Fault Diagnosis Of Power System Based On Adaptive Fuzzy Petri Nets And Colored Petri Nets

Modern power system networks have been expanding with UHV transmission and large generators, so inter-regional power grid interconnection more closely. When a fault occurs, the fault elements must be fast identified, in order to restoration of electricity and ensure the stable operation of power system to avoid greater losses. As more and more power monitoring device, a fault of power system will cause the influx of a large number of alarm information to dispatch center, and it is a huge workload for fault diagnosis by artificial. Therefore need to develop an effective method can be based on various sources of information quickly and accurately to power system fault diagnosis.Petri nets have a good capacity of parallel processing and graphical representation, ideal for dealing with discrete problems. Adaptive fuzzy Petri nets (AFPN) and colored Petri nets (CPN) are based on Petri net evolved. AFPN has the advantages both from Petri nets and fuzzy logic theory, which can form of probability to represent uncertainty and incomplete problem and through weight training closer to the real system. CPN has the ability to simplify the Petri nets model is a high-level Petri nets. This study is based on AFPN and CPN's power grid fault diagnosis method to help dispatchers quickly determine fault, improve the diagnostic efficiency, and ensure the safe operation of power supply stability and power.In this paper, establish a suitable layered fault diagnosis system base on adaptive fuzzy Petri nets, for solve the problems that the faulted data from SCADA system to dispatch center may uncertainty or incomplete. The system is divided into three layers. The first layer is the fault area search for use of circuit breaker status information; include quick to judge simple fault and delineation of the diagnostic area. The second layer establish the AFPN fault diagnosis model which use protection and circuit breakers data, the value of the output weight is calculate by using statistical data so that the system can enhance the explanatory; the value of the input weight using a large number of fault samples have been training to make the model more closely simulated the real system. The third layer establish the expansion AFPN model which can use PMU data, to address the situation of results of the second layer can not diagnose or diagnostic results are not satisfactory. Simulation results show that the accuracy of the algorithm. By comparison with the existing FPN model, show that the model has effectively improved the confidence factor of the diagnosis and the diagnostic level of fault tolerance. In addition, there establish a fault diagnosis model based on colored Petri net to solve universal framework for Petri net model is not strong, and Petri net model complex problems. Then define applicable to the model rules of colored Petri nets, and discussed when the network topology changes, how many initial set data of the model may changes. Simulation results show that the CPN model structure is simple and versatile, easy to transplant; and able to adapt network topology changes.