小波包分解与神经网络相结合的变速箱齿轮故障识别  

Fault Recognition of Automotive Transmission Based on Wavelet Packet Decomposition and Neural Network

作　　者：羊拯民[1] 朱忠奎[1] 

机构地区：[1]合肥工业大学机械与汽车工程学院

出　　处：《农业机械学报》2001年第5期78-81,共4页Transactions of the Chinese Society for Agricultural Machinery

摘　　要：提出了一种识别变速箱齿轮故障的新方法。通过对小波包分解的分析研究 ,将基于小波包能量的小波包分解特征提取方法用于提取齿轮运行状态的特征向量 ;并以此作为 BP神经网络的输入对神经网络进行训练 ,建立了基于 BP神经网络的齿轮运行状态分类器 ,用以识别齿轮的运行状态 ;最后 ,以变速箱齿轮故障识别为例 ,用文中所述方法对变速箱齿轮的正常状态、磨损状态、断齿状态进行识别验证。验证结果表明该方法的效果良好。A new method of recognizing the working state of the gear based on wavelet packet decomposition and BP neural network is presented in this paper. Wavelet packet decomposition was studied and feature parameter vector extraction method based on wavelet packet energy was applied to feature parameter vector extraction of the gear working state. Then neural network was introduced into gear fault diagnosis. The feature parameter vectors were used to train BP network and the trained BP network acted as the gear working state classifier. With the example of automotive transmission gear fault diagnosis, this method was used to recognize normal state, wear state and faulty state of transmission gear. The result shows that this gear working state recognition method based on wavelet packet decomposition and BP neural networks is quite effective.

关 键 词：小波包分解 神经网络 齿轮 故障 识别 

分 类 号：U279.323[机械工程—车辆工程]