The Research On Backward Reasoning Method Of Power Grid Fault Diagnosis

Power grid fault diagnosis is an important part of intelligent schedule system, after the power grid failure, it provides decision support for operation. With the development of dispatching automation technology, available data sources of power grid fault diagnosis system include traditional SCADA system, fault information system, WAMS etc. How to use the information resources effectively is the research direction of the development of power grid fault diagnosis. This paper researches power grid fault diagnosis method based on multi-data sources by backward reasoning methods.Firstly, with a suspicious fault equipment, this paper studied the method collecting fault information in all date sources initiatively and effectively, and reached the function finding all the information related with fault in failure process in each data sources by local piece of information. Using all logical judgement method improved, it reasons backward the features generated by this equipment, so as to determine targets of information collecting in each data sources. This method which is not affected by operation mode has the real-time characteristics.Secondly, the paper studied and developed the automatic modeling function of power grid fault diagnosis database based on the EMS data. It reduces the quantity of data maintenance greatly, and meets practical needs.Thirdly, with a suspicious fault equipment, this paper researched and developed the function of confirmating fault equipment and reproducing occurrence process of fault information in each data sources. By actual action order of switches, this paper reasoned backward the protections and switches that should act and the fault current duration on the switches, then analysed with fault information (evidence), especially wave record data, finally confirmed fault equipment. Fault information (switch and protection movement) caused by the fault equipment in various data sources were inserted time series by the logic relationship. It can form evidence chain of protection and switch acting in different stations, and reproduce fault process. The diagnosis method could position fault equipment accurately, and the form of diagnosis result organizes fault information by the causal logical in the entire fault process in all data sources, so it provides a better support for scheduling staff knowing, reading and processing fault.Finally, with programming by VC, this paper realized the functions collecting information actively and reproducing fault process, and verified the correctness of the software functions by testing actual grid fault cases.