引用本文:林海军,滕召胜,迟海,吴阳平,易钊.基于信息融合的汽车衡称重传感器故障诊断[J].控制理论与应用,2010,27(1):25~31.[点击复制]
LIN Hai-jun,TENG Zhao-sheng,CHI Hai,WU Yang-ping,YI Zhao.Diagnosis for load cells in truck scale based on information fusion[J].Control Theory and Technology,2010,27(1):25~31.[点击复制]
基于信息融合的汽车衡称重传感器故障诊断
Diagnosis for load cells in truck scale based on information fusion
摘要点击 2156  全文点击 1370  投稿时间：2009-01-16  修订日期：2009-04-24
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/j.issn.1000-8152.2010.1.CCTA090057
  2010,27(1):25-31
中文关键词  汽车衡  称重传感器  故障诊断  信息融合  径向基神经网络
英文关键词  truck scale  load sensor  fault-diagnosis  information fusion  radial-basis-function-neural-network
基金项目  商务部优化机电和高新技术产品进出口结构资金资助项目(财企[2007]301号); 湖南省自然科学基金资助项目(00JJY2061).
作者单位E-mail
林海军* 湖南大学 电气与信息工程学院 智能仪器研究所 linhaijun801028@126.com 
滕召胜 湖南大学 电气与信息工程学院 智能仪器研究所  
迟海 湖南大学 电气与信息工程学院 智能仪器研究所  
吴阳平 湖南大学 电气与信息工程学院 智能仪器研究所  
易钊 湖南大学 电气与信息工程学院 智能仪器研究所  
中文摘要
      传统汽车衡不具备故障诊断功能, 任一称重传感器发生故障都将导致称重系统失效. 为此提出了一种基于信息融合的汽车衡称重传感器故障诊断方法, 利用径向基函数神经网络(RBFNN)逼近汽车衡多路称重传感器之间的函数关系, 预测各传感器的输出, 并给出RBFNN的训练算法; 以各传感器的预测信号与实测信号为输入, 建立了融合检测模型, 采用表决融合检测准则, 完成故障传感器寻址、故障类型识别、故障程度判决和故障传感器正常输 出估计等故障诊断. 大量实验与现场检定证明, 采用这种方法的汽车衡准确实现了称重传感器故障诊断, 任一称重传感器失效后的汽车衡性能优于正常状态下4级秤的指标, 其最大称重误差≤60.7%, 提高了系统可靠性.
英文摘要
      Conventional truck scale without fault diagnosis will be disabled when anyone load cell is going wrong in operation. A fault-diagnosis method for load cells is proposed based on information-fusion technique. The radial-basisfunction- neural-network(RBFNN) with a training algorithm is employed to approximately model the internal relations among load cells for predicting their outputs. The prediction outputs together with the real outputs of the load cells are sent to a fusion-detection model developed by us. This model employs the criterion of voting-fusion-diagnosis to generate the fusion-diagnosis results, which include locations of faulty load cells, the types and the degrees of faults, the estimated outputs of faulty load cells in normal operating condition. Field tests show that the truck scale installed with the proposed diagnostic facilities discriminates load cells precisely. In the case of one faulty cell, its maximum weighing error is less than 0.7%, exhibiting a performance better than that of a 4th class scale under normal operating condition.