectly related data in the traffic big data cloud computing platform through the established traffic management and control service index evaluation index system , Forming functional elements serving the evaluation of traffic management and control service index, forming a visual display of evaluation indexes of traffic management and control service index through the deep learning technology of artificial intelligence, establishing traffic monitoring, dynamic modeling, online simulation, management evaluation Online deduction platform environment to assist decision support.

本申请实施例采取的技术方案还包括：在所述步骤d中，根据城市的人口与GDP产值的大小，将城市划分成五大类型，包括：特大型、直辖级、省会级、地级市、县级市。The technical solution adopted in the embodiments of the present application further includes: in the step d, according to the size of the city’s population and GDP, the city is divided into five types, including: super large, municipal level, provincial capital level, prefecture-level city, County-level city.

本申请实施例采取的技术方案还包括：在所述步骤d中，形成大数据环境交通管理与控制服务指数评价体系主要涵盖基础指标和附加指标，所述基础指标对于所有的城市都适用，所述附加指标只对某些类型的城市适用。The technical solutions adopted in the embodiments of the present application also include: in the step d, forming a big data environment traffic management and control service index evaluation system mainly covers basic indicators and additional indicators. The basic indicators are applicable to all cities. The additional indicators mentioned are only applicable to certain types of cities.

相对于现有技术，本申请实施例产生的有益效果在于：本申请实施例的交通管理与控制服务指数的评价指标体系建立及发布方法根据城市交通与区域交通整体评价指标体系任务，建立必要的指数评价指标与相关交通***属性进行关联，每一种关联模式所建立的评价指标与交通大数据云计算平台信息源池里的直接与间接关联数据建立对应关系，城市交通与区域交通***的管理与控制服务指数就可以通过选取的关联数据的阈值及变化，评价不同城市、不同区域的交通***管理与控制服务水平及能力，面向政府部门、行业企业、公众市民提供科学决策的技术手段与可视化工具；本申请实施例的交通管理与控制服务指数的评价指标体系建立及发布方法能加速促进城市交通与社会经济快速发展、优化提升城市交通能力与居民生活水平，科学决策城市交通***建设发展目标，使城市交通发展与对策更加精准有效。本发明能够建立全面的、***的交通管理与控制方向服务指数指标体系，与传统交通管理与控制评价仅依赖于交通模型分析方法存在很大的差异，将实时动态的交通大数据云计算平台的数据进行关联，完成高可视化的在线推演当前交通管理与控制现状与未来演变趋势，改变以往城市交通领域用“昨天的指标”评估“今天的问题”的不能接受明显错误做法。Compared with the prior art, the beneficial effects of the embodiments of the present application are as follows: the establishment and release of the evaluation index system of the traffic management and control service index of the embodiments of the present application establishes the necessary Index evaluation indicators are associated with the relevant transportation system attributes. The evaluation indicators established by each association mode correspond to direct and indirect related data in the information source pool of the transportation big data cloud computing platform. Urban transportation and regional transportation system management With the control service index, you can evaluate the management and control service levels and capabilities of transportation systems in different cities and different regions through the selection of thresholds and changes in related data, and provide scientific decision-making technical means and visualization for government departments, industry enterprises, and the public. Tools; the establishment and release method of the evaluation index system of the traffic management and control service index in the embodiment of this application can accelerate the rapid development of urban transportation and social economy, optimize and improve urban transportation capacity and residents' living standards, and make scientific decisions on the development goals of urban transportation system construction To make urban transportation development and countermeasures more precise and effective. The invention can establish a comprehensive and systematic traffic management and control direction service index indicator system, which is very different from the traditional traffic management and control evaluation only relying on the traffic model analysis method. It will be a real-time dynamic traffic big data cloud computing platform. Correlate the data, complete the high-visibility online deduction of the current traffic management and control status and future evolution trends, and change the previous urban traffic field to use "yesterday's indicators" to evaluate "today's problems". Obviously wrong practices cannot be accepted.

附图说明BRIEF DESCRIPTION

图1是本申请实施例的交通管理与控制服务指数的评价指标体系建立及发布方法的流程图；1 is a flowchart of a method for establishing and issuing an evaluation index system for a traffic management and control service index according to an embodiment of the present application;

图2是区域交通与城市交通大数据云计算平台的结构示意图；Figure 2 is a schematic diagram of the structure of the regional transportation and urban transportation big data cloud computing platform;

图3是大数据环境交通管理与控制服务指数方法及***体系结构示意图。Figure 3 is a schematic diagram of the big data environment traffic management and control service index method and system architecture.

具体实施方式detailed description

为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。In order to make the purpose, technical solutions and advantages of the present application more clear, the present application will be described in further detail in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

请参阅图1，是本申请实施例的交通管理与控制服务指数的评价指标体系建立及发布方法的流程图。本申请实施例的交通管理与控制服务指数的评价指标体系建立及发布方法包括以下步骤：Please refer to FIG. 1, which is a flowchart of a method for establishing and issuing an evaluation index system for a traffic management and control service index according to an embodiment of the present application. The method for establishing and releasing the evaluation index system of the traffic management and control service index in the embodiment of the present application includes the following steps:

步骤100：根据选取的交通管理服务指数及交通控制服务指数构建城市交通***的交通管理与控制服务指数指标体系；Step 100: According to the selected traffic management service index and traffic control service index, construct the traffic management and control service index index system of the urban traffic system;

在步骤100中，城市交通***主要由城市运输***(交通行为的运作***)、城市道路***(交通行为的通道***)和城市交通管理***(交通行为的控制与保障***)所组成，城市道路***是为城市运输***完成交通行为而服务的，城市交通管理***则是整个城市交通***正常、高效运转的保证。城市交通***是城市社会、经济和物质结构的基本组成部分，城市交通***把分散在城市各处的城市生产活动、生活活动连接起来，在组织生产、安排生活、提高城市客流与货流的有效运转以及促进城市经济发展方面起着十分重要的作用。城市的布局结构、规模大小、乃至城市的生活方式都需要一个城市交通***的支撑。构建的交通管理与控制服务指数评价指标***具体见下表：In step 100, the urban transportation system is mainly composed of an urban transportation system (traffic behavior operation system), an urban road system (traffic behavior passage system), and an urban traffic management system (traffic behavior control and security system). The system serves the urban transportation system to complete the traffic behavior, and the urban transportation management system is the guarantee for the normal and efficient operation of the entire urban transportation system. The urban transportation system is a basic part of the urban social, economic and material structure. The urban transportation system connects the urban production activities and life activities scattered throughout the city, and is effective in organizing production, arranging life, and improving urban passenger flow and freight flow. It plays a very important role in the operation and promotion of urban economic development. The city's layout structure, scale, and even the city's lifestyle all require the support of an urban transportation system. The construction of the evaluation index system of the traffic management and control service index is shown in the following table:

选取的交通管理服务指数主要包括18个种类：交通行政管理服务指数、交通秩序管理服务指数、交通运行管理服务指数、交通优先管理服务指数、交通***管理服务指数、交通需求管理服务指数、交通事件管理服务指数、交通拥挤管理服务指数、道路交通***运行指数、轨道交通***服务指数、常规公交***服务指数、出租车公交服务指数、自行车公交服务指数、快速公交BRT服务指数、公交专用道服务指数、交通信息发布服务指数、交通设施管养服务指数、停车管理***服务指数。The selected traffic management service index mainly includes 18 categories: traffic administrative management service index, traffic order management service index, traffic operation management service index, traffic priority management service index, traffic system management service index, traffic demand management service index, traffic incident Management service index, traffic congestion management service index, road traffic system operation index, rail transit system service index, conventional bus system service index, taxi bus service index, bicycle bus service index, BRT BRT service index, bus lane service index , Traffic information release service index, traffic facility management service index, parking management system service index.

选取的交通控制服务指数主要包括18个种类：交叉口信号控制运行指数、交叉口通行能力服务指数、交叉口延误控制服务指数、交叉口排队长度监控服务指数、交叉口潮汐车道诱导服务指数、交叉口渠化设计服务指数、交叉口相位设计服务指数、交叉***通流控制服务指数、关键路段协调控制服务指数、公交优先信号控制服务指数、城市快速路控制服务指数、交叉口智能控制指挥机器人、智能车路协同管控云机器人、区域交通信号控制服务指数、城市交通事件控制服务指数、高速公路交通控制服务指数、交通行为管理控制服务指数、城市安全交通控制服务指数。The selected traffic control service index mainly includes 18 categories: intersection signal control operation index, intersection capacity service index, intersection delay control service index, intersection queue length monitoring service index, intersection tide lane guidance service index, intersection Intersection channelization design service index, intersection phase design service index, intersection traffic flow control service index, key road coordination control service index, bus priority signal control service index, urban expressway control service index, intersection intelligent control command robot, Intelligent vehicle-road collaborative management and control cloud robot, regional traffic signal control service index, urban traffic event control service index, highway traffic control service index, traffic behavior management control service index, and urban safety traffic control service index.

步骤200：设计区域交通与城市交通大数据云计算平台；Step 200: Design a big data cloud computing platform for regional transportation and urban transportation;

在步骤200中：面对交通管理与控制服务指数方法及***，需要依托交通大数据云计算平台信息源池环境建设，区域交通与城市交通大数据云计算平台如图2所示。设计区域交通与城市利用交通大数据云计算引擎，构建分布式计算结构、交通管理与控制服务指数关联的NOSQL数据库、采用人工智能的深度学***台即服务PaaS/软件即服务SaaS/容器即服务CaaS的交通大数据云计算平台。实现区域交通与城市交通的管理与控制服务指数建立规模更大、种类更多、结构更复杂的关联性数据；构建交通大数据的资源化模型；促进交通管理与控制服务指数的评价参与更多学科、更复合型、更交叉化、更跨行业的融合；是交通管理与控制服务指数更面向交通大数据、更趋于实时动态在线的分析研判技术发展；基于交通大数据云计算平台更有效地实现交通管理与控制服务指数的可视化描述与展示；完成更多维度、更多视角、更多模式的交通管理与控制服务能力的提升，提高以人为本的分析研判评价交通管理与控制服务指数水平和能力。In step 200: Facing the traffic management and control service index method and system, it is necessary to build on the information source pool environment construction of the transportation big data cloud computing platform. The regional transportation and urban transportation big data cloud computing platform is shown in FIG. 2. Design regional transportation and cities to use the transportation big data cloud computing engine to build a distributed computing structure, a NOSQL database associated with traffic management and control service index, deep learning technology using artificial intelligence, and build Web2 under virtualization and Docker container mode. 0Traffic management and control service index information release environment; meanwhile, architecture design infrastructure as a service IaaS/platform as a service PaaS/software as a service SaaS/container as a service CaaS traffic big data cloud computing platform. Realize the management and control service index of regional traffic and urban traffic. Establish larger scale, more types, and more complex structure of related data; construct a resourceful model of traffic big data; promote the evaluation of traffic management and control service index to participate more Discipline, more complex, more cross-cutting, and more cross-industry integration; the traffic management and control service index is more oriented to the development of traffic big data and more inclined to real-time dynamic online analysis and judgment technology; the cloud computing platform based on traffic big data is more effective Realize the visual description and display of the traffic management and control service index; complete the improvement of traffic management and control service capabilities in more dimensions, more perspectives, and more models, and improve the level of human-oriented analysis and evaluation of the traffic management and control service index and ability.

