WO2016004775A1 - 一种基于云计算的轨道交通信号综合运维方法及*** - Google Patents

一种基于云计算的轨道交通信号综合运维方法及*** Download PDF

Info

Publication number
WO2016004775A1
WO2016004775A1 PCT/CN2015/075007 CN2015075007W WO2016004775A1 WO 2016004775 A1 WO2016004775 A1 WO 2016004775A1 CN 2015075007 W CN2015075007 W CN 2015075007W WO 2016004775 A1 WO2016004775 A1 WO 2016004775A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
railway
station
index
real
Prior art date
Application number
PCT/CN2015/075007
Other languages
English (en)
French (fr)
Inventor
鲍侠
Original Assignee
北京泰乐德信息技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京泰乐德信息技术有限公司 filed Critical 北京泰乐德信息技术有限公司
Publication of WO2016004775A1 publication Critical patent/WO2016004775A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism

Definitions

  • the invention provides a railway signal comprehensive operation and maintenance method and system based on cloud computing, relating to railway signal data, railway communication data, railway knowledge data, system alarm data, machine learning, vehicle equipment, station equipment, central equipment, trackside equipment And other technical fields to solve the problems faced by the comprehensive operation and maintenance of railway signals.
  • CSM Signal Centralized Monitoring System
  • equipment maintenance machines and communication network management systems for ground signal equipment.
  • TJWX-I and TJWX-2000 and other continuously upgraded signal centralized monitoring CSM systems In order to improve the modern maintenance level of China's railway signal system equipment, since the 1990s, it has independently developed TJWX-I and TJWX-2000 and other continuously upgraded signal centralized monitoring CSM systems.
  • most stations use a signal centralized monitoring system to realize the real-time monitoring of the status of the station signal equipment, and provide the electrical department with the current status of the equipment and the accident analysis by monitoring and recording the main operational status of the signal equipment.
  • the basic basis has played an important role.
  • centralized monitoring CSM system is also widely deployed in urban rail centralized stations/vehicle sections, etc., for urban rail operation and maintenance.
  • vehicle signal equipment there is a DMS system for dynamic monitoring of the vehicle equipment.
  • high-speed railway-specific RBC system, etc. is also faced with the need to incorporate the signal centralized monitoring system, and also faces the need to improve its monitoring capabilities, operation and maintenance capabilities, and equipment self-diagnosis capabilities.
  • the railway system Faced with more and more communication data, signal data, and current network conditions, the railway system generates a large amount of various types of data, but does not make good use of this data. Most of the data is currently stored at each station, and the station simply stores the data, and when the fault occurs, view the relevant data to analyze the cause of the failure. This situation has led to the inability of various types of data generated by the railway to be effectively stored and utilized, and the analysis of faults still relies on manual experience analysis. In many cases, faults can be found in the event of an accident, which not only leads to the manual diagnosis of the railway. Technical problems such as large workload, fault monitoring and low diagnostic efficiency in signal system failures also increase the risk of driving. Therefore, it is an urgent need in the field of rail transit to improve the comprehensive management and operation capability of railway signals, to identify hidden dangers, to cure hidden dangers, to promote the repair of fault repair status, and to ensure safe driving and improve transportation capacity.
  • the invention provides a cloud computing based signal integrated operation and maintenance method and system.
  • the system includes three types of hardware: data acquisition machine, terminal and data center.
  • the software includes a data acquisition subsystem, a data storage subsystem, a data preprocessing subsystem, a data real-time analysis subsystem, an offline data mining subsystem, a presentation subsystem, and a data query system.
  • a cloud computing-based integrated operation and maintenance system for rail transit signals comprising:
  • a data center connected to the data collection machine, including a cloud security layer, a basic resource layer, a virtualization layer, a data storage layer, a computing engine layer, a component layer, and a Web cluster;
  • the cloud security layer passes data backup, data restoration, and access Control, guarantee the security of the system;
  • the basic resource layer is the hardware platform of the data center;
  • the virtualization layer realizes the virtualization of the hardware through the virtualization software, and shields the underlying hardware differences;
  • the data storage layer includes the distributed file system, the column database, Relational database, which is used to store unstructured, semi-structured, and structured data;
  • the computing engine layer uses cloud computing processing;
  • the component layer is a component that specifically handles various services; and
  • the Web cluster is a load-balanced service publishing layer.
  • the terminal connects the data collector and the data center to display data analysis results.
  • the data center, the data acquisition machine, and the terminal are separately deployed in the railway company and the railway bureau, and only the data acquisition machine and the terminal are deployed in the electric power station and the station.
  • Real-time data analysis components are deployed in stations, power stations, railway bureaus, and railway companies.
  • the real-time data analysis components of stations and power stations are deployed in data acquisition machines.
  • the real-time data analysis components of railway bureaus and railway companies are deployed in data. Inside the center.
  • the computing engine layer includes an offline data processing architecture MapReduce, a real-time distributed processing architecture spark, a data movement engine sqoop, and an underlying resource management system yarn, Yarn for allocating storage resources and computing resources according to tasks submitted by the user for processing the task.
  • MapReduce offline data processing architecture MapReduce
  • real-time distributed processing architecture spark a real-time distributed processing architecture spark
  • data movement engine sqoop a data movement engine sqoop
  • underlying resource management system yarn Yarn for allocating storage resources and computing resources according to tasks submitted by the user for processing the task.
  • the component layer includes a data management component, a data analysis component, a fault diagnosis component, and the like.
  • the first comprehensive signal operation and maintenance platform needs to solve the problem of data acquisition.
  • railway bureaus, electric power stations and stations used by railways data collectors are also distributed at these four levels. Collect different types of signal data and perform different processing on the data; deploy data centers, data acquisition machines and terminals in railway companies and railway bureaus respectively, and only deploy data acquisition machines and terminals in the electricity service stations and stations;
  • the electric power segment, the railway bureau, and the railway company all deploy real-time data analysis components.
  • the real-time data analysis components of the station and the electric service segment are deployed in the data acquisition machine, and the real-time data analysis components of the railway bureau and the railway company are deployed in the data center;
  • each railway internal signal system uses a 2M bandwidth wired network for data transmission.
  • the signal data collected per second at a single station.
  • an electric service segment has about tens to hundreds of stations; this makes it impossible for the station data to be completely transmitted to the electricity section; according to this situation, the monitoring data is stored locally by the collector.
  • Strategy only part of the data is uploaded to the electricity section and the railway bureau, railway company;
  • the data acquisition machine of the station collects the real-time data of the station, and uses the real-time data analysis component deployed locally to perform pre-processing, feature extraction and feature selection on the signal data, and then analyze the feature data in real time by using the analysis model to obtain the current signal system. Operation status, and will transmit corresponding data (including extracted partial features and analysis results, including raw data of each station) to the electricity segment;
  • the data acquisition machine of the electric service segment accepts the data transmitted from the internal stations, and analyzes the data of the different stations received by the real-time data analysis component deployed locally, and obtains the analysis result of the entire electric service segment, and analyzes The result is transmitted to the data acquisition machine of the railway bureau;
  • the data acquisition machine of the railway bureau receives the data from each electrical segment and transfers the data to the data center. Then, the data is processed and analyzed by the real-time data analysis component deployed in the data center, and the analysis results and some features are analyzed. Data acquisition machine transmitted to the railway company;
  • the data acquisition machine of the railway company receives the data sent by the railway bureau, and transfers the data to the data center, and analyzes and processes the real-time data analysis component of the data center to obtain the global analysis result of the railway company;
  • the method further includes a step 7), which indexes the data stored by the railway company, the railway bureau, the electric power station and the station, and quickly acquires various types of signal data of the relevant time period and position according to the index when the fault occurs, so as to Quickly analyze, locate and resolve faults.
  • the data collector can reuse existing machines, and only needs to install corresponding data storage components, data preprocessing components, real-time data analysis components, and data transmission components.
  • the data collection machine of the railway company and the railway bureau only needs to deploy the data transmission component, and does not need to complete the functions of data storage and processing.
  • the data storage of the data acquisition machine adopts a hierarchical storage strategy. When the bandwidth of the railway network is sufficient, the extracted feature data can be directly transmitted to the upper layer, so that the collection machine does not need to store the signal data locally.
  • a major innovation of the present invention is a hierarchical data distributed storage method, which is not a simple distributed storage, but a new hierarchical data storage architecture designed based on the network restriction of the railway, that is, the original data is stored separately.
  • the collected layer in addition to storing the data of this layer, also needs to store the metadata of the lower layer (the metadata here mainly refers to the index data, that is, the storage information of the lower layer real data, such as where some data is stored) It is used to quickly locate the specific location of the data, and the upper layer manages the underlying data through metadata.
  • the electricity segment includes two types of data, one is the data collected by the electricity segment itself, and the other is the metadata of each station in the pipe area, including the station number, equipment number, time interval, storage location, etc. Positioning data. This can be pushed up.
  • the data acquisition machine of the station in step 2) obtains the signal data through the signal station machine of the station, and the station machine first determines whether the transmission needs to be performed through the existing rules, thereby performing compression processing on the signal data. If the value of the voltage (or other signal value) does not exceed 20% (or other threshold), the data does not need to be transmitted if it is considered that the data has not changed.
  • the data acquisition machine establishes a socket connection with the signal station machine of the station, and the signal station machine of the station transmits the compressed data to the data acquisition machine, and the data acquisition machine uses the existing rules to restore the data.
  • the real-time data analysis component described in steps 2) to 5) mainly performs fault prediction based on the obtained feature data, and classifies the feature by using the classification model to determine whether a fault occurs.
  • the real-time data analysis is based on the historical signal data that has been marked for mining to obtain a classification model for various types of faults.
