Distribution Network Fault Diagnosis System

With the economic and social development and people’s living standards improve, power users put forward higher requirements for security and reliability of the power system. However, due to the explosive growth of the electricity load, expanding the grid-scale grid topologies are making more and more complex, the fault diagnosis become more difficult. Distribution network is the last sector of power system Generation-Transmission-Distribution sectors, directly related to the vital interests of users. In the natural, human factors (such as the construction site is not standardized and natural disasters) and other issues interfering with the grid security issues have become increasingly prominent, that result in frequent accidents in distribution network. After the fault, the rapid confirmation of the source and cause of the malfunction failure is a prerequisite to restore sexual power for the power company, so the research of effective distribution network fault diagnosis system, auxiliary" operators to quickly and accurately identify the faulty position and repair faults, it is significantly for improving the distribution network reliability and security. Today large-scale investment in remote communication and remote sensing equipment, improving automation level of the grid, which also provided the foundation for the distribution network fault diagnosis system. The current difficulties in grid fault diagnosis method is to study how to overcome the interference uncertainties troubleshooting process (device and circuit breakers to protect the reliability, accuracy and completeness of the dispatch center received the alarm message), how to carry out the large-scales networks of power diagnosis online and how to deal with network topology changes brought diagnose problems and other issues.In this paper, the current mainstream method of fault diagnosis and abroad were summarized and analyzed. For the main problems exist in fault diagnosis method, this paper is proposed a method that combining elements using FOA-GRNN (Fruit fly optimization algorithm-general regression neural network) neural network the fault diagnosis model and fault diagnosis method causal network alarm timing based screening model, and so the development of the distribution network based fault diagnosis expert system. When the distribution network failure, the system automatically recognizes the first blackout area and establish a set of suspicious faulty components; alarm message followed by SCADA (Supervisory Control And Data Acquisition) and SOE (sequence of events) alert messages received through the screening system removed false alarms, and finally the protection devices and circuit breakers upload information to FOA-GRNN neural network fault diagnosis, failure to obtain credibility distribution network fault element sets, integrated fault diagnosis expert system and ultimately given by the solution.Cases verified by calculation, this method can effectively improve the efficiency of operating personnel troubleshooting quick and accurate in fault diagnosis of distribution network. Compared to other fault diagnosis method, the method can better deal with a variety of uncertainties that the existence in fault diagnosis process and distribution network topology changes, the complexity of the distribution network failures have better diagnostic capabilities.