基于粒子群优化LS-WSVM的旋转机械故障诊断

Rotating machinery fault diagnosis based on LS-WSVM with particle swarm optimization

为了更好地进行旋转机械故障诊断,提出一种粒子群优化(particle swarm optimization,PSO)最小二乘小波支持向量机(least square wavelet support vector machine,LS-WSVM)的故障诊断模型。先将故障信号经验模式分解(empirical mode decomposition,EMD)为多个内禀模态分量(intrinsic mode function,IMF)之和,再提取表征故障特征的IMF分量能量构造特征向量输入到PSO优化的LS-WSVM进行故障模式识别。EMD分解可自适应提取故障特征信号,PSO参数优化可快速准确得到LS-WSVM的全局最优参数,提高LS-WSVM的故障诊断精度和自适应诊断能力。通过滚动轴承的故障模拟实验验证了该方法的有效性。

In order to identify the fault of rotating machinery better, a model of least square wavelet support vector machine （LS-WSVM） optimized by particle swarm optimization （PSO） algorithm is proposed. Fault vibration signals are decomposed into several stationary intrinsic mode functions （ IMFs）, then the instantaneous amplitudes of the IMFs that have the fault characteristics are computed and regarded as the input characteristic vector of the LS-WSVM optimized by PSO algorithm for fault classification. EMD decomposition can extract fault vibration signals from original signals adaptively. PSO algorithm has more superior performance on global optimization and convergence speed. The fault diagnosis experiments of rolling-bearings show the effectiveness of this novel model. Compared with other methods, the experiment results show that the proposed approach converges faster and produces better hyper-parameters.