摘要
为了解决用单一(振动,压力,温度)传感器对液压泵故障诊断时效率低的问题,在PSO-BP诊断层的基础上,利用D-S证据理论对多传感器信号进行融合处理,建立了一种基于PSO-BP诊断层与D-S决策层融合的液压泵故障诊断模型,并针对液压泵正常状态以及五中典型故障(漏油,柱塞磨损,配流盘磨损,松靴磨损,轴承磨损)开展测试分析。研究结果表明:利用本故障诊断模型能够更准确判断柱塞磨损程度与松靴磨损状态,柱塞磨损诊断效率为98.6%,松靴磨损诊断效率为98.4%,单一传感器诊断精度没有超多90%,通过D-S决策层把数据融合后精度都在98%以上,证明了PSO-BP诊断层与D-S决策层融合模型的可行性。本研究具有很高的液压泵故障诊断效率,尤其适用于一些微弱的故障信息,对提前侦测故障危险具有很好的价值。
In order to solve the problem of low efficiency in fault diagnosis of a hydraulic pump using a single(vibration,pressure or temperature)sensor,a fault diagnosis model of a hydraulic pump based on the fusion of PSO-BP diagnosis layer and D-S decision layer is established using D-S evidence theory.And diagnosis of the normal operation state and five typical faults(oil leakage,plunger wear,valve plate wear,loose shoe wear,bearing wear)is carried out.The results show that:the diagnosis efficiency of plunger wear is 98.6%,and the diagnosis efficiency of loose shoe wear is 98.4%.The diagnostic accuracy using a single sensor is improved from below 90%to over 98%after data fusion through the D-S decision layer.Therefore,the feasibility of the fusion model of PSO-BP diagnosis layer and D-S decision layer is validated.This study has a high efficiency of fault diagnosis of hydraulic pumps,especially with some weak fault information,and is useful in detecting fault danger in advance.
作者
崔四芳
宋慧啟
李峰
卢治功
CUI Sifang;SONG Huiqi;LI Feng;LU Zhigong(Department of Mechanical Manufacturing,Xinxiang Vocational and Technical College,Xinxiang Henan 453006,China;Department of Automotive Technology,Xinxiang Vocational and Technical College,Xinxiang Henan 453006,China;School of Mechanical Engineering,Henan Polytechnic University,Xinxiang Henan 453006,China;China Power Technology Information Industry Co.,Ltd.,Xinxiang Henan 453006,China)
出处
《机械设计与研究》
CSCD
北大核心
2022年第2期155-157,173,共4页
Machine Design And Research
基金
河南省科技厅科技攻关项目(212102210163)
河南省教育厅河南省职业教育教学改革项目(ZJC17040)。
关键词
柱塞泵
故障诊断
多源传感器
神经网络
数据融合
诊断输出
plunger pump
fault diagnosis
multi-source sensor
neural network
data fusion
diagnostic output
