Research On Typical Fault Early Warning And Diagnosis Method Of Reciprocating Machinery

As a typical representative of reciprocating machinery,diesel engines and reciprocating compressors are widely applied in many fields such as industry,agriculture,petroleum,petrochemical and military industries.Their safe and stable operation plays a very important role in guaranteeing the normal operation of these fields.However,the diesel engine and reciprocating compressor have the characteristics of many vulnerable parts and easy to fail.At present,most of the diesel engine units of the company adopt the method of correction maintenance.The monitoring methods are mostly thermal parameter monitoring,and it is difficult to achieve automatic diagnosis of diesel engine typical faults.Reciprocating compressor piston rod failures occurred frequently,but there is still no effective way for piston rod condition monitoring and fault diagnosis.Therefore,it is of great practical significance to study how to effectively monitor the operating state of the reciprocating machinery and to realize fault early warning and diagnosis.According to the characteristics that there is many information in the thermal parameters and vibration signals of the reciprocating machine unit,the early warning and common fault diagnosis methods of diesel engine and the diagnosis method of reciprocating compressor piston rod fault based on sensor technology and artificial intelligence diagnosis method were studied.The proposed characteristic parameters were applied to the actual engineering case to verify.The main contents of the paper are as follows:(1)The performance parameters of the diesel engine unit will be changed under different load states.By extracting relevant characteristic parameters,the BP neural network model based on parameter prediction residuals was applied to predict the load of the diesel engine.The "load stability parameter"was proposed,and the effectiveness of the method for fault warning was verified.(2)For the cylinder head bearing wear fault,misfire fault and cylinder scuffing fault of diesel engine,a fault diagnosis method based on ReliefF-PCA algorithm and SVM algorithm is proposed.The experimental case has showed that the method can realize the typical fault diagnosis of diesel engine.(3)Aiming at the piston rod loosening failure,piston rod fracture failure and piston support ring wear fault of reciprocating compressor,a method based on harmonic wavelet for piston rod axis orbit purification and fault diagnosis method based on manifold learning and neural network was proposed.The effectiveness of the method for the fault diagnosis of piston rod has been proved by the experimental case and the actual project case.