摘要
滚动轴承振动信号是非平稳信号,而且往往是由许多分量的调频调幅信号组成;轴承故障特征频率容易被高频的调频调幅信号淹没;局域均值分解技术是近年来新的时频分析方法,能够有效的处理非平稳信号;它能将多分量的调制信号自适应的分解为若干个生产函数(PF)分量,每个PF分量都是一个单分量的调幅调频信号,包含了原始信号的局部特征;利用共振解调技术对每个PF分量进行包络分析,能够有效地提取高频载波中的故障特征频率;同时,为了实现智能诊断,提取各PF分量包络信号的能量值作为特征向量,训练BP神经网络,从而达到智能诊断目的。
Rolling bearing fault vibration signal is nonstationary signal, and consist of multi-component AM-FM signal. The fault characteristic frequency of rolling bearing is easily submerged by high frequency AM-FM signal. Local mean decomposition （LMD） is a new time-frequency analysis method in recent years, which can effectively processing nonstationary signal. By using LMD, multi-compo- nent modulated signals could be adaptively decomposed into a set of product functions （PF）, each PF component is single-component modu- lated signals that involves the local characteristic of the original signal. Apply Demodulated Resonance Technique to each PF component, the fault characteristic frequency can be efficiently extracted from high frequency carrier through envelopment analysis. Furthermore, in order to achieve intelligent diagnosis, extract the energy value of envelope signal of each PF as the feature vectors to train BP Neural Network, final- Iv achieve the ouroose of intelligent diagnosis.
出处
《计算机测量与控制》
北大核心
2013年第7期1762-1765,共4页
Computer Measurement &Control
基金
国家自然基金项目资助(51176028)
吉林省自然科学基金项目资助(201115179)
关键词
滚动轴承
局域均值分解
共振解调
BP神经网络
故障诊断
rolling bearing
local mean decomposition
resonance demodulated
BP neural network
fault diagnosis
