论文部分内容阅读
针对机械故障信号经常是多种故障信号的混合(它们相互作用、相互干扰),给正确的故障识别造成很大困难的实际情况,提出基于神经网络非线性主分量分析的机械故障信号分离方法。阐述故障信息的分离与主分量分析的关系,并将二者统一起来,从理论上证明应用主分量分析方法进行故障分离的有效性;介绍神经网络非线性主分量分析;提出基于神经网络非线性主分量分析的故障分离方法。利用实际故障信号进行分离,取得令人满意的结果。
Aiming at the fact that the mechanical fault signal is often a mixture of many kinds of fault signals (they interact and interfere with each other), it is very hard for the correct fault recognition. Based on the nonlinear principal component analysis of neural network, a mechanical fault signal separation method is proposed. The relationship between the separation of fault information and principal component analysis is expounded and the two are unified to prove the validity of fault isolation by using principal component analysis (PCA). The nonlinear principal component analysis of neural networks is introduced. The neural network based on nonlinearity Fault separation method of principal component analysis. Use the actual fault signal separation, and achieved satisfactory results.