数据引擎层完成对交通大数据进行收集、存储、计算、挖掘和管理。针对交通管理与控制服务指数评价应用中具体数据处理策略、算法设计和性能提升进行优化和实现。根据城市大数据的特性，研究时空大数据及智能交通领域中MapReduce、流式计算、内存计算、图计算等并行计算模式，搭建面向时空大数据的实时处理和非实时处理相结合的混合架构；实现挖掘算法的并行化，提升计算速度；实现对动态、多源、多尺度时空数据的可量化、可视化展现，服务于实际应用。The data engine layer completes the collection, storage, calculation, mining and management of transportation big data. Optimize and implement specific data processing strategies, algorithm design and performance improvement in traffic management and control service index evaluation applications. According to the characteristics of urban big data, study parallel computing modes such as MapReduce, streaming computing, memory computing, graph computing in the field of spatio-temporal big data and intelligent transportation, and build a hybrid architecture that combines real-time processing and non-real-time processing for spatio-temporal big data; Realize the parallelization of mining algorithms and improve the calculation speed; realize the quantifiable and visual display of dynamic, multi-source, and multi-scale spatio-temporal data to serve practical applications.

交通大数据云计算平台支持在线分析和离线分析、支持多种分析算法的实现、支持地理信息数据的导入导出以及查询计算，可完成相关可量化、可视化展现和应用服务。支持多种数据的导入和接入。包括出租车轨迹与运营数据、公交车和地铁智能卡数据、公交车轨迹数据、货车轨迹数据、手机话单数据等。数据性质包括实时数据和历史数据。支持数据的清洗和预处理，融合挖掘分析。支持包括地理信息数据的匹配、地理图层叠加、区域划分等相关可视化方法，为算法研究和结果验证提供了可视化平台基础。The transportation big data cloud computing platform supports online analysis and offline analysis, supports the implementation of multiple analysis algorithms, supports the import and export of geographic information data and query calculations, and can complete related quantifiable, visual display and application services. Support the import and access of multiple data. Including taxi trajectory and operation data, bus and subway smart card data, bus trajectory data, truck trajectory data, mobile phone bill data, etc. The nature of data includes real-time data and historical data. Support data cleaning and pre-processing, fusion mining analysis. It supports related visualization methods including geographic information data matching, geographic layer overlay, region division, etc., and provides a visual platform foundation for algorithm research and result verification.

针对传统数据中心架构面临的主要挑战以及对数据中心的理解，分布式云平台数据中心的总体架构设计主要包括两类数据中心：第一是策略与备份节点类数据中心，第二是分布式业务节点类数据中心。在物理上两者可以合设，也可以分开。In response to the main challenges faced by the traditional data center architecture and the understanding of the data center, the overall architecture design of the distributed cloud platform data center mainly includes two types of data centers: the first is the strategy and backup node data center, and the second is the distributed business Node data center. Physically, the two can be co-located or separated.

策略与备份类节点数据中心，负责承载整个数据数据中心的统一管理、备份及全局数据共享：统一Portal入口，提供全局跨多个分布式数据中心基础设施资源状态信息统一管理Portal，以及全局容灾与业务请求路由策略的管理界面；在多个数据中心之间共享的公共数据(如用户签约预认证数据、内部子网间结算数据、运营商间结算数据等)；在线业务应用主用数据的备份数据、历史归档以及日志数据等，以及依托这些数据进行初步BI分析与挖掘的平台和应用逻辑；为加速数据访问，分布式节点数据中心会缓存频繁访问的策略与备份的共享数据镜像拷贝，同时将策略与备份节点的数据变更同步到分布式节点侧。Policy and backup node data center, responsible for carrying the unified management, backup and global data sharing of the entire data data center: unified Portal portal, providing global unified management of Portal resource status information across multiple distributed data centers, and global disaster recovery Management interface with business request routing strategy; public data shared between multiple data centers (such as user signing pre-authentication data, inter-subnet settlement data, inter-operator settlement data, etc.); online business application main data Backup data, historical archives, log data, etc., as well as platforms and application logic that rely on these data for preliminary BI analysis and mining; to speed up data access, distributed node data centers will cache frequently accessed strategies and backup shared data mirror copies, At the same time, the policy and backup node data changes are synchronized to the distributed node side.

分布式业务节点类数据中心，负责承载在线业务应用(Online Carrier Applications)，以及在线内部IT办公自动化及ERP/CRM/SCM/PLM/HRM类等应用，支撑应用的各类中间件(数据库、web框架、SDP等)，及上述应用所需读写与访问的***配置数据，用户签约数据以及交通领域媒体类数据(如个人邮箱、电子书、相片、视频、博客内容等)。Distributed business node data center, responsible for carrying online business applications (Online Carrier Applications), as well as online internal IT office automation and ERP/CRM/SCM/PLM/HRM and other applications, supporting various types of middleware (database, web) Framework, SDP, etc.), and system configuration data for reading, writing, and accessing required by the above applications, user subscription data, and media data in the transportation field (such as personal mailboxes, e-books, photos, videos, blog content, etc.).

步骤300：根据交通管理与控制服务指数指标关联与评价进行建模；Step 300: Modeling based on the correlation and evaluation of traffic management and control service index indicators;

在步骤300中，交通管理与控制服务指数评价指标体系主要是建立在交通管理服务指数方面和交通控制服务指数两个方面进行关联设计，两个方面的服务指数评价指标分别建立在交通管理与控制领域内容。交通管理与控制服务指数评价的指标体系种类设计，关系到交通管理与控制服务的有效性和服务水平的科学性评价，基于交通管理与控制学科的专业性评价指标含义，建立基于交通大数据云计算的服务指数评价指标体系，详见如下表所示。In step 300, the traffic management and control service index evaluation index system is mainly based on the traffic management service index and the traffic control service index. The two aspects of the service index evaluation index are based on traffic management and control. Field content. The design of the index system of traffic management and control service index evaluation is related to the scientific evaluation of the effectiveness and service level of traffic management and control services. Based on the meaning of the professional evaluation indicators in the discipline of traffic management and control, the establishment of a traffic-based big data cloud The calculated service index evaluation index system is shown in the following table.

交通管理与控制服务指数评价指标体系关联结构设计，基于这一评价指标体系关联结构，可以建立交通管理与控制服务指数可量化、可视化的评价指标体系。交通管理与控制服务指数评价模型和评价方法是在交通***服务指数评价指标体系建立后，基于交通大数据云计算平台，建立交通管理与控制服务数评价指标与交通大数据的匹配关联性评价建模，形成对交通***服务水平和服务能力的评价方法，这是衡量交通***管理与控制高品质提升的可量化、可视化重要方法，主要包括多目标决策方法、层次分析法、广义函数法及评价方法的应用与选择。上述的交通管理与控制服务指数评价的指标体系种类内容中，没有哪一个指标能够单独作为评价整个交通***的依据。因此，还需要建立一个综合性指标，它能够全面地反映交通***状况，以利于各个城市交通***之间或同一城市交通***不同时期之间比较。多目标决策是具有两个以上的决策目标，并且需用多种标准来评价和优选方案的决策。大多是交通管理与控制服务指数评价决策中最重要的战略决策。例如一个重大交通***项目的决策，就要考虑经济效益、社会效益、安全施工与环境保护等多方面的目标，需要用多种标准进行评价方案和优选方案。其特点是：第一，由于目标和标准的多样性，造成方案比较优劣的工作比较复杂，难以找到使所有目标达到最佳的方案；第二，决策过程是从淘汰较差方案开始，在剩下的方案中选取满意的方案，用满意标准取代最优标准。多目标决策法的基本原理：从人们在多目标条件下合理进行决策的过程和机制，多目标决策的理论主要有：多目标决策过程的分析和描述；冲突性的分解和理想点转移的理论；多属性效用理论；需求的多重性和层次性理论等，它们是构成多目标决策分析方法的理论基础。在多目标决策中，有一部分方案经比较后可以淘汰，称为劣解；但还有一批方案既不能淘汰，又不能互相比较，从多目标上考虑又都不是最优解，称为“非劣解”(或“有效解”、“帕累托解”)。决策分析是在交通***规划、建设、管理、运行阶段为解决当前或未来可能发生的问题，在若干可选的方案中选择和决定最佳方案的一种分析过程。在交通经济***的研究管理与控制过程中我们所面临的交通***决策问题常常是多目标的，例如我们在研究交通运输组织生产过程的组织决策时，既要考虑交通***的运输能力最大，又要使运输服务质量高，运输成本低等。这些目标之间相互作用和矛盾，使决策过程相当复杂，使决策者常常很难轻易作出决策。这类具有多个目标的决策总称就是多目标决策，多目标决策方法现已广泛地应用于交通运输工程、智能交通工程等领域。The associative structure design of the evaluation index system of the traffic management and control service index. Based on this associative structure of the evaluation index system, a quantifiable and visual evaluation index system of the traffic management and control service index can be established. The traffic management and control service index evaluation model and evaluation method is based on the establishment of the traffic system service index evaluation index system, based on the traffic big data cloud computing platform, to establish a matching correlation evaluation evaluation index of traffic management and control service number and traffic big data. Model to form an evaluation method for the service level and service capability of the transportation system. This is an important method to measure the high-quality improvement of transportation system management and control. It mainly includes multi-objective decision-making methods, analytic hierarchy process, generalized function method and evaluation. Application and selection of methods. Among the types of index systems for the above traffic management and control service index evaluation, no single index can be used as the basis for evaluating the entire transportation system. Therefore, it is also necessary to establish a comprehensive indicator that can comprehensively reflect the status of the transportation system, so as to facilitate comparison between various urban transportation systems or between different periods of the same urban transportation system. Multi-objective decision-making is a decision that has more than two decision-making goals, and needs to use a variety of criteria to evaluate and optimize the program. Most of them are the most important strategic decisions in traffic management and control service index evaluation decisions. For example, in the decision-making of a major transportation system project, it is necessary to consider the economic, social benefits, safe construction and environmental protection and other objectives. It is necessary to use a variety of criteria for evaluation and optimization. Its characteristics are: first, due to the diversity of goals and standards, the work of comparing the advantages and disadvantages of the plan is more complicated, and it is difficult to find the best plan for all goals; second, the decision-making process starts from eliminating the poorer plan. Among the remaining plans, select the satisfactory plan and replace the optimal standard with the satisfactory standard. The basic principle of multi-objective decision-making method: from the process and mechanism of people making decisions rationally under multi-objective conditions, the theory of multi-objective decision-making mainly includes: analysis and description of multi-objective decision-making process; the theory of conflicting decomposition and ideal point transfer ; Multi-attribute utility theory; multiplicity and hierarchical theory of demand, etc., which constitute the theoretical basis of the multi-objective decision analysis method. In multi-objective decision-making, a part of the schemes can be eliminated after comparison, which is called inferior solution; but there are a batch of schemes that can neither be eliminated or compared with each other, nor considered optimal in terms of multi-objectives, called "non- Inferior solution" (or "effective solution", "Pareto solution"). Decision analysis is an analysis process that selects and decides the best one among several alternatives in order to solve the current or future problems in the planning, construction, management and operation of the transportation system. The transportation system decision-making problems we face in the research management and control process of the transportation economic system are often multi-objective. For example, when we study the organization and decision-making of the transportation organization production process, we must consider the maximum transportation capacity of the transportation system, and It is necessary to make transportation services of high quality and low transportation costs. The interaction and contradiction between these goals make the decision-making process quite complicated and make it difficult for decision-makers to make decisions easily. This type of decision-making with multiple objectives is generally called multi-objective decision-making. Multi-objective decision-making methods are now widely used in transportation engineering, intelligent transportation engineering and other fields.

