基于经验模态分解和散度指标的风力发电机滚动轴承故障诊断方法

Fault diagnosis of bearing in wind turbine based on empirical mode decomposition and divergence index

提出了一种基于经验模态分解(Empirical Mode Decomposition,EMD)和散度指标的滚动轴承故障诊断方法。该方法首先对原始振动信号进行经验模态分解,再以峭度为准则,选取包含故障信息的特征模态函数(Intrinsic Mode Function,IMF)进行信号重构,并对重构后的信号进行Hilbert包络谱分析,然后提取故障特征量,最后通过计算故障特征量的J-散度(J-divergence)和KL-散度(Kullback-Leibler divergence)来判断故障类型和描述故障程度。通过从试验台采集的无故障和具有内环故障、外环故障和滚动体故障样本,以及从某风电场风力发电机齿轮箱高速输出端采集的近一年的监测数据分析结果,证明了所选故障特征量的准确性,同时也验证了所提出的基于经验模态分解和散度指标的滚动轴承故障诊断方法的有效性和准确性。

This paper presents a novel fault diagnosis of bearings based on Empirical Mode Decomposition（EMD） and divergence index. The reconstructed signal can be obtained by some set of IMF components based on the rule of kurtosis after the initial vibration signal is processed by EMD. The Hilbert envelope analysis of the reconstructed signal is conducted. From the power spectrum of Hilbert envelope signal, we can identify the amplitude of the characteristic fault frequency and its integer multiples, which are used to compute the J-divergence and KL-divergence to identify the style and severity of fault. The result of non-fault sample, inner-ring fault sample, outer-ring fault sample and rolling bearing fault sample collected from experiment test and the condition monitoring analysis result acquired from high speed output terminal of gear case of wind turbines demonstrate that the characteristic fault frequency and its multiple can reflect fault information precisely and the proposed solution is effective for characterizing and detecting a range of rolling bearing faults.