  • the data analysis algorithms used by the real-time data analysis component include support vector machine, Bayesian classifier, rough set, decision tree, neural network, etc.
  • the algorithm uses the algorithm to uniformly process the collected signal data in the data center to mine fault identification.
  • the model is provided for use in real-time analysis.
  • step 6) the terminal of the station is mainly used to display various analysis results inside the station, including real-time operating status, and various failure analysis results; the terminal of the electric service segment is mainly used to display the The overall operating state in the electricity section, when the station in the pipeline area fails, the terminal also needs to display the corresponding fault information; the terminal of the railway bureau is used to display the operating status of the railway bureau level, various fault alarm information, etc.
  • the terminal of the railway company is used to show the operating status of the entire railway company, and it needs to be quickly displayed and tracked when the station fails.
  • step 7 needs to first establish an index according to the stored data at each station, and then transfer the index file of the station to the electricity segment, and the electricity segment establishes the locally collected data and the received station index file.
  • the secondary index is then transmitted to the data center of the railway bureau.
  • the data center of the railway bureau establishes a three-level index on the various types of data received, and then transmits the index to the data center of the railway company, and the data center of the railway company receives the data.
  • Various types of data establish a four-level index to form a four-level index of the railway company, railway bureau, electricity section, and station.
  • the index can be established according to the railway bureau, the electricity section, the station, and the equipment. Further, by means of the established index file, specific signal data can be directly obtained according to the time and location information of the fault.
  • the invention unifies the comprehensive management of railway signal data, including data collection, data storage, data analysis and Fruit display.
  • the life cycle of the entire signal data is managed uniformly to form an organic whole, which improves the utilization of signal data.
  • the invention collects and processes various signal data dispersed in each electric power station and station through the data acquisition system, thereby changing the status quo of isolation of various signal data. It provides a good acquisition platform for further mining of signal data in the future.
  • the invention adapts to the four-level management mode of the railway company, the railway bureau, the electricity section and the station through the hierarchical deployment mode, and the hierarchical deployment mode is also applicable to the current network status of the railway, and the data is optimized.
  • the centralized approach greatly reduces the amount of data transferred to the electricity segment and the railway company.
  • the invention saves a large amount of labor cost by using the model to identify the fault, no need to manually observe the monitoring information and then perform fault identification and analysis; can improve the accuracy of the fault identification of the rail transit monitoring data, shorten the fault repair time, and greatly improve The efficiency of fault handling of rail transit improves the operation and maintenance capabilities.
  • the invention can realize real-time analysis of the signal data, improve the real-time performance of the data analysis, and the analysis result can be directly displayed locally.
  • the invention realizes the storage requirement for the increasing signal data through the hierarchical storage and indexing strategy, and realizes the rapid positioning of the fault through the four-layer index structure, and the relevant personnel can quickly obtain the information through the time, type, location and the like of the fault occurrence.
  • the signal data corresponding to the occurrence of the fault is retrieved. This allows technicians and managers to locate faults more quickly and directly, and propose solutions.
  • Figure 1 is a schematic diagram of the organization structure of the railway signal system.
  • FIG 2 is an overall architectural diagram of the integrated operation and maintenance system of the present invention.
  • FIG 3 is an overall deployment diagram of the data acquisition machine, data center, and terminal of the present invention.
  • FIG. 4 is a diagram of a data center architecture of the present invention.
  • Figure 5 is a schematic diagram of the operation of the integrated operation and maintenance system of the present invention in a digital railway "integrated" platform.
  • Figure 6 is a flow chart showing a 25 Hz phase sensitive track circuit for distinguishing between indoor and outdoor faults in the embodiment.
  • the integrated operation and maintenance scheme of the railway signal of the invention is used for solving the technical problems of the signal data dispersion processing, the large workload, the low efficiency, the high risk, and the difficulty of fault inquiry in the prior art.
  • the user objects of the present invention are mainly state-owned railways and domestic Large enterprise railways, such as urban rail transit.
  • the railway signal system is divided into four levels: the railway head office, the railway bureau, the electric power station and the station. The following describes the internal departments of the national railway signal system and their corresponding functions in conjunction with Figure 1.
  • the railway company's electrical department is in charge of the railway signal equipment of all roads in the country. It is the final gathering point of all monitoring data. It has the widest coverage, the most comprehensive types of equipment, and the largest amount of data for analysis.
  • the main tasks include: (1) Equipment angle: Keep track of the working status of all signal equipment of each road station; use the statistical report function to understand the failure rate, fault type, fault impact and other information of various signal equipment in actual application; Maintenance of signal equipment of each road station; through the statistical report function, compare the comprehensive performance of different manufacturers, different types, different types of signal equipment, provide data support and quantitative reference for equipment procurement, operation, maintenance, etc.; through data integration Analysis and discovery of weak links in signal equipment, providing targeted recommendations for equipment upgrades and introduction of new systems and new equipment; (2) Operation and maintenance perspective: mastering the operation and maintenance of railway bureaus, and arranging inspection priorities in a targeted manner, Focus on rectification and prevent problems before they happen.
  • the types of signal equipment of each road bureau are relatively uniform.
  • the main tasks include: (1) equipment angle: check the alarm information of each signal equipment; use the statistical report function to understand the failure rate of various signal equipment in practical applications, Information such as the type of fault and the impact of the fault; timely understand the fault handling situation; monitor the working status of the signal equipment in charge; when the fault occurs, comprehensively compare the alarm information and status monitoring information of the vehicle equipment and the ground equipment to determine the root cause of the fault.
  • equipment angle check the alarm information of each signal equipment
  • use the statistical report function to understand the failure rate of various signal equipment in practical applications, Information such as the type of fault and the impact of the fault; timely understand the fault handling situation; monitor the working status of the signal equipment in charge; when the fault occurs, comprehensively compare the alarm information and status monitoring information of the vehicle equipment and the ground equipment to determine the root cause of the fault.
  • Through statistical reports and data analysis functions compare the equipment of different manufacturers, and provide corresponding requirements and opinions to the equipment manufacturers
  • Operation and maintenance angle master the operation and maintenance of each electrical service
  • the station includes ordinary workshops, vehicle workshops, and transportation workshops, and some stations also include high-speed workshops. They are used to manage different monitoring devices, including in-station signal equipment and interval signal equipment, vehicle equipment, CTC, RBC, TDCS and other equipment.
  • the fault is detected and the cause of the fault is identified through the integrated monitoring system. Identify the location, equipment, and other information of the fault through various statistical reports, perform key monitoring on the location where the fault occurrence rate is high, and perform statistical analysis on the equipment that is prone to failure, providing a basis for equipment replacement and procurement.
  • the hierarchical data storage and processing method of the invention is very suitable for the classification characteristics of the national railway and the data analysis of each level. demand.
  • the invention adopts a four-level mode of a railway company, a railway bureau, an electric power section and a station, and respectively deploys a data center, a data acquisition machine and a terminal in a railway company and a railway bureau, and only deploys a data acquisition machine and a terminal in an electric power section and a station.
  • Real-time data analysis components are deployed at stations, power stations, railway bureaus, and railway companies.
  • the real-time data analysis components of stations and power stations are deployed in data acquisition machines.
  • the real-time data analysis components of railway bureaus and railway companies are deployed in data centers.
  • the railway Administration and the railway Corporation also need to deploy an offline data mining system in the data center to analyze and mine the models used in real-time data analysis.
  • the solution of the present invention mainly includes the following six parts:
  • Railway signals include many types, such as communication data, vehicle data, trackside data, station data, etc.
  • the resulting location also includes stations, electricity stations, railway bureau management centers, railway company centers, etc., so the data acquisition subsystem also needs It is deployed at the station, the electricity section, the railway bureau and the railway company signal center.
  • the data acquisition subsystem needs to receive and parse the signal data according to the specifications of the signal data.
  • the signal acquisition subsystem transmits the collected signal data to the data preprocessing subsystem to analyze and process the signal data.
  • the data preprocessing subsystem receives the signal data collected from the data acquisition subsystem, finds the signal data related to the fault according to different faults, and then performs preprocessing such as deduplication and denoising on the data to ensure the validity of the data.
  • the signal data the data is transformed into a space vector model, and feature extraction algorithms such as information gain algorithms are used to extract, de-duplicate and select features according to the needs, and find features useful for fault classification, and finally generate suitable for model training and real-time.
  • the eigenvector of the analysis is used to extract, de-duplicate and select features according to the needs, and find features useful for fault classification, and finally generate suitable for model training and real-time.
  • the collected data is first manually labeled, and various types of fault data are obtained. Through the training of these data, the corresponding classification model and parameters can be generated for the next step of signal data analysis.
  • the subsystem analyzes the historically monitored data by manual data, firstly pre-processes the data already marked, generates feature vectors with labeled categories, and then selects appropriate features and initial parameters to train the data, thereby Get the model of the fault classification.
  • the subsystem receives the real-time signal data processed by the data pre-processing subsystem, and uses the classification model obtained by the classification model mining subsystem to analyze and calculate the data to obtain the current operating state of the system. If there is no fault, the normal display will be performed. When the fault occurs, the relevant personnel should be reminded by warning.