多目标决策的原则：交通管理与控制服务指数评价的多目标决策原则是在多目标决策实践中应遵循的行为准则。主要包括：The principle of multi-objective decision-making: The multi-objective decision-making principle of traffic management and control service index evaluation is the code of conduct that should be followed in the practice of multi-objective decision-making. mainly includes:

在满足决策需要的前提下，尽量减少交通管理与控制服务指数评价目标个数；可采用剔除从属性目标，并通过交通大数据关联性匹配把类似的目标合并为一个目标，或者把那些只要求达到起码标准而不要求达到最优的次要目标降为约束条件；以及通过同度量求和、求平均值或构成综合函数的方法，用综合评价指标来代替单项评价指标的办法达到目的。Under the premise of meeting the needs of decision-making, try to reduce the number of evaluation targets of traffic management and control service index; you can use the elimination of dependent targets, and merge similar targets into one through traffic big data correlation matching, or combine those that only require Reaching the minimum standard without requiring the achievement of the optimal secondary goal is reduced to constraints; and the method of summing, averaging or composing a comprehensive function with the same measure, using a comprehensive evaluation index instead of a single evaluation index to achieve the goal.

按照交通管理与控制服务指数评价目标的轻重缓急，决定目标的取舍。为此，就要将目标按重要程度排列出一个顺序，并规定出重要性系数，以便在选优决策时有所遵循。According to the priorities of the traffic management and control service index evaluation goals, the choice of goals is determined. To this end, it is necessary to arrange the goals in order of importance and specify the importance coefficients so that they can be followed in the selection of optimal decisions.

对交通管理与控制服务指数评价相互矛盾的目标，应以总目标为基准进行协调，力求对各目标全面考虑，统筹兼顾。For the conflicting objectives of the traffic management and control service index evaluation, the overall objectives should be used as the basis for coordination, and all objectives should be considered in a comprehensive manner.

交通管理与控制服务指数评价模型：当交通管理与控制服务指数评价决策对象具有多个评价目标时，从若干可行方案(也称解)中，选择一个满意方案(解)的决策方法。进行多目标决策时，根据服务指数评价事前确定的评价标准，从一组非劣解中，通过“辨优”和“权衡”找出一个令人满意的解。Traffic management and control service index evaluation model: When the traffic management and control service index evaluation decision-making object has multiple evaluation objectives, from a number of feasible solutions (also called solutions), choose a satisfactory solution (solution) decision method. When making multi-objective decision-making, according to the evaluation criteria determined in advance by the service index evaluation, from a set of non-inferior solutions, find a satisfactory solution through "distinguishing excellence" and "trade-offs".

多目标决策问题的某一可行方案与其他可行方案两两比较时，其结果有三种可能：第一，所有目标都是最优的方案，称为完全最优解，这种情况极少出现；第二，所有目标都是最劣的方案，称为劣解，立即可以淘汰；第三，目标有优有劣，既不能肯定方案为最优，也不能立即予以淘汰，这种方案称为非劣解，又称有效解或帕雷托最优解。多目标最优问题的数学模型为：设交通管理与控制服务指数评价***有m个目标f1(x)，f2(x)，...，fm(x)，要求评价由n个变量组成的方案x＝(x1，x2，...，xn)T，如果这些目标都要求最大(或最小)，并要求解满足约束条件集合R，则数学模型可表达成如下形式：When a feasible solution of a multi-objective decision-making problem is compared with other feasible solutions in pairs, the result has three possibilities: First, all goals are the optimal solution, called the complete optimal solution, which rarely occurs; Second, all goals are the worst solutions, called inferior solutions, and can be eliminated immediately; third, the goals have advantages and disadvantages, and neither can be sure that the solution is optimal, nor can it be eliminated immediately. Inferior solutions, also known as effective solutions or Pareto optimal solutions. The mathematical model of the multi-objective optimal problem is: suppose that the traffic management and control service index evaluation system has m objectives f1 (x), f2 (x), ..., fm (x), and requires evaluation of n variables Scheme x = (x1, x2, ..., xn) T, if these goals require maximum (or minimum), and the solution is required to satisfy the set of constraints R, then the mathematical model can be expressed as follows:

或

or

式(1)中F(x)＝(f1(x)，f2(x)，...，fm(x))为目标向量。In formula (1), F(x)=(f1(x), f2(x), ..., fm(x)) is the target vector.

多目标决策方法应用模式分析：在大数据环境交通管理与控制服务指数方法及***发明中，用于多目标交通管理与控制服务指数决策的综合评价方法很多，如层次分析法、单纯矩阵评价法、模糊分析法、广义函数法、加权相对偏差距离最小法、集合分析法、模糊综合评判发、主成分分析法、因子分析法等，各种方法所需要的前提条件不同，各自的用途也不同。本发明中只是介绍相关理论方法的应用模式，具体应用时可以结合不同城市交通***的特点以及同一城市交通***不同时期的管理与控制服务指数的不同评价模式进行选择。Multi-objective decision-making method application mode analysis: In the invention of the big data environment traffic management and control service index method and system, there are many comprehensive evaluation methods for multi-objective traffic management and control service index decision-making, such as AHP and simple matrix evaluation method , Fuzzy analysis method, generalized function method, weighted relative deviation distance minimum method, set analysis method, fuzzy comprehensive evaluation, principal component analysis method, factor analysis method, etc., various methods require different prerequisites, and their respective uses are also different . The present invention only introduces the application mode of related theoretical methods. In specific application, it can be selected by combining the characteristics of different urban transportation systems and the different evaluation modes of the management and control service index of the same urban transportation system at different periods.

层次分析法用于建立交通管理与控制服务指数评价决策体系的分层评价结构，并利用服务指数评价指标与交通大数据关联性匹配所得到的判断矩阵求出各项评价指标的权重。The analytic hierarchy process is used to establish the hierarchical evaluation structure of the traffic management and control service index evaluation decision system, and the weight of each evaluation index is obtained by using the judgment matrix obtained by matching the correlation between the service index evaluation index and the traffic big data.

单纯矩阵评价法是利用服务指数评价指标与交通大数据关联性匹配所得到的判断矩阵确定各个城市交通***服务指数评价指标得分，用于可量化的定性指标确定。The simple matrix evaluation method is to use the judgment matrix obtained by matching the correlation between the service index evaluation index and the traffic big data to determine the score of each urban transportation system service index evaluation index, which is used to determine the quantifiable qualitative index.

模糊分析法是利用判断矩阵对各个城市交通管理与控制服务指数评价指标排序，实际上是前两种方法的简化处理。The fuzzy analysis method is to use the judgment matrix to rank the evaluation indicators of each urban traffic management and control service index, which is actually a simplified treatment of the first two methods.

广义函数法是在已知权重和所有城市交通管理与控制服务指数的各项指标值后，再经过分级标定，把指标值转化为得分，然后采用加权求和的方法得到总分。The generalized function method is to know the weights and all the index values of all urban traffic management and control service indexes, and then pass the grading calibration to convert the index values into scores, and then use the weighted sum method to get the total score.

加权相对偏差距离最小法是在已知权重和所有城市交通管理与控制服务指数评价的各项指标值后，构造“虚拟最佳城市交通***”，以各个实际城市与“虚拟最佳城市交通***”的加权相对偏差距离大小来判断各个城市交通管理与控制服务指数的优劣。The weighted relative deviation distance minimum method is to construct a "virtual optimal urban transportation system" after knowing the weights and all index values of all urban traffic management and control service index evaluations. The weighted relative deviation distance is used to judge the pros and cons of each urban traffic management and control service index.

集合分析发也是在已知交通管理与控制服务指数评价指标值矩阵和权重后，由排序矩阵、指数矩阵得到城市交通管理与控制服务指数的排序。The set analysis is also based on the matrix and weight of the evaluation index value matrix and weight of the traffic management and control service index, and then the ranking of the urban traffic management and control service index is obtained from the ranking matrix and the index matrix.

模糊综合评判发是依次确定因素(评价指标)集、判断集，并通过单因素评判得到模糊矩阵，用模糊矩阵与权重向量共同得到交通管理与控制服务指数综合评判结果。The fuzzy comprehensive evaluation is to determine the set of factors (evaluation indicators) and judgments in sequence, and obtain the fuzzy matrix by single factor evaluation, and use the fuzzy matrix and the weight vector to obtain the comprehensive evaluation result of the traffic management and control service index.

