WO2019196869A1 - Method for determining list of patrolling base stations, and patrolling apparatus - Google Patents

Abstract

Disclosed are a method for determining a list of patrolling base stations, and a patrolling apparatus, for obtaining a list of patrolling base stations and improving the probability of patrolling a problematic base station. The method of embodiments of the present application comprises: acquiring current operation and maintenance data of a first base station set; according to the current operation and maintenance data, constructing features of base stations, so as to obtain current feature data of the first base station set; according to the current feature data and the pre-constructed different types of patrolling models, obtaining a score of each patrolling model corresponding to each of the base stations in the first base station set; according to the score of each patrolling model corresponding to each of the base stations, obtaining, by means of calculation, a total score of each of the base stations; and according to the total score of each of the base stations, obtaining, by means of calculation, the list of patrolling base stations.

Description

一种确定巡检基站列表的方法以及巡检装置Method for determining list of inspection base stations and inspection device

本申请要求于2018年04月12日提交中国专利局、申请号为201810326933.8、申请名称为“一种确定巡检基站列表的方法以及巡检装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。This application claims the priority of the Chinese patent application filed on April 12, 2018, the Chinese Patent Office, the application number is 201810326933.8, and the application name is "a method for determining the list of inspection base stations and the inspection device". The citations are incorporated herein by reference.

技术领域Technical field

本申请涉及计算机领域，尤其涉及一种确定巡检基站列表的方法以及巡检装置。The present application relates to the field of computers, and in particular, to a method for determining a list of patrol base stations and a patrol device.

背景技术Background technique

优质的电信基站维护是电信网络高效安全运行的重要保障。为了保证通讯质量良好就要求做好线路维护工作，保持线路设备经常处于完整状态和标准性能，并在障碍形成前及时发现并消除影响线路安全的因素。巡检是日常代维工作中的重要组成部分，也是隐患排查、故障发现的主要手段。现有的巡检模式多是代维人员人为将所有基站按比例分为不同的固定频次进行巡检。其中，固定频次例如月、季、半年、年。而实际情况中各个巡检对象因网络运行环境、所处的地理环境、社会环境不同，实际的现场情况是存在差异的，此种巡检方式下，对基站进行固定频次的巡检，可能对网络质量好、无投诉的基站和传输线路段进行无用的巡检，而存在隐患的基站和传输线路段并没有及时发现，没有很好的评价标准，且对代维人员巡检的真实性、完整性和执行效果缺乏掌控。High-quality telecommunication base station maintenance is an important guarantee for efficient and safe operation of telecommunication networks. In order to ensure good communication quality, it is required to do line maintenance work, keep line equipment constantly in full state and standard performance, and find and eliminate the factors affecting line safety before obstacles are formed. Inspection is an important part of daily generation and maintenance work, and it is also the main means of hidden trouble investigation and fault detection. Most of the existing inspection modes are performed by the Devi personnel to divide all the base stations into different fixed frequencies for inspection. Among them, the fixed frequency is for example, month, season, half year, year. In actual situations, due to different network operating environments, geographical environment, and social environment, actual on-site conditions are different. Under such inspection mode, fixed-frequency inspections of base stations may be performed. Base stations and transmission line segments with good network quality and no complaints are used for patrol inspections. The base stations and transmission line segments with hidden dangers are not discovered in time, there is no good evaluation standard, and the authenticity and integrity of the inspections of the maintenance personnel are And lack of control over the execution.

发明内容Summary of the invention

本申请实施例提供了一种确定巡检基站列表的方法以及巡检装置，用于得到巡检基站列表，提高对问题基站进行巡检的概率。The embodiment of the present application provides a method for determining a patrol base station list, and a patrol device, which is used to obtain a patrol base station list, and improve the probability of performing a patrol on the problem base station.

有鉴于此，本申请实施例第一方面提供了一种确定巡检基站列表的方法，可以包括：巡检装置可以获取第一基站集合的当前运维数据；当前运维数据可以可以是第一基站集合中各个基站的设备等信息、以及各个基站相应的运行状态信息等。根据该当前运维数据进行基站特征构建，得到该第一基站集合的当前特征数据；根据该当前特征数据和该预先构建的不同类型的多个巡检模型，得到该第一基站集合中每个基站对应的各个巡检模型的得分；根据该每个基站对应的各个巡检模型的得分，计算得到每个基站的总得分；根据该每个基站的总得分，计算得到该巡检基站列表。In view of this, the first aspect of the present application provides a method for determining a patrol base station list, which may include: the patrol device may obtain current operation and maintenance data of the first base station set; the current operation and maintenance data may be first. Information such as devices of each base station in the base station set, and corresponding operational status information of each base station. Performing base station feature construction according to the current operation and maintenance data to obtain current feature data of the first base station set; and obtaining each of the first base station sets according to the current feature data and the pre-built multiple types of different inspection models a score of each patrol model corresponding to the base station; calculating a total score of each base station according to the scores of the respective patrol models corresponding to each base station; and calculating the patrol base station list according to the total score of each base station.

在本申请实施例中，先可以获取第一基站集合的当前运维数据，然后，再进行基站特征构建，得到当前特征数据。再根据当前特征数据和该预先构建的不同类型的多个巡检模型，分别得到每个基站对应的各个巡检模型的得分，然后得到每个基站的总得分。从而，可以根据每个基站的总得分，进行排列，得到巡检基站列表。提高对问题基站的巡检概率。通过组合巡检模型的方式，将基站业务指标进行细分，先分别建模，再进行组合，从而提高了巡检模型的可解释性以及准确性。In the embodiment of the present application, the current operation and maintenance data of the first base station set may be acquired first, and then the base station feature is constructed to obtain current feature data. Then, according to the current feature data and the plurality of pre-built different types of patrol models, the scores of the respective patrol models corresponding to each base station are respectively obtained, and then the total score of each base station is obtained. Therefore, it is possible to perform an arrangement according to the total score of each base station, and obtain a list of the patrol base stations. Improve the probability of inspection of the problem base station. By combining the inspection models, the base station service indicators are subdivided and modeled separately and then combined to improve the interpretability and accuracy of the inspection model.

可选的，在本申请的一些实施例中，该获取基站集合的当前运维数据之前，该方法还可以包括：巡检装置先可以获取第二基站集合的历史运维数据；历史运维数据可以是第二 基站集合中各个基站的设备等信息、以及各个基站相应的运行状态信息等。然后再根据该历史运维数据得到该第二基站集合的历史特征数据，以及该第二基站集合中每个基站的运行状态信息；根据该历史特征数据以及该第二基站集合中每个基站的运行状态信息，训练得到该不同类型的多个巡检模型。Optionally, in some embodiments of the present application, before acquiring the current operation and maintenance data of the base station set, the method may further include: the patrol device may first obtain historical operation and maintenance data of the second base station set; and historical operation and maintenance data. It may be information such as devices of each base station in the second base station set, and corresponding operational status information of each base station. And then obtaining historical feature data of the second base station set and running state information of each base station in the second base station set according to the historical operation and maintenance data; and according to the historical feature data and each base station in the second base station set Running state information, training to obtain multiple inspection models of different types.

需要说明的是，该巡检模型可以深度神经网络，也可以是决策树，也可以是随机森林等，具体不做限定。在本申请实施例中，可以根据第二基站集合的历史运维数据，进行基站特征数据的抽取，获取每个基站的运行状态信息，来训练得到不同类型的多个巡检模型，这多个不同类型的巡检模型，就可以作为参考巡检模型使用了。而且，对于多个不同类型的巡检模型，也可以周期性的重新训练更新。对于同一个基站来说，可以得到不同类型的巡检模型的得分，进而，来得到一个总得分，即先分别建模，再进行组合，从而提高了巡检模型的可解释性以及准确性。It should be noted that the inspection model may be a deep neural network, a decision tree, or a random forest, and is not limited. In the embodiment of the present application, the base station feature data may be extracted according to the historical operation and maintenance data of the second base station set, and the running state information of each base station is obtained to train different types of multiple inspection models. Different types of inspection models can be used as reference inspection models. Moreover, for a number of different types of inspection models, it is also possible to periodically retrain the updates. For the same base station, the scores of different types of inspection models can be obtained, and then, a total score is obtained, that is, separately modeled and then combined, thereby improving the interpretability and accuracy of the inspection model.

可选的，在本申请的一些实施例中，该根据该当前运维数据进行基站特征构建，得到该第一基站集合的当前特征数据，可以包括：使用预置的抽取方式从该当前运维数据中，抽取预置的特征数据类型对应的特征数据，得到该第一基站集合的当前特征数据。本申请实施例对根据当前运维数据，进行基站特征构建，得到当前特征数据，提供了一个具体的实现方式，增加了方案的可行性。Optionally, in some embodiments of the present application, performing the base station feature construction according to the current operation and maintenance data to obtain the current feature data of the first base station set may include: using the preset extraction manner from the current operation and maintenance In the data, the feature data corresponding to the preset feature data type is extracted, and the current feature data of the first base station set is obtained. In the embodiment of the present application, the base station feature is constructed according to the current operation and maintenance data, and the current feature data is obtained, which provides a specific implementation manner and increases the feasibility of the solution.

可选的，在本申请的一些实施例中，该当前特征数据可以包括基站类型、基站区域、主设备服务年限、外设备服务年限。Optionally, in some embodiments of the present application, the current feature data may include a base station type, a base station area, a primary device service age, and an outer device service age.

