Research On Fault Diagnosis Of Active Distribution Network Based On BP And Time Constrained Hierarchical Fuzzy Petri Net

With the increasing complexity of distribution network structure,the performance of distribution network fault diagnosis methods plays a vital role for the safe and stable operation of a power system.When the distribution network breaks down,especially when complex failures happen,alarm information will overwhelm the dispatching center.The hierarchical fuzzy Petri net can better describe the relationship between the fault and the action of the protection and the circuit breaker and has simple reasoning and visual expression.With the scaling of the distribution network.The distributed power connected to the distribution network keeps increasing.Therefore it is urgent to develop the fault diagnosis scheme for the distribution network with distributed power.This paper reviews the concept of distribution network fault diagnosis and the current research status.Then a fault diagnosis model is constructed for an active distribution network by Petri net.The hierarchical fuzzy Petri nets have an excellent adaptivity of the distribution network topology,and the Gaussian function is optimized.The model is combined into the Petri net fault diagnosis model according to the time sequence relationship between the protection and the circuit breaker.The action time of the corresponding circuit breaker of the primary protection.The primary protection is used to obtain the cause event action starting point.The initial probability of the library is modified by the time sequence reasoning relationship.The reliability of hierarchical fuzzy Petri net fault diagnosis is demonstrated by numerical simulation.The unreasonable confidence in the ultimate database for the process of Petri net fault diagnosis is corrected by optimizing the input and output weights of Petri net with the neural network,whose structure is similar to the back propagation(BP)neural network.Such correction results in better accuracy of fault diagnosis.To get better adaptability for multiple signal loss,this paper rebuilds the inference process of fault diagnosis,and optimizes the time sequence inference process according to the situation of numerous signal loss in fault information,and sets different diagnosis probability.By comparing and verifying several examples,our proposal demonstrates solid reliability and accuracy.