Fault Diagnosis Method Of Reciprocating Compressor Based On MRSSD And CMSDE

The reciprocating compressor is a vital equipment in the petroleum chemical industry field,which is mainly used for the compression and transportation of ethylene,natural gas and other inflammable and explosive gases,if it fails to work,it could cause a catastrophic accident.Therefore,it is particularly necessary to conduct fault diagnosis research on reciprocating compressors,and it is also one of the hot spots of domestic and foreign research in recent years.There are many parts of reciprocating compressors,and the damage of a small part may often lead to the shutdown of the whole machine,which will bring huge economic losses to the production of enterprises.If it is possible to combine advanced vibration signal testing technology with the powerful data analysis and processing capabilities of the computer,collect and analyze the vibration signal of the fault location of the reciprocating compressor,find out the cause,and accurately and timely predict and diagnose the hidden fault,can greatly save the loss.However,reciprocating compressors are not only complicated in structure and harsh in working environment,but their vibration signals usually show strong non-stationary and non-linear characteristics.It is difficult for traditional signal processing techniques to separate useful fault information from such complex signals.In this paper,according to the characteristics of reciprocating compressor signals,research and analysis of relevant research results at home and abroad,and a fault diagnosis method for reciprocating compressors based on multiple resonance-based sparse signal decomposition(MRSSD)and composite multi-scale symbolic dynamic entropy(CMSDE)is proposed.This method uses MRSSD to decompose the fault signal according to the signal characteristics of the reciprocating compressor,Get the low resonance component that finally contains the fault feature,then perform the MSDE quantitative calculation of the low resonance component,Extract the fault feature vector,and finally input the fault feature vector into the Support Vector Machine(SVM)to identify and classify the fault type.Experimental results show that this method can accurately extract fault feature information,diagnose the type of fault,and can effectively improve the accuracy of fault diagnosis compared with other methods.First,consult relevant literature,understand the structure and failure mechanism of reciprocating compressors,grasp the status and development trends of fault diagnosis of reciprocating compressors at home and abroad,summarize and analyze the commonly used feature extraction methods and intelligent identification methods of reciprocating compressors.Then,the signal resonance sparse signal decomposition theory and algorithm are analyzed,and it is found that the traditional resonance sparse decomposition method requires artificial selection of high and low quality factor values,resulting in inaccurate decomposition results,although genetic algorithms can be used to optimize the selection of high,Low quality factor,but the efficiency of genetic algorithm is not high.For these problems,a method of multiple resonance sparse signal decomposition is proposed,this method determines whether the high resonance component kurtosis value obtained by each resonance sparse signal decomposition meets the requirements by setting a threshold,Then decide whether to continue to decompose the low resonance component to obtain the final low resonance component.The experimental results of analog signals and actual vibration signals of reciprocating compressors show that this method can effectively achieve the effective separation of fault information.Next,in the process of studying multi-scale symbolic dynamic entropy,it was found that the traditional multi-scale coarse-grain analysis method shortened the length of the time series,for the analysis of short time series,multi-scale symbolic dynamic entropy may often produce inaccurate and suspicious entropy estimates at larger time scale,for this problem,composite multi-scale symbolic dynamic entropy is proposed.By changing the coarse-grained method,the entropy values of all symbols are calculated under the same scale factor,and then averaged.This method can significantly improve the accuracy of multi-scale symbolic dynamic entropy.Finally,the 2D12-70 reciprocating compressor is taken as the research object,this paper introduces the structure and working principle of the reciprocating compressor,the fault mechanism of bearing and gas valve and the arrangement of measuring points,and gives the fault diagnosis method and specific process.Firstly,set the value range of high and low quality factors to find out the low quality factors that can represent the impact components,and the signal is decomposed by resonance sparse signal decomposition,obtaining the high and low resonance components.Then,the decomposition is judged according to the kurtosis value of the high resonance component at this time.If it is less than the given threshold value,the high quality factor value will be changed,and the resonance sparse decomposition of the low resonance components is continued.When it is larger than the threshold value,the decomposition is terminated.The composite multi-scale symbolic dynamic entropy of the final low resonance component is calculated,the fault feature vector is constructed.Finally,use support vector machines to classify and identify and diagnose the fault type.The results prove that the fault diagnosis method based on MRSSD and CMSDE proposed in this paper can accurately diagnose the common faults of bearings and gas valves of reciprocating compressors.