主成分分析法和因子分析法都是在已知多个样本数据条件下，计算各个指标的相关矩阵，得到主成分或主因子，从而确定交通管理与控制服务指数综合评价指标的计算，因子分析法是主成分分析法的推广。Principal component analysis method and factor analysis method both calculate the correlation matrix of each indicator under the condition of known multiple sample data to obtain the principal component or principal factor, so as to determine the calculation of the comprehensive evaluation index of traffic management and control service index, factor analysis method It is the promotion of principal component analysis.

步骤400：根据交通管理与控制服务指数大数据进行深度学习；Step 400: Perform deep learning based on the big data of the traffic management and control service index;

在步骤400中，交通管理与控制服务指数评价模型和评价方法建立之后，需要开展基于服务指数评价模型和评价方法的交通大数据深度学习算法--多任务学习运用。多任务学习(Multi-task learning)是迁移学习算法的一种，迁移学习可理解为定义一个一个源领域source domain和一个目标领域(target domain)，在源领域深度学习，并把学习到的知识迁移到目标领域，提升目标领域的学习效果(performance)。多任务学习(Multi-task learning)：由于我们的关注点集中在单个任务上，我们忽略了可能帮助优化度量评价指标的其它信息。具体来说，这些信息来自相关任务的评价训练信号。通过共享相关任务之间的表征，可以使我们的评价模型更好地概括原始任务。这种方法被称为多任务学习(MTL)。其也是一种归纳迁移机制，主要目标是利用隐含在多个相关任务的训练信号中的特定交通管理与控制领域信息来提高泛化能力，多任务学习通过使用共享表示并行训练多个任务来完成这一目标。归纳迁移是一种专注于将解决一个问题的知识应用到相关问题的方法，从而提高学习的效率。比如，学习交通出行的行走时掌握的能力可以帮助学会跑，我们可以在相关的学习任务之间迁移通用的知识。此外，由于使用共享表示，多个任务同时进行预测时，减少了数据来源的数量以及整体模型参数的规模，使预测更加高效。因此，在交通管理与控制多个应用领域中，可以利用多任务学习来提高效果或性能。In step 400, after the establishment of the traffic management and control service index evaluation model and evaluation method, it is necessary to develop a multi-task learning application of the traffic big data deep learning algorithm based on the service index evaluation model and evaluation method. Multi-task learning (Multi-task learning) is a kind of transfer learning algorithm. Transfer learning can be understood as defining a source domain and a target domain, deep learning in the source domain, and the learned knowledge Move to the target area and improve the learning performance of the target area. Multi-task learning (Multi-task learning): Because our focus is on a single task, we ignore other information that may help optimize the evaluation metrics. Specifically, this information comes from the evaluation training signals of related tasks. By sharing the representations between related tasks, our evaluation model can better summarize the original tasks. This method is called Multitask Learning (MTL). It is also an inductive migration mechanism. The main goal is to use the specific traffic management and control domain information implicit in the training signals of multiple related tasks to improve the generalization ability. Multi-task learning is to train multiple tasks in parallel by using shared representations. Accomplish this goal. Inductive transfer is a method that focuses on applying knowledge to solve a problem to related problems, thereby improving learning efficiency. For example, learning the ability to learn when walking in traffic can help learn to run, and we can transfer common knowledge between related learning tasks. In addition, due to the use of shared representations, when multiple tasks simultaneously make predictions, the number of data sources and the size of the overall model parameters are reduced, making predictions more efficient. Therefore, in multiple application areas of traffic management and control, multi-task learning can be used to improve the effect or performance.

归纳偏执(inductive bias)：归纳迁移的目标是利用额外的信息来源来提高当前任务的学***所需要的训练样本数量。归纳偏执会导致一个归纳学习器更偏好一些假设，多任务学习正是利用隐含在相关任务训练信号中的信息作为一个归纳偏执来提高泛化能力。归纳偏置的作用就是用于指导学习算法如何在评价模型空间中进行搜索，搜索所得评价模型的性能优劣将直接受到归纳偏置的影响，而任何一个缺乏归纳偏置的学习***都不可能进行有效的学习。不同的学习算法(决策树，神经网络，支持向量机等)具有不同的归纳偏置，人们在解决交通***实际问题时需要人工地确定采用何种学习算法，实际上也就是主观地选择了不同的归纳偏置策略。一个很直观的想法就是，是否可以将归纳偏置的确定过程也通过学习过程来自动地完成，也就是采用“学习如何去学(learning to learn)”的思想。多任务学习恰恰为上述思想的实现提供了一条可行途径，即利用相关任务中所包含的有用信息，为所关注任务的学习提供更强的归纳偏置。Inductive biases: The goal of inductive migration is to use additional sources of information to improve the learning performance of the current task, including improving generalization accuracy, learning speed, and the intelligibility of the learned model. Providing stronger inductive paranoia is a method for migration to improve generalization ability, which can produce better generalization ability on a fixed training set, or reduce the number of training samples required to achieve the same performance level. Inductive paranoia will cause an inductive learner to prefer some assumptions. Multitask learning is to use the information implicit in the training signals of related tasks as an inductive paranoia to improve the generalization ability. The role of inductive bias is to guide the learning algorithm how to search in the evaluation model space. The performance of the search evaluation model will be directly affected by the inductive bias, and any learning system lacking inductive bias is impossible. Conduct effective learning. Different learning algorithms (decision trees, neural networks, support vector machines, etc.) have different inductive biases. People need to manually determine which learning algorithm to use when solving practical problems in the transportation system. In fact, they choose different subjectively. Inductive bias strategy. A very intuitive idea is whether the process of determining the inductive bias can also be completed automatically through the learning process, that is, the idea of "learning to learn" is adopted. Multi-task learning provides a feasible way for the realization of the above ideas, that is, to use the useful information contained in the relevant tasks to provide a stronger induction bias for the learning of the tasks concerned.

多任务学习(Multi-Task Learning，MTL)是一种采用归纳迁移机制的深度学习方法，主要目标是利用隐含在多个交通管理与控制服务指数评价任务提炼数据中的特定交通***领域信息来提高泛化能力。多任务学习通过使用共享表示并行训练多个任务来完成这一目标。因此，交通管理与控制服务指数评价指标体系的多任务学习算法是在学习一个问题的同时，可以通过使用共享表示来获取其他相关问题的知识，这其中采用的归纳迁移学习(Inductive Transfer Learning)算法是一种专注于将解决一个问题的知识应用到相关问题上的方法，从而提高深度学习效率。Multi-Task Learning (MTL) is a deep learning method that uses an inductive transfer mechanism. The main goal is to use specific transportation system domain information implicit in the data extracted from multiple traffic management and control service index evaluation tasks Improve generalization ability. Multi-task learning accomplishes this goal by training multiple tasks in parallel using shared representations. Therefore, the multi-task learning algorithm of the traffic management and control service index evaluation index system is to learn a problem and can use the shared representation to obtain knowledge of other related problems. The inductive transfer learning (Inductive Transfer Learning) algorithm is used. It is a method that focuses on applying knowledge to solve a problem to related problems, thereby improving the efficiency of deep learning.

交通管理与控制服务指数评价指标体系多任务学习可以学到多个任务的共享表示，这个共享表示具有较强的抽象能力，能够适应多个不同但相关的目标，通常可以使交通管理与控制主任务获得更好的泛化能力。此外，由于使用了共享表示，多个任务同时进行预测时，减少了数据来源的数量，以及整体评价模型参数的规模，使预测更加高效。因此，在交通管理与控制服务指数评价指标体系等多个应用领域中，可以利用多任务学习来提高效果和性能。Traffic management and control service index evaluation index system Multi-task learning can learn a shared representation of multiple tasks. This shared representation has strong abstract capabilities and can adapt to many different but related goals. It can usually make the traffic management and control master The task gets better generalization ability. In addition, due to the use of shared representations, when multiple tasks simultaneously make predictions, the number of data sources and the size of the overall evaluation model parameters are reduced, making predictions more efficient. Therefore, in multiple application areas such as traffic management and control service index evaluation index system, multi-task learning can be used to improve the effect and performance.

如果交通管理与控制服务指数评价指标中的两个任务处理时如果相同函数，但是在任务信号中加入交通影响因素处理，那么很明显这两个任务是相关的。If the two tasks in the traffic management and control service index evaluation indicators are processed with the same function, but the traffic influencing factors are added to the task signal, then the two tasks are obviously related.

如果交通管理与控制服务指数评价指标中的两个任务用于预测同一个个体属性的不同方面，那么这些任务比预测不同个体属性的不同方面更相关。If two tasks in the traffic management and control service index evaluation index are used to predict different aspects of the same individual attribute, then these tasks are more relevant than predicting different aspects of different individual attributes.

两个任务共同训练时能相互帮助并不意味着它们是相关的。例如通过在后向交通流传播网络的一个额外输出中加入交通影响因素可以提高泛化能力，但是这个影响因素任务与其他任务不相关。Being able to help each other when two tasks are jointly trained does not mean that they are related. For example, by adding traffic influencing factors to an additional output of the backward traffic flow propagation network, the generalization ability can be improved, but this influencing factor task is not related to other tasks.

交通管理与控制服务指数大数据深度学习网络是具有多个隐层的神经网络，逐层将输入数据转化成非线性的、更抽象的特征表示，并且在深度学习网络中各层的评价模型参数不是人为设定的，而是给定学习器的参数后在训练过程中学到的，这给了多任务学习施展拳脚的空间，具备足够的能力在训练过程中学习多个任务的共同特征。Traffic management and control service index big data deep learning network is a neural network with multiple hidden layers, which converts the input data into non-linear, more abstract feature representation layer by layer, and the evaluation model parameters of each layer in the deep learning network It is not artificially set, but learned in the training process after giving the parameters of the learner, which gives the multi-task learning room to show its punches and has the ability to learn the common characteristics of multiple tasks during the training process.

多任务学***所需要的训练样本数量。归纳偏向会导致一个归纳学习器更好一些假设，多任务学习正是利用隐含在相关任务训练信号中的信息作为一个归纳偏向来提高泛化能力的。Multi-task learning is an inductive transfer method that makes full use of traffic management and control specific field information implicit in the training signals of multiple related tasks. In the backward transmission process, multi-task learning allows the shared hidden layer to be dedicated to a certain The features of each task are used by other tasks, so multi-task learning can be applied to the special evidence of several different tasks. Such features are often not easy to learn in single-task learning networks. The objective of the inductive transfer of the evaluation index system of the traffic management and control service index is to use additional information sources to improve the learning performance of the current task, including improving the accuracy of Pan-China, the learning speed, and the interpretability of the evaluation model. Increasing stronger induction bias is a method to improve generalization ability, which can produce better generalization ability on the specified training set, or reduce the number of training samples needed to achieve the same performance level. Inductive bias will lead to an assumption that an inductive learner is better. Multitask learning is to use the information implicit in the training signals of related tasks as an inductive bias to improve the generalization ability.

