小波和能量特征提取的旋转机械故障诊断方法

转子系统和轴承是旋转机械中的关键零部件,其长期处于高速、满负荷运行极易出现故障。基于振动信号处理的诊断方法具有可在线、实时诊断的特点,针对频谱分析对非线性振动信号故障特征提取的不足,研究小波包对振动信号进行特征提取。由于传统软、硬阈值量化方法在阈值处分别存在恒定偏差和不连续的问题,设计了一种参数可调的改进连续函数对阈值进行量化。系统首先对振动信号进行小波包分解与去噪,然后采用小波包能量特征提取方法完成对旋转机械的转子不平衡故障、不对中故障、转子动静碰摩故障进行有效诊断。测试结果表明,轴承出现不同故障时,通过小波包分解后不同子带能量的不同,可用模式识别方法有效进行故障识别。

Fault diagnosis of rotating machinery based on wavelet and energy feature extraction

Rotor and bearing are the key part of rotating machinery, its long term operation in high speed and full load appears fault easily. The diagnostic method based on vibration signal processing has the characteristics of diagnosing online and real time. Aiming at the shortage of extracting fault feature of the nonlinear vibration signal, the feature extraction of vibration signal by using wavelet packet is researched. Considering that the traditional soft and hard threshold quantization methods would have constant deviation and be discontinuous in the threshold, an improved continuous function whose parameters were adjustable to quantify the threshold was designed. First, the system makes wavelet packet decomposition and denoising to vibration signal. Then, it diagnoses the faults of rotating machinery effectively, including rotor unbalance fault, misalignment fault and rotor movement rubbing fault, by the way of extracting feature of wavelet packet energy. The test results show that when the bearing shows different failure, pattern recognition method could be used to identify the fault effectively based on the fact that energy vary from one sub band to another after the decomposition of wavelet packet.