摘要：With the development and utilization of clean energy from wind power and the increasing number of wind turbines, its safety diagnosis technology is becoming more and more important. The traditional diagnostic methods are insufficient in accuracy and convergence speed. In this paper, we analyzed in de-tail the shortcomings of wavelet neural network technology in wind turbine and circuit safety diag-nosis. On this basis, an improved particle swarm op-timization algorithm is proposed to further optimize the parameters and structure of the wavelet neural network. The four methods of statistical analysis,BP neural network, wavelet neural network and particle swarm wavelet neural network are used for safety di-agnosis and analysis of simulated circuit test signals. The results show that the convergence accuracy of the particle swarm wavelet neural network method is 99%, which is higher than the other three methods, so its convergence speed is significantly improved. In addition, three algorithms of BP neural network, wavelet neural network and particle swarm wavelet neural network are used to diagnose the safety of wind turbines with clean energy. The results show that under the same accuracy, the particle swarm wavelet neural network method has the least number of iterations and the fastest convergence speed. This further proves that the particle swarm wavelet neural network method studied in this paper is more accu-rate than other methods for wind turbine safety diag-nosis.
