Abstract: In view of the Local Characteristic-scale Decomposition（LCD） in serious endpoint effect, the BP neural network is applied to the signal in the extension, and the method based on BP neural network endpoint extension Local Characteristic-scale Decomposition（BP-LCD） is proposed. The BP neural network is used to predict the end of the signal and then the curve is decomposed by LCD in this method. Through the analysis of the simulation signal, the method can effectively suppress the endpoint effect in the LCD method, then the method is applied to the actual fault diagnosis of rolling bearing. The results show that the method is effective.

Key words: Back Propagation（BP） neural network endpoint extension, Local Characteristic-scale Decomposition（LCD）, endpoint effect, rolling bearing, fault diagnosis

摘要： 针对局部特征尺度分解（Local Characteristic-scale Decomposition，LCD）方法中严重的端点效应，将BP神经网络应用到信号的延拓中，提出了一种提出基于BP神经网络延拓局部特征尺度分解（BP neural network endpoint extension Local Characteristic-scale Decomposition，BP-LCD）方法。该方法首先利用BP神经网络将待分解信号进行两端预测延拓，然后对延拓后的曲线进行LCD分解。通过仿真信号的分析，验证了该方法可以有效地抑制LCD方法中的端点效应；将该方法应用到实际滚动轴承的故障诊断中，结果表明了该方法的有效性。

关键词: 反向传播（BP）神经网络延拓, 局部特征尺度分解, 端点效应, 滚动轴承, 故障诊断