基于小波包分解和神经网络集成群的滚动轴承故障诊断  

Rolling bearing fault diagnosis based on wavelet packet decomposition and neural network integrated swarm

作　　者：柴立平[1] 孟壮壮 石海峡 李强[1] CHAI Liping;MENG Zhuangzhuang;SHI Haixia;LI Qiang(School of Mechanical Engineering,Hefei University of Technology,Hefei 230009,China)

机构地区：[1]合肥工业大学机械工程学院,安徽合肥230009

出　　处：《合肥工业大学学报(自然科学版)》2025年第4期447-454,共8页Journal of Hefei University of Technology:Natural Science

基　　金：安徽省科技重大专项资助项目(202203a05020026);安徽省高校协同创新资助项目(GXXT-2019-004)。

摘　　要：文章提出一种将多个神经网络相结合的神经网络集成群算法进行滚动轴承故障诊断。首先对原始振动信号进行小波包变换,分别采用小波包能量和小波包样本熵作为特征向量;其次采用多个粒子群优化反向传播(particle swarm optimization-back propagation,PSO-BP)神经网络分别对轴承进行故障诊断,比较分析小波包能量和小波包样本熵作为特征向量的适配程度;再以多个神经网络作为神经网络集成群的基础子网络,通过统计耦合、输出耦合和统计输出耦合形成神经网络集成群的二级网络;最后通过最终统计耦合输出神经网络集成群的分类结果。研究结果表明,该方法可获得理想的滚动轴承故障诊断准确率,在负载变化时具有良好的泛化性能。A neural network integrated swarm algorithm that combines multiple neural networks is proposed for rolling bearing fault diagnosis.Firstly,wavelet packet transform was performed on the original vibration signal,and wavelet packet energy(WPE)and wavelet packet sample entropy(WPSE)were used as feature vectors’respectively.Secondly,multiple particle swarm optimization-back propagation(PSO-BP)neural network pairs were used to diagnose the bearing faults separately and compare the adaptability of WPE and WPSE as feature vectors.Then,by using these neural networks as the basic sub-networks of the neural network integrated swarm,the secondary networks of the neural network integrated swarm were formed through statistical coupling,output coupling,and statistical and output coupling.Finally,the classification results of the neural network integrated swarm were output through the final statistical coupling.The results show that the method can obtain the desired accuracy of rolling bearing fault diagnosis and has good generalization performance when the load changes.

关 键 词：滚动轴承 故障诊断 小波包变换 粒子群优化反向传播神经网络 神经网络集成群 

分 类 号：TH17[机械工程—机械制造及自动化]