可选的，在本申请的一些实施例中，该第二基站集合中每个基站的运行状态信息可以包括：该第二基站集合中每个基站在第一预置时长内的告警数量或者断电时长。可以理解的是，每个基站的运行状态信息可以作为标识基站的巡检优先级，在标识过程中，首先根据基站的业务指标对基站进行排序，并且需要根据巡检量确定相关的基站的数量，相较于人工标识的方法，这种业务指标直接标识的方式同时也是大大减低了人力的投入，同时摒弃了人为因素的干扰，直接使用业务结果作为指标，更为贴近业务目标。Optionally, in some embodiments of the present application, the running status information of each base station in the second set of base stations may include: the number of alarms of each base station in the second set of base stations in the first preset duration or Electricity time. It can be understood that the running status information of each base station can be used as the patrol priority of the identified base station. In the identification process, the base station is first sorted according to the service indicator of the base station, and the number of related base stations needs to be determined according to the patrol quantity. Compared with the method of manual identification, the direct identification of such business indicators also greatly reduces the input of manpower, and at the same time discards the interference of human factors, directly uses the business results as indicators, and is closer to the business objectives.

可选的，在本申请的一些实施例中，该方法还可以包括：若该第一基站集合中的各个基站的总得分中存在总得分相同的基站，则确定该总得分相同的基站中在第二预置时长内未巡检的基站为待巡检基站列表。在本申请实施例中，当出现总得分相同的基站时，可以看下，这些基站近期有没有被巡检过，如果有被巡检过，那么，这次，可以先不用巡检了，可以对近期未巡检的基站进行巡检。尽可能的保证，提高问题基站的巡检概率。Optionally, in some embodiments of the present application, the method may further include: if there is a base station with the same total score among the total scores of the base stations in the first base station set, determining that the base station with the same total score is in the base station The base station that is not inspected within the second preset duration is a list of base stations to be patrolled. In the embodiment of the present application, when there are base stations with the same total score, it can be seen that these base stations have not been inspected recently. If there is a patrol, then this time, the patrol may not be used first. The base station that has not been inspected recently is inspected. As much as possible to ensure that the probability of inspection of the problem base station.

本申请实施例第二方面提供一种巡检装置，该巡检装置具有实现上述第一方面和第一方面的任一种可能的设计中方法的功能。所述功能可以通过硬件实现，也可以通过硬件执行相应的软件实现。所述硬件或软件包括一个或多个与上述功能相对应的模块。A second aspect of the embodiments of the present application provides a patrol device having the function of implementing any of the possible design methods of the first aspect and the first aspect described above. The functions may be implemented by hardware or by corresponding software implemented by hardware. The hardware or software includes one or more modules corresponding to the functions described above.

本申请实施例第三方面提供一种巡检装置，包括：A third aspect of the embodiments of the present application provides a patrol device, including:

存储器，用于存储程序；Memory for storing programs;

处理器，用于执行所述存储器存储的所述程序，当所述程序被执行时，所述处理器用于执行第一方面或第一方面所述的任意一种可能的实现方式。a processor for executing the program stored by the memory, the processor for performing any one of the possible implementations of the first aspect or the first aspect when the program is executed.

可选的，上述存储器可以是物理上独立的单元，也可以与处理器集成在一起。Optionally, the above memory may be a physically separate unit or may be integrated with the processor.

在第四方面的一种实现方式中，该巡检装置可以是芯片。In an implementation of the fourth aspect, the inspection device can be a chip.

本申请的又一方面提了供一种计算机可读存储介质，所述计算机可读存储介质中存储有指令，当其在计算机上运行时，使得计算机执行上述各方面所述的方法。Yet another aspect of the present application is directed to a computer readable storage medium having instructions stored therein that, when executed on a computer, cause the computer to perform the methods described in the various aspects above.

需要说明的是，本申请技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产口的形式体现出来，该计算机软件产品存储在一个存储介质中，用于储存为上述巡检装置所用的计算机软件指令，其包含用于执行上述各方面为巡检装置所设计的程序。It should be noted that the part of the technical solution of the present application or the contribution to the prior art or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium. And for storing computer software instructions for use in the above-mentioned inspection device, including programs for performing the above-described aspects for the inspection device.

该存储介质包括：U盘、移动硬盘、只读存储器(read-only memory，ROM)、随机存取存储器(random access memory，RAM)、磁碟或者光盘等各种可以存储程序代码的介质。The storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like, which can store program codes.

本申请的又一方面提供了一种包含指令的计算机程序产品，当其在计算机上运行时，使得计算机执行上述各方面所述的方法。Yet another aspect of the present application provides a computer program product comprising instructions which, when run on a computer, cause the computer to perform the methods described in the various aspects above.

本申请的又一方面提供了一种计算机程序，当其在计算机上运行时，使得计算机执行上述各方面所述的方法。Yet another aspect of the present application provides a computer program that, when run on a computer, causes the computer to perform the methods described in the various aspects above.

从以上技术方案可以看出，本申请实施例具有以下优点：As can be seen from the above technical solutions, the embodiments of the present application have the following advantages:

在本申请实施例中，首先获取第一基站集合的当前运维数据；根据所述当前运维数据进行基站特征构建，得到所述第一基站集合的当前特征数据；根据所述当前特征数据和所述预先构建的不同类型的多个巡检模型，得到所述第一基站集合中每个基站对应的各个巡检模型的得分；根据所述每个基站对应的各个巡检模型的得分，计算得到每个基站的总得分；根据所述每个基站的总得分，计算得到所述巡检基站列表。就可以第一基站集合的当前运维数据，及时的得到巡检基站列表，而不是像现有固定频次的巡检。更能满足用户的需求，提高对问题基站进行巡检的概率。In the embodiment of the present application, first acquiring current operation and maintenance data of the first base station set, performing base station feature construction according to the current operation and maintenance data, and obtaining current feature data of the first base station set; and according to the current feature data and Calculating the scores of the respective inspection models corresponding to each base station in the first base station set by the plurality of different types of inspection models of the different types of the pre-constructed; calculating the scores of the respective inspection models corresponding to each of the base stations Obtaining a total score of each base station; calculating the list of the patrol base stations according to the total score of each base station. It is possible to obtain a list of patrol base stations in time for the current operation and maintenance data of the first base station set, instead of the patrol inspection like the existing fixed frequency. It can better meet the needs of users and improve the probability of conducting inspections on problem base stations.

附图说明DRAWINGS

为了更清楚地说明本申请实施例技术方案，下面将对实施例和现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，还可以根据这些附图获得其它的附图。In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings used in the embodiments and the prior art description will be briefly described below. Obviously, the drawings in the following description are only some implementations of the present application. For example, other drawings can also be obtained from these drawings.

图1为本申请实施例所应用的场景示意图；FIG. 1 is a schematic diagram of a scenario applied to an embodiment of the present application;

图2为本申请实施例所应用的***架构图；2 is a system architecture diagram of an application according to an embodiment of the present application;

图3为本申请实施例中巡检模型构建模块中组合模型的一个示意图；3 is a schematic diagram of a combined model in a patrol model building module according to an embodiment of the present application;

图4为本申请实施例中确定巡检基站列表的方法的一个实施例示意图；4 is a schematic diagram of an embodiment of a method for determining a list of a patrol base station according to an embodiment of the present application;

图5为本申请实施例中区域A基站的巡检模型结构的示意图；5 is a schematic diagram of a structure of a patrol inspection model of a regional A base station according to an embodiment of the present application;

图6为本申请实施例中断电告警决策树的一个示意图；6 is a schematic diagram of a circuit for interrupting an electrical alarm according to an embodiment of the present application;

图7为本申请实施例中巡检任务生成模块的流程示意图；FIG. 7 is a schematic flowchart of a patrol task generation module according to an embodiment of the present application;

图8为本申请实施例中基站故障密度的示意图；8 is a schematic diagram of a base station fault density in an embodiment of the present application;

图9为本申请实施例中巡检装置的一个实施例示意图；FIG. 9 is a schematic diagram of an embodiment of a patrol device according to an embodiment of the present application;

图10为本申请实施例中巡检装置的一个实施例示意图。FIG. 10 is a schematic diagram of an embodiment of a patrol inspection apparatus according to an embodiment of the present application.

具体实施方式detailed description

本申请实施例提供了一种确定巡检基站列表的方法以及巡检装置，用于得到巡检基站列表，提高对问题基站进行巡检的概率。The embodiment of the present application provides a method for determining a patrol base station list, and a patrol device, which is used to obtain a patrol base station list, and improve the probability of performing a patrol on the problem base station.

为了使本技术领域的人员更好地理解本申请方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行描述，显然，所描述的实施例仅仅是本申请一部分的实施例，而不是全部的实施例。基于本申请中的实施例，都应当属于本申请保护的范围。In the following, the technical solutions in the embodiments of the present application are described in conjunction with the accompanying drawings in the embodiments of the present application. It is obvious that the described embodiments are only a part of the present application. Embodiments, rather than all of the embodiments. Based on the embodiments in the present application, they should all fall within the scope of protection of the present application.

在其中一种实现方式中，采用的是例行巡检的方式，即根据基站类型对基站进行月度或者季度的例行巡检，如下表1所示：In one implementation manner, a routine inspection method is adopted, that is, a monthly or quarterly routine inspection of the base station according to the type of the base station, as shown in Table 1 below:

表1Table 1

在这种实现方式中，巡检的周期一般都是固定化的，没有考虑到网络质量这一因素，与网络质量脱节，而且有大量的无效巡检，以及存在隐患的基站，也可称问题基站不能及时发现等缺陷。In this implementation mode, the period of the inspection is generally fixed. The network quality is not taken into consideration, and the network quality is out of line. There are a large number of invalid inspections and base stations with hidden dangers. The base station cannot find defects such as in time.