  • the subsystem is a variety of display terminals, deployed in the signal center of the station, the electricity section, the railway bureau and the railway company, for real-time display of the operational status of the jurisdiction, timely reminding of alarm information, query signal data, etc., provided to the site Used by technicians and managers. It is convenient for relevant personnel to understand the current running status of the system more intuitively.
  • the results and scope of the signal data analysis for different levels are also different.
  • the station is real-time display of the real-time analysis of the station; the electricity section shows the jurisdiction Part of the analysis results of all stations and all fault information; the railway company station section shows the operating status and analysis results of all the electric power stations and stations in the company.
  • steps (2) ⁇ (4) because there is no data center deployed at the station and the electricity section, it is completed in the collection machine (ie real-time data analysis component); in the railway bureau and railway company Completed in the data center.
  • the railway bureau and the railway company can also directly deploy the data acquisition component on the central server without the need of a special data acquisition machine to realize the function of the data acquisition machine.
  • the railway signal system departments only query and analyze the signal data in the jurisdiction, so the data requirements have obvious locality.
  • the signal data storage is carried out in a hierarchical manner.
  • a hierarchical data indexing strategy is adopted. That is to say, different data index structures are used in stations, electric power stations, railway bureaus and railway companies to ensure the speed of data query.
  • the concept of the data block (Block) is introduced here, which has a fixed storage space of 64M, and the signal data storage devices of the entire network are uniformly numbered by the data center server of the railway company.
  • Deviceid the number of the monitoring equipment in the workshop
  • Blockid the data block number, this is the global unique number, the specific location of the data storage, can be directly mapped to the corresponding physical storage location through this id;
  • the signal data is at the beginning of the data block
  • Length the size of the signal data, which is also the storage length occupied by the data
  • the index data of the station is generated during data storage and stored in the local machine.
  • the station terminal performs data query, it first reads the index file into the local machine, and then reads the actual data according to the data block and the offset.
  • the electrical segment stores two kinds of data, which are the actual signal data collected and the index data.
  • Index data is also divided into two categories: index of station index data; index of local data.
  • the local data index is similar to the station data, except that the workid becomes the depotid representing the electricity segment, as shown in Table 2.
  • Table 3 is the data structure of the index of the station index data:
  • the number of the station is unique within the entire road;
  • the index adds a station number to the station index to identify the station.
  • An index record represents the specific location of the index file storage of a station work area, and the length of the index file.
  • the railway bureau is similar to the electricity section.
  • the stored data is divided into two categories: the data collected by the railway bureau itself and the index data.
  • the index data is divided into a local data index and an index of the electrical segment index in the pipe area.
  • the local data index data structure is shown in Table 4:
  • Officeid is the number of the railway bureau used to identify the unique code of the railway bureau.
  • the index data structure of the electrical segment index is shown in Table 5:
  • the index is added with an electrical segment number relative to the index of the electrical segment to identify the electrical segment.
  • An index record represents the specific location of the index file storage of an electrical segment, and the length of the index file.
  • the railway company is similar to the railway bureau.
  • the stored data is divided into two categories: the data collected by the railway company itself and the number of indexes. according to.
  • the index data is divided into a local data index and an index of each railway station index data.
  • the local data index data structure is shown in Table 6:
  • the index data structure of the railway bureau index is shown in Table 7:
  • the index adds a railway bureau number to the railway bureau's index to identify the railway bureau.
  • An index record represents the specific location of the index file storage of a railway bureau, and the length of the index file.
  • the invention realizes centralized management of railway signal data by using the integrated operation and maintenance platform, and realizes unified collection and centralized storage of various signal data of the railway.
  • the data preprocessing subsystem performs unified preprocessing on the collected monitoring data to extract useful features for data mining and real-time analysis.
  • the classification model mining subsystem analyzes and processes the labeled data, finds the appropriate classification model and parameters for the specific fault problem, and transmits the mining result to the real-time data analysis subsystem.
  • the data analysis subsystem uses the classification model to classify the real-time signal feature data after pre-processing, obtains the fault analysis result, and transmits the result to the result display subsystem, and the result display subsystem displays the corresponding analysis result according to the position where it is located. This includes the interior of the station, the interior of the electricity section and the operation of the entire railway company.
  • the specific process of the method is specifically illustrated by the following examples and examples:
  • the 2 is an overall architecture diagram of the system, which includes a data acquisition system (ie, a data acquisition machine), a data center, and a terminal.
  • the data center includes structured data storage, semi-structured data storage, unstructured data storage, and offline data processing. , real-time big data processing, statistical analysis, data mining, fault warning, query engine, push engine and other modules.
  • the terminal can be various types of receiving terminals, including PC computers, notebooks, pads, smart phones, and the like.
  • Figure 3 is a schematic diagram of the deployment of the system, deployed in four levels, namely the railway company, the railway bureau, the electricity section and the station.
  • the specific architecture of the data center refers to Figure 3. Because this data center needs to manage and process all kinds of signal data of all the electric power stations and stations in the entire railway company, it requires large storage and computing power.
  • the data acquisition machine is mainly used to receive data sent by various railway bureaus. To deploy the data acquisition subsystem and the data preprocessing subsystem, it is responsible for pre-processing the collected signal data and transmitting it to the data center; the terminal is connected with the data center to display the running status of the entire company, including real-time signal display and signal. Data query and fault alarm.
  • the railway bureau needs to deploy a small data center, data collector and terminal.
  • the data center of the railway bureau is used to receive and process the signal data uploaded by the internal electricity section. It needs to deploy components including distributed file system, parallel processing architecture, real-time data analysis, etc.
  • the data acquisition machine is mainly used to collect the various generated at the railway bureau level. Class signal data, and preprocess the data and transmit it to the data center.
  • the terminal is used to display, query and alert various types of signal data in the railway bureau.
  • the electricity section and the station do not need to deploy a data center, only data collectors and terminals are required. In addition to the functions of signal acquisition, these collectors also have the ability of data preprocessing, real-time data analysis, data storage and data transmission. After the data acquisition machine collects the data, the data is pre-processed and analyzed in real time, and the analysis result is displayed through the local terminal, and part of the analysis result and signal data are transmitted to the data center of the railway bureau.
  • the data acquisition system of the electrical segment receives signal data and analysis results transmitted from various stations in the pipe area.
  • the terminal is used to display and query various signal data and analysis results.
  • the station only needs to deploy the data acquisition machine and terminal as the electricity service segment. Its function is similar, except that the data acquisition machine of the station only needs to collect the signal data of the station, and pre-process and analyze the data in real time, and according to the demand. Part of the data and analysis results are transmitted to the collector of the electricity segment.
  • the terminal is mainly used to display various signal data and real-time analysis of data in the station.
  • the cloud security layer ensures system security through data backup, data restoration and access control.
  • the base resource layer builds the hardware platform of the data center, which can reuse existing server devices.
  • the virtualization layer virtualizes the hardware through virtualization software, shields the underlying hardware differences, and forms a highly available, scalable, and scalable cluster.
  • the data storage layer is composed of a distributed file system, a columnar database, and a relational database for storing unstructured, semi-structured, and structured data, respectively.
  • the computing engine layer includes the offline data processing architecture MapReduce, the real-time distributed processing architecture spark, the data movement engine sqoop and the underlying resource management system yarn. Yarn is used to allocate storage resources and computing resources for processing tasks based on tasks submitted by users.
  • the component layer is a component that specifically handles various services, including data management components, data analysis components, and fault diagnosis components. The results of various data collection and analysis need to be published through the architecture of B/S or C/S.
  • the Web cluster is a service publishing layer based on load balancing.
  • the cloud computing-based integrated operation and maintenance system of the present invention can be integrated with other processing platforms.
  • a digital railway integration platform is built using "general components” and “business plug-ins”.
  • the "data center and big data management & data mining analysis platform” can be realized based on the invention of a common component, which is the core of data acquisition, storage and data processing, and with the equipment comprehensive monitoring platform, transportation platform, operation and maintenance. Platforms, scheduling and collaboration platforms are integrated to form a digital railway integration platform.
  • Figure 6 is a flow chart of a 25 Hz phase sensitive track circuit that distinguishes between indoor and outdoor faults.
  • the analysis of the cause of track circuit failure is a classification problem, which is very suitable for data mining.
  • the signal acquisition associated with it comes from the station equipment.
  • the first is the model training phase, where the existing data is manually labeled to identify those that are faulty. In this way, a training set is formed.
  • the data mining subsystem is used to train these data sets, and the Bayesian classifier is selected as the training model to obtain corresponding model parameters.
  • These classification models are then deployed in the station's data acquisition machine.
  • the data acquisition machine first receives the monitoring data from each monitoring device and then preprocesses the data to obtain the feature vector for analysis.
  • the real-time data analysis component uses the obtained classification model to analyze and calculate the real-time data to obtain whether the system is currently faulty.
  • the results of the failure analysis are then presented through the display system, and the relevant analysis results and features are transmitted to the electricity section, the railway bureau, and the railway center for analysis at a higher level.
  • Junction box receiving terminal voltage Open the outdoor side voltage of the cable terminal Send terminal voltage Data acquisition timestamp 25.00 25.00 25.00 521365 24.00 24.00 25.00 521365 27.00 27.00 27.00 521365 0.00 0.00 0.00 521365 ... ... ... ... ...
  • the first column of numbers represents the type of fault:
  • ⁇ 1 indicates that the fault is indoors
  • ⁇ 2 indicates that the fault is outdoors

