基于SAPSO优化的小波神经网络在滚动轴承故障诊断中的应用

针对滚动轴承振动信号复杂及故障类型难以预知的问题,提出一种基于模拟退火粒子群算法(SAPSO)优化小波神经网络来诊断滚动轴承故障的新方法,并将其应用于滚动轴承故障诊断。实验表明,该方法能减少迭代次数、提高收敛精度。

Wavelet Neural Network Base on SAPSO Algorithm and Application in Fault Diagnosis of Rolling Bearing

According to the questions,the vibration signal is complex and the fault mode foreknew is difficult a novel method of unknown exception diagnosis in rolling bearing based on the wavelet neural networks(WNN)is proposed based on particle swarm optimiza-tion based on simulated annealing(SAPSO) is proposed to optimize parameters of WNN.As this new algorithm has some virtues such as high convergence speed and not easily trapping local mini-ma.Then,the algorithm is applied to neural network’s training in fault diagnosis of rolling bearing.The experimental result indicates that the algorithm can reduce the number of training and error.