Research On Decision Support System Of Railway Signaling Equipment Maintenance Based On Case-Based Reasoning

With the rapid development of China railway construction, the task of railway signaling equipment maintenance is getting more difficult. The harsh environment along the railway has a adverse impact on the maintenance of signaling equipment. The presence of traditional maintenance decision method is difficult to obtain domain knowledge. In this thesis, the design and implement of the railway signaling equipment Maintenance Decision Support System has been completed by researching the Case-Based Reasoning(CBR) railway signaling equipment maintenance decision support technology. Finally, the system can make use of a similar fault case to provide equipment maintenance decision-making service. The decision support system of railway signaling equipment maintenance plays a very significant role in ensuring the reliable and safe operation of railway systems.The CBR railway signaling equipment maintenance decision support system has been researched, in this thesis. After elaborating the purpose and significance of the research, current situation and development trend of CBR maintenance decision support system is analyzed. The method of case classification and case retrieval is an emphasis. The CBR signaling equipment maintenance decision support system has been designed and built. The main contents include the following four aspects.(1) According to the characteristics and needs of railway signaling equipment maintenance, the railway signaling equipment maintenance decision model based on CBR is designed. Researching the fault case representation method, the railway signaling equipment fault case base is constructed.(2) Due to the fault description text has the characteristics of feature sparse and context dependent, a classification method based on Chi-square feature and BTM(Biterm Topic Model) has been proposed. The method combines mathematical statistics and semantic information of fault description text, which reduces the impact of the characteristic sparse text description and the context-dependent effectively, and improves the accuracy of classification and the efficiency of fault case retrieval.(3) After the fault case is classified, the most similar cases is retrieved by the text description of the current device problem. There are three points should be taken into consideration. They are the statistics information, semantic information and topic information of the fault description text. The case retrieve method based on multi-feature fusion matching text similarity has been proposed. The similarity is calculated by VSM-based similarity algorithm, HowNet similarity algorithm and BTM algorithm. Finally, it is combined with the similarity of text description to obtain a comprehensive text similarity. The experimental results demonstrate the effectiveness of the method.(4) Based on the above key technologies, the railway signaling equipment maintenance decision support system is designed and implemented. It has a user-friendly interface, convenient function, higher practical value.