Landscapes

  • Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Economics (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Train Traffic Observation, Control, And Security (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本发明涉及一种基于云计算的轨道交通信号综合运维方法及***。采用铁路公司、铁路局、电务段、车站的四级模式，在铁路公司、铁路局分别部署数据中心、数据采集机和终端，在电务段、车站只部署数据采集机和终端；同时部署实时数据分析组件，车站、电务段的实时数据分析组件部署于数据采集机内，铁路局、铁路公司的实时数据分析组件部署于数据中心内。本发明采用层次性的部署方式，通过在采集机上直接部署实时数据分析组件实现对信号数据的实时分析，通过分级存储和索引策略实现了对不断增长的信号数据的存储需求，并且通过四层索引结构实现对故障的快速定位，可以通过故障发生的时间、类型、位置等信息快速的检索到故障发生时对应的信号数据。

Description

一种基于云计算的轨道交通信号综合运维方法及*** 技术领域
本发明提供一种基于云计算的铁路信号综合运维方法及***，涉及铁路信号数据、铁路通信数据、铁路知识数据、***报警数据、机器学习、车载设备、车站设备、中心设备、轨旁设备等技术领域，用以解决铁路信号综合运维所面临的问题。
背景技术
目前，轨道交通(国有铁路、企业铁路和城市轨道交通)通信、信号领域的监测维护产品主要有三类：针对地面信号设备的有CSM(信号集中监测***)、各设备维护机、通信网管***。为了提高我国铁路信号***设备的现代化维修水平，从90年代开始，先后自主研制了TJWX-I型和TJWX-2000型等不断升级中的信号集中监测CSM***。目前大部分车站都采用了信号集中监测***，实现了对车站信号设备状态的实时监测，并通过监测与记录信号设备的主要运行状态，为电务部门掌握设备的当前状态和进行事故分析提供了基本依据，发挥了重要作用。并且，对城市轨道交通信号设备，集中监测CSM***也被广泛部署在城轨集中站/车辆段等处，供城轨运维使用。针对车载信号设备的有DMS***，对车载设备动态监测。此外，伴随我国高速铁路的建设发展，高铁特有的RBC***等，也面临着纳入信号集中监测***的需求，也面临着提高其监测能力、运维能力，以及设备自诊断能力的需求。
面对越来越多的通信数据、信号数据，以及目前的网络状况，使得铁路***虽然产生了大量的各类数据，但是却没有很好的利用这些数据。绝大多数数据目前是存储在各个车站，并且车站也只是对数据做了简单的存储，当出现故障的时候查看相关的数据以便于分析故障原因。这种情况导致了铁路产生的各类数据不能够有效的存储和利用，并且故障的分析目前仍需依靠人工经验分析判断，很多情况下在已经出现事故时才能发现故障，不仅导致了人工诊断铁路信号***故障时工作量大、故障监测与诊断效率低下等技术问题，也增加了行车的危险。因此，提高铁路信号综合管理运维能力，查隐患，治隐患，推动故障修向状态修发展，从而保障行车安全、提高运力，是轨道交通领域的迫切需求。
发明内容
为了解决现有技术中信号数据量大、效率低、风险高、人工分析故障劳动强度大等问题， 本发明提供了一种基于云计算的信号综合运维方法及***。***包括数据采集机、终端和数据中心三种硬件。软件包括数据采集子***、数据存储子***、数据预处理子***、数据实时分析子***、离线数据挖掘子***、展示子***、数据查询***等。
本发明采用的技术方案如下：
一种基于云计算的轨道交通信号综合运维***，其包括：
数据采集机，用于采集轨道交通信号数据；
数据中心，连接所述数据采集机，包括云安全层、基础资源层、虚拟化层、数据存储层、计算引擎层、组件层及Web集群；其中，云安全层通过数据备份、数据还原和访问控制，保障***的安全；基础资源层是数据中心的硬件平台；虚拟化层通过虚拟化软件实现对硬件的虚拟化，屏蔽底层的硬件差异；数据存储层包括分布式文件***、列式数据库、关系数据库，分别用于存储非结构化、半结构化和结构化的数据；计算引擎层采用云计算处理架；组件层是具体处理各种业务的组件；Web集群是基于负载均衡的服务发布层；
终端，连接所述数据采集机和所述数据中心，用于展示数据分析结果。
进一步地，采用铁路公司、铁路局、电务段、车站的四级模式，在铁路公司、铁路局分别部署数据中心、数据采集机和终端，在电务段、车站只部署数据采集机和终端。在车站、电务段、铁路局、铁路公司均部署实时数据分析组件，其中车站、电务段的实时数据分析组件部署于数据采集机内，铁路局、铁路公司的实时数据分析组件部署于数据中心内。
进一步地，计算引擎层包括离线数据处理架构MapReduce、实时分布式处理架构spark以及数据移动引擎sqoop和底层资源管理***yarn，Yarn用于根据用户提交的任务来分配存储资源和计算资源用于处理该任务。
进一步地，组件层包括数据管理组件、数据分析组件、故障诊断组件等。
一种采用上述***的基于云计算的轨道交通信号综合运维方法，其步骤包括：
1)信号综合运维平台首先需要解决的是数据采集的问题，针对当前铁路采用的铁路公司、铁路局、电务段、车站的四级管理模式，数据采集机也分布在这四个层次用于采集不同类型的信号数据，并对数据做不同的处理；在铁路公司、铁路局分别部署数据中心、数据采集机和终端，在电务段、车站只部署数据采集机和终端；在车站、电务段、铁路局、铁路公司均部署实时数据分析组件，其中车站、电务段的实时数据分析组件部署于数据采集机内，铁路局、铁路公司的实时数据分析组件部署于数据中心内；
鉴于铁路目前的网络现状，各个铁路内部信号***均是采用2M带宽的有线网络进行数据传输，随着车站采集的信号数据种类和频率的不断提高，单个车站每秒采集到的信号数据 约为200-500kb之间，一个电务段有大约几十到上百个车站；这样就导致车站的数据不可能完全传输到电务段；根据这种情况对监测数据采用采集机本地存储的策略，仅把部分数据上传电务段及铁路局、铁路公司；
2)车站的数据采集机采集车站的实时数据，利用在本地部署的实时数据分析组件对信号数据进行预处理、特征提取、特征选择，然后利用分析模型对特征数据进行实时分析，得到信号***当前的运行状态，并会将相应的数据(包括提取的部分特征以及分析结果，也可包括各车站的原始数据)传输至电务段；
3)电务段的数据采集机接受来自内部的各个车站传输过来的数据，利用本地部署的实时数据分析组件对接收的不同车站的数据进行分析，得到整个电务段的分析结果，并将分析结果传输至铁路局的数据采集机；
4)铁路局的数据采集机接收来自各个电务段的数据，并将数据转存到数据中心，然后利用数据中心部署的实时数据分析组件对数据进行处理和分析，并将分析结果及部分特征传输到铁路公司的数据采集机；
5)铁路公司的数据采集机接收铁路局发送过来的数据，并将数据转存至数据中心，利用数据中心的实时数据分析组件进行分析处理，得到铁路公司全局的分析结果；
6)利用部署于铁路公司、铁路局、电务段和车站的终端，进行数据分析结果的展示。
进一步地，还包括步骤7)，该步骤对铁路公司、铁路局、电务段及车站存储的数据建立索引，在出现故障时根据该索引快速获取相关时间段及位置的各类信号数据，以便于快速的进行故障分析、定位和解决。
进一步地，所述数据采集机可以复用已有的机器，只需要安装相应的数据存储组件、数据预处理组件、实时数据分析组件和数据传输组件。铁路公司、铁路局的数据采集机只需要部署数据传输组件，不需要完成数据存储、处理的功能。数据采集机的数据存储采用分级存储策略，当铁路网络带宽足够的情况下可以直接将抽取的特征数据传输到上层，这样采集机本地就不需要存储信号数据。
进一步地，所述数据采集机对于采集到的本地信号数据保存在本地***中，然后在上层以元数据的形式来管理数据的存储结构。本发明的一个主要的创新是层次型数据分布式存储方式，并不是简单的分布式存储，而是基于铁路的网络限制而设计的一种新的层次数据存储架构，即：原始数据分别存储在被采集的层次，上层除了存储本层的数据外，还需要存储下层的元数据(元数据这里主要指的是索引数据，也就是下层真实数据的存储信息，比如某些数据存储在什么位置)，用于快速的定位数据的具***置，上层通过元数据来管理下层的数据。 在数据计算时，也不是简单的将下层分析结果向上传输，除了需要传输分析结果外，还需要传输部分原始数据作为上层数据分析的依据。从而形成一种层次型的数据存储、处理架构。如电务段包括两类数据，一类是电务段本身采集的数据，另外是管区内的各个车站的元数据，包括车站编号、设备编号、时间区间、存储位置等信息，以保障可以快速的定位数据。向上可以此类推。
进一步地，步骤2)中车站的数据采集机通过车站的信号站机获取信号数据，站机首先通过已有的规则判断是否需要进行传输，从而对信号数据进行压缩处理。如电压(也可以是其它信号值)的取值在标准值波动不超过20％(也可以是其它阈值)的情况下，认为数据没有发生变化，则不需要传输。数据采集机通过与车站的信号站机建立socket连接，车站的信号站机将压缩处理后的数据传输到数据采集机，数据采集机利用已有的规则对数据进行还原。
进一步地，步骤2)～5)所述的实时数据分析组件主要是根据得到的特征数据进行故障预测，利用分类模型对特征进行分类，判断是否产生故障。具体地，实时数据分析是根据已经标注的历史信号数据进行分析挖掘得到针对各类故障的分类模型。实时数据分析组件使用的数据分析算法包括支持向量机、贝叶斯分类器、粗糙集、决策树、神经网络等，利用算法对采集到的信号数据在数据中心进行统一的处理，挖掘出故障识别的模型，提供给实时分析组建使用。
