摘要
为提升故障诊断效果,设计基于多智能体的光传感器故障诊断系统。采集光传感器信号,利用小波变换的多尺度加权排列熵特征提取方法,提取光传感器故障特征;智能体根据提取的故障特征生成故障诊断总任务,并分解成故障诊断智能体可直接诊断的故障子任务;根据子任务为故障诊断智能体分配子任务;故障诊断智能体利用径向基函数神经网络,完成光传感器故障诊断。实验结果表明:该系统可有效提取光传感器故障特征;在不同故障程度下,该系统均可精准诊断光传感器故障。其故障诊断准确率为98%左右、诊断耗时为0.5 s左右。
In order to improve the fault diagnosis effect,an optical sensor fault diagnosis system based on multi-a-gent is designed.The signal of the optical sensor is collected,and the fault characteristics of the optical sensor are ex-tracted by using the multi-scale weighted permutation entropy feature extraction method of wavelet transform.Accord-ing to the extracted fault features,the general fault diagnosis task is generated and decomposed into fault sub-tasks that can be directly diagnosed by the fault diagnosis agent.Assign sub-tasks to fault diagnosis agents according to sub-tasks;the fault diagnosis agent uses radial basis function neural network to diagnose the fault of optical sensor.Ex-perimental results show that the system can extract the fault characteristics of optical sensor effectively.Under different fault degree,the system can accurately diagnose the fault of optical sensor.The accuracy of fault diagnosis is about 98%and the time of diagnosis is about 0.5 s.
作者
李焕贞
皮珣珣
LI Huanzhen;PI Xunxun(Information Engineering College,Jiangxi University of Technology,Nanchang 330098,China)
出处
《激光杂志》
CAS
北大核心
2023年第12期240-245,共6页
Laser Journal
基金
江西省教育厅科学技术研究项目(No.GJJ14766)
江西省高等学校研究项目(No.JXJG-21-24-1)。
关键词
多智能体
光传感器
故障诊断
小波变换
多尺度
神经网络
multi agent
optical sensor
fault diagnosis
wavelet transform
multi scale
neural network