在另一种实现方式中，是一种依托专家知识建立基站的动态巡检方案。即根据专家经验，通过结合基站级别、覆盖区，根据基站外部因素、网络特性、设备故障等动态特性评估点建立基站打分模型，定期计算基站得分，将巡检任务分配给得分低于阈值的基站。其中，基站打分模型采用的打分公式如下所示：In another implementation manner, it is a dynamic inspection scheme for establishing a base station based on expert knowledge. That is, according to the expert experience, by combining the base station level and the coverage area, the base station scoring model is established according to the dynamic characteristic evaluation points such as the external factors of the base station, the network characteristics, and the equipment failure, the base station score is periodically calculated, and the inspection task is assigned to the base station whose score is lower than the threshold. . Among them, the scoring formula adopted by the base station scoring model is as follows:

基站巡检评估计＝满分-（Σ外部因素扣分+Σ重要程度扣分+Σ故障类型扣分+Σ性能指标扣分)Base station inspection and evaluation meter = full score - (Σ external factor deduction + Σ important degree deduction + Σ fault type deduction + Σ performance index deduction)

这种实现方式是一种基于规则的方法，基站打分模型的建立完全依靠专家知识(例如阈值的选择、各个因素分数的定义等)，具有非常强的主观性。而且，对于数据量较大、特征较多的情况，领域专家很难充分发现数据中隐藏的信息(例如特征之间的相互作用等)。该技术的可重用性较差，如有新的特征引用，则打分公式需要人工进行调整，人力和时间花费较高。This implementation is a rule-based approach. The establishment of the base station scoring model relies entirely on expert knowledge (such as threshold selection, definition of individual factor scores, etc.) and is very subjective. Moreover, for large data volumes and many features, it is difficult for domain experts to fully discover the hidden information in the data (such as interactions between features, etc.). The reusability of the technology is poor. If there is a new feature reference, the scoring formula needs to be manually adjusted, and the manpower and time are expensive.

根据当前高效的巡检策略需求，需要可以根据基站的运行状态迅速而且准确判断基站的运行状态，选择性的对风险更高的基站群及时进行巡检。为实现基站的动态以及智能运维，本申请实施例提出了一种基于机器学习算法的动态巡检方案，摒弃了传统的静态运维方式或者是基于人工打分规则的巡检(preventive maintenance，PM)模式，通过机器学习的方法，即根据机器学习算法和基站的相关信息定期动态生成巡检基站列表，实现了基于基站运维数据对基站运行状态的排序，将传统的按里程巡检转化为按需巡检，优化巡检 周期和管理，同时通过实时数据流进行模型的自我学习和迭代，不断挖掘业务价值，降低运维成本，提高运维效果。According to the requirements of the current high-efficiency patrol policy, it is required to quickly and accurately determine the running status of the base station according to the running status of the base station, and selectively perform patrol inspection on the base station group with higher risk in time. In order to realize the dynamics and intelligent operation and maintenance of the base station, the embodiment of the present application proposes a dynamic inspection scheme based on the machine learning algorithm, which abandons the traditional static operation and maintenance method or the inspection based on the manual classification rule (preventive maintenance, PM). The mode, through the machine learning method, that is, dynamically generating the patrol base station list according to the machine learning algorithm and the related information of the base station, realizing the ordering of the base station operating state based on the base station operation and maintenance data, and converting the traditional mileage inspection into On-demand inspection, optimize inspection cycle and management, and self-learn and iterate through real-time data flow, continuously mining business value, reducing operation and maintenance costs, and improving operation and maintenance.

如图1所示，图1为本申请实施例所应用的场景示意图。本申请主要针对电信基站巡检的智能规划，图1展示了某区域的基站分布，现有的巡检任务安排主要将区域进行划分，对于同一个区域内的基站按照历程进行每月或者每季度的巡检。然而由于基站的使用年限、外部环境、设备状态不同等因素，基站的运行状态会有较大的出入，传统的巡检方式无法对高风险基站及时进行巡检，不能预先对问题基站进行排查。本申请主要是对基站大数据(也称为基站运维数据)进行分析，通过机器学习算法对基站的运行状态进行分析以及预测，从而及时将巡检任务分配给有需要的基站。As shown in FIG. 1 , FIG. 1 is a schematic diagram of a scenario applied to an embodiment of the present application. This application is mainly for the intelligent planning of telecommunication base station inspection. Figure 1 shows the distribution of base stations in a certain area. The existing inspection task arrangement mainly divides the area. For the base stations in the same area, the monthly or quarterly basis is followed. Inspection. However, due to factors such as the age of the base station, the external environment, and the state of the device, the running status of the base station may be greatly different. The traditional patrol mode cannot perform timely patrols on the high-risk base station, and the problem base station cannot be checked in advance. This application mainly analyzes the base station big data (also called base station operation and maintenance data), analyzes and predicts the running status of the base station through the machine learning algorithm, and then distributes the inspection task to the base station in need in time.

如图2所示，图2为本申请实施例所应用的***架构图。在图2所示中，可以包括但不限于：基站运维数据管理***、数据预处理模块、基站特征构造模块、巡检模型构建模块、巡检任务生成模块。As shown in FIG. 2, FIG. 2 is a system architecture diagram of an application according to an embodiment of the present application. As shown in FIG. 2, the base station operation and maintenance data management system, the data preprocessing module, the base station feature construction module, the inspection model construction module, and the inspection task generation module may be included.

其中，基站运维数据管理***，主要的功能作用是采集以及存储基站的运维数据。运维数据为运行以及维护过程中的数据，包括但不限定于基站的属性、运行过程中的告警信息、断站信息、维护过程中的维护记录等。Among them, the base station operation and maintenance data management system, the main function is to collect and store the operation and maintenance data of the base station. The operation and maintenance data is the data in the running and maintenance process, including but not limited to the attributes of the base station, the alarm information during the running process, the disconnection information, and the maintenance record during the maintenance process.

数据预处理模块主要负责从运维数据管理***读取基站运维数据后，对基站运维数据进行清洗，去除异常数据对定性数据进行等量化等等。The data preprocessing module is mainly responsible for reading the operation and maintenance data of the base station from the operation and maintenance data management system, cleaning the operation and maintenance data of the base station, removing the abnormal data, and quantifying the qualitative data.

下面对图2所示中的基站特征构造模块进行详细说明，如下所示：The base station feature construction module shown in FIG. 2 will be described in detail below, as follows:

基站特征构造模块主要包含基站特征数据抽取以及基站运行状态信息的获取，在模型训练阶段，需要进行基站特征数据抽取以及基站运行状态信息的获取，在巡检任务生成模块，可以根据基站运维数据以及设定的巡检日期，构建基站的特征向量。The base station feature structure module mainly includes base station feature data extraction and base station operation state information acquisition. In the model training phase, base station feature data extraction and base station operation state information acquisition are required, and the patrol task generation module can perform base station operation and maintenance data according to the base station. And set the patrol date to construct the feature vector of the base station.

(1)基站特征数据抽取(1) Base station feature data extraction

在基站特征数据抽取的部分，主要结合基站运维数据以及专家知识提取基站的静态特征和动态特征。其中，可以理解的是，基站的静态特征和动态特征，都属于预置的特征数据类型。基站的静态特征指的是基站中长期不会改变的一些属性，例如基站的类型、基站的供电类型、基站所处的地域类型等。反之，基站的动态特征是基站在不同的时间会有较大改变的一些属性，例如基站空调年限、电池年限、电池备电时长、市电的断电时长等等。对于不同地域的巡检问题，基站相应会有不同的属性，在建模的过程中，如果有需要的话需要进行持续的筛选。In the part of the base station feature data extraction, the static characteristics and dynamic characteristics of the base station are extracted mainly by combining the base station operation and maintenance data and the expert knowledge. It can be understood that the static features and dynamic features of the base station belong to the preset feature data type. The static characteristics of the base station refer to some attributes that the base station does not change in the long term, such as the type of the base station, the type of power supply of the base station, and the type of the area where the base station is located. Conversely, the dynamic characteristics of the base station are some attributes that the base station may change greatly at different times, such as the base station air conditioning age, battery life, battery backup duration, utility power off duration, and the like. For the inspection of different regions, the base station will have different attributes. In the process of modeling, continuous screening is needed if necessary.

示例性的，如下述表2所示，为基站的特征数据类型的示例说明。Illustrative, as shown in Table 2 below, is an illustrative illustration of the feature data type of the base station.

表2Table 2

(2)基站运行状态信息的获取(2) Acquisition of base station operating status information

需要说明的是，获取的基站运行状态信息可以作为基站巡检优先级的标识信息。基站巡检的目标是通过巡检，对基站潜在的问题进行排查，从而保障基站的安全运行。由于基站巡检的特殊性，可以同时采用两种不同的标识方式，在单个模型训练中，可以将基站未来一段时间的基站指标(例如告警数量、断站时长)作为基站的标识信息进行模型训练。在组合模型中，可以采用排序学习中的数据标识方式给基站巡检的优先级进行标识。It should be noted that the obtained base station running status information may be used as the identification information of the base station patrol priority. The goal of the base station inspection is to check the potential problems of the base station through patrol inspection to ensure the safe operation of the base station. Due to the particularity of the base station inspection, two different identification modes can be used at the same time. In a single model training, the base station indicators (such as the number of alarms and the length of the disconnection) of the base station in the future may be used as the identification information of the base station for model training. . In the combined model, the priority of the patrol inspection of the base station may be identified by using the data identification manner in the order learning.

示例性的，基站巡检优先级的标识可以分为二元标识以及五级标识或者其他等级方式的标识，具体不做限定。二元标识即区分基站中和巡检日期相关(需要去巡检)以及不相关(不需要去巡检)的基站，并分别标识1和0。在五级标识中，基站巡检的相关度可以进一步的细分，可以分为：完美(4)、出色(3)、好(2)、一般(1)、差(0)，其中，后面两档对应于不相关。For example, the identifier of the base station patrol priority may be classified into a binary identifier and a five-level identifier or an identifier of another level, which is not limited. The binary identifier distinguishes the base stations in the base station that are related to the inspection date (need to go to the inspection) and unrelated (do not need to go to the inspection), and identify 1 and 0 respectively. In the five-level identification, the correlation of the base station inspection can be further subdivided into: perfect (4), excellent (3), good (2), general (1), poor (0), among which The two files correspond to irrelevance.

