Trandformer Fault Diagnosis Based On The Neural Network Of Improved Sparrow Search Algorithm

As a key equipment in the power system,once the power transformer fails,it will cause paralysis of the power system and have a serious impact on the social economy and people ’s lives.In order to avoid transformer faults,scholars have proposed a variety of methods to diagnose transformers.However,the traditional transformer fault diagnosis method has certain limitations,and can not judge the fault type of the transformer well.In this thesis,an improved sparrow search algorithm is proposed and combined with BP neural network for transformer fault diagnosis.Chaos mapping,dynamic adaptive weight,sine and cosine algorithm,differential mutation and reverse learning strategy are introduced to improve and optimize,which improves the accuracy and reliability of transformer fault diagnosis.The main research contents of this thesis are as follows:1.The BP neural network model is studied in detail,and the principle of BP neural network is introduced.The forward propagation and back propagation of BP algorithm are deduced by detailed mathematical formulas.The BP neural network is integrated into the transformer fault diagnosis,and the fault results are obtained by training the transformer fault data.The advantages and disadvantages of BP neural network are analyzed,which is prepared for the optimization and improvement of BP neural network.2.The BP neural network model is studied in detail,and the principle of BP neural network is introduced.The two propagation modes of forward propagation and back propagation are deduced by detailed mathematical formulas.The BP neural network is integrated into the transformer fault diagnosis,and the fault results are obtained by training the transformer fault data.The advantages and disadvantages of BP neural network are analyzed,which is prepared for the optimization and improvement of BP neural network.3.The combination of sparrow search algorithm and BP neural network is applied to the field of transformer fault diagnosis.The position vector of the sparrow individual is mapped to the weight and threshold of the BP neural network.The best position vector of the sparrow individual is found by simulating the sparrow ’s behavior of foraging and avoiding natural enemies,and assigned to the weight and threshold of the BP neural network,which overcomes the shortcomings of slow convergence speed and local extremum of the BP neural network.4.The sparrow search algorithm has the disadvantages of slow convergence speed and easy to fall into local extremum.A hybrid strategy is proposed to improve the sparrow search algorithm.The chaotic map is used to initialize the sparrow population to increase the diversity of the population and solve the problem of uneven population distribution.The sine-cosine strategy and dynamic adaptive weight factor are used to improve the update method of the discoverer position,so that it can better jump out of the local optimal solution and increase the global search ability of the algorithm.Differential mutation perturbation and reverse learning strategy are used to perform optimal solution perturbation mutation to improve the ability of sparrow search algorithm to fall into local areas.The effectiveness of the improved sparrow algorithm is verified by standard test functions,and the transformer fault simulation is carried out in combination with BP neural network.The results show that the improved algorithm has higher diagnostic accuracy.