基于LCD降噪和VPMCD的滚动轴承故障诊断方法

提出了一种基于局部特征尺度分解(LCD)降噪和多变量预测模型(VPMCD)的滚动轴承故障诊断方法。该方法首先采用LCD对滚动轴承振动信号进行降噪;然后计算降噪后信号在不同维数下的模糊熵,并以模糊熵为特征值,采用VPMCD方法建立模糊熵的预测模型;最后用所建立的模型来预测待分类样本的特征值,把预测结果作为分类依据进行模式识别。实验分析结果表明，采用LCD方法降噪可以有效地提高VPMCD的分类性能,与神经网络、支持向量机等分类器相比，VPMCD方法可以更准确、更有效地识别滚动轴承的工作状态和故障类型。

A Rolling Bearing Fault Diagnosis Method Based on LCD De-noising and VPMCD

A fault diagnosis of rolling bearing based on LCD de-noising and VPMCD was proposed. Firstly, using the LCD on the rolling bearing vibration signals to reduce noise signals, then the fuzzy entropy of the de-noising signals in the different dimensions was calculated and as characteristic values. Using the VPMCD method to establish the fuzzy entropy prediction model, and finally the characteristic values of those unclassified signals samples were predicted by the model. The results of the prediction would be recognized by the model as accordance to classify. The experimental results prove that the LCD de-noising can effectively increase the VPMCD classification performance, compared with neural network and support vector machine classifier, the VPMCD methods can identify the work states and fault patterns of the rolling bearing more accurately and more effectively.