需要说明的是，这里基站巡检优先级的标识，采用的是***数字，还可以是字母、文字，或者其任一组合等标识的方式，具体不做限定。It should be noted that the identifier of the base station inspection priority is an Arabic numeral, and may be a letter, a text, or any combination thereof, and is not limited.

模拟排序学习中的人工标识模式，基站巡检优先级的标识可以由业务专家根据基站的特征信息，给出基站的对应的巡检优先级，然而对于人工标识的方法，首先会存在较大的主观因素与分析，同时业务专家标识的基站巡检优先级与最终的业务指标结果可能存在较大的差异。基于以上问题，本申请提出了一种基于业务指标的标识方法，即直接用基站的可用率或者告警数量等业务指标标识基站的巡检优先级。在标识过程中，首先根据基站的业务指标对基站进行排序，并且需要根据巡检量确定相关的基站的数量，相较于人工标识的方法，这种业务指标直接标识的方式同时也是大大减低了人力的投入，同时摒弃了人为因素的干扰，直接使用业务结果作为指标，更为贴近业务目标。The artificial identification mode in the simulation sorting learning, the identifier of the base station inspection priority can be given by the service expert according to the characteristic information of the base station, and the corresponding inspection priority of the base station is given. However, for the manual identification method, there is a large first Subjective factors and analysis, and the base station inspection priority and the final business indicator result of the business expert identification may be quite different. Based on the above problem, the present application proposes an identification method based on the service indicator, that is, directly using the service indicator such as the availability rate or the number of alarms of the base station to identify the patrol priority of the base station. In the identification process, the base station is first sorted according to the service indicator of the base station, and the number of related base stations needs to be determined according to the patrol quantity. Compared with the manual identification method, the direct identification manner of the service indicator is also greatly reduced. The input of manpower, while abandoning the interference of human factors, directly using business results as indicators, is closer to business objectives.

下面对图2所示中的巡检模型构建模块进行详细说明，如下所示：The following is a detailed description of the inspection model building module shown in Figure 2, as follows:

基站告警及断站受多种因素影响，如高温、断电等。在巡检模型构建模块本申请提出了一种多因素集成模型对基站未来业务指标进行预测的基站巡检模型。示例性的，如图3所示，图3为巡检模型构建模块中组合模型的一个示意图。Base station alarms and disconnections are affected by many factors, such as high temperature and power failure. In the inspection model building module, the present application proposes a base station inspection model for predicting future service indicators of a base station by a multi-factor integration model. Exemplarily, as shown in FIG. 3, FIG. 3 is a schematic diagram of a combined model in a patrol model building module.

在图3所示中，受不同因素的影响，基站告警及断站可以分为不同的类型，例如高温、断电、传输、传感器等等。在巡检模型构建模块本申请提出了一种多因素集成模型对基站未来运行状态进行预测的巡检模型。以基站的告警作示例，如某区域基站告警主要分为A(高温)、B(断电)、C(传输)三类，即可以对三类告警分别建立相应的预测模型，即因子A模型，因子B模型以及因子C模型。告警一般为连续的数值，因此预测模型一般为回归模型。三个模型相应的加和即为基站的总得分，也是基站未来一段时间的告警的预测值即基站未来业务指标的预测值。进一步可根据基站的总得分，进行基站倒序排序，对于其中的总得分相同的基站，可以结合过去一段时间的告警数量或者断电时长进行二次排序，从而得到基站最终的排序列表。In Figure 3, due to different factors, base station alarms and disconnects can be divided into different types, such as high temperature, power off, transmission, sensors and so on. In the inspection model building module, the present application proposes a multi-factor integration model to predict the future operation state of the base station. Take the alarm of the base station as an example. For example, the alarms of a certain base station are mainly classified into three types: A (high temperature), B (power off), and C (transmission). That is, the corresponding prediction models can be established for the three types of alarms, that is, the factor A model. , factor B model and factor C model. Alarms are generally continuous values, so the prediction model is generally a regression model. The corresponding sum of the three models is the total score of the base station, and is also the predicted value of the alarm of the base station in the future, that is, the predicted value of the future service indicator of the base station. Further, the base station may be sorted in reverse order according to the total score of the base station. For the base stations with the same total score, the base station may be combined with the number of alarms in the past period or the power-off duration to obtain a final sorted list of the base stations.

在巡检任务生成模块中，根据巡检模型给出的站点排序结果，运维人员根据实际的运维能力，例如每周的运维站点数量，可以选取相应的排名较高同时巡检间隔较长的站点，安排相应的巡检任务。需要补充说明的，在完成相应站点的巡检任务后，运维人员将巡检 记录上传至运维数据管理***。In the patrol task generation module, according to the site ranking result given by the patrol model, the operation and maintenance personnel can select the corresponding higher ranking and the patrol interval according to the actual operation and maintenance capabilities, such as the number of weekly operation and maintenance sites. Long site, arrange the corresponding inspection tasks. If additional explanations are required, the operation and maintenance personnel upload the inspection records to the operation and maintenance data management system after completing the inspection tasks of the corresponding sites.

在本申请中，首先，可以根据基站运维数据，即基站的设备等信息以及基站相应的运行状态，构建巡检模型从而实现对基站的排序，是一种监督学习，因此在模型训练过程中需要构建基站的特征(也称为Feature)，以及对训练数据进行标识，这个是基站特征构造模块的主要任务。当完成基站特征抽取后，即可进行进一步的巡检模型构建。巡检模型构建模块主要负责模型训练，主选定假设函数和损失函数，基于现有训练数据，调整参数，使得损失函数最小。In the present application, firstly, according to the operation and maintenance data of the base station, that is, the information of the base station, and the corresponding running state of the base station, the patrol inspection model is constructed to realize the ordering of the base station, which is a supervised learning, and thus in the model training process. It is necessary to construct the characteristics of the base station (also called Feature) and to identify the training data, which is the main task of the base station feature construction module. After the base station feature extraction is completed, a further inspection model construction can be performed. The inspection model building module is mainly responsible for model training. The main selection hypothesis function and loss function are based on the existing training data, and the parameters are adjusted to minimize the loss function.

在传统的排序或者信息检索中，训练的结果往往是一个分类或者回归函数，在之后的检索中，直接利用这个分类或者回归函数对基站进行打分，根据分数进行排序从而选择相应的巡检基站。在本申请实施例中，针对基站的特点，采用组合模型的方式先多个回归函数对基站不同维度分别打分，综合之后再对基站进行排序的方式选择需要巡检的基站。巡检任务生成模块中，主要是根据新的巡检数据，以及给出的巡检时间，利用巡检模型对基站数据进行评估，得到基站列表排序，并结合业务要求，选择基站进行巡检。In traditional sorting or information retrieval, the result of training is often a classification or regression function. In the subsequent retrieval, the base station is directly scored by using this classification or regression function, and the corresponding patrol base stations are selected according to the scores. In the embodiment of the present application, for the characteristics of the base station, a plurality of regression functions are used to score different dimensions of the base station by using a combined model, and then the base station that needs to be inspected is selected in a manner of sorting the base stations after synthesis. In the patrol task generation module, the base station data is evaluated by using the patrol model according to the new patrol data and the patrol time given, and the base station list is sorted, and the base station is selected for patrol inspection according to the service requirements.

下面以实施例的方式，对本申请技术方案做进一步的说明，如图4所示，图4为本申请实施例中确定巡检基站列表的方法的一个实施例示意图。The technical solution of the present application is further described in the following embodiments. As shown in FIG. 4, FIG. 4 is a schematic diagram of an embodiment of a method for determining a list of patrol base stations according to an embodiment of the present application.

401、获取第一基站集合的历史运维数据。401. Obtain historical operation and maintenance data of the first base station set.

在本申请实施例中，数据预处理模块可以主动从基站运维***获取第一基站集合的历史运维数据；也可以是被动接收基站运维***发送的第一基站集合的历史运维数据。历史运维数据可以是第一基站集合中各个基站的设备等信息以及第一基站集合中各个基站相应的运行状态等信息，此处不做具体限定。In the embodiment of the present application, the data pre-processing module may actively obtain the historical operation and maintenance data of the first base station set from the base station operation and maintenance system, or may be the historical operation and maintenance data of the first base station set sent by the passive receiving base station operation and maintenance system. The historical operation and maintenance data may be information such as the device of each base station in the first base station set, and the corresponding running status of each base station in the first base station set, and is not specifically limited herein.

402、根据历史运维数据得到第二基站集合的历史特征数据，以及所述第二基站集合中每个基站的运行状态信息。402. Obtain historical feature data of the second base station set according to historical operation and maintenance data, and running state information of each base station in the second base station set.

示例性的，数据预处理模块获取第一基站集合的历史运维数据之后，可以对历史运维数据进行数据预处理，即可以对历史运维数据进行清洗，去除异常数据，得到第一定性数据；然后，可以对第一定性数据进行等量化等处理。For example, after obtaining the historical operation and maintenance data of the first base station set, the data pre-processing module may perform data pre-processing on the historical operation and maintenance data, that is, the historical operation and maintenance data may be cleaned, the abnormal data is removed, and the first qualitative is obtained. Data; then, the first qualitative data can be equally quantized and the like.

