基于改进LMD方法的风电机组齿轮箱故障诊断研究

局部均值函数的求取是局部均值分解(LMD)的关键环节。针对局部均值函数求取存在偏差进而造成模态混叠的问题,提出了一种基于局部积分均值的LMD风电机组齿轮箱故障诊断方法。该方法改变了对相邻两极值点求平均值的思路,采用求取相邻两极值点的局部积分均值,再通过滑动平均法进行平滑处理,最终得到局部均值函数。为实现风电机组齿轮箱故障诊断,首先采用改进LMD方法对信号进行降噪处理,然后采用多尺度熵提取降噪处理后信号的特征向量,最后采用极限学习机进行故障诊断。通过仿真分析,证明了该方法能有效解决模态混叠现象,提高了LMD的分解精度。试验验证分析表明,该方法的故障诊断准确率为100%,通过对比分析表明,该方法优于其他故障诊断方法,具有工程应用价值。

Study on Fault Diagnosis of Wind Turbine Gearbox Based on Improved LMD Method

Finding the local mean function is the key link of the local mean decomposition(LMD).A fault diagnosis for the gearbox of the LMD wind turbine based on the local integral mean is proposed considering its deviation and mode mixing.This method applied local integral mean value of the adjacent two extreme points then smoothing by the moving average method to finally obtains the local mean function instead of averaging the two adjacent extreme points.In order to realize the fault diagnosis of wind turbine gearbox,the improved LMD method for de-noising the signal and multi-scale entropy for extracting the feature vector of the de-noised signal,and finally the limit learning machine is used.Simulation analysis proves that this method can effectively solve the mode mixing phenomenon and improve the decomposition accuracy of LMD.Through experimental verification,this method has a fault diagnosis accuracy rate of 100%.Through comparative analysis,it is of engineering applicatin value superior to other fault diagnosis methods.