交通管理与控制服务指数评价指标体系多任务深度学习，给出多个监测标签信息，利用任务之间的关联性相互促进。Traffic management and control service index evaluation index system multi-task deep learning, gives multiple monitoring label information, and uses the correlation between tasks to promote each other.

DeepID2中共有两个损失函数，第一个是交通管理分类损失函数，对应于交通管理与控制服务指数评价指标体系：There are two loss functions in DeepID2. The first is the traffic management classification loss function, which corresponds to the traffic management and control service index evaluation index system:

式(2)中：Ident(f，t，θ _id)表示交通管理与控制服务指数分类评价任务的损失函数，f为DeepID2特征向量，θ _id为软最大值层参数，p _i为属于第i类的先验概率，即对于所有类别i，p _i＝0，除了p _t＝1，t为目标类别。 In equation (2): Ident(f, t, θ _id ) represents the loss function of the traffic management and control service index classification evaluation task, f is the DeepID2 feature vector, θ _id is the soft maximum layer parameter, and p _i is the i The prior probability of the class, that is, for all categories i, p _i =0, except for p _t =1, t is the target category.

第二个是交通控制分类损失函数，对应于交通控制服务指数评价指标中的对比损失：The second is the traffic control classification loss function, which corresponds to the comparative loss in the traffic control service index evaluation index:

式(3)中：Verif(f _i，f _j，y _ij，θ _ue)为评价指标体系度量学习任务的损失函数，f _i，f _j为每层对(i，j)的特征向量；y _ij＝1表示(i，j)属于同一类。 In formula (3): Verif (f _i , f _j , y _ij , θ _ue ) is the evaluation index system to measure the loss function of the learning task, f _i , f _j are the feature vectors of each layer pair (i, j); y _ij = 1 indicates that (i, j) belongs to the same category.

训练过程如下：The training process is as follows:

输入：训练集X＝{(xi，li)}，初始化参数θ _c，θ _id，θ _ue，超参数λ，学习率η(t)，t←0，而不收敛做t←t+1，两个训练样本的样本(xi，li)和(xj，lj)格式X。 Input: training set X={(xi, li)}, initialization parameters θ _c , θ _id , θ _ue , hyperparameter λ, learning rate η(t), t←0, without convergence t←t+1, The format X of the samples (xi, li) and (xj, lj) of the two training samples.

f _i＝Conv(x _i，θ _c)和f _j＝Conv(x _i，θ _c)， f _i = Conv(x _i , θ _c ) and f _j = Conv(x _i , θ _c ),

那里y _ij＝1，如果l _i＝l _j，和y _ij＝-1，否则

Where y _ij = 1, if l _i = l _j, and y _ij = -1, otherwise

校正θ _id＝θ _id-η(t)·θ _id，θ _ue＝θ _ue-η(t)·θ _ue，以及θ _c＝θ _c-η(t)·θ _c Correct θ _id =θ _id -η(t)·θ _id , θ _ue =θ _ue -η(t)·θ _ue , and θ _c =θ _c -η(t)·θ _c

输出θ _c；总之，要扩大类间距离和缩小类内距离。 Output θ _c ; in short, the distance between classes and the distance between classes should be reduced.

将多任务学习视为一种归约迁移(inductive transfer)：归约迁移(inductive transfer)通过引入归约偏置(inductive bias)来改进模型，使得模型更倾向于某些假设。如常见的一种归约偏置(Inductive bias)是L1正则化，它使得模型更偏向于那些稀疏的解。在多任务学习场景中，归约偏置(Inductive bias)是由辅助任务来提供的，这会导致模型更倾向于那些可以同时解释多个任务的解，这样做会使得模型的泛化性能更好。Consider multi-task learning as an inductive transfer: Inductive transfer improves the model by introducing inductive biases, making the model more inclined to certain assumptions. A common reduction bias (Inductive biases) is L1 regularization, which makes the model more biased towards those sparse solutions. In a multi-task learning scenario, reduction bias (Inductive Bias) is provided by auxiliary tasks, which will cause the model to prefer solutions that can explain multiple tasks at the same time. Doing so will make the generalization performance of the model more it is good.

参数的硬共享机制：参数的硬共享机制是神经网络的多任务学习中最常见的一种方式，在交通管理与控制服务指数评价实际应用中，通常通过在所有任务之间共享隐藏层，同时保留几个特定任务的输出层来实现。硬共享机制降低了过拟合的风险。这些共享参数过拟合风险的阶数是N，其中N为任务的数量，比任务相关参数的过拟合风险要小。评价指标体系中越多任务同时学习，我们的模型就能捕捉到越多任务的同一个表示，从而导致在我们原始任务上的过拟合风险越小。Parameter hard sharing mechanism: The parameter hard sharing mechanism is the most common way of neural network multi-task learning. In the practical application of traffic management and control service index evaluation, usually by sharing the hidden layer among all tasks, at the same time Keep the output layer of several specific tasks to achieve. The hard sharing mechanism reduces the risk of overfitting. The order of the over-fitting risk of these shared parameters is N, where N is the number of tasks, which is smaller than the over-fitting risk of task-related parameters. The more tasks in the evaluation index system learn at the same time, our model can capture the same representation of the more tasks, resulting in less risk of overfitting on our original tasks.

参数的软共享机制(基于约束的共享regularization based)：每个评价指标任务都由自己的模型，自己的参数。在共享软的参数时，每个任务都有自己的参数和模型。模型参数之间的距离是正则化的，以便鼓励参数相似化。我们对模型参数的距离进行正则化来保障参数的相似，如使用L2距离正则化，或使用迹正则化(trace norm)。用于深度神经网络中的软共享机制的约束很大程度上是受传统多任务学习中正则化技术的影响。约束深度神经网络软的参数共享的思想受到了MTL正则化技术的极大启发，这种思想已经用于其它交通管理与控制服务指数评价模型开发。Parameter soft sharing mechanism (constrained sharing based regularization): Each evaluation index task has its own model and its own parameters. When sharing soft parameters, each task has its own parameters and models. The distance between model parameters is regularized to encourage parameter similarity. We regularize the distance of the model parameters to ensure the similarity of the parameters, such as using L2 distance regularization, or using trace regularization. The constraints of the soft sharing mechanism used in deep neural networks are largely affected by the regularization techniques in traditional multi-task learning. The idea of constrained deep neural network soft parameter sharing has been greatly inspired by MTL regularization technology. This idea has been used in the development of other traffic management and control service index evaluation models.

多任务学***均，可以使得模型获得更好表示F；若任务噪音严重，数据量小，数据维度高，则对于模型来说区分相关与不相关特征变得困难。多任务有助于将模型注意力集中在确实有影响的那些特征上，是因为其他任务可以为特征的相关与不相关性提供额外的证据；对于任务B来说很容易学习到某些特征G，而这些特征对于任务A来说很难学到。这可能是因为任务A与特征G的交互方式更复杂，或者因为其他特征阻碍了特征G的学习。通过交通管理与控制服务指数评价指标体系的多任务学习，我们可以允许模型窃听(eavesdrop)，即使用任务B来学习特征G；多任务学习更倾向于学习到一类模型，这类模型更强调与其他任务也强调的那部分表示。由于一个对足够多的训练任务都表现很好的假设空间，对来自于同一环境的新任务也会表现很好，所以这样有助于模型展示出对新任务的泛化能力；单任务学习时，梯度的反向传播倾向于陷入局部极小值。多任务学习中不同任务的局部极小值处于不同的位置，通过相互作用，可以帮助隐含层逃离局部极小值；添加的任务可以改变权值更新的动态特性，可能使网络更适合多任务学习。多任务并行学习，提升了浅层共享层(shared representation)的学习速率，可能较大的学习速率提升了学习效果；正则化机制多任务学习通过引入归纳偏置(inductive bias)起到与正则化相同的作用。它减小了模型过拟合的风险，同时降低了模型的Rademacher复杂度，即拟合随机噪音的能力。Effectiveness of multi-task learning: Since all tasks have more or less noise, for example, when we train the model on task A, our goal is to get a good representation of task A, while ignoring data-related Noise and generalization performance. Since different tasks have different noise patterns, learning two tasks at the same time can get a more generalized representation. If only learning task A has to bear the risk of overfitting task A, while learning task A and task B at the same time average the noise patterns, the model can get a better representation of F; if the task is serious, the data volume is small, and the data dimension High, it becomes difficult for the model to distinguish between relevant and uncorrelated features. Multitasking helps to focus the model on those features that do have an impact, because other tasks can provide additional evidence for the correlation and irrelevance of the features; for task B, it is easy to learn certain features G , And these features are difficult for Task A to learn. This may be because the interaction between task A and feature G is more complicated, or because other features hinder the learning of feature G. Through multi-task learning of the traffic management and control service index evaluation index system, we can allow model eavesdropping (ie, use task B to learn feature G); multi-task learning tends to learn a class of models, which emphasizes more That part with other tasks also emphasized. Since a hypothesis space that performs well for enough training tasks, it also performs well for new tasks from the same environment, so this helps the model show the ability to generalize to new tasks; when single task learning , The back propagation of the gradient tends to fall into a local minimum. In multi-task learning, the local minima of different tasks are in different positions. Through interaction, they can help the hidden layer escape from the local minima; the added tasks can change the dynamic characteristics of the weight update, which may make the network more suitable for multi-tasking. Learn. Multi-task parallel learning improves the learning rate of the shallow shared layer (shared representation), and the larger learning rate may improve the learning effect; the regularization mechanism multi-task learning works by introducing inductive bias (inductive bias) to play with regularization The same effect. It reduces the risk of overfitting the model, while reducing the Rademacher complexity of the model, that is, the ability to fit random noise.

多任务学习MTL的多种形式：联合学习(joint learning)、自主学习(learning to learn)和带有辅助任务的学习(learning with auxiliary task)等都可以指MTL。优化多个损失函数就等同于进行多任务学习(与单任务学习相反)。这些情况有助于交通管理与控制服务指数评价指标体系明确地思考如何在多任务学习方面做尝试并从中获得启发。Various forms of multi-task learning MTL: joint learning (joint learning), autonomous learning (learning to learning) and learning with auxiliary tasks (learning with auxiliary tasks) can all refer to MTL. Optimizing multiple loss functions is equivalent to multi-task learning (as opposed to single-task learning). These circumstances help the traffic management and control service index evaluation index system to clearly think about how to try and get inspiration from multi-task learning.