(1)基站特征数据抽取：(1) Base station feature data extraction:

根据专家知识以及第一基站集合的历史运维数据，进行第一基站集合的历史特征数据构建，以区域A基站为例来进行说明。即使用预置的抽取方式，从历史运维数据中，抽取预置的特征数据类型对应的特征数据。对区域A中基站的运维数据从基站类型、基站传输类型、外部设备等几个维度共选取了23维特征，如下表3所示，为区域A基站特征数据的示例说明：Based on the expert knowledge and the historical operation and maintenance data of the first base station set, the historical feature data of the first base station set is constructed, and the area A base station is taken as an example for description. That is, using the preset extraction method, the feature data corresponding to the preset feature data type is extracted from the historical operation and maintenance data. For the operation and maintenance data of the base station in the area A, 23-dimensional features are selected from the base station type, the base station transmission type, and the external device, as shown in Table 3 below, which is an example of the regional A base station feature data:

特征标号Feature label	特征名称Feature name	用途use	特征属性Feature attribute
11	is_hubIs_hub	用于判断是否是HUB基站Used to determine whether it is a HUB base station	静态Static
22	is_indoorIs_indoor	基站类型Base station type	静态Static
33	is_outdoorIs_outdoor	基站类型Base station type	静态Static
44	is_ibsIs_ibs	基站类型Base station type	静态Static

55	is_smallCellis_smallCell	基站类型Base station type		静态Static
66	is_cowIs_cow	基站类型Base station type	静态Static
77	is_mvIs_mv	基站传输类型Base station transmission type	静态Static
88	is_fiberIs_fiber	基站传输类型Base station transmission type	静态Static
99	gen_onlyGen_only	是否只有油机Is it only oil machine?	静态 Static
1010	Stolen_or_VandalismStolen_or_Vandalism	是否存在被偷和破坏的情况Whether there are cases of being stolen and destroyed	静态 Static
1111	noCivilWorknoCivilWork	是否有市政施工Is there municipal construction?	静态Static
1212	desertSitedesertSite	是否是沙漠基站Is it a desert base station?	动态dynamic
1313	seaSiteseaSite	是否是海边基站Is it a seaside base station?	静态Static
1414	siteTypesiteType	基站Base station	静态Static
1515	airConBrandairConBrand	空调品牌Air conditioning brand	静态Static
1616	is_summerIs_summer	当前气候Current climate	动态dynamic
2020	grid_levelGrid_level	市电的稳定性Mains stability	动态dynamic
21twenty one	batteryAgebatteryAge	电池服务时长Battery service duration	动态dynamic
22twenty two	airConAgeairConAge	空调服务时长Air conditioning service duration	动态dynamic
23twenty three	backupTimebackupTime	电池备电时长Battery backup time	动态dynamic

表3table 3

其中，部分基站特征数据抽取方式如下所示：Among them, some base station feature data extraction methods are as follows:

1)静态特征1) Static features

空调品牌(airConBrand)对应如下：The air-conditioning brand (airConBrand) corresponds to the following:

DAIKIN：1DAIKIN: 1

LIEBERT：2LIEBERT: 2

TOPCOOL：3TOPCOOL: 3

Others：4Others:4

基站特征数据类型(siteType)的构造方式如下：The base station feature data type (siteType) is constructed as follows:

如果一个基站同时是2g、3g和4g基站，则该基站特征数据类型标记为3；If a base station is 2g, 3g, and 4g base stations at the same time, the base station feature data type is marked as 3;

如果一个基站同时是2g和3g基站，则该基站特征数据类型标记为2；If a base station is both 2g and 3g base stations, the base station feature data type is marked as 2;

如果一个基站同时是3g和4g基站，则该基站特征数据类型标记为2；If a base station is a 3g and 4g base station at the same time, the base station feature data type is marked as 2;

如果一个基站同时为2g和4g基站，则该基站特征数据类型标记为2；If a base station is both 2g and 4g base stations, the base station feature data type is marked as 2;

如果一个基站为2g基站，则该基站特征数据类型标记为1；If a base station is a 2 g base station, the base station feature data type is marked as 1;

如果一个基站为3g基站，则该基站特征数据类型标记为1；If a base station is a 3g base station, the base station feature data type is marked as 1;

如果一个基站为4g基站，则该基站特征数据类型标记为1。If a base station is a 4g base station, the base station feature data type is marked as 1.

2)动态特征2) Dynamic features

其中的动态变量airConAge和batteryAge取设备安装日期到‘巡检日期’(例如2016/5/27)的天数，backupTime表示的是电池的备电时长，取PM报告中最近电池备电测试结果。grid_level为过去一周的市电断电时长(小时)。The dynamic variables airConAge and batteryAge take the number of days from the installation date of the device to the patrol date (for example, 2016/5/27). The backupTime indicates the battery backup time and the latest battery backup test result in the PM report. Grid_level is the length of the mains power outage (hours) in the past week.

(2)基站的运行状态信息：(2) Operation status information of the base station:

需要说明的是，基站的运行状态信息可以标识基站的巡检优先级。根据区域A的每周巡检量，对于每个巡检日期(Date)，可以将基站列表中总告警数量的前7％的基站标记为1(即需要巡检的基站)，其他的标记为0。It should be noted that the running status information of the base station may identify the patrol priority of the base station. According to the weekly patrol volume of area A, for each patrol date (Date), the first 7% of the total number of alarms in the base station list can be marked as 1 (ie, the base station requiring patrol), and the other labels are 0.

对于区域A基站的数据，可以将告警分为空调相关、市电相关以及其他因素相关(在不同的区域，告警的类别可根据专家经验做其他细分)。在训练数据构造时，可以假设每周生成巡检基站的模式，共构建了2016/5/27至2017/4/28共49天的数据。数据示例如下，为区域A基站训练数据的示例。For the data of the area A base station, the alarms can be classified into air-conditioning related, mains-related, and other factors (in different areas, the types of alarms can be subdivided according to expert experience). In the training data construction, it can be assumed that the mode of the patrol base station is generated every week, and a total of 49 days of data from 2016/5/27 to 2017/4/28 is constructed. An example of the data is as follows, an example of training data for the area A base station.

site_idSite_id	S131S131	S128S128	S137S137	S168S168
DateDate	2016052720160527	2016052720160527	2016052720160527	2016052720160527
is_hubIs_hub	11	11	11	11
is_indoorIs_indoor	11	11	11	11
is_outdoor Is_outdoor	00	00	00	00
is_ibs Is_ibs	00	00	00	00
is_smallCell is_smallCell	00	00	00	00
is_cow Is_cow	00	00	00	00
is_mvIs_mv	11	11	11	11
is_fiber Is_fiber	00	00	00	00
gen_onlyGen_only	11	11	11	11
Stolen_or_VandalismStolen_or_Vandalism	11	11	11	11
noCivilWork noCivilWork	00	00	00	00
desertSitedesertSite	11	11	11	11
seaSite seaSite	00	00	11	00
siteTypesiteType	33	33	33	33

airConBrandairConBrand	22	22	33	22
is_summerIs_summer	11	11	11	11
grid_levelGrid_level	2.6572.657	00	00	00
batteryAgebatteryAge	468468	352352	461461	461461
airConAgeairConAge	27342734	26682668	23972397	27172717
backupTimebackupTime	100100	140140	120120	100100
pmIntervalpmInterval	500500	7373	5555	500500
alarmOtheralarmOther	33	77	11	22
alarmPoweralarmPower	22	00	00	00
alarmACalarmAC	22	00	00	00
alarmAlarm	77	77	11	22
labelLabel	11	11	00	00

表4Table 4

其中，alarmOther、alarmPower和alarmAC分别为基站未来一周的其他告警、市电断电告警和空调相关告警。alarm则为未来一周基站的总告警数目。Among them, alarmOther, alarmPower and alarmAC are other alarms, power-off alarms and air-conditioning related alarms of the base station in the coming week. Alarm is the total number of alarms for the base station in the coming week.

403、根据第一基站集合的历史特征数据以及第一基站集合中每个基站的运行状态信息，训练得到不同类型的多个巡检模型。403. Train, according to the historical feature data of the first base station set and the running state information of each base station in the first base station set, to obtain multiple patrol models of different types.

可以理解的是，巡检模型也可以称为巡检排序模型、排序模型等。巡检模型可以是深度神经网络，也可以是决策树，也可以是随机森林等，不做具体限定。运行状态信息可以为每个基站在第一预置时长内的告警数量或者断电时长。这里的第一预置时长可以是设置的一段时长，这里的第一预置时长，可以根据实际需求，灵活调整。It can be understood that the inspection model can also be called a polling sorting model, a sorting model, and the like. The inspection model may be a deep neural network, a decision tree, or a random forest, and is not specifically limited. The running status information may be the number of alarms or the power-off duration of each base station within a first preset duration. The first preset duration can be set for a period of time, and the first preset duration here can be flexibly adjusted according to actual needs.

示例性的，下面以巡检模型为决策数据为例进行说明，对于组合模型的训练过程如下所示：Exemplarily, the following is an example of a decision model for the decision data. The training process for the combined model is as follows:

根据区域A基站的区域特性以及专家知识，可以将基站的告警细化分为空调相关、电源相关和其他三类，并对这三类别分别构建决策树模型，对这三类相关的告警进行预测，每类决策树模型都对应得到一个得分。这三类决策树的得分之和可以作为基站的总得分，然后根据每个基站的总得分对基站进行排序，整体的框架见图5所示，图5为区域A基站的巡检模型结构的示意图。According to the regional characteristics and expert knowledge of the regional base station, the base station's alarm refinement can be divided into three categories: air-conditioning correlation, power supply correlation and other three categories. The decision tree model is constructed for each of the three categories, and the three types of related alarms are predicted. Each type of decision tree model corresponds to a score. The sum of the scores of the three types of decision trees can be used as the total score of the base station, and then the base stations are sorted according to the total score of each base station. The overall framework is shown in FIG. 5, and FIG. 5 is the structure of the inspection model of the area A base station. schematic diagram.