进一步地，步骤6)中：所述的车站的终端主要用于展示该车站内部的各类分析结果，包括实时的运行状态，以及各种故障分析结果；电务段的终端主要用于展示该电务段内整体的运行状态，当管区内的车站出现故障的时候，该终端也需要展示对应的故障信息；铁路局的终端，用于展示铁路局层面的运行状态，各种故障报警信息等；铁路公司的终端，用于展示整个铁路公司的运行状态，当车站出现故障时也需要快速的展示和跟踪。
进一步地，步骤7)所述的过程需要首先在各个车站根据存储的数据建立索引，然后将车站的索引文件传输到电务段，电务段对本地采集的数据及接收到的车站索引文件建立二级索引，然后将索引传输至铁路局的数据中心，铁路局的数据中心对接收的各类数据建立三级索引，然后将索引传输至铁路公司的数据中心，铁路公司的数据中心对接收的各类数据建立四级索引，形成一个铁路公司、铁路局、电务段、车站的四级索引。具体实施时可以按照铁路局、电务段、车站、设备等方式建立索引。进而借助于建立的索引文件，可以根据故障的时间、位置信息直接获取到具体的信号数据。
与现有技术相比，本发明的优点是：
本发明统一了铁路信号数据的综合管理，包括数据的采集、数据存储、数据分析以及结 果展示。将整个信号数据的生命周期统一管理，形成一个有机的整体，提高了信号数据的利用率。
本发明通过数据采集***，将分散在各个电务段、车站的各种信号数据进行统一的采集和处理，从而改变了各种信号数据相互隔离的现状。为未来对信号数据进行进一步挖掘提供了良好的采集平台。
本发明通过层次性的部署方式，适应了铁路的铁路公司、铁路局、电务段及车站的四级管理模式，并且这种层次性的部署方式也是适用于铁路目前的网络现状，优化了数据集中的方式，大大减少了传输到电务段和铁路公司的数据量。
本发明通过使用模型识别故障，节省了大量的人力成本，不再需要人工的去观察监测信息然后进行故障识别和分析；能够提高轨道交通监测数据故障识别的准确率，缩短故障修复时间，大大提高轨道交通的故障处理效率,提高运维能力。
本发明通过在采集机上直接部署实时数据分析组件，可以实现对信号数据的实时分析，提高了数据分析的实时性，分析结果可以直接在本地进行展示。
本发明通过分级存储和索引的策略实现了对不断增长的信号数据的存储需求，并且通过四层索引结构实现对故障的快速定位，相关人员可以通过故障发生的时间、类型、位置等信息快速的检索到故障发生时对应的信号数据。这样技术人员和管理人员都可以更加快速直接的定位故障，并提出解决方案。
附图说明
图1是铁路信号***组织结构示意图。
图2是本发明的综合运维***的整体架构图。
图3是本发明的数据采集机、数据中心和终端的整体部署图。
图4是本发明数据中心架构图。
图5本发明的综合运维***在数字铁路“一体化”平台中的作用示意图。
图6是实施例中一个25Hz相敏轨道电路区分室内室外故障的流程图。
具体实施方式
下面通过具体实施例和附图，对本发明做详细的说明。
本发明的铁路信号的综合运维方案用于解决现有技术中信号数据分散处理、工作量大、效率低下、风险性高、故障查询困难等技术问题。本发明的用户对象主要是国有铁路和国内 的大型企业铁路，如城市轨道交通等。从图1可以看出，铁路信号***内部分为铁路总公司、铁路局、电务段和车站四级。下面结合图1来介绍国铁信号***内部各级部门以及它们对应的职能。
1)铁路总公司
铁路总公司电务部门主管全国各路局的铁路信号设备，是所有监控数据的最终汇集点，设备覆盖面最广、涉及设备种类最全面、用于分析的数据量最大。主要的任务包括：(1)设备角度：随时掌握各路局全部信号设备的工作状态；通过统计报表功能，了解各种信号设备在实际应用中的故障率、故障种类、故障影响等信息；掌握各路局信号设备的维护情况；通过统计报表功能，对不同厂家、不同种类、不同型号的信号设备的综合性能进行比较，为设备采购、运营、维护等提供数据支持及量化参考；通过数据整合分析，发现信号设备的薄弱环节，为设备的升级改造及新***、新设备的引入提供针对性建议；(2)运维角度：掌握各铁路局的运维情况，有针对地安排检查重点、整改重点，防患于未然。
2)铁路局
各路局的信号设备种类相对来说比较统一，主要的任务包括：(1)设备角度：查看各信号设备的报警信息；通过统计报表功能，了解各种信号设备在实际应用中的故障率、故障种类、故障影响等信息；及时了解故障的处理情况；监控所主管信号设备的工作状态；发生故障时，综合对比车载设备和地面设备的报警信息、状态监控信息等，判断导致故障的根本原因；通过统计报表及数据分析等功能，对不同厂商设备进行比较，并向设备厂家提供相应需求及意见；(2)运维角度：掌握所属各个电务段的运维情况，有针对地安排检查重点、整改重点，防患于未然。
3)电务段
对其管内的所有信号设备进行总体监控，对其下属的各车间进行工作指导。进行设备的具体维护工作，又关心各类数据的总体分析处理及比较结果。
4)车站/工区
车站包括普通车间、车载车间、运调车间，有些车站还包括高速车间等。分别用于管理不同的监测设备，包括站内信号设备及区间信号设备、车载设备、CTC、RBC、TDCS等设备。通过综合监控***及时发现故障并识别故障的原因。通过各种统计报表识别故障发生的地点、设备等信息，针对故障发生率较高的位置进行重点的监控，对易出现故障的设备进行统计分析，为设备更换和采购提供依据。
本发明的分级式的数据存储和处理方式非常适用于国铁的分级特点以及各级的数据分析 需求。本发明即采用铁路公司、铁路局、电务段、车站的四级模式，在铁路公司、铁路局分别部署数据中心、数据采集机和终端，在电务段、车站只部署数据采集机和终端。在车站、电务段、铁路局、铁路公司部署实时数据分析组件，其中车站、电务段的实时数据分析组件部署于数据采集机内，铁路局、铁路公司的实时数据分析组件部署于数据中心内。铁路局、铁路总公司还需要在数据中心内部署离线数据挖掘***，用于分析挖掘实时数据分析所使用的模型。本发明的方案主要包括下面六个部分的内容：
(1)数据采集：
需要将铁路产生的各类信号数据采集起来，包括通信数据、车载设备数据、车站设备数据、中心设备数据、轨旁设备数据等。
铁路信号包括多种类型，如通信数据、车载数据、轨旁数据、车站数据等，产生的位置也包括车站、电务段、铁路局管理中心、铁路公司中心等，因此数据采集子***也需要部署在车站、电务段、铁路局和铁路公司信号中心。数据采集子***需要根据信号数据的规范，接收并解析信号数据。信号采集子***将采集到的信号数据传输给数据预处理子***，对信号数据进行分析处理。
(2)数据预处理：
数据预处理子***接收来自数据采集子***采集到的信号数据，根据不同的故障，找到与该故障相关的信号数据，然后对数据首先进行去重、去噪等预处理，保证数据的有效性。根据信号数据将数据转换为空间向量模型，并根据需要，利用特征选择算法比如信息增益算法等对特征进行抽取、去重和选择，找到对故障分类有用的特征，最终生成适于模型训练和实时分析的特征向量。
(3)分类模型挖掘：
对于采集到的数据首先进行人工标注，得到标注的各类故障数据，通过这些数据进行训练，可以产生相应的分类模型及参数，用于下一步的信号数据分析。
该子***针对通过对人工标注的历史监测数据的分析，首先对这些已经标注的数据进行数据预处理，产生带标注类别的特征向量，然后选择适当的特征和初始参数对这些数据进行训练，从而得到故障分类的模型。
选择符合规律的模型，对模型进行训练，找到针对具体的故障找到相应的分类模型以及参数，使其对该类故障具有最好的分类效果。
(4)信号数据实时分析：
包括数据过滤、去重、特征提取、特征选择、归一化等步骤，然后利用故障模型对这些 数据进行故障预测和分类。该子***接收经过数据预处理子***处理之后的实时信号数据，利用分类模型挖掘子***得到的分类模型对这些数据进行分析计算，得到当前***的运行状态。没有故障的时候进行正常的显示，当出现故障的时候需要以警告的方式来提醒相关人员。
(5)分析结果展示：
该子***即各种显示终端，部署于车站、电务段、铁路局和铁路公司的信号中心，用于实时的显示管辖区的运行状态、及时提醒报警信息、查询信号数据等，提供给现场技术人员和管理人员使用。便于相关人员可以更加直观的了解***当前的运行状态。针对不同层次(车站、电务段、铁路局、铁路公司)的信号数据分析的结果的展示方式和范围也有所不同，车站是实时展示该车站内部的实时分析情况；电务段展示的是管辖所有车站的部分分析结果及所有的故障信息；铁路公司站段展示的是公司内所有电务段、车站的运行状态和分析结果。
对于上面的步骤(2)～(4)，在车站和电务段两级因为没有部署数据中心，因此就是在采集机中完成的(即实时数据分析组件)；而在铁路局和铁路公司则在数据中心完成。具体实施时，铁路局和铁路公司也可以不需要专门的数据采集机，直接把数据采集组件部署在中心服务器上，实现数据采集机的功能。
(6)对铁路公司、铁路局、电务段及车站存储的数据建立索引，在出现故障时根据该索引快速获取相关时间段及位置的各类信号数据，以便于快速的进行故障分析、定位和解决。
a)分级索引的结构
铁路信号***各级部门只对管辖区内的信号数据进行查询和分析，因此数据需求具有明显的局部性。据铁路***目前的网络现状以及数据需求的特点，采用分级的方式进行信号数据存储，为了快速高效的进行数据查询，采用分级式的数据索引策略。也就是在车站、电务段、铁路局、铁路公司分别采用不同的数据索引结构来保证数据查询的速度。这里引入数据块(Block)的概念，是具有64M固定大小的存储空间，由铁路公司的数据中心服务器对全网的信号数据存储设备进行统一的编号。
b)车站数据索引结构，如表1所示。
表1.车站数据索引结构
workid deviceid blockid offset length
workid：车间的编号，在车站内部是唯一的；
deviceid：车间内监测设备的编号；