多任务学习与其他学习算法之间的关联行。迁移学习自1995年以来，迁移学习吸引了众多的研究者的目光，迁移学习有很多其他名字学习去学习(Learning to learn)、终身学习(life-long learning)、推导迁移(inductive transfer)、知识强化(knowledge consolidation)、上下文敏感性学习(context-sensitive learning)、基于知识的推导偏差(knowledge-based inductive bias)、累计/增量学习(increment/cumulative learning)等。在机器学习、深度学习和数据挖掘的大多数评价指标任务中，会假设training和inference时，采用的数据服从相同的分布(distribution)、来源于相同的特征空间(feature space)。在现实交通管理与控制服务指数评价指标体系应用中，这个假设很难成立，往往遇到一些问题，包括：带标记的训练样本数量有限：处理A领域(target domain)的分类问题时，缺少足够的训练样本。同时与A领域相关的B(source domain)领域，拥有大量的训练样本，但B领域与A领域处于不同的特征空间或样本服从不同分布；数据分布会发生变化：交通管理与控制服务指数评价指标数据分布与时间、地点或其他动态因素相关，随着动态因素的变化，数据分布会发生变化，以前收集的数据已经过时，需要重新收集数据，重建模型。多任务学习是针对数据给出多个交通管理与控制监测信息(标签)进行学习。多标签学习(Multilabel learning)是多任务学习中的一种，建模多个label之间的相关性，同时对多个label进行建模，多个类别之间共享相同的数据/特征。多类别学习(Multiclass learning)是多标签学习任务中的一种，对多个相互独立的类别(classes)进行建模。Association line between multi-task learning and other learning algorithms. Transfer learning Since 1995, transfer learning has attracted the attention of many researchers. Transfer learning has many other names to learn (learning to learn), life-long learning (life-long learning), inductive transfer (inductive) transfer, knowledge Reinforcement (knowledge-consolidation), context-sensitive learning (context-sensitive learning), knowledge-based inference bias (knowledge-based inductive biases), cumulative/incremental learning (increment/cumulative learning), etc. In most evaluation index tasks of machine learning, deep learning, and data mining, it is assumed that when training and inference, the data used follow the same distribution and come from the same feature space. In the application of the actual traffic management and control service index evaluation index system, this assumption is difficult to hold, and often encounters some problems, including: the number of labeled training samples is limited: when dealing with the classification problem of the A domain (target) domain, there is not enough Training samples. At the same time, the B (source domain) domain related to the A domain has a large number of training samples, but the B domain and the A domain are in different feature spaces or samples follow different distributions; the data distribution will change: traffic management and control service index evaluation indicators The data distribution is related to time, place or other dynamic factors. As the dynamic factors change, the data distribution will change. The previously collected data is outdated, and the data needs to be collected again to rebuild the model. Multi-task learning is to give multiple traffic management and control monitoring information (tags) for data to learn. Multilabel learning (Multilabel learning) is a kind of multi-task learning, modeling the correlation between multiple labels, modeling multiple labels at the same time, and sharing the same data/features among multiple categories. Multiclass learning (Multiclass learning) is a type of multi-label learning task that models multiple independent classes.

步骤500：通过已建立的交通管理与控制服务指数评价指标体系，与交通大数据云计算平台中的直接关联数据或者间接关联数据进行匹配；Step 500: Through the established traffic management and control service index evaluation index system, match with the directly related data or indirectly related data in the transportation big data cloud computing platform;

在步骤500中，交通管理与控制服务指数评价指标涉及在36个服务指数发明中的不同内容，形成基于交通大数据的直接或间接关联的数据体系，这些数据可以直接或者间接地表达评价交通管理与控制服务指数的意义，并通过交通大数据云计算平台的信息源池支撑，建立定义交通管理与控制服务指数的指标体系，评价交通管理与控制服务指数的技术水平和能力。如对于城市交通***的交叉口信号控制服务指数评价指标的建立，就包括典型评价指标、行人评价指标、自行车评价指标、信号控制交叉口服务水平指标、交通控制评价指标方法、过饱和交通流状态等。建立交通管理与控制服务指数的评价指标关联性匹配，主要是通过已建的交通管理与控制服务指数评价指标体系，与交通大数据云计算平台中的直接关联数据或者间接关联数据进行匹配，形成服务于交通管理与控制服务指数评价的功能要素，基于交通管理与控制学科的理论与方法，通过人工智能的深度学***台环境。In step 500, the traffic management and control service index evaluation index involves different content in the 36 service index inventions, forming a directly or indirectly related data system based on traffic big data, which can directly or indirectly express and evaluate traffic management With the significance of the control service index, and supported by the information source pool of the transportation big data cloud computing platform, an index system that defines the traffic management and control service index is established to evaluate the technical level and capabilities of the traffic management and control service index. For example, the establishment of evaluation indicators for the signal control service index of intersections in urban transportation systems includes typical evaluation indicators, pedestrian evaluation indicators, bicycle evaluation indicators, signal control intersection service level indicators, traffic control evaluation indicator methods, and oversaturated traffic flow status. Wait. Establish the correlation matching of the evaluation indicators of the traffic management and control service index, mainly through the established traffic management and control service index evaluation index system, matching with the directly related data or indirect related data in the traffic big data cloud computing platform to form The functional elements serving the evaluation of the traffic management and control service index are based on the theory and methods of the traffic management and control discipline, and the real-time, online, quantitative, and visual display of the traffic management and control service index evaluation indicators are formed through the deep learning and other technologies of artificial intelligence. , Establish an online deduction platform environment for auxiliary decision support of traffic operation monitoring, dynamic modeling, online simulation, and management evaluation.

步骤600：匹配交通管理与控制服务指数评价指标对应城市，形成大数据环境交通管理与控制服务指数评价***。Step 600: Match the corresponding index of the traffic management and control service index evaluation index to form a big data environment traffic management and control service index evaluation system.

在步骤600中，建立交通管理与控制服务指数匹配后的评价指标，面向城市交通与区域交通***，开展相关基于交通大数据的交通管理与控制服务指数的分析研判，形成评价交通管理与控制服务水平与能力的量化、可视化动态评价指标不同范围的评价值。面向政府部门、行业企业、公众市民提供城市交通***管理与控制的服务指数评价决策支持的科学方法与技术工具、现代化手段。交通管理与控制服务指数评价指标对应域值详见下表所示：In step 600, the evaluation index after matching the traffic management and control service index is established. For the urban traffic and regional traffic system, the analysis and judgment of the traffic management and control service index based on traffic big data is carried out to form an evaluation of the traffic management and control service. Quantify the level and ability, and visualize the evaluation values of different ranges of dynamic evaluation indicators. It provides scientific methods, technical tools, and modernization methods for providing service index evaluation and decision support for urban transportation system management and control for government departments, industry enterprises, and the public. The corresponding domain values of the traffic management and control service index evaluation indicators are shown in the table below:

大数据环境下交通管理与控制服务指数方法是客观地、公平地评价不同特点的城市交通***与区域交通***，其交通管理及控制服务水平和能力是该城市与区域交通规划、建设、管理一体化高品质服务的具体体现。根据城市的人口与GDP产值的大小，将城市划分成五大类型，即特大型、直辖级、省会级、地级市、县级市，如当前的特大型城市包括四个：北京、上海、深圳、广州。The traffic management and control service index method in a big data environment is an objective and fair evaluation of urban transportation systems and regional transportation systems with different characteristics. Its traffic management and control service levels and capabilities are the integration of urban, regional transportation planning, construction and management. The embodiment of high-quality service. According to the size of the city's population and GDP, the city is divided into five types, namely, super-large, municipal, provincial, prefecture-level, and county-level cities. For example, the current super-large cities include four: Beijing, Shanghai, and Shenzhen , Guangzhou.

城市分类：A类城市：地市级以上城市。市区上年度国内生产总值500亿元以上(不含市辖县国内生产总值)，或市区总人口200万以上(不含市辖县人口，县级市为城关镇人口，下同)的城市；B类城市：地、市(不含县级市)、州、盟。市区上年度国内生产总值250(不含市辖县国内生产总值)亿元以上，或市区总人口50万以上且市区上年度国内生产总值170亿元以上的城市；C类城市：县级市、县上一年度国内生产总值90亿元以上，或市(城)区总人口100万以上但上年度国内生产总值不足170亿元的城市；D类城市：其他城市。City classification: Class A cities: cities above prefecture-level cities. The GDP of the urban area in the previous year is more than 50 billion yuan (excluding the GDP of the municipal counties), or the total population of the urban area is more than 2 million (excluding the population of the municipal counties, the county-level cities are the population of Chengguan Township, the same below) ) Cities; Type B cities: prefectures, cities (excluding county-level cities), states, and alliances. Urban areas with a GDP of more than 250 million yuan (excluding the GDP of counties under the jurisdiction of the city) of more than 100 million yuan in the previous year, or a city with a population of more than 500,000 and a GDP of more than 17 billion yuan in the previous year; Cities: County-level cities and counties with a GDP of more than 9 billion yuan in the previous year, or cities with a population of more than 1 million but a GDP of less than 17 billion yuan in the previous year; Category D cities: other cities .

统计范围：建成区：城市行政区内实际已成片开发建设、市政公用设施和公共设施基本具备的区域。对核心城市，它包括集中连片的部分以及分散的若干个已经成片建设起来，市政公用设施和公共设施基本具备的地区；对一城多镇来说，它包括由几个连片开发建设起来的，市政公用设施和公共设施基本具备的地区组成。因此，建成区范围，一般是指建成区外轮廓线所能包括的地区，也就是这个城市实际建设用地所达到的范围。市区：指城市行政区域内的全部土地面积(包括水域面积)，地级以上城市行政区不包括市辖县(市)，按国务院批准的行政区划面积为准。市区人口：指城市行政区域内有常住户口和未落常住户口的人，以及被注销户口的在押犯、劳改、劳教人员；未落常住户口人员是指持出生、迁移、复员转业、劳改释放、解除劳教等证件未落常住户口的、无户口的人员以及户口情况不明且定居一年以上的流入人口；地级以上城市行政区不包括市辖县(市)；按公安部门的户籍统计为准。Statistical scope: built-up area: an area within the urban administrative area that has actually been developed and constructed, with municipal public facilities and basic facilities. For the core city, it includes a centralized contiguous part and several scattered areas that have been built into pieces, and municipal public facilities and public facilities are basically available; for a city and a number of towns, it includes several contiguous development and construction It is composed of municipal public facilities and areas where public facilities are basically available. Therefore, the built-up area generally refers to the area that can be included in the outer contour of the built-up area, that is, the area that the city actually uses for construction. Urban area: refers to the total land area (including water area) within the city's administrative area. The city's administrative area above the prefecture level does not include the counties (cities) under the jurisdiction of the city. The area of the administrative division approved by the State Council shall prevail. Urban population: Refers to people with permanent residence and unregistered permanent residence in the city’s administrative area, as well as detainees, labor reform, and re-education personnel whose household registration has been cancelled; non-resident permanent residence registration refers to birth, relocation, demobilization, job transfer, and labor reform release 、Removal of re-education through labor and other documents that have not been registered as permanent residents, persons without household registration, and influx of population whose residence status is unknown and who have settled for more than one year; city administrative districts above the prefecture level do not include counties (cities) under municipal jurisdiction; .