单个模型的训练过程可以以断电相关告警模型为例进行说明：The training process of a single model can be illustrated by taking the power-related alarm model as an example:

首先利用决策树模型对变量(因素)重要性进行分析，标准化后的各个因素基尼系数如下，数值越高，表示该因素的变化会更多影响断电导致的告警数量，即数值越高，影响越大。Firstly, the decision tree model is used to analyze the importance of variables (factors). The Gini coefficient of each factor after standardization is as follows. The higher the value, the more the change of the factor will affect the number of alarms caused by power failure, that is, the higher the value, the influence The bigger.

表5table 5

如表5所示，为断电告警各因素的重要性。由上表5可见，断电告警主要和市电断电时长相关，是否是沙漠基站、电池年限和备电时长也会有一定的影响。再次，还可以选取最为重要的4个因素建立决策树，模型如图6所示，图6为断电告警决策树的一个示意图。As shown in Table 5, the importance of various factors for power failure alarms. It can be seen from Table 5 above that the power failure alarm is mainly related to the length of the mains power failure, and whether it is the desert base station, the battery life and the backup power duration will also have a certain impact. Again, the most important four factors can be selected to establish a decision tree. The model is shown in Figure 6. Figure 6 is a schematic diagram of the power-off alarm decision tree.

整体而言，市电断电时间较长的基站其断电告警量更高，同时基站的备电电池年限较高，备电时长较小的基站其断电告警也会更高，这些结果基本符合专家知识。Overall, the base station with a longer power-off period has a higher power-off alarm, and the base station has a higher battery life, and the base station with a smaller backup time has a higher power-off alarm. These results are basically Meet expert knowledge.

另外两类告警：空调相关告警以及其他告警模型也可以采用以上方式分别获得，此处不再赘述。The other two types of alarms: air-conditioning-related alarms and other alarm models can also be obtained in the above manner, and are not described here.

需要说明的是，在巡检模型构建好之后，就可以使用多个巡检模型计算得到基站的巡检顺序了。It should be noted that after the patrol model is constructed, the patrol sequence of the base station can be calculated using multiple patrol models.

404、获取第二基站集合的当前运维数据。404. Obtain current operation and maintenance data of the second base station set.

数据预处理模块可以主动从基站运维***获取第二基站集合的当前运维数据；也可以是被动接收基站运维***发送的第二基站集合的当前运维数据。当前运维数据可以是当前需要待巡检区域中各个基站的设备等信息以及待巡检区域中各个基站相应的运行状态等信息，此处不做具体限定。这里的第二基站集合可以和上述的第一基站集合相同，也可以不同。The data pre-processing module may actively acquire the current operation and maintenance data of the second base station set from the base station operation and maintenance system; or may be the current operation and maintenance data of the second base station set sent by the passive receiving base station operation and maintenance system. The current operation and maintenance data may be information such as the information of the devices that need to be in the patrol area and the corresponding running status of each base station in the patrol area, and is not specifically limited herein. The second set of base stations herein may be the same as or different from the first set of base stations described above.

可选的，数据预处理模块获取第二基站集合的当前运维数据之后，可以对第二基站集合的当前运维数据进行数据预处理，即可以对当前运维数据进行清洗，去除异常数据，得到第二定性数据；然后，可以对第二定性数据进行等量化等。Optionally, after obtaining the current operation and maintenance data of the second base station set, the data pre-processing module may perform data pre-processing on the current operation and maintenance data of the second base station set, that is, the current operation and maintenance data may be cleaned to remove abnormal data. The second qualitative data is obtained; then, the second qualitative data can be equally quantized and the like.

405、根据当前运维数据进行基站特征构建，得到第二基站集合的当前特征数据。405. Perform base station feature construction according to current operation and maintenance data, and obtain current feature data of the second base station set.

具体的，可以包括：使用预置的抽取方式从当前运维数据中，抽取预置的特征数据类型对应的特征数据，得到第二基站集合的当前特征数据。可以理解的是，当前特征数据可以包括但不限于基站类型、基站区域、主设备服务年限、外设备服务年限。Specifically, the method may include: extracting feature data corresponding to the preset feature data type from the current operation and maintenance data by using a preset extraction manner, to obtain current feature data of the second base station set. It can be understood that the current feature data may include, but is not limited to, a base station type, a base station area, a primary device service age, and an outer device service age.

详细过程可参考前文中对基站特征数据抽取部分的说明，此处不再赘述。For a detailed procedure, reference may be made to the description of the base station feature data extraction part in the foregoing, and details are not described herein again.

406、根据当前特征数据和预先构建的不同类型的多个巡检模型，得到第一基站集合中每个基站对应的各个巡检模型的得分。406. Obtain a score of each patrol model corresponding to each base station in the first base station set according to the current feature data and the pre-built multiple types of patrol models.

示例性的，预先构建巡检模型为高温告警模型、断电告警模型、主设备告警模型。根据当前特征数据，以及这三个巡检模型，对每个基站来说，可以分别得到3个得分。Exemplarily, the patrol model is pre-built into a high temperature alarm model, a power failure alarm model, and a master device alarm model. According to the current feature data, and the three patrol models, for each base station, three scores can be obtained respectively.

407、根据每个基站对应的各个巡检模型的得分，计算得到每个基站的总得分。407. Calculate a total score of each base station according to scores of each patrol model corresponding to each base station.

示例性的，每个基站的总得分，为各个巡检模型的得分加起来的分数。Exemplarily, the total score of each base station is the sum of the scores of the individual inspection models.

408、根据每个基站的总得分，计算得到巡检基站列表。408. Calculate a list of the patrol base stations according to the total score of each base station.

假设，第二基站集合中有10个基站，每个基站都会有一个总得分，那么，根据每个基站的总得分，就可以计算得到巡检基站列表了。Assume that there are 10 base stations in the second base station set, and each base station will have a total score. Then, according to the total score of each base station, the list of the patrol base stations can be calculated.

可选的，若第二基站集合中的各个基站的总得分中存在总得分相同的基站，则确定总得分相同的基站中在第二预置时长内未巡检的基站为待巡检基站列表。即当出现总得分相同的基站时，可以看下，这些基站近期有没有被巡检过，如果有被巡检过，那么，这次，可以先不用巡检了，可以对近期未巡检的基站进行巡检。尽可能的保证，提高问题基站的巡检概率。Optionally, if there is a base station with the same total score in the total score of each base station in the second base station set, determining that the base station that is not inspected within the second preset time period of the base station with the same total score is the list of the base station to be patrolled . That is, when there are base stations with the same total score, you can see if these base stations have been inspected recently. If they have been patrolled, then this time, you can use the patrol first. The base station performs a patrol. As much as possible to ensure that the probability of inspection of the problem base station.

需要说明的是，第二预置时长，可以和前文中的第一预置时长相同，也可以不同，也可以根据实际需求而灵活调整。It should be noted that the second preset duration may be the same as the first preset duration in the foregoing, or may be different, or may be flexibly adjusted according to actual needs.

409、根据巡检基站列表对对应的基站进行巡检。409. Perform a patrol on the corresponding base station according to the patrol base station list.

可以理解的是，步骤409为可选的步骤，可以在巡检装置上显示巡检基站列表，然后，用户可以根据显示的巡检基站列表，可以对总得分较高的基站进行巡检。It can be understood that step 409 is an optional step, and the patrol base station list can be displayed on the patrol device. Then, the user can perform patrol on the base station with a higher total score according to the displayed patrol base station list.

在模型应用模块，主要是根据第二基站集合的当前运维数据，提取第二基站集合的当前特征数据，通过巡检模型(多个不同类型的巡检模型)得到基站的总得分，并由此给基站进行排序，进一步，对于得分相同的基站，可以根据基站过去一段时间(例如一个月)的基站告警量进行再次的排序。根据排序结果，结合基站的巡检日期，选择巡检基站。如图7所示，图7为巡检任务生成模块的流程示意图。In the model application module, the current feature data of the second base station set is extracted according to the current operation and maintenance data of the second base station set, and the total score of the base station is obtained by the inspection model (a plurality of different types of inspection models), and This sorts the base stations. Further, for base stations with the same score, the base station alarms of the base station in a past period of time (for example, one month) can be sorted again. According to the sorting result, the patrol base station is selected in combination with the patrol date of the base station. As shown in FIG. 7, FIG. 7 is a schematic flowchart of a patrol task generation module.

将空调、电源以及其他告警的决策树模型结果进行叠加即为基站的预测告警值，对该告警值进行排序，其中前k个基站即为选定的巡检基站。对于区域A基站而言，每周的巡检量为8％左右，即若采用按历程巡检的方式，准确率为7％，采用巡检模型，其准确率为20％，总结如表6所示，表6为试验数据结果的一个示例。The result of the decision tree model of the air conditioner, the power supply, and other alarms is superimposed as the predicted alarm value of the base station, and the alarm value is sorted, wherein the first k base stations are the selected patrol base stations. For the regional A base station, the weekly inspection quantity is about 8%, that is, if the accuracy is 7% according to the history inspection method, the inspection model is used, and the accuracy rate is 20%, as summarized in Table 6. As shown, Table 6 is an example of the results of the test data.

表6Table 6

从结果可见，精准巡检的查准率约为按里程巡检的3倍左右，即利用本申请的巡检方案，问题基站有更大的可能性被及时巡检。It can be seen from the results that the accuracy of the precision inspection is about 3 times that of the mileage inspection. That is to say, with the inspection scheme of the present application, the problem base station has a greater possibility of being inspected in time.

目前算法已经在区域A实验局进行上线部署运行，示例性的，如图8所示，图8为基站故障密度的示意图。在图8所示中，给出了区域A在8月1日-11月4日基站故障密度，试验局开始前后(10月1日开始)，区域A中的试验区域和非试验区域故障周密度降低为44％vs34％，说明试验区域即使频次降低了20％，网络质量相比其他区域没有降低。At present, the algorithm has been deployed in the regional A experimental office for online deployment. As an example, as shown in FIG. 8, FIG. 8 is a schematic diagram of the base station fault density. In Figure 8, the base station fault density of area A from August 1st to November 4th is given. Before and after the start of the test station (starting on October 1st), the test area and the non-test area failure week in area A are shown. The density reduction was 44% vs 34%, indicating that the test area was reduced by 20% even if the frequency was reduced, and the network quality was not lower than other areas.

