基于细菌觅食决策和深度置信网络的滚动轴承故障诊断  

Rolling bearing fault diagnosis based on bacterial foraging algorithm and deep belief network

作　　者：陶洁[1,2] 刘义伦[1,3] 杨大炼 宾光富[4] 

机构地区：[1]中南大学机电工程学院,长沙410083 [2]湖南科技大学知识处理与网络化制造实验室,湖南湘潭411201 [3]中南大学轻合金研究院,长沙410083 [4]湖南科技大学机械设备健康维护湖南省重点实验室,湖南湘潭411201

出　　处：《振动与冲击》2017年第23期68-74,共7页Journal of Vibration and Shock

基　　金：国家自然科学基金(51575176;11702091);湖南科技大学机械设备健康维护湖南省重点实验室开发(201605)

摘　　要：在利用深度置信网络进行滚动轴承故障诊断时,网络结构的设置不仅影响诊断的结果,还影响计算效率。为提高滚动轴承故障诊断的准确率,提出基于细菌觅食决策和深度置信网络的滚动轴承故障诊断方法。该方法利用采集的样本数据对深度置信网络进行训练,以构造细菌觅食决策算法的适应度函数,通过计算各个细菌的适应度来衡量模型的优劣。由于细菌觅食决策算法具有并行搜索能力,能有效选取深度置信网络各隐节点数、学习率、动量等参数,生成合适的分类器提高滚动轴承故障诊断的准确率。实验中,与SVM(Support Vector Machines)、BPNN(Back Propagation Neural Network)、KNN(k-Nearest Neighbor)等方法做比较,所提方法对滚动轴承故障诊断的准确率达到97.83%,能更加高效、准确的识别滚动轴承故障。When studying rolling bearing fault diagnosis with the deep belief network method,parameters in the deep belief network have a great effect on fault diagnosis results and it is hard to obtain suitable parameters. Here,the fault diagnosis method based on the bacterial foraging algorithm and the deep belief network was proposed to improve the correct rate of bearing fault diagnosis. The parallel search ability of the bacterial foraging algorithm was adopted to effectively choose the number of hidden layer,the number of hidden nodes,the learning rate in a deep belief network.The deep belief network ＇s training data classification error was used to calculate the fitness function of the bacterial foraging algorithm to build an appropriate fault classifier and finish rolling bearing fault diagnosis. The test results showed that the correct rate of the proposed method for rolling bearing fault diagnosis reaches 98. 5%; compared with BPNN,SVM and KNN methods,the proposed method can more stably and more accurately identify rolling bearing faults.

关 键 词：深度置信网络 细菌觅食决策算法 滚动轴承 故障诊断 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]