交通管理与控制服务指数的评价指标体系的形成与发展，这一评价指标体系的建立主要涵盖基础指标和附加指标，基础指标对于所有的城市都适用，附加指标只对某些类型的城市适用。如大城市的主要问题是交通拥堵，而小城市的主要问题是交通秩序，因此交叉口的相关评价指数对于小城市的意义就不大。根据需要本发明专利还增加了部分指标，根据党的“十九大”后，中国城市快速城市化进程建设与发展的需要，增加了交叉口让行标志设置情况和交通诱导等相关指标，引导交通管理与控制向更高的水平发展。对于部分指标不同城市类型将赋予不同的内涵的调整，基于考虑不同类型城市要求不同，如交通管理与控制的规划、公交优先政策等。The formation and development of the evaluation index system of the traffic management and control service index. The establishment of this evaluation index system mainly covers basic indicators and additional indicators. The basic indicators are applicable to all cities, and the additional indicators are only applicable to certain types of cities. For example, the main problem of large cities is traffic congestion, and the main problem of small cities is traffic order, so the relevant evaluation index of intersections has little significance for small cities. According to the need, the invention patent has also added some indicators. According to the needs of the construction and development of the rapid urbanization process of Chinese cities after the Party’s “Nineteenth National Congress”, relevant indicators such as the setting of intersection yield signs and traffic guidance have been added to guide Traffic management and control are developing to a higher level. For some indicators, different types of cities will be given different connotation adjustments, based on considering the different requirements of different types of cities, such as traffic management and control planning, public transport priority policies, etc.

本申请实施例的交通管理与控制服务指数的评价指标体系建立及发布方法、在深圳市城市交通与区域交通***的规划、建设、管理、运行一体化应用中得到了实验验证。基于深圳市综合交通运行指挥中心的交通大数据信息源池，深圳市建设了城市交通大数据云计算平台体系环境，在深圳市城市综合交通公共信息平台中，开展数据分析、数据挖掘、数据交换、数据存储、数据共享等研究实践，实现了“一个网络、四个平台”的***体系结构，即交通信息通信与传输网络，交通信息采集子平台、城市综合交通公共信息子平台、交通仿真平台、交通信息服务平台。大数据环境交通管理与控制服务指数方法及***体系结构详见图3所示。The establishment and release method of the evaluation index system of the traffic management and control service index in the embodiment of the present application has been experimentally verified in the integrated application of planning, construction, management and operation of the Shenzhen urban traffic and regional traffic system. Based on the traffic big data information source pool of the Shenzhen Comprehensive Traffic Operation Command Center, Shenzhen has built an urban traffic big data cloud computing platform system environment. In the Shenzhen City Comprehensive Transportation Public Information Platform, it conducts data analysis, data mining, and data exchange , Data storage, data sharing and other research practices have realized the "one network, four platforms" system architecture, that is, traffic information communication and transmission network, traffic information collection sub-platform, urban comprehensive traffic public information sub-platform, traffic simulation platform 3. Transportation information service platform. Big data environment traffic management and control service index method and system architecture are shown in Figure 3.

本申请实施例的交通管理与控制服务指数的评价指标体系建立及发布方法与当前国内外相关的最好的现有技术相比，特别是在城市交通与经济发展、城市交通与居民生活、城市交通***的建设与发展战略、城市交通管理体制与政策、城市交通评价体系的功能与构成、城市交通管理与控制***建设、城市交通管理措施与队伍建设、城市交通管理与控制现代化服务等方面，都具有突出优点，具有重要的商业价值与社会价值，包括以下优点：加速促进城市交通与社会经济快速发展：党的十九大提出交通强国建设与交通引领支撑社会经济发展战略，实施从供给侧结构性改革、需求侧精细化管理措施，交通***如何落到实处，体现在城市交通与区域交通领域，就是采用交通大数据云计算新一代信息技术，从交通***的核心领域交通管理与控制内容入手，进行可量化、可视化地分析评价交通运行态势并找出问题，反馈到城市交通与区域交通规划设计中来，形成闭环提升优化交通***规划、建设、管理、运行一体化服务水平和能力；而这一服务水平和能力的评价指标体系，依托于交通管理与控制服务指数方法的发明。社会经济发展离不开交通的引领与支撑，城市是经济发展的中心，是各种交通运输方式的集中地和总枢纽，城市交通***具有特别重要的地位。在社会经济发展中产生了越来越大的交通需求，给城市交通***提出了更高的要求；社会经济发展对城市交通***的通畅性和高效性提出了更高的要求；城市经济发展中产业结构的变化影响着交通需求产生的内在机制；城市经济发展中产业结构布局与城市人口就业布局紧密关联，并调整影响和改变交通需求的时空分布特性。优化提升城市交通能力与居民生活水平：在城市交通中交通管理与控制服务直接与城市居民生活息息相关，表现出的优点在于通过交通大数据云计算平台引擎的支撑，可以直接量化、可视化影响城市居民生活质量水平指标；衡量城市交通运行状对物资丰富程度的影响比例；对城市居民居住地形成的出行方式与出行工具布局的生活方式影响状态细分；评价城市交通运行环境，判断交通拥堵、交通安全、交通污染等量化技术指标的共享与发布；科学决策城市交通***建设发展目标：由于城市交通与经济发展、居民生活之间的密切关系，城市交通的建设不仅要尽量满足经济发展和生活质量提高的需求，还要充分发挥交通对经济发展、城市化和居民生活方式的引导作用，变追随性发展为引导性发展。交通***的建设目标主要包括交通功能目标、资源利用目标、环境保护目标三个方面。交通功能目标是城市交通***的基本目标，主要涵盖舒适性、安全性、高效性、可达性等；交通环境保护目标要求交通行为应尽可能减小对空气、声环境、生态及其他人类生活环境要素的负面影响；交通资源利用目标要求城市交通***能够有效地利用土地、能源、人力等资源。大数据环境交通管理与控制服务指数方法可以量化、可视化递精准掌控与把握评价指标体系，使其决策支持更加科学化；使城市交通发展与对策更加精准有效：解决城市交通问题的关键主要包括关注供求两个方面的关系、采取综合措施两个内容。大数据环境交通管理与控制服务指数方法是综合解决城市交通问题的关键分析研判的工具及手段。它可以精准有效地建立保证科学决策、规划实施、具有综合协调能力的组织管理体制的评价；做好交通与土地利用的协调规划分析；指定城市交通发展战略计划技术指标量化；进行城市开发时倒入交通影响分析研判；落实优先发展公共交通的政策与措施考核；整合交通规划提高交通整体效率的可视化手段；监测具有合理性层次秩序的城市道路交通网络；加速推进道路交通管理的科学化和现代化进程；实施交通需求管理模式建立；有针对性地开展智能交通***的研究与应用；加强城市停车***规划与管理；完善城市道路交通设施监管等优点。The establishment and release method of the evaluation index system of the traffic management and control service index according to the embodiment of the present application is compared with the current best domestic and international related existing technologies, especially in urban transportation and economic development, urban transportation and resident life, urban Construction and development strategies of transportation systems, urban transportation management systems and policies, functions and components of urban transportation evaluation systems, construction of urban transportation management and control systems, urban transportation management measures and team construction, modern services of urban transportation management and control, etc. All have outstanding advantages, have important commercial value and social value, including the following advantages: accelerating the rapid development of urban transportation and socio-economic development: the 19th National Congress of the Communist Party proposed a strategy for the construction of a powerful transportation country and transportation support to support social and economic development. Structural reform, demand-side refined management measures, how the transportation system is implemented, reflected in the field of urban transportation and regional transportation, is the use of transportation big data cloud computing new generation information technology, from the core field of transportation system traffic management and control content Start with a quantifiable and visual analysis and evaluation of the traffic operation situation and find out the problems, and feed back to the urban transportation and regional transportation planning and design to form a closed loop to improve the integrated service level and capacity of the optimized transportation system planning, construction, management and operation; And this service level and capability evaluation index system relies on the invention of the traffic management and control service index method. Socio-economic development is inseparable from the guidance and support of transportation. The city is the center of economic development, the concentration and general hub of various transportation modes, and the urban transportation system has a particularly important position. In the socio-economic development, there is an increasing demand for transportation, which puts forward higher requirements for the urban transportation system; socio-economic development imposes higher requirements on the smoothness and efficiency of the urban transportation system; in the urban economic development The change of industrial structure affects the internal mechanism of traffic demand; the layout of industrial structure in urban economic development is closely related to the layout of urban population employment, and adjusts to affect and change the spatial and temporal distribution characteristics of traffic demand. Optimizing and improving urban transportation capacity and residents' living standards: In urban transportation, traffic management and control services are directly related to the lives of urban residents. The advantage shown is that through the support of the transportation big data cloud computing platform engine, it can directly quantify and visually affect the urban residents. Quality-of-life indicators; measure the proportion of urban traffic operations affecting the abundance of materials; subdivide the lifestyle impact status of travel modes and travel tool layouts formed by urban residents' residences; evaluate the urban traffic operating environment, determine traffic congestion, traffic Sharing and release of quantitative technical indicators such as safety and traffic pollution; scientific decision-making and development goals of urban transportation system construction: due to the close relationship between urban transportation and economic development and residents' life, the construction of urban transportation must not only meet economic development and quality of life as much as possible The increased demand must also give full play to the guiding role of transportation in economic development, urbanization and the way of life of residents, and change the follow-up development into a guide development. The construction goals of the transportation system mainly include three aspects: transportation function goals, resource utilization goals, and environmental protection goals. The traffic function goal is the basic goal of the urban transportation system, mainly covering comfort, safety, efficiency, accessibility, etc.; the traffic environmental protection goal requires that the traffic behavior should be as small as possible to the air, acoustic environment, ecology and other human life Negative impact of environmental factors; the goal of transportation resource utilization requires that the urban transportation system can effectively use land, energy, human resources and other resources. The big data environment traffic management and control service index method can quantify, visualize and accurately control and grasp the evaluation index system, making its decision support more scientific; making urban transportation development and countermeasures more accurate and effective: the key to solving urban transportation problems mainly includes attention The relationship between supply and demand and comprehensive measures are taken. The big data environment traffic management and control service index method is a key analysis tool for comprehensively solving urban traffic problems. It can accurately and effectively establish the evaluation of an organization management system that guarantees scientific decision-making, planning and implementation, and has comprehensive coordination capabilities; do a good job in the analysis of coordinated planning of transportation and land use; quantify the technical indicators of designated urban transportation development strategic plans; Incorporate traffic impact analysis and judgment; implement policies and measures for prioritizing public transportation; integrate visual means for improving transportation efficiency by integrating transportation planning; monitor urban road transportation networks with reasonable levels of order; accelerate the scientific and modernization of road traffic management Process; implement the establishment of traffic demand management model; carry out targeted research and application of intelligent transportation system; strengthen the planning and management of urban parking system; improve the supervision of urban road traffic facilities and other advantages.

