基于1DDCNN和PCA信息融合的滚动轴承FLHI智能提取方法

An intelligent extraction method of the full life health indicator of rolling bearings based on one-dimensional deep convolutional neural network and principal component analysis

滚动轴承故障预测方法的核心在于健康指数(HI)的构建,绝大部分已经提出的HI都是基于专家经验人工构造的,且往往只能适用于部件某一特定退化阶段的趋势分析。为解决上述问题,结合振动信号的一维特性,提出一种基于一维深度卷积神经网络(1DDCNN)结合主成分分析(PCA)的滚动轴承全寿命健康指数(FLHI)智能提取法;利用1DDCNN对原始时域信号自适应提取特征,深度挖掘能够表征研究对象健康状态的退化特征矩阵,而后利用PCA法对提取的特征矩阵进行融合,从而实现研究对象的FLHI智能提取。滚动轴承试验振动信号实测结果表明,相较于传统健康指数,FLHI在趋势性、鲁棒性和单调性方面更具有优势。

The core of rolling bearing fault prognosis methods lies in the construction of a health indicator(HI).Most of the proposed HI is constructed artificially based on expert experience and it can only be applied to the trend analysis of a specific degradation stage of components.To solve the above problems,combined with the one-dimensional characteristics of vibration signals,a full life health indicator(FLHI)of rolling bearing intelligent extraction method based on one-dimensional deep convolutional neural network(1DDCNN)and principal component analysis(PCA)was proposed.1DDCNN was used to extract features adaptively from the original signals,and it can deeply mine the degradation feature matrix that can represent the health state of the research object.And then,the extracted feature matrix was fused by the PCA method,so as to realize the FLHI intelligent extraction.The experimental results show that FLHI is more advantageous in terms of tendency,robustness,and monotonicity than the traditional HI.