基于WOA-LSSVM算法的风机齿轮箱故障诊断

Fault diagnosis of wind turbine's gearbox based on WOA-LSSVM

针对风机齿轮箱振动信号非线性、非平稳性的特点,提出了一种基于改进的集合经验模态分解(MEEMD)、样本熵、鲸鱼优化算法优化的最小二乘支持向量机(WOA-LSSVM)算法的风机齿轮箱故障诊断方法。采用MEEMD对采集的齿轮箱振动信号进行分解,选择合适的本征模态函数(IMF)分量,并计算IMF分量的样本熵构造特征向量;使用WOA-LSSVM对其进行故障识别。实验数据对比表明,基于MEEMD样本熵和WOA-LSSVM算法的风机齿轮箱故障诊断方法可以更有效地实现风电机组齿轮箱的故障诊断。

Aimed at the characteristics of the non-linear and non-stationary vibration signals of wind turbine's gearbox,a fault diagnosis method of wind turbine's gearbox is proposed based on the modified ensemble empirical mode decomposition(MEEMD)method,the sample entropy and the least squares support vector machine optimized by the whale optimization algorithm(WOA-LSSVM).The MEEMD is used to decompose the vibration signals of the gearbox.Then the appropriate intrinsic mode function(IMF)components are selected,and the sample entropy of the IMF components is calculated to construct the feature vector.Finally,the WOA-LSSVM is used to identify the fault.The comparison of experimental data shows that the fault diagnosis method based on the MEEMD sample entropy and the WOA-LSSVM can realize the fault diagnosis of wind turbine's gearbox more effectively.