虽然本发明参照当前的较佳实施方式进行了描述，但本领域的技术人员应能理解，上述较佳实施方式仅用来说明本发明，并非用来限定本发明的保护范围，任何在本发明的精神和原则范围之内，所做的任何修饰、等效替换、改进等，均应包含在本发明的权利保护范围之内。Although the present invention has been described with reference to current preferred embodiments, those skilled in the art should understand that the above preferred embodiments are only used to illustrate the present invention and are not intended to limit the scope of protection of the present invention. Within the scope of the spirit and principle, any modifications, equivalent replacements, improvements, etc., should be included in the scope of protection of the present invention.

Claims

一种交通管理与控制服务指数的评价指标体系建立及发布方法，包括：A method for establishing and publishing an evaluation index system for a traffic management and control service index, including:

步骤a：根据选取的交通管理服务指数及交通控制服务指数构建城市交通***的交通管理与控制服务指数指标体系；Step a: According to the selected traffic management service index and traffic control service index, construct the traffic management and control service index index system of the urban transportation system;

步骤b：根据交通管理与控制服务指数指标关联与评价进行建模；Step b: Modeling based on the correlation and evaluation of traffic management and control service index indicators;

步骤c：通过所述交通管理与控制服务指数评价指标体系，与交通大数据云计算平台中的数据进行匹配；Step c: Through the traffic management and control service index evaluation index system, match with the data in the traffic big data cloud computing platform;

步骤d：匹配交通管理与控制服务指数评价指标对应城市，形成大数据环境交通管理与控制服务指数评价体系。Step d: Match the corresponding index of the traffic management and control service index evaluation index to form a big data environment traffic management and control service index evaluation system.
根据权利要求1所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，在所述步骤a中，选取的交通管理服务指数包括：交通行政管理服务指数、交通秩序管理服务指数、交通运行管理服务指数、交通优先管理服务指数、交通***管理服务指数、交通需求管理服务指数、交通事件管理服务指数、交通拥挤管理服务指数、道路交通***运行指数、轨道交通***服务指数、常规公交***服务指数、出租车公交服务指数、自行车公交服务指数、快速公交BRT服务指数、公交专用道服务指数、交通信息发布服务指数、交通设施管养服务指数、停车管理***服务指数。The establishment and release method of the evaluation index system of the traffic management and control service index according to claim 1, characterized in that, in the step a, the selected traffic management service index includes: traffic administration management service index, traffic order management Service Index, Transportation Operation Management Service Index, Transportation Priority Management Service Index, Transportation System Management Service Index, Transportation Demand Management Service Index, Traffic Incident Management Service Index, Traffic Congestion Management Service Index, Road Transportation System Operation Index, Rail Transportation System Service Index , Index of conventional bus system service, Index of taxi bus service, Index of bicycle bus service, Index of BRT BRT service, Index of bus lane service, Index of traffic information release service, Index of transportation facility maintenance service, Index of parking management system service.
根据权利要求1或2所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，在所述步骤a中，选取的交通控制服务指数包括：交叉口信号控制运行指数、交叉口通行能力服务指数、交叉口延误控制服务指数、交叉口排队长度监控服务指数、交叉口潮汐车道诱导服务指数、交叉口渠化设计服务指数、交叉口相位设计服务指数、交叉***通流控制服务指数、关键路段协调控制服务指数、公交优先信号控制服务指数、城市快速路控制服务指数、交叉口智能控制指挥机器人、智能车路协同管控云机器人、区域交通信号控制服务指数、城市交通事件控制服务指数、高速公路交通控制服务指数、交通行为管理控制服务指数、城市安全交通控制服务指数。The establishment and release method of the evaluation index system of the traffic management and control service index according to claim 1 or 2, characterized in that in step a, the selected traffic control service index includes: intersection signal control operation index, Intersection Capacity Service Index, Intersection Delay Control Service Index, Intersection Queue Length Monitoring Service Index, Intersection Tide Lane Guidance Service Index, Intersection Channelization Design Service Index, Intersection Phase Design Service Index, Intersection Traffic Flow Control Service Index, Key Road Coordination Control Service Index, Bus Priority Signal Control Service Index, Urban Expressway Control Service Index, Intersection Intelligent Control Command Robot, Intelligent Vehicle and Road Coordination Control Robot, Regional Traffic Signal Control Service Index, Urban Traffic Event Control Service Index, Expressway Traffic Control Service Index, Traffic Behavior Management Control Service Index, City Safety Traffic Control Service Index.
根据权利要求1或2所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，所述步骤b还包括：设计区域交通与城市交通大数据云计算平台，所述设计区域交通与城市交通大数据云计算平台具体包括：根据交通大数据云计算引擎，构建分布式计算结构、交通管理与控制服务指数关联的NOSQL数据库；采用人工智能的深度学***台即服务PaaS、软件即服务SaaS、容器即服务CaaS的交通大数据云计算平台。The method for establishing and publishing an evaluation index system for a traffic management and control service index according to claim 1 or 2, wherein step b further comprises: designing a big data cloud computing platform for regional transportation and urban transportation, the design The regional transportation and urban transportation big data cloud computing platform specifically includes: according to the transportation big data cloud computing engine, build a distributed computing structure, traffic management and control service index NOSQL database; deep learning technology using artificial intelligence, in the virtualization and Under the Docker container model, build a traffic management and control service index information publishing environment; architecture design infrastructure as a service IaaS, platform as a service PaaS, software as a service SaaS, container as a service CaaS traffic big data cloud computing platform.
根据权利要求4所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，在所述步骤b中，根据交通管理与控制服务指数指标关联与评价进行建模具体为：基于区域交通与城市交通大数据云计算平台，建立交通管理与控制服务数评价指标与交通大数据的匹配关联性评价建模，所述建模采用：多目标决策方法、层次分析法、单纯矩阵评价法、模糊分析法、广义函数法、加权相对偏差距离最小法、集合分析法、模糊综合评判发、主成分分析法和/或因子分析法。The method for establishing and releasing the evaluation index system of the traffic management and control service index according to claim 4, wherein in step b, modeling based on the association and evaluation of the traffic management and control service index index specifically includes: Based on the regional transportation and urban transportation big data cloud computing platform, the establishment of a matching relevance evaluation model of traffic management and control service number evaluation indicators and transportation big data, the modeling uses: multi-objective decision-making method, AHP, simple matrix Evaluation method, fuzzy analysis method, generalized function method, weighted relative deviation distance minimum method, set analysis method, fuzzy comprehensive evaluation, principal component analysis method and/or factor analysis method.
根据权利要求1所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，所述步骤b还包括：根据交通管理与控制服务指数大数据进行多任务学习，所述多任务学习是采用归纳迁移机制的深度学习方法，利用隐含在多个交通管理与控制服务指数评价任务提炼数据中的特定交通*** 领域信息来提高泛化能力。The method for establishing and publishing an evaluation index system for a traffic management and control service index according to claim 1, wherein step b further comprises: performing multi-task learning based on the big data of the traffic management and control service index. Task learning is a deep learning method that uses an inductive transfer mechanism, and uses specific traffic system domain information hidden in data in multiple traffic management and control service index evaluation tasks to improve generalization capabilities.
根据权利要求6所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，所述多任务学习的形式包括：联合学习、自主学习和带有辅助任务的学习。The method for establishing and releasing the evaluation index system of the traffic management and control service index according to claim 6, wherein the forms of the multi-task learning include: joint learning, autonomous learning and learning with auxiliary tasks.
根据权利要求1所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，所述步骤c具体包括：通过已建的交通管理与控制服务指数评价指标体系，与交通大数据云计算平台中的直接关联数据或者间接关联数据进行匹配，形成服务于交通管理与控制服务指数评价的功能要素，通过人工智能的深度学***台环境。The method for establishing and publishing an evaluation index system for a traffic management and control service index according to claim 1, wherein step c specifically comprises: passing the established traffic management and control service index evaluation index system Match directly or indirectly related data in the data cloud computing platform to form functional elements that serve the evaluation of the traffic management and control service index, and form a visual display of the evaluation index of the traffic management and control service index through the deep learning technology of artificial intelligence. Establish an online deduction platform environment with auxiliary decision support for traffic operation monitoring, dynamic modeling, online simulation, and management evaluation.
根据权利要求1所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，在所述步骤d中，根据城市的人口与GDP产值的大小，将城市划分成五大类型，包括：特大型、直辖级、省会级、地级市、县级市。The establishment and release method of the evaluation index system of the traffic management and control service index according to claim 1, characterized in that in step d, cities are divided into five types according to the size of the city’s population and GDP, Including: extra large, municipal level, provincial capital level, prefecture level city, county level city.
根据权利要求1所述的交通管理与控制服务指数的评价指标体系建立及发布方法，其特征在于，在所述步骤d中，形成大数据环境交通管理与控制服务指数评价体系主要涵盖基础指标和附加指标，所述基础指标对于所有的城市都适用，所述附加指标只对某些类型的城市适用。The method for establishing and publishing an evaluation index system for a traffic management and control service index according to claim 1, wherein in step d, forming a big data environment traffic management and control service index evaluation system mainly covers basic indexes and Additional indicators, the basic indicators are applicable to all cities, and the additional indicators are only applicable to certain types of cities.