Research On Fault Diagnosis Method Of Reciprocating Compressor Based On SPA And SQPE

Reciprocating compressor,as an important component equipment in industrial production,has complex structure and numerous parts.Once a fault occurs,it may cause economic losses,or even casualties.Therefore,it is extremely important to study its status detection and fault diagnosis.Aiming at the nonlinear and non-stationary characteristics of reciprocating compressor signal,this paper combines smooth prior approach(SPA)and sample quantile permutation entropy(SQPE)to extract its features.At the same time,the fault identification method is used to classify the faults.The results show that the method proposed in this paper can effectively identify the faults of reciprocating compressor.In this paper,while improving the effectiveness and accuracy of diagnosis and identification of typical faults of reciprocating compressor,the research on bearing fault performance degradation evaluation is carried out.Based on Particle Filter(PF)and Long Short Term Memory(LSTM),an effective prediction model is established,and together with the underlying research of fault diagnosis,the equipment life cycle analysis is formed,Realize the transformation of reciprocating compressor from "regular maintenance" to "preventive maintenance",and provide important theoretical basis and technical support for the safe operation of reciprocating compressor.First of all,this paper briefly describes the research significance of fault diagnosis and residual life prediction for reciprocating compressors,and studies and summarizes the commonly used fault diagnosis methods and life prediction methods for reciprocating compressors at present.On this basis,this paper puts forward the research concept and main work of this paper.Secondly,the working principle and common faults of the reciprocating compressor are analyzed.Taking the existing 2D-90 MG reciprocating compressor in the laboratory as the research object,a signal acquisition system is built using Lab VIEW software to collect the vibration signals of the reciprocating compressor fault state.Through research and analysis,it can be seen that the gas valve and bearing are the most vulnerable parts of reciprocating compressor.Therefore,this paper takes the gas valve and bearing as the research object to carry out fault diagnosis analysis and life prediction research.Thirdly,in order to solve the problem that the vibration signal of reciprocating compressor is nonlinear and non-stationary due to the impact,friction and other factors in its work,which leads to its unclear fault characteristics and is not conducive to extraction,this paper proposes a fault feature extraction method for reciprocating compressor based on smooth prior analysis(SPA)and sample quantile permutation entropy(SQPE).In this process,to solve the problem that SPA parameter selection is not universal,sparrow search algorithm(SSA)is used for parameter optimization.In order to obtain vibration signals with prominent fault characteristics,variable mode decomposition(VMD)is used to denoise the signals,and a feature extraction method of multi-scale sample quantile permutation entropy(MSQPE)is proposed to extract fault characteristics of reciprocating compressor.Then,combined with SPA-VMD-MSQPE and SVM methods,a fault diagnosis method for reciprocating compressor based on SPA-VMD-MSQPE and SVM is proposed,and experimental verification is carried out with the valve fault signal data.The experimental results show that the recognition rate of the fault diagnosis method proposed in this paper for the state of the reciprocating compressor valve reaches 97.5%,and the accurate recognition of the state of the valve is successfully achieved.Under the same conditions,through the comparison and analysis with other common fault diagnosis methods,it is confirmed that the method proposed in this paper can obtain better recognition effect.Finally,combined with the feature extraction method,the life prediction research was carried out with the bearing as the research object.On the basis of SQPE features,16time-frequency domain features that can represent the bearing degradation were obtained.After removing the features irrelevant to or redundant with the bearing degradation using the feature selection algorithm,the principal component analysis(PCA)was used to reduce the dimensions of the features twice to build a health factor that can reflect the bearing degradation.Take continuous 3 δ Methods The first failure point of bearing data was determined,and the life prediction was started from the first failure point.A combined mathematical and analog residual life prediction method based on particle filter(PF)and long short memory network(LSTM)was proposed.Finally,the effectiveness of the proposed method was proved by comparing with the actual bearing regression curve.The accuracy of the method proposed in this paper for bearing residual life prediction is proved by the two indexes of root mean square error(RMSE)and mean absolute error(MAE),and the superiority of the method proposed in this paper for bearing residual life prediction is further proved by comparing with the indexes of RMSE and MAE of other life prediction methods.