基于双层人工免疫网络的皮带秤故障诊断  

The belt weigher's fault diagnosis based on multilayer artificial immune model

作　　者：高森祺 童一飞[1] 李东波[1] 袁延强[1] 

机构地区：[1]南京理工大学机械工程学院,江苏南京210094

出　　处：《机械设计与制造工程》2017年第1期107-111,共5页Machine Design and Manufacturing Engineering

基　　金：江苏省前瞻性联合研究项目(BY2015004-05)

摘　　要：当前基于神经网络等传统算法的皮带秤的故障诊断方法对样本数量需求多且易出现局部最优解,为提高皮带秤故障诊断效率及使用精度,引入了人工免疫网络模型。传统的ai Net网络模型对已知故障的识别效率高,但难以有效识别未知故障,为弥补这一缺陷,基于生物免疫机制设计了双层免疫网络,以克隆选择算法为核心搭建了适应性诊断层实现对未知故障的学习,并运用在皮带秤的故障检测中。该方法对已有故障的识别率保持在95%以上,对新故障的识别率也高达90%以上,实际运行效果良好。The current intelligent weighing system of belt weigher mainly uses neural network for fault diagnosis. This method demands too many samples and is easier to generate local optimum results. In order to improve the fault diagnosis efficiency of the belt weigher and further ensure the accuracy, it proposes the artificial immune network model into the intelligent weighing. The traditional aiNet network model has high recognition efficiency to the known problems, but could not effectively identify the unknown faults. Aiming at this limitation, it intro- duces the muhilayer immune network based on biological immune mechanism, designs the clonal selection algo- rithm for the self- adaptive immune layer, realizes the unknown fault study. It applies the proposed model into the practical engineering of the belt weigher fault diagnosis, remains the recognition rate of existing fault at more than 95%, improves the recognition rate of the new fault as high as 90% or more, achieves good results.

关 键 词：故障诊断 人工免疫网络模型 双层免疫网络 适应性诊断层 

分 类 号：TH165[机械工程—机械制造及自动化]