本申请提出的一种基于排序的巡检动态调度方案，根据排序结果及时调整巡检频次以及巡检安排，具有以下的有益效果：可以及时解决网络故障及隐患，保证网络安全高效运 行同时减少故障检修的频次；提升巡检的针对性，聚焦智能化处理，降低运维成本，提升代维能力以及运维效率；提供针对性的主动运维方案，实现例行维护到按需维护的转变，提升方案的竞争力。A sorting-based patrol dynamic scheduling scheme proposed by the present application adjusts the patrol frequency and the patrol arrangement according to the sorting result in time, and has the following beneficial effects: the network fault and the hidden danger can be solved in time, and the network is safely and efficiently operated while reducing the fault. The frequency of overhaul; enhance the pertinence of inspections, focus on intelligent processing, reduce operation and maintenance costs, improve generation and maintenance capabilities, and improve operation and maintenance efficiency; provide targeted active operation and maintenance solutions to achieve the transition from routine maintenance to on-demand maintenance. Improve the competitiveness of the program.

本方法可用于各类的基站评估以及巡检安排，例如电网基站或者设备巡检。The method can be used for various types of base station evaluation and inspection arrangements, such as power grid base stations or equipment inspection.

本申请技术方案所要解决的问题是提出新型巡检动态调度机制和框架，人为制定的“静态”巡检周期调整为算法自动判断的“动态”巡检周期，从按规定巡检转化为按需求巡检，把每一次巡检任务分配给需要的基站。The problem to be solved by the technical solution of the present application is to propose a new dynamic inspection mechanism and framework for patrol inspection, and the "static" inspection cycle set by the human being is adjusted to the "dynamic" inspection cycle automatically determined by the algorithm, from the patrol inspection to the demand-based inspection. Patrol, assign each inspection task to the required base station.

在本申请实施例中，将巡检机制调度问题转化为对预置时长内基站的优先级排序问题，通过机器学习以及运维大数据，实现对基站运行的评估，从而把巡检任务安排给更需要的基站。结合专家知识，构建基站特征数据，利用基站未来一段时间的业务指标对基站进行标识。通过组合模型的方式，将基站业务指标进行细分，先分别建模，再进行组合，从而提高了模型的可解释性以及准确性。In the embodiment of the present application, the scheduling problem of the patrol mechanism is converted into a priority ordering problem for the base station within the preset duration, and the evaluation of the operation of the base station is implemented through machine learning and operation and maintenance of big data, thereby arranging the patrol task to More needed base stations. Combining expert knowledge, the base station feature data is constructed, and the base station is identified by using the service indicator of the base station for a certain period of time. By combining the models, the base station service indicators are subdivided, and then separately modeled and then combined to improve the interpretability and accuracy of the model.

上面对本申请实施例中的方法实施例进行了说明，下面对本申请实施例中的装置部分进行说明，如图9所示，图9为本申请实施例中巡检装置的一个实施例示意图。可以包括：The method embodiment of the present application is described above. The following is a description of the device in the embodiment of the present application. As shown in FIG. 9, FIG. 9 is a schematic diagram of an embodiment of a patrol device according to an embodiment of the present application. Can include:

获取模块901，用于获取第一基站集合的当前运维数据；The obtaining module 901 is configured to acquire current operation and maintenance data of the first base station set.

处理模块902，用于根据当前运维数据进行基站特征构建，得到第一基站集合的当前特征数据；根据当前特征数据和预先构建的不同类型的多个巡检模型，得到第一基站集合中每个基站对应的各个巡检模型的得分；根据每个基站对应的各个巡检模型的得分，计算得到每个基站的总得分；根据每个基站的总得分，计算得到巡检基站列表。The processing module 902 is configured to perform base station feature construction according to current operation and maintenance data, to obtain current feature data of the first base station set, and obtain each of the first base station set according to the current feature data and the pre-built multiple types of different inspection models. The scores of the respective patrol models corresponding to the base stations; the total score of each base station is calculated according to the scores of the respective patrol models corresponding to each base station; and the patrol base station list is calculated according to the total score of each base station.

可选的，在本申请的一些实施例中，Optionally, in some embodiments of the present application,

获取模块901，还用于获取第二基站集合的历史运维数据；The obtaining module 901 is further configured to acquire historical operation and maintenance data of the second base station set;

处理模块902，还用于根据历史运维数据得到第二基站集合的历史特征数据，以及第二基站集合中每个基站的运行状态信息；根据历史特征数据以及第二基站集合中每个基站的运行状态信息，训练得到不同类型的多个巡检模型。The processing module 902 is further configured to obtain historical feature data of the second base station set according to the historical operation and maintenance data, and operation state information of each base station in the second base station set; and according to the historical feature data and each base station in the second base station set Running state information, training to get multiple inspection models of different types.

可选的，在本申请的一些实施例中，Optionally, in some embodiments of the present application,

处理模块902，具体用于使用预置的抽取方式从当前运维数据中，抽取预置的特征数据类型对应的特征数据，得到第一基站集合的当前特征数据。The processing module 902 is specifically configured to extract feature data corresponding to the preset feature data type from the current operation and maintenance data by using a preset extraction manner to obtain current feature data of the first base station set.

可选的，在本申请的一些实施例中，Optionally, in some embodiments of the present application,

当前特征数据包括基站类型、基站区域、主设备服务年限、外设备服务年限。The current feature data includes the base station type, the base station area, the service life of the primary device, and the service life of the external device.

可选的，在本申请的一些实施例中，Optionally, in some embodiments of the present application,

第二基站集合中每个基站的运行状态信息包括：第二基站集合中每个基站在第一预置时长内的告警数量或者断电时长。The running status information of each base station in the second base station set includes: the number of alarms or the power-off duration of each base station in the second base station set in the first preset duration.

可选的，在本申请的一些实施例中，Optionally, in some embodiments of the present application,

处理模块902，还用于若第一基站集合中的各个基站的总得分中存在总得分相同的基站，则确定总得分相同的基站中在第二预置时长内未巡检的基站为待巡检基站列表。The processing module 902 is further configured to: if there is a base station with the same total score among the total scores of the base stations in the first base station set, determine that the base station that is not inspected within the second preset time period of the base stations with the same total score is to be patrolled. Check the list of base stations.

如图10所示，图10为本申请实施例中巡检装置的另一个实施例示意图。可以包括：As shown in FIG. 10, FIG. 10 is a schematic diagram of another embodiment of a patrol inspection apparatus according to an embodiment of the present application. Can include:

处理器1001和存储器1002；The processor 1001 and the memory 1002;

其中，存储器1002用于存储程序，当程序被处理器1001调用时，用于执行以下步骤：The memory 1002 is configured to store a program, and when the program is called by the processor 1001, is used to perform the following steps:

获取第一基站集合的当前运维数据；Obtaining current operation and maintenance data of the first base station set;

根据当前运维数据进行基站特征构建，得到第一基站集合的当前特征数据；Performing base station feature construction according to current operation and maintenance data, and obtaining current feature data of the first base station set;

根据当前特征数据和预先构建的不同类型的多个巡检模型，得到第一基站集合中每个基站对应的各个巡检模型的得分；Obtaining scores of respective patrol models corresponding to each base station in the first base station set according to the current feature data and the plurality of patrol models of different types pre-built;

根据每个基站对应的各个巡检模型的得分，计算得到每个基站的总得分；Calculating the total score of each base station according to the scores of the respective patrol models corresponding to each base station;

根据每个基站的总得分，计算得到巡检基站列表。Based on the total score of each base station, a list of patrol base stations is calculated.

可选的，在本申请的一些实施例中，处理器1001还用于执行以下步骤：Optionally, in some embodiments of the present application, the processor 1001 is further configured to perform the following steps:

获取第二基站集合的历史运维数据；Obtaining historical operation and maintenance data of the second base station set;

根据历史运维数据得到第二基站集合的历史特征数据，以及第二基站集合中每个基站的运行状态信息；根据历史特征数据以及第二基站集合中每个基站的运行状态信息，训练得到不同类型的多个巡检模型。Obtaining historical feature data of the second base station set and running state information of each base station in the second base station set according to the historical operation and maintenance data; and training is different according to the historical feature data and the running state information of each base station in the second base station set Multiple inspection models of the type.

可选的，在本申请的一些实施例中，处理器1001还用于执行以下步骤：Optionally, in some embodiments of the present application, the processor 1001 is further configured to perform the following steps:

使用预置的抽取方式从当前运维数据中，抽取预置的特征数据类型对应的特征数据，得到第一基站集合的当前特征数据。The feature data corresponding to the preset feature data type is extracted from the current operation and maintenance data by using a preset extraction method to obtain current feature data of the first base station set.

可选的，在本申请的一些实施例中，当前特征数据包括基站类型、基站区域、主设备服务年限、外设备服务年限。Optionally, in some embodiments of the present application, the current feature data includes a base station type, a base station area, a primary device service age, and an outer device service age.

可选的，在本申请的一些实施例中，第二基站集合中每个基站的运行状态信息包括：第二基站集合中每个基站在第一预置时长内的告警数量或者断电时长。Optionally, in some embodiments of the present application, the running status information of each base station in the second base station set includes: the number of alarms or the power-off duration of each base station in the second set of base stations in the first preset duration.

