摘要
为了准确获取往复泵故障规律和特征量用以监控和诊断往复泵工作状态。本文通过人为实验模拟往复泵液力端泵阀漏失、弹簧断裂、柱塞磨损等6种典型故障,得到了往复泵在典型故障状态下的示功图并分析其产生原因和规律;利用MATLAB编程分别对6种故障状态下的示功图提取6组灰度矩阵特征向量,所得特征量样本数据通过支持向量机训练。结果表明,其故障自动识别率能达到95%以上,具有较高的诊断准确性,可作为往复泵在线监控和故障自动诊断的数据基础。
In order to accurately obtain the failure characteristics of a reciprocating pump so as to monitor and diagnose its working conditions, this article artificially simulated 6 kinds of typical failure at its hydraulic end such as the leakage of pump valve, spring break, plunger wear, obtaining the indicator diagram and law of the reciprocating pump under typical fault conditions. It extracted the feature of its gray matrix based on the indicator diagram and then pretreated them to extract 6 groups of characteristics with the MATLAB. The results obtained with support vector machine trained sample data show that the automatic fault recognition rate of the method reaches more than 95%, having a higher diagnostic accuracy. Therefore the method can be used to establish the automatic online monitoring and fault diagnosis data base of a reciprocating pump.
出处
《机械科学与技术》
CSCD
北大核心
2016年第2期279-284,共6页
Mechanical Science and Technology for Aerospace Engineering
基金
省部共建"石油天然气装备"教育部重点实验室(西南石油大学)项目(2013sts03)资助
关键词
往复泵
示功图
特征量
故障规律
诊断
MATLAB
data acquisition, design of experiments, eigenvalues and eigenfunctions, failure analysis, fault detection, feature extraction, flow rate, fracture, MATLAB, matrix algebra, mesh generation, monitoring, power control, reciprocating pumps wcar of mate rials, schematic diagrams, stochastic models, support vector machines
fault diagnosis, indicator diagram
