基于改进小波神经网络的旋转机械故障诊断

Fault Diagnosis of Rotating Machinery Based on Improved Wavelet Neural Network

根据旋转机械复杂的故障特点,提出了结合第2代小波分析、模糊理论和神经网络形成的改进小波神经网络。该诊断方法利用第2代小波分析对信号进行预处理-特征向量的提取,然后用训练好的模糊神经网络进行故障识别,并对相应算法进行了优化。将改进小波神经网络运用于旋转机械的故障诊断,通过计算机仿真和试验的结果表明,该方法在处理多故障耦合的情况时优势明显,不仅可以正确识别故障的类型,还可以进一步对故障的严重程度及其发展趋势进行可靠的诊断。这表明改进的诊断方法是行之有效的,为旋转机械的故障诊断提供了理论支持。

According to the complicated characteristic of rotating machinery faults,a fault diagnosis method based on an improved wavelet neural network is put forward,which combines a second generation wavelet and a fuzzy neural network. The method uses the wavelet to extract eigenvectors of signals and the fuzzy neural network to realize fault diagnosis,the corresponding algorithm is optimized.The improved wavelet neural network is applied to fault diagnosis of rotating machinery,the result of simulation and test indicates that the method has obvious advantage for dealing with multi-coupled faults.Using the method not only can identify the fault type,but also can get the degree and development trend of faults.So the improved diagnosis method is effective and provids a theoretical basis for the fault diagnosis of rotating machinery.