Reliability Evaluation Of Reciprocating Compressor Based On The Running State Information

Reciprocating compressor is widely used for petroleum and chemical industry and so on,which plays a significant role in the area of petroleum and chemical industry. Reciprocatingcompressor whose structure is very complicated and which consists of many moving parts,once a component failure will be immeasurable influenced on production, which not onlycauses huge economic losses, but also brings security risk. Therefore, According to therunning conditions of reciprocating compressor, to launch the reliability assessment work,which bring important actual value and huge economic benefits. Currently, Fourier transform,wavelet transform, etc., which are widely used for non-stationary signal decomposition, butthose methods are fixed type of decomposition and no adaptability and already aren’t satisfiedwith the needs of reciprocating compressor signal processing. Reciprocating compressorvibration signals are typical nonlinear and non-stationary, According to the characteristics ofreciprocating compressor signals, the technology of EMD-Information Entropy was extractedthe original vibration signal feature, information entropy value tables were set up underdifferent operating conditions,“reliability evaluation system of reciprocating compressorbased on the running state information” was built by support vector machine (SVM).First of all, review of the literature revealed reliability evaluation of reciprocatingcompressor, discussing the background and significance of the subject, summarizing theresearch current situation and difficulties of reliability evaluation of reciprocating compressor.Second, summarizing the knowledge of2D12reciprocating compressor failure mechanismand data analysis. According to the characteristics of reciprocating compressor vibrationsignal，EMD which is suitable for nonlinear and non-stationary signal decomposition, wasextracted vibration signal feature, calculating the information entropy value which cancharacterize reciprocating compressor operating condition, building SVR model which referto reliability evaluation of reciprocating compressor. Finally, Matlab software which is thepowerful mathematical and has visual function, was chosen to build the reciprocatingcompressor reliability assessment system. The system which can be used to reliabilityevaluation based on the running state information not only on valve, cylinder, big end bearingshell and small end bearing shell, but also on complete reciprocating compressor, and meansquare error (MSE) and correlation coefficient were chosen as indicators to evaluate thereliable operation of the reciprocating compressor reliability evaluation system.