基于SFLA改进卷积神经网络的滚动轴承故障诊断  

Fault diagnosis of rolling bearing based on an improved convolutional neural network using SFLA

作　　者：李益兵[1,2] 马建波 江丽 LI Yibing;MA Jianbo;JIANG Li(School of Mechanical and Electronic Engineering,Wuhan University of Technology,Wuhan 430070,China;Hubei Digital Manufacturing Key Laboratory,Wuhan 430070,China)

机构地区：[1]武汉理工大学机电工程学院,武汉430070 [2]数字制造湖北省重点实验室,武汉430070

出　　处：《振动与冲击》2020年第24期187-193,共7页Journal of Vibration and Shock

基　　金：湖北省自然科学基金(2015CFB698)。

摘　　要：针对卷积神经网络(CNN)用于滚动轴承故障诊断时,训练次数比较多,网络结构不容易确定等问题,设计了一种基于混合蛙跳(SFLA)优化CNN的算法(SFLA-CNN),以及基于该算法的滚动轴承故障诊断模型。该模型利用SFLA强大的全局寻优能力和局部深度搜索能力来优化CNN结构,随后运用具有最优结构的CNN模型直接从原始振动信号中提取低维故障特征,并将其输入到Softmax分类器中进行故障识别。与BP神经网络、CNN等方法对比分析,试验结果表明,SFLA-CNN算法具有更高的准确率以及更少的训练次数。In order to solve the problems of the convolutional neural network(CNN)used in rolling bearings fault diagnosis,such as more training times and uncertain network structure,etc.,an algorithm for optimizing CNN based on shuffled frog leaping algorithms(SFLA-CNN)was designed in this paper,and the rolling bearing fault diagnosis model based on this algorithm was proposed.The model uses SFLA’s powerful global optimization ability and local depth search ability to optimize the CNN structure,and then uses the CNN model with optimal structure to extract low-dimensional fault features directly from the original vibration signal and inputs it into the Softmax classifier for faults identification.Compared with the BP neural network and the CNN,the experimental results show that the SFLA-CNN algorithm has higher accuracy and less training time.

关 键 词：卷积神经网络(CNN) 混合蛙跳算法(SFLA) 滚动轴承 故障诊断 

分 类 号：THA133.3[机械工程]