blockid：数据块编号，这个是全局唯一的编号，数据存储的具***置，可以通过这个id直接映射到对应的物理存储位置；
offset：信号数据在该数据块的起始位置；
length：信号数据的大小，也是数据占用的存储长度；
车站的索引数据在数据存储时生成，并存储在本地机器中，车站终端进行数据查询时，首先到本地机器中读取索引文件，然后根据数据块、偏移量去读取实际数据。
c)电务段数据索引结构
电务段存储两种数据，分别是采集到的实际的信号数据以及索引数据。索引数据也分为两类：车站索引数据的索引；本地数据的索引。本地数据索引与车站数据类似，只是workid变成代表电务段的depotid，如表2所示。
表2.电务段的本地数据索引结构
depotid deviceid blockid offset length
表3是车站索引数据的索引的数据结构：
表3.电务段的车站索引数据的索引结构
Stationid workid blockid offset length
Stationed：车站的编号，在全路范围内是唯一的；
该索引相对于车站的索引增加了一个车站编号，用于识别车站，一条索引记录就代表了一个车站工区的索引文件存储的具***置，以及索引文件的长度等。
d)铁路局数据索引结构
铁路局与电务段类似，存储的数据分为两类：铁路局本身采集到的数据以及索引数据。其中索引数据分为本地数据索引以及管区内电务段索引的索引。
本地数据索引数据结构如表4所示：
表4.铁路局的本地数据索引数据结构
officeid deviceid blockid offset length
Officeid：是铁路局的编号，用于识别该铁路局的唯一编码。
电务段索引的索引数据结构如表5所示：
表5.铁路局的电务段索引的索引数据结构
deoptid Stationid blockid offset length
该索引相对于电务段的索引增加了一个电务段编号，用于识别电务段，一条索引记录就代表了一个电务段的索引文件存储的具***置，以及索引文件的长度等。
e)铁路公司数据索引结构
铁路公司与铁路局类似，存储的数据分为两类：铁路公司本身采集到的数据以及索引数 据。其中索引数据分为本地数据索引以及各个铁路局索引数据的索引。
本地数据索引数据结构如表6所示：
表6.铁路公司的本地数据索引数据结构
companyid deviceid blockid offset length
铁路局索引的索引数据结构如表7所示：
表7.铁路公司的铁路局索引的索引数据结构
officeid deoptid blockid offset length
该索引相对于铁路局的索引增加了一个铁路局编号，用于识别铁路局，一条索引记录就代表了一个铁路局的索引文件存储的具***置，以及索引文件的长度等。
通过上述的各级数据存储及索引结构，保证了各级各部门能够根据自身的索引文件快速的查询到管区内的所有信号数据，兼顾了铁路信号***当前网络、存储空间实际情况的情况下保证了信号数据的访问速度。
本发明利用综合运维平台实现了对铁路信号数据的集中管理，实现了对铁路各类信号数据的统一采集和集中存储。数据预处理子***对采集到的各类监测数据进行统一的预处理，抽取出有用的特征以提供给数据挖掘和实时分析使用。分类模型挖掘子***通过对标注的数据进行分析处理，针对具体的故障问题找出合适的分类模型及参数，并将挖掘结果传输给实时数据分析子***使用。数据分析子***利用分类模型对预处理之后实时信号特征数据进行分类，得到故障分析结果，并将该结果传输给结果展示子***，结果展示子***根据本身所处的位置展示相应的分析结果，包括车站内部、电务段内部以及整个铁路公司的运行情况。下面通过图示及实例来具体说明该方法的具体流程：
图2是***的整体架构图，该***包括数据采集***(即数据采集机)、数据中心和终端，数据中心包括结构化数据存储、半结构化数据存储、非结构化数据存储、离线数据处理、实时大数据处理、统计分析、数据挖掘、故障预警、查询引擎、推送引擎等模块。终端可以是各种类型的接收终端，包括PC电脑、笔记本、Pad、智能手机等。
图3是***的部署示意图，分四个层次进行部署，分别为铁路公司、铁路局、电务段和车站。
a)铁路公司
该层次需要部署数据中心、数据采集机和终端。数据中心的具体架构参考图3，因为这个数据中心需要管理和处理整个铁路公司内的所有电务段、车站的各类信号数据，因此需要较大的存储、计算能力。数据采集机主要用于接收各个铁路局发送过来的数据，该机器上需 要部署数据采集子***、数据预处理子***，负责将采集到的信号数据预处理之后传输到数据中心即可；终端与数据中心连接用于展示整个公司的运行状态，包括实时信号展示、信号数据查询及故障报警。
b)铁路局
铁路局需要部署一个小型的数据中心、数据采集机以及终端。铁路局数据中心用于接受和处理管辖内电务段上传的信号数据，需要部署包括分布式文件***、并行处理架构、实时数据分析等组件；数据采集机主要用于采集铁路局层面产生的各类信号数据，并对数据进行预处理后传输给数据中心。终端用于展示、查询和预警铁路局内的各类信号数据。
c)电务段
电务段及车站不需要部署数据中心，只需要数据采集机和终端。这些采集机除了具有信号采集的功能外，还具有数据预处理、实时数据分析、数据存储和数据传输的能力。数据采集机采集数据后，对数据进行预处理及实时分析，将分析结果通过本地的终端进行展示，并将部分分析结果及信号数据传输到铁路局数据中心中。电务段的数据采集***接收来自管区内各个车站传输的信号数据和分析结果。终端用于显示和查询各种信号数据和分析结果。
d)车站
车站与电务段一样只需要部署数据采集机及终端，其功能也类似，只是车站的数据采集机只需要采集本车站的信号数据，并将这些数据进行预处理和实时分析，并根据需求将部分数据及分析结果传输给电务段的采集机。终端主要用于展示车站内各种信号数据以及数据的实时分析情况。
图4是数据中心的架构图，主要包括云安全层、基础资源层、虚拟化层、存储***层、计算引擎层、组件层及web集群。云安全层通过数据备份、数据还原和访问控制，保障了***的安全。基础资源层构建了数据中心的硬件平台，该部分可以复用现有的服务器设备。虚拟化层通过虚拟化软件实现对硬件的虚拟化，屏蔽底层的硬件差异，形成一个高可用、可伸缩、可扩展的集群。数据存储层通过分布式文件***、列式数据库、关系数据库构成，分别用于存储非结构化、半结构化和结构化的数据。计算引擎层包括离线数据处理架构MapReduce、实时分布式处理架构spark以及数据移动引擎sqoop和底层资源管理***yarn。Yarn用于根据用户提交的任务来分配存储资源和计算资源用于处理该任务。组件层是具体处理各种业务的组件，包括数据管理组件、数据分析组件、故障诊断组件等。各种数据采集、分析的结果都需要通过B/S或者C/S的架构来发布出去，Web集群是基于负载均衡的服务发布层。
本发明的基于云计算的综合运维***，可以与其它处理平台集成在一起。如图5所示为一种数字铁路一体化平台，采用“通用组件”和“业务插件”的方式构建。其中“数据中心与大数据管理&数据挖掘分析平台”可以以通用组件的方式一本发明为基础实现，是数据采集、存储和数据处理的核心，并与设备综合监控平台、运输平台、运维平台、调度协同平台等集成在一起，共同形成数字铁路一体化平台。
下面通过一个具体的实例来来详细描述平台的工作流程：
图6是一个25Hz相敏轨道电路区分室内室外故障的流程图。轨道电路故障原因分析是一个分类问题，非常适合使用数据挖掘的方法进行分析挖掘。与其相关的信号采集来自于车站设备。首先是模型训练阶段，对已有的数据进行人工标注，标注出那些是有故障的数据。这样就形成了一个训练集，使用数据挖掘子***对这些数据集进行训练，选择贝叶斯分类器作为训练模型，得到对应的各种模型参数。然后将这些分类模型部署在车站的数据采集机中，数据采集机首先接收来自各个监测设备的监测数据，然后对这些数据进行预处理，得到用于分析的特征向量。实时数据分析组件利用得到的分类模型对实时数据进行分析计算，得到***目前是否有故障。然后将故障分析的结果通过展示***进行展示，并将相关的分析结果及特征传输到电务段、铁路局及铁路中心中，用于在更高层面上的分析。
通过数据预处理和特征选择之后，完成的特征提取结果为：
分线盒受端电压 甩开电缆端子室外侧电压 送端电压 数据采集时间戳
25.00 25.00 25.00 521365
24.00 24.00 25.00 521365
27.00 27.00 27.00 521365
0.00 0.00 0.00 521365
… … … …
为了简化说明，上表中三个测试点的正常电压值均设置为25v。故障的类型分为三类：
(1)无故障；
(2)故障在室内；
(3)故障在室外；
(4)室内短路；
(5)室内开路；
将上述数据进行向量化，以便于改进贝叶斯分类器进行计算：
实例数据位：
0 1:25.0 2:25.0 3:25.0
0 1:25.0 2:25.0 3:25.0
0 1:25.0 2:25.0 3:25.0
4 1:30.0 2:25.0 3:25.0
4 1:30.0 2:35.0 3:20.0
1 1:0.0 2:0.0 3:0.0
2 1:0.0 2:25.0 3:25.0
3 1:0.0 2:50.0 3:25.0
3 1:15.0 2:50.0 3:25.0
1 1:0.0 2:0.0 3:0.0
1 1:0.0 2:0.0 3:0.0
最前面的一列数字代表故障的类型：
● 0表示没有故障
● 1表示故障在室内
● 2表示故障在室外
● 3表示室内短路
● 4表示室内开路
以上实施例仅用以说明本发明的技术方案而非对其进行限制，本领域的普通技术人员可以对本发明的技术方案进行修改或者等同替换，而不脱离本发明的精神和范围，本发明的保护范围应以权利要求所述为准。

