周超;张进杰;张旭东;孙旭;王瑶;
Zhou Chao;Zhang Jinjie;Zhang Xudong;Sun Xu;Wang Yao;Beijing Key Laboratory of Health Monitoring Control and Fault Self-recovery for High-end Machinery Beijing University of Chemical Technology;State Key Laboratory of Compressor Technology(Anhui Provincial Laboratory of Compressor Technology);

• 1:高端机械装备健康监控与自愈化北京市重点实验室北京化工大学
• 2:压缩机技术国家重点实验室(压缩机技术安徽省实验室)

摘要(Abstract)：

活塞杆是往复压缩机的核心运动部件之一,一旦发生松动、断裂等故障易导致恶性事故。压缩机运行负荷与故障均会导致活塞杆运行状态的改变,从活塞杆运行数据中提取故障特征需排除压缩机运行负荷的影响。本文分别采用改进的离散点包络与信息熵评估方法,从活塞杆轴心轨迹中提取特征参数,采用流形学习方法进行特征降维后构建负荷敏感特征集,与常规包络方法提取特征进行对比,证明了本文方法提取的活塞杆运行状态特征具有更强的敏感性;进一步应用神经网络构建负荷识别模型,实现了压缩机运行负荷的自动识别,为消减负荷影响,实现变负荷工况下压缩机故障特征提取与诊断奠定了基础。
Piston rod is one of the core moving parts of reciprocating compressor.When the piston rod becomes loose or breaks,it tends to cause malignant accidents.Both compressor operating load and fault will lead to the change of operating state of piston rod.It is necessary to extract fault characteristics from operating data of piston rod to eliminate the influence of compressor operating load.In this paper,the improved method of discrete point envelope and information entropy was used to extract the feature parameters of axis orbit from the piston rod.The manifold learning method was used to construct sensitive feature set of load after feature dimension reduction.Compared with the conventional envelope method,it is proved that the operating state characteristics of piston rod extracted by this paper are more sensitive.Furthermore,the neural network was used to construct the load identification model and the automatic identification of compressor operating load was realized,which lays a foundation for reducing the influence of load and realizing fault feature extraction and diagnosis of the compressor under variable load condition.

关键词(KeyWords)： 活塞杆;往复压缩机;离散点包络;信息熵;变负荷工况
piston rod;reciprocating compressor;discrete point envelope;information entropy;variable load condition

Abstract:

Keywords:

基金项目(Foundation): 国家重点研发计划项目(2016YFF0203305);; 中央高校基本科研业务费专项资金资助项目(JD1912);; 压缩机技术国家重点实验室(压缩机技术安徽省实验室)开放基金项目(SKL-YSJ201808)

作者(Authors): 周超;张进杰;张旭东;孙旭;王瑶;
Zhou Chao;Zhang Jinjie;Zhang Xudong;Sun Xu;Wang Yao;Beijing Key Laboratory of Health Monitoring Control and Fault Self-recovery for High-end Machinery Beijing University of Chemical Technology;State Key Laboratory of Compressor Technology(Anhui Provincial Laboratory of Compressor Technology);

参考文献(References)：

• [1]张宗珍.活塞压缩机变工况热力与动力特性的仿真研究[D].阜新:辽宁工程技术大学,2007.
• [2]Dhumale R B,Lokhande S D.Neural ne-twork fault diagnosis of voltage source inverter under variable load conditions at different frequencies[J].Measurement,2016,91:S026322411630118X.
• [3]Zhu H,Hu J,Gao L,et al.Fault diagnosis of incipient broken rotor bar-s for squirrelcage induction motor under continuous variable load condition[J].Chinese Journal of Scientific Instrument,2014,35(7):1646-1653.
• [4]赵梦芸,程贵健,李强,等.往复压缩机变负荷工况阀片运动规律及气缸内状态仿真研究[J].机电工程,2019,36(5):480-484.
• [5]马波,高金吉,江志农.往复压缩机活塞杆断裂早期预警技术的研究[J].机械强度,2008,30(3):445-449.
• [6]马晋,江志农,高金吉.基于活塞杆轴心位置轨迹的往复压缩机故障诊断技术研究[J].振动工程学报,2012,25(4):453-459.
• [7]Jiang Z N,Zhu L N,Zhang J J,et al.Research on fault diagnosis method of piston rod based on harmonic wavelet and manifold learning[J].高技术通讯(英文版),2018(3):232-240.
• [8]王庆锋,董良遇,张新,等.一种基于活塞杆动态能量指数的故障监测诊断方法[J].流体机械,2016,44(9):47-52.
• [9]刘彬,戴桂平,林洪彬.一种改进的基于小波变换的包络提取算法研究[J].仪器仪表学报,2006,27(1):34-37.
• [10]林京,刘红星,屈梁生.信号包络特征识别在故障诊断中的应用[J].振动、测试与诊断,1998(1):34-38.
• [11]何小高,张庆,贾林山,等.变速重载设备齿轮箱轴承故障的改进阶次包络分析方法[J].西安交通大学学报,2019,53(8):98-106.
• [12]黄传金,孟雅俊,雷文平,等.复局部均值分解全矢包络技术及其在转子故障特征提取中的应用[J].机械工程学报,2016,52(7):69-78
• [13]李泽滔,何梦林,Dahhou B.一种用于故障隔离的参数区间包络线计算方法[J].现代机械,2018(1):31-34.