滚动轴承振动诊断的BP神经网络方法

归纳和总结了BP神经网络多参数诊断法的实施步骤,阐述了故障轴承振动与信号的关系以及BP神经网络的工作原理和实现过程,并以滚动轴承故障诊断为例,提取了反映滚动轴承故障类型的振动信号特征参数,用这些特征参数训练神经网络,利用MATLAB6.5神经网络工具箱模拟和仿真BP神经网络,然后用训练后的BP神经网络对故障模式进行识别。结果表明,如果神经网络设计合理,训练适当,并且在MATLAB仿真的过程中合理的选择训练函数和各种参数,则具有很强的故障识别能力。说明了利用MATLAB仿真BP神经网络进行滚动轴承振动诊断是可行而且有效的。

Vibrating Diagnosis of Rolling Bearings Based on Backpropagation Neural Network

This paper summarizes the steps of the multi-parameter diagnosis method based on backpropagation neural network and illuminate the connection between fault diagnosis of rolling bearings and signal and principle of BP neural network.On this basis,the characteristic which represent operation state of the rolling bearing are extract to construct characteristic vector,and a rolling bearings fault diagnosis experiment is designed to verify this method with ANN toolbox of MATLAB6.5 simulating BP neural network.The conclusion indicates that ANN possess good capability of identification with reasonable design and proper training and rational choice of training function an concerned parameter in process of simulation of BP neural network with MATLAB.It is demonstrated that it is feasible and impactful to carry through fault vibrating diagnosis of rolling bearings with BP neural netword simulating in MATLAB.