Abstract:The disadvantages of the traditional troubleshooting method for secondary loop faults are: first, it takes a lot of manpower and time; second, it is difficult to fully consider various factors at the fault site; third, it is highly dependent on the technical skills of relay protection personnel. This paper proposes a method based on optimized neural network for big data expert system, and builds relay protection secondary loop fault location system (RSFLS). Based on the abstract classification of the relay protection secondary loop, the defect database model of the entity-relationship model (E-R model) is established, and the optimized neural network is used for randomization and self-learning. In addition, establish a library of relay protection personnel experience. An expert system architecture based on optimized neural networks is formed to locate fault points. The method effectively solves the problems of low efficiency and high randomness in the troubleshooting of the relay protection secondary circuit. The RSFLS system developed based on this method has been put into practical use, which can effectively shorten the troubleshooting time of the relay protection secondary circuit and improve the troubleshooting accuracy. The test results prove that the method and system have significant effects in improving the reliability of the relay protection system.