LMS滤波法在笼型异步电机转子断条诊断中的特征

The features of LMS adaptive filter in rotor bar broken diagnosis of squirrel cage asynchronous machines

在诊断笼型异步电机转子断条故障时,可以通过对定子电流的频谱分析来提取代表故障特征的边频分量。为克服实际诊断中直接做快速傅里叶变换(fast fourier transformation,FFT)频谱分析存在的边频分量被主频量掩盖的缺陷,采用最小均方误差(least mean square,LMS)自适应滤波法滤除主频量再做FFT频谱分析,以突出边频分量。给出了转子断条故障诊断中LMS滤波原理及滤波器参数确定,详细分析了电机不同运行转差率下LMS滤波对边频量的采样规律和等效采样周期。结果表明,当依主频设定的固定采样周期采样电机电流时,LMS滤波是通过对边频分量在不同周波的不同位置采样拟合出完整周波,等效采样周期随转差率变化,且一般至少比设定的固定采样周期小一个数量级。

For the diagnosis of rotor broken bar fault in squirrel cage asynchronous machines, spectrum a- nalysis of stator current is often used to extract the side frequency components that represent the typical fault characteristics. To remedy the drawbacks existed in doing FFT spectrum analysis for the curent di- rectly,which could make the side frequency components be overcovered by main frequency signal, the Least Mean Square （LMS） error adaptive filter algorithm was used prior to F~＇I＂ analysis to filter main fre- quency signal and highlight the side frequency components. The LMS adaptive filter algorithm and the pa- rameters determination in rotor broken bar fault diagnosis were given, and the sample features and the ef- fective sample period for side frequency component at different operation slips were analyzed in detaile. The results show that, with LMS filter algorithm, when the stator current was sampled at a fixed sample pe- riod chosen according to main frequency, the side frequency component is actually sampled at the different of its different individual waveform, and a full waveform is comopsed accrodingly. The effective sam- pie period for side frequency component varies with slip and is smaller than the fixed one by an order of magnitude at least.