基于MFCCS和改进VPMCD的滚动轴承故障诊断

将梅尔倒谱和系数(MFCCS)与改进的基于变量预测模型的模式识别算法(VPMCD)相结合,提出了一种滚动轴承故障的诊断方法.将语音信号识别中最常用的特征参数梅尔倒谱系数(MFCC)应用到轴承故障诊断领域,提出了适用于滚动轴承故障识别的特征参数梅尔倒谱和系数.同时,采用主成分分析(PCA)方法来解决VPMCD方法中求解得到的预测模型方程系数与理想系数存在偏差的问题.然后,使用改进的VPMCD算法对特征参数进行训练,再利用预测模型对待诊断样本数据进行模式识别和诊断,并用实验室模拟试验台的数据,对该方法进行了验证,实验结果能够有效区分轴承的故障种类,证明了方法的有效性.

Fault Diagnosis of Rolling Bearing Based on MFCCS and Improved VPMCD

A diagnosis method for rolling bearing based on MFCCS and improved VPMCD is proposed.The concept of MFCCS is put forward based on MFCC,which is commonly used in voice recognition as feature parameter.And it is applied to fault diagnosis of rolling bearing as feature parameter.At the same time,PCA is used to solve the problem that the coefficients of the predictive model are not accurate.Then the feature parameters are trained by improved VPMCD to get the VPMs.Through predicting the test data with the VPMs,the result of fault type is obtained.Finally,the method is validated by experiment using the data from lab''s rolling bearing test bench.The experimental results show that the method can accurately identify the fault type of rolling bearing and prove that the method can be used in fault diagnosis of rolling bearing.