VPMCD和改进ITD的联合智能诊断方法研究

Research on Combined intelligent diegnostic method based on VPMCD and improved ITD

提出了一种基于变量预测模型(Variable predictive model based class discriminate,简称VPMCD)和改进固有时间尺度分解(Intrinsic time-scale decomposition,简称ITD)算法的滚动轴承故障诊断方法。VPMCD方法充分利用了特征值之间的相互内在关系来建立预测模型,并以这些模型对待诊断样本的特征值的预测结果作为分类依据来进行模式识别。ITD方法能自适应地将非平稳信号分解成为若干单分量信号(固有旋转分量,Proper rotation component,简称PRC)之和。首先对ITD算法进行了改进;接着采用改进ITD算法对原始振动信号进行分解得到多个内禀尺度分量(Intrinsic scale component,简称ISC);然后对包含主要故障信息的若干内禀尺度分量建立对数正态分布模型,并提取其对数均值和对数标准差作为故障特征值;最后采用VPMCD模式识别方法得到各故障特征值的预测模型,并利用预测模型对待诊断样本的故障类型和工作状态进行分类和识别。对滚动轴承正常、外圈故障和内圈故障振动信号的分析结果表明了该方法的有效性。

A fault diagnosis approach of roller bearing based on variable predictive model based class discriminate （VPMCD） and improved intrinsic time-scale decomposition （ITD） algorithm is proposed.The method of VPMCD makes full use of the internal variable association between the feature parameters of training samples to establish variable predictive models,by which the variable predictive models can be used to identify the feature parameters of test samples.The ITD algorithm can decompose a non-stationary signal into the sum of many single component signals （Proper rotation component,PRC）.In this paper,the ITD algorithm was improved first.Then original vibration signal was decomposed into intrinsic scale components （ISC） by improved ITD algorithm.And then logarithmic normal distribution models were established by some ISCs which include the most dominant fault information and whose logarithmic mean and logarithmic standard deviation were extracted to serve as feature parameters.Finally,variable predictive models of different feature parameters were established by VPMCD,which were used to classify and recognize different fault types and working states of samples to be diagnosed.The analysis results from the roller bearing vibration signals of normal,outer fault and inner fault demonstrate the effectiveness of the proposed method.