Claims (10)

  1. 一种基于云计算的轨道交通信号综合运维***，其特征在于，包括：
    数据采集机，用于采集轨道交通信号数据；
    数据中心，连接所述数据采集机，包括云安全层、基础资源层、虚拟化层、数据存储层、计算引擎层、组件层及Web集群；其中，云安全层通过数据备份、数据还原和访问控制，保障***的安全；基础资源层是数据中心的硬件平台；虚拟化层通过虚拟化软件实现对硬件的虚拟化，屏蔽底层的硬件差异；数据存储层包括分布式文件***、列式数据库、关系数据库，分别用于存储非结构化、半结构化和结构化的数据；计算引擎层采用云计算处理架；组件层是具体处理各种业务的组件；Web集群是基于负载均衡的服务发布层；
    终端，连接所述数据采集机和所述数据中心，用于展示数据分析结果。
  2. 如权利要求1所述的***，其特征在于：采用铁路公司、铁路局、电务段、车站的四级模式，在铁路公司、铁路局分别部署数据中心、数据采集机和终端，在电务段、车站只部署数据采集机和终端；在车站、电务段、铁路局、铁路公司均部署实时数据分析组件，其中车站、电务段的实时数据分析组件部署于数据采集机内，铁路局、铁路公司的实时数据分析组件部署于数据中心内。
  3. 如权利要求1所述的***，其特征在于：所述计算引擎层包括离线数据处理架构MapReduce、实时分布式处理架构spark以及数据移动引擎sqoop和底层资源管理***yarn，Yarn用于根据用户提交的任务来分配存储资源和计算资源用于处理该任务。
  4. 如权利要求1所述的***，其特征在于：所述组件层包括数据管理组件、数据分析组件、故障诊断组件。
  5. 一种采用权利要求1所述***的基于云计算的轨道交通信号综合运维方法，其步骤包括：
    1)采用铁路公司、铁路局、电务段、车站的四级模式，在铁路公司、铁路局分别部署数据中心、数据采集机和终端，在电务段、车站只部署数据采集机和终端；在车站、电务段、铁路局、铁路公司均部署实时数据分析组件，其中车站、电务段的实时数据分析组件部署于数据采集机内，铁路局、铁路公司的实时数据分析组件部署于数据中心内；
    2)车站的数据采集机采集车站的实时数据，利用在本地部署的实时数据分析组件对信号数据进行预处理、特征提取和特征选择，然后利用分析模型对特征数据进行实时分析，得到信号***当前的运行状态，并会将相应的数据传输至电务段；
    3)电务段的数据采集机接受来自内部的各个车站传输过来的数据，利用本地部署的实时数据分析组件对接收的不同车站的数据进行分析，得到整个电务段的分析结果，并将分析结 果传输至铁路局的数据采集机；
    4)铁路局的数据采集机接收来自各个电务段的数据，并将数据转存到数据中心，然后利用数据中心部署的实时数据分析组件对数据进行处理和分析，并将分析结果及部分特征传输到铁路公司的数据采集机；
    5)铁路公司的数据采集机接收铁路局发送过来的数据，并将数据转存至数据中心，利用数据中心的实时数据分析组件进行分析处理，得到铁路公司全局的分析结果；
    6)利用部署于铁路公司、铁路局、电务段和车站的终端，进行数据分析结果的展示。
  6. 如权利要求5所述的方法，其特征在于：还包括步骤7)，该步骤对铁路公司、铁路局、电务段及车站存储的数据建立索引，在出现故障时根据该索引快速获取相关时间段及位置的各类信号数据，以便于快速的进行故障分析、定位和解决。
  7. 如权利要求6所述的方法，其特征在于，所述建立索引的方法是：首先在各个车站根据存储的数据建立索引，然后将车站的索引文件传输到电务段，电务段对本地采集的数据及接收到的车站索引文件建立二级索引，然后将索引传输至铁路局的数据中心，铁路局的数据中心对接收的各类数据建立三级索引，然后将索引传输至铁路公司的数据中心，铁路公司的数据中心对接收的各类数据建立四级索引，形成一个铁路公司、铁路局、电务段、车站的四级索引。
  8. 如权利要求7所述的方法，其特征在于，所述四级索引中：
    车站数据索引结构包括车间的编号、车间内监测设备的编号、数据块编号、信号数据在数据块的起始位置、信号数据的大小；车站的索引数据在数据存储时生成，并存储在本地机器中，车站终端进行数据查询时，首先到本地机器中读取索引文件，然后根据数据块、偏移量去读取实际数据；
    电务段数据索引结构包括本地数据索引和车站索引数据的索引；
    铁路局数据索引结构包括本地数据索引和管区内电务段索引数据的索引；
    铁路公司数据索引结构包括本地数据索引和各个铁路局索引数据的索引。
  9. 如权利要求5所述的方法，其特征在于：所述数据采集机对于采集到的本地信号数据保存在本地***中，上层通过元数据来管理下层的数据。
  10. 如权利要求5所述的方法，其特征在于：步骤2)中车站的数据采集机通过车站的信号站机获取信号数据，车站的信号站机通过已有的规则判断是否需要进行传输，从而对信号数据进行压缩处理；车站的数据采集机与车站的信号站机建立socket连接，车站的信号站机将压缩处理后的数据传输到数据采集机，数据采集机利用所述已有的规则对数据进行还原。
PCT/CN2015/075007 2014-07-07 2015-03-25 一种基于云计算的轨道交通信号综合运维方法及*** WO2016004775A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201410321145.1 2014-07-07
CN201410321145.1A CN104077552B (zh) 2014-07-07 2014-07-07 一种基于云计算的轨道交通信号综合运维方法及***

