基于混沌分形理论的特征提取技术在气阀故障诊断中应用  

Feature extraction method based on chaotic fractal theory and its application in fault diagnosis of gas valves

作　　者：马晋[1] 江志农[1] 高金吉[1] 

机构地区：[1]北京化工大学诊断与自愈工程研究中心,北京100029

出　　处：《振动与冲击》2012年第19期26-30,共5页Journal of Vibration and Shock

基　　金：国家自然科学基金重点项目(51135001);国家重点基础研究发展计划"973"计划项目(2012CB026000)

摘　　要：气阀故障是往复压缩机的故障率最高的部件,如果气阀故障发现不及时很容易造成气阀串气,进而引起往复压缩机的压缩效率大大降低,如果阀片断裂脱落甚至会造成拉缸的恶性事故。针对目前往复压缩机气阀故障诊断无法评估损坏程度问题,结合混沌分形理论,提出了采用基于混沌分形理论的自动识别方法对气阀早期故障冲击变化信号进行特征提取,实现了气阀故障的早期预警。Faults of gas valve often happen for reciprocating compressors. If the fault is not found and solved in time, the compressed gas leaks from gas valves and the efficiency of a reciprocating compressor is greatly reduced. If the valve plate breaks and falls into the cylinder, the serious accident of piston scraping occurs. Aiming at that the level of valve fault could not be evaluated with the current fault diagnosis technique for gas values in reciprocating compressors, an automatic identification method to extract early shock signals of gas valves based on chaotic fractal theory was presented. Early warning of gas valve fault was implemented with chaotic entropy analysis method.

关 键 词：气阀 混沌 G—P算法 特征提取 往复压缩机 早期预警 

分 类 号：TH165.3[机械工程—机械制造及自动化] TP391[自动化与计算机技术—计算机应用技术]