Research On Fault Diagnosis Of Distribution Network With Improved Directional Weighted Fuzzy Petri Net Based On GA-BP Optimization

In the fault diagnosis of the distribution network,because the authenticity of the fault alarm information received is difficult to distinguish,there are not only the protection and circuit breaker maloperation,refusing to operate,but also the protection and circuit breaker action information there are missing alarms,false alarms.Moreover,the parameters of the common power system fault diagnosis model based on Petri net are often set according to expert experience.Although there are some optimization methods used to optimize the model parameters,the existing optimization methods may make the optimization results fall into the local optimal solution for the complex structure of the distribution network,thus leading to low efficiency and accuracy of the fault diagnosis of the distribution network.In view of the above situation in the distribution network,this paper considers the GA-BP optimization of the improved direction-weighted fuzzy Petri net fault diagnosis model for the topological structure of the variable distribution network fault diagnosis,this model can not only effectively adapt to the change of the topology of the distribution network,but also when the protection and circuit breakers refuse or maloperation,still has a high reliability of fault diagnosis.Firstly,this paper introduces directional-weighted fuzzy Petri net from the basic definition and reasoning process of directional-weighted fuzzy Petri net,and considers the improvement of directional-weighted fuzzy Petri net from the model structure and the setting of the initial value of the initial library.Then make use of the main protection,near backup protection,far backup protection of each suspicious fault element and the corresponding circuit breaker action information to conduct timing information inference,screen out the alarm information that does not conform to the time constraints,and determine the weight parameters of the model,and use the Gaussian function to correct the output function of the model.At the same time,Simulink module in MATLAB is used to build the simulation model.Two examples are used to verify the correctness of the distribution network fault diagnosis based on Petri net in this paper.Meanwhile,the improved model is compared with the original model through four cases.The results show that the reliability of the obtained fault diagnosis results is as high as 0.8279 when the main protection action information is lost,which has a high fault diagnosis accuracy.Secondly,in the process of distribution network fault diagnosis based on Petri Net,there exists the problem of unreasonable input arc weights of input warehouse and corresponding changes,and the traditional intelligent optimization algorithms based on BP neural network,differential evolution algorithm,particle swarm optimization and other intelligent optimization algorithms in the optimization process of input arc weights,which make the optimization results easily fall into the local optimal solution.Furthermore,the certainty of the final fault diagnosis result is affected.Therefore,GA-BP optimization algorithm is considered to optimize the input arc weight of the model.The simulation results show that after about 58 generations of training,the parameters to be optimized gradually converge,the convergence accuracy reaches 1.17×10-5,and the optimization algorithm has a good global optimization ability.For complex faults such as protection or circuit breaker rejection in the distribution network,7 cases were simulated to verify the accuracy of the proposed GA-BP optimization based on the improved directional-weighted fuzzy Petri net fault diagnosis model and its adaptability to the topology change of the distribution network.Finally,a 110 k V high-voltage distribution network is used for simulation verification.The proposed distribution network fault diagnosis method based on GA-BP optimization can still identify fault components when the circuit breaker corresponding to the main protection is rejected,and the reliability of the fault diagnosis results is as high as 0.7375.It has high fault tolerance and adaptability.