Research On Fault Diagnosis Method Of Reciprocating Compressor Based On Swarm Decomposition Algorithm And MEMDE

Reciprocating compressors are mechanical equipment used for the compression and transportation of unstable gases such as ethylene and natural gas in the petrochemical industry.When they fail,they will not only cause production stagnation and economic damage,but may also cause catastrophic accidents and endanger production personnel.life safety.In order to detect failures of reciprocating compressors in time,avoid economic losses,and ensure personal safety,the implementation of condition monitoring and fault diagnosis of reciprocating compressors has become one of the key issues that petroleum and petrochemical companies pay attention to.Due to the non-linear and non-stationary characteristics of the reciprocating compressor signal,when the traditional time-frequency analysis method represented by Fourier transform is used to analyze and process the vibration signal of the reciprocating compressor,the decomposition effect is not ideal.Group decomposition algorithm is an adaptive decomposition algorithm,which has outstanding performance in processing non-stationary signals.This paper introduces it into the field of reciprocating compressor fault diagnosis,and based on the parameter optimization of the group decomposition algorithm,combined with the multi-scale morphological envelope dispersion entropy to judge the uncertainty of the signal,a method that can accurately extract the fault characteristics of the vibration signal is proposed.,So as to effectively identify the method of reciprocating compressor failure.First of all,in-depth investigation and study of the working principle,key component structure and common failure mechanism of reciprocating compressors.Based on a detailed review of related literature on reciprocating compressor fault diagnosis,the current development status and future development trends in the field of reciprocating compressor fault diagnosis are explained.,Summarizes and analyzes the commonly used feature extraction methods and intelligent identification methods in the field of reciprocating compressor fault diagnosis.Then,based on the in-depth study of the group decomposition algorithm,the threshold parameter in the group decomposition algorithm affects the decomposition accuracy,affects the decomposition speed,and causes the decomposition distortion.The differential evolution genetic algorithm is used to optimize the threshold parameters.This method judges whether the envelope entropy of the signal components obtained from each group decomposition meets the requirements by setting the threshold range,so as to obtain the optimal solution of the signal group decomposition parameters.The verification of simulated signals and measured signals shows that parameter optimization can effectively improve the decomposition accuracy of the group decomposition algorithm and achieve accurate acquisition of fault information.Subsequently,a quantitative analysis method of sequence complexity of multi-scale morphological envelope spreading entropy is introduced.The calculation speed of the multi-scale morphological envelope dispersion entropy is fast,which reflects the complexity of the signal amplitude,and can accurately measure the uncertainty of the signal at different scales.Compared with the multi-scale dispersion entropy,the fault feature information extracted by this method can be more comprehensive Reflect the nature of the fault.Two entropy algorithms are used to compare and analyze the simulated signals.The results show that the multi-scale morphological envelope dispersion entropy can better characterize the randomness and dynamic mutation of the signal,and the entropy value is more stable.The measured vibration signal of the reciprocating compressor verifies the effectiveness of the method.Finally,in view of the non-linear and non-stationary characteristics of reciprocating compressor signals,a fault diagnosis method for reciprocating compressors based on group decomposition and multi-scale morphological envelope entropy dispersion is proposed.First,the method uses differential evolution algorithm to optimize the parameters of the original group decomposition algorithm,and then uses the parameter optimized group decomposition algorithm to decompose the vibration signals of the reciprocating compressor in different states to obtain multiple oscillation components with fixed modes.Calculate the entropy value of the multi-scale morphological envelope dispersion of these components,and input it as a feature vector into the support vector machine for training,testing,classification and recognition,and finally realize the accurate recognition of the reciprocating compressor fault.Experiments have verified that this method can effectively extract fault characteristic information,and can accurately diagnose different fault types of reciprocating compressors.