Publications (1)

Publication Number Publication Date
WO2016004775A1 true WO2016004775A1 (zh) 2016-01-14

Family

ID=51598802

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/075007 WO2016004775A1 (zh) 2014-07-07 2015-03-25 一种基于云计算的轨道交通信号综合运维方法及***

Country Status (2)

Country Link
CN (1) CN104077552B (zh)
WO (1) WO2016004775A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018090605A1 (zh) * 2016-11-18 2018-05-24 中兴通讯股份有限公司 一种数据中心的管理方法及***

Families Citing this family (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10282676B2 (en) * 2014-10-06 2019-05-07 Fisher-Rosemount Systems, Inc. Automatic signal processing-based learning in a process plant
CN104077552B (zh) * 2014-07-07 2017-08-25 北京泰乐德信息技术有限公司 一种基于云计算的轨道交通信号综合运维方法及***
CN104504956A (zh) * 2014-12-22 2015-04-08 中国神华能源股份有限公司 一种基于虚拟现实的铁路电务训练***
CN104908783B (zh) * 2015-05-27 2017-01-18 中国铁路总公司 铁路电务综合监测维护***体系架构
CN104890702A (zh) * 2015-05-27 2015-09-09 中国铁路总公司 一种铁路信号车地综合分析监测***
CN105574593B (zh) * 2015-12-18 2020-05-05 中南大学 基于云计算和大数据的轨道状态静态检控***及方法
CN105676842B (zh) * 2016-03-14 2019-06-18 中国铁路总公司 一种高铁列控车载设备故障诊断方法
CN106250429A (zh) * 2016-07-26 2016-12-21 浪潮软件股份有限公司 一种基于sqoop的数据抽取方法
CN106341467B (zh) * 2016-08-30 2019-11-29 国网江苏省电力公司电力科学研究院 基于大数据并行计算的用电信息采集设备状态分析方法
CN108268023B (zh) * 2016-12-30 2022-04-26 上海嘉成轨道交通安全保障***股份公司 一种轨道交通站台门远程故障诊断方法及***
CN107316158A (zh) * 2017-07-05 2017-11-03 云能服(北京)科技有限公司 一种多维作业数据处理方法和装置
CN110254471A (zh) * 2018-03-12 2019-09-20 上海铁鑫电气科技有限公司 一种铁路信号轨道电路引接线断线实时监测设备
CN109377090A (zh) * 2018-11-22 2019-02-22 湖南铁路科技职业技术学院 一种基于云服务的铁路运输数据通讯支撑平台
CN109889601A (zh) * 2019-03-12 2019-06-14 湖南铁路科技职业技术学院 铁路信号微机监测***
CN110096383B (zh) * 2019-04-10 2022-08-30 卡斯柯信号有限公司 一种信号设备维护信息自动分类方法
CN110032090A (zh) * 2019-04-15 2019-07-19 深圳众维轨道交通科技发展有限公司 一种基于bim+ai实现有轨电车远程监控和授权的方法
CN110182244A (zh) * 2019-04-23 2019-08-30 深圳众维轨道交通科技发展有限公司 一种基于云计算和ai智能的有轨电车云平台
CN110329319B (zh) * 2019-06-28 2021-09-03 卡斯柯信号有限公司 一种面向智慧城轨的全自动运行***
CN110481605B (zh) * 2019-08-21 2022-02-08 沈阳风驰软件股份有限公司 一种铁路运输生产管控***
CN111106968A (zh) * 2019-12-31 2020-05-05 国网山西省电力公司信息通信分公司 一种构建信息通信智能调度指挥沙盘的方法
CN111332340A (zh) * 2020-03-06 2020-06-26 东莞理工学院 一种轨道交通监测数据的存储和处理方法及***
CN111369022A (zh) * 2020-03-10 2020-07-03 上海申铁信息工程有限公司 一种铁路车站运维监控平台与装置
CN111292218A (zh) * 2020-03-10 2020-06-16 上海申铁信息工程有限公司 一种铁路车站设备智能化监控***的构建方法与装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012119879A1 (de) * 2011-03-07 2012-09-13 Siemens Aktiengesellschaft Eisenbahnleitsystem
CN103264717A (zh) * 2013-05-21 2013-08-28 北京泰乐德信息技术有限公司 一种轨道交通综合监控调度协同与运维信息化***
CN103338261A (zh) * 2013-07-04 2013-10-02 北京泰乐德信息技术有限公司 一种轨道交通监测数据的存储和处理方法及***
CN103391185A (zh) * 2013-08-12 2013-11-13 北京泰乐德信息技术有限公司 一种轨道交通监测数据的云安全存储和处理方法及***
CN104077552A (zh) * 2014-07-07 2014-10-01 北京泰乐德信息技术有限公司 一种基于云计算的轨道交通信号综合运维方法及***

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101917237B (zh) * 2010-07-27 2013-02-20 北京全路通信信号研究设计院有限公司 一种铁路信号监测方法和***

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012119879A1 (de) * 2011-03-07 2012-09-13 Siemens Aktiengesellschaft Eisenbahnleitsystem
CN103264717A (zh) * 2013-05-21 2013-08-28 北京泰乐德信息技术有限公司 一种轨道交通综合监控调度协同与运维信息化***
CN103338261A (zh) * 2013-07-04 2013-10-02 北京泰乐德信息技术有限公司 一种轨道交通监测数据的存储和处理方法及***
CN103391185A (zh) * 2013-08-12 2013-11-13 北京泰乐德信息技术有限公司 一种轨道交通监测数据的云安全存储和处理方法及***
CN104077552A (zh) * 2014-07-07 2014-10-01 北京泰乐德信息技术有限公司 一种基于云计算的轨道交通信号综合运维方法及***

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018090605A1 (zh) * 2016-11-18 2018-05-24 中兴通讯股份有限公司 一种数据中心的管理方法及***
CN108075917A (zh) * 2016-11-18 2018-05-25 中兴通讯股份有限公司 一种数据中心的管理方法及***

Also Published As

Publication number Publication date
CN104077552B (zh) 2017-08-25
CN104077552A (zh) 2014-10-01

Similar Documents

Publication Publication Date Title
WO2016004775A1 (zh) 一种基于云计算的轨道交通信号综合运维方法及***
CN110059631B (zh) 接触网非接触式监测缺陷识别方法
CN108591104B (zh) 一种基于云平台的风机故障预测与健康管理***、方法
US10756809B1 (en) Emergency communication satellite terminal management system
CN103500173B (zh) 一种轨道交通监测数据的查询方法
CN104503399B (zh) 一种集团级风电机组状态监测及故障诊断平台
CN105574593B (zh) 基于云计算和大数据的轨道状态静态检控***及方法
RU2546320C2 (ru) Интеллектуальная сеть
CN103699698A (zh) 一种基于改进贝叶斯的轨道交通故障识别方法及***
CN103745229A (zh) 一种基于svm的轨道交通故障诊断方法及***
WO2019196869A1 (zh) 一种确定巡检基站列表的方法以及巡检装置
CN103338261A (zh) 一种轨道交通监测数据的存储和处理方法及***
CN104777813A (zh) 综合气象观测运行监控***及其监控方法
CN103631788B (zh) 基于共享数据库的车辆制造质量问题诊断***
CN108860223B (zh) 一种数据处理***及方法
CN103760901A (zh) 一种基于关联规则分类器的轨道交通故障识别方法
CN103986758A (zh) 一种高速动车组运行故障数据远程实时传输和智能分析判断***
CN112948457A (zh) 客运索道检测监测与健康诊断***、方法、介质、设备
CN107705054A (zh) 满足复杂数据的新能源并网发电远程测试诊断平台及方法
CN112734256A (zh) 轨道交通车辆智能运维***
CN112883001A (zh) 一种基于营配贯通数据可视化平台的数据处理方法、装置及介质
CN107563686B (zh) 一种铁路运输十八点统计数据校验方法、***和存储介质
CN112278012B (zh) 一种动车组轮对管理***
CN108258802B (zh) 一种配电网中配电设备的运行状况的监测方法和装置
CN113379314B (zh) 一种基于推理算法的年度生产计划智能监管方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15819040

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15819040

Country of ref document: EP

Kind code of ref document: A1