可选的，在本申请的一些实施例中，处理器1001还用于执行以下步骤：Optionally, in some embodiments of the present application, the processor 1001 is further configured to perform the following steps:

若第一基站集合中的各个基站的总得分中存在总得分相同的基站，则确定总得分相同的基站中在第二预置时长内未巡检的基站为待巡检基站列表。If there is a base station with the same total score among the total scores of the base stations in the first base station set, it is determined that the base stations that are not patrolled within the second preset time period of the base stations with the same total score are the list of base stations to be patrolled.

在上述实施例中，可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行所述计算机程序指令时，全部或部分地产生按照本申请实施例所述的流程或功能。所述计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。所述计算机指令可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一个计算机可读存储介质传输，例如，所述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质，(例如，软盘、硬盘、磁带)、光介质(例如，DVD)、或者半导体介质(例如固态硬盘Solid State Disk(SSD))等。In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, it may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in accordance with embodiments of the present application are generated in whole or in part. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer readable storage medium or transferred from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be from a website site, computer, server or data center Transfer to another website site, computer, server, or data center by wire (eg, coaxial cable, fiber optic, digital subscriber line (DSL), or wireless (eg, infrared, wireless, microwave, etc.). The computer readable storage medium can be any available media that can be accessed by a computer or a data storage device such as a server, data center, or the like that includes one or more available media. The usable medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium (such as a solid state disk (SSD)).

所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的***，装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。A person skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the system, the device and the unit described above can refer to the corresponding process in the foregoing method embodiment, and details are not described herein again.

在本申请所提供的几个实施例中，应该理解到，所揭露的***，装置和方法，可以通 过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个***，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。In the several embodiments provided herein, it should be understood that the disclosed systems, devices, and methods may be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.

所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.

另外，在本申请各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

所述集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(read-only memory，ROM)、随机存取存储器(random access memory，RAM)、磁碟或者光盘等各种可以存储程序代码的介质。The integrated unit, if implemented in the form of a software functional unit and sold or used as a standalone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application, in essence or the contribution to the prior art, or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium. A number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like, which can store program code. .

以上所述，以上实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围。The above embodiments are only used to explain the technical solutions of the present application, and are not limited thereto; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still The technical solutions described in the embodiments are modified, or the equivalents of the technical features are replaced by the equivalents. The modifications and substitutions of the embodiments do not depart from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims (15)

1. 一种确定巡检基站列表的方法，其特征在于，包括：A method for determining a list of patrol base stations, comprising:

   获取第一基站集合的当前运维数据；Obtaining current operation and maintenance data of the first base station set;

   根据所述当前运维数据进行基站特征构建，得到所述第一基站集合的当前特征数据；Performing base station feature construction according to the current operation and maintenance data to obtain current feature data of the first base station set;

   根据所述当前特征数据和所述预先构建的不同类型的多个巡检模型，得到所述第一基站集合中每个基站对应的各个巡检模型的得分；And obtaining, according to the current feature data and the plurality of pre-built different types of patrol models, scores of respective patrol models corresponding to each base station in the first base station set;

   根据所述每个基站对应的各个巡检模型的得分，计算得到每个基站的总得分；Calculating a total score of each base station according to the scores of the respective patrol models corresponding to each of the base stations;

   根据所述每个基站的总得分，计算得到所述巡检基站列表。The list of the patrol base stations is calculated according to the total score of each base station.
2. 根据权利要求1所述的方法，其特征在于，所述获取基站集合的当前运维数据之前，所述方法还包括：The method according to claim 1, wherein before the acquiring the current operation and maintenance data of the base station set, the method further includes:

   获取第二基站集合的历史运维数据；Obtaining historical operation and maintenance data of the second base station set;

   根据所述历史运维数据得到所述第二基站集合的历史特征数据，以及所述第二基站集合中每个基站的运行状态信息；Obtaining, according to the historical operation and maintenance data, historical feature data of the second base station set, and running state information of each base station in the second base station set;

   根据所述历史特征数据以及所述第二基站集合中每个基站的运行状态信息，训练得到所述不同类型的多个巡检模型。And obtaining, according to the historical feature data and the running state information of each base station in the second base station set, the plurality of patrol models of the different types are trained.
3. 根据权利要求1或2所述的方法，其特征在于，所述根据所述当前运维数据进行基站特征构建，得到所述第一基站集合的当前特征数据，包括：The method according to claim 1 or 2, wherein the performing the base station feature construction according to the current operation and maintenance data to obtain the current feature data of the first base station set includes:

   使用预置的抽取方式从所述当前运维数据中，抽取预置的特征数据类型对应的特征数据，得到所述第一基站集合的当前特征数据。The feature data corresponding to the preset feature data type is extracted from the current operation and maintenance data by using a preset extraction manner to obtain current feature data of the first base station set.
4. 根据权利要求1-3任一项所述的方法，其特征在于，所述当前特征数据包括基站类型、基站区域、主设备服务年限、外设备服务年限。The method according to any one of claims 1-3, wherein the current feature data includes a base station type, a base station area, a primary device service age, and an outer device service age.
5. 根据权利要求2所述的方法，其特征在于，所述第二基站集合中每个基站的运行状态信息包括：所述第二基站集合中每个基站在第一预置时长内的告警数量或者断电时长。The method according to claim 2, wherein the operating state information of each of the second base station sets comprises: the number of alarms of each base station in the second set of base stations within a first preset duration or Power off time.
6. 根据权利要求1-5任一项所述的方法，其特征在于，所述方法还包括：The method according to any one of claims 1 to 5, wherein the method further comprises:

   若所述第一基站集合中的各个基站的总得分中存在总得分相同的基站，则确定所述总得分相同的基站中在第二预置时长内未巡检的基站为待巡检基站列表。If there is a base station with the same total score among the total scores of the base stations in the first base station set, determine that the base stations that are not patrolled within the second preset time period of the base stations with the same total score are the list of base stations to be patrolled. .
7. 一种巡检装置，其特征在于，包括：A patrol device, comprising:

   获取模块，用于获取第一基站集合的当前运维数据；An obtaining module, configured to acquire current operation and maintenance data of the first base station set;

   处理模块，用于根据所述当前运维数据进行基站特征构建，得到所述第一基站集合的当前特征数据；根据所述当前特征数据和所述预先构建的不同类型的多个巡检模型，得到所述第一基站集合中每个基站对应的各个巡检模型的得分；根据所述每个基站对应的各个巡检模型的得分，计算得到每个基站的总得分；根据所述每个基站的总得分，计算得到所述巡检基站列表。a processing module, configured to perform base station feature construction according to the current operation and maintenance data, to obtain current feature data of the first base station set; and according to the current feature data and the pre-built multiple types of multiple inspection models Obtaining a score of each patrol model corresponding to each base station in the first base station set; calculating a total score of each base station according to a score of each patrol model corresponding to each base station; The total score is calculated and the list of the inspection base stations is calculated.
8. 根据权利要求7所述的巡检装置，其特征在于，A patrol device according to claim 7, wherein

   所述获取模块，还用于获取第二基站集合的历史运维数据；The acquiring module is further configured to acquire historical operation and maintenance data of the second base station set;

   所述处理模块，还用于根据所述历史运维数据得到所述第二基站集合的历史特征数据，以及所述第二基站集合中每个基站的运行状态信息；根据所述历史特征数据以及所述第二 基站集合中每个基站的运行状态信息，训练得到所述不同类型的多个巡检模型。The processing module is further configured to obtain, according to the historical operation and maintenance data, historical feature data of the second base station set, and running state information of each base station in the second base station set; according to the historical feature data and The running state information of each base station in the second base station set is trained to obtain multiple inspection models of the different types.
9. 根据权利要求7或8所述的巡检装置，其特征在于，A patrol device according to claim 7 or 8, wherein

   所述处理模块，具体用于使用预置的抽取方式从所述当前运维数据中，抽取预置的特征数据类型对应的特征数据，得到所述第一基站集合的当前特征数据。The processing module is configured to extract feature data corresponding to the preset feature data type from the current operation and maintenance data by using a preset extraction manner to obtain current feature data of the first base station set.
10. 根据权利要求7-9任一项所述的巡检装置，其特征在于，所述当前特征数据包括基站类型、基站区域、主设备服务年限、外设备服务年限。The patrol device according to any one of claims 7-9, wherein the current feature data comprises a base station type, a base station area, a service life of the main device, and an service life of the external device.
11. 根据权利要求8所述的巡检装置，其特征在于，所述第二基站集合中每个基站的运行状态信息包括：所述第二基站集合中每个基站在第一预置时长内的告警数量或者断电时长。The patrol device according to claim 8, wherein the operation status information of each base station in the second base station set includes: an alarm that each base station in the second base station set is within a first preset time period The number or length of power outage.
12. 根据权利要求7-11任一项所述的巡检装置，其特征在于，A patrol device according to any one of claims 7 to 11, wherein

    所述处理模块，还用于若所述第一基站集合中的各个基站的总得分中存在总得分相同的基站，则确定所述总得分相同的基站中在第二预置时长内未巡检的基站为待巡检基站列表。The processing module is further configured to: if there is a base station with the same total score among the total scores of the base stations in the first base station set, determine that the base stations with the same total score are not inspected within the second preset time period. The base station is a list of base stations to be patrolled.
13. 一种巡检装置，其特征在于，包括：A patrol device, comprising:

    处理器和存储器；Processor and memory;

    其中，所述存储器用于存储程序，当所述程序被所述处理器调用时，用于执行如权利要求1-6任意一项所述的方法。The memory is used to store a program for performing the method of any of claims 1-6 when the program is invoked by the processor.
14. 一种计算机可读存储介质，包括指令，当其在计算机上运行时，使得计算机执行如权利要求1-6任意一项所述的方法。A computer readable storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of any of claims 1-6.
15. 一种包含指令的计算机程序产品，当其在计算机上运行时，使得计算机执行如权利要求1-6任意一项所述的方法。A computer program product comprising instructions which, when run on a computer, cause the computer to perform the method of any of claims 1-6.

Images