Research On Fault Diagnosis Of Reciprocating Compressor Based On Resonance-based Sparsity Decomposition

With the continuous development of modern society,the status of petroleum and chemical industry in the national economy have been constantly improved.As the core equipment in this field,reciprocating compressor is widely used for the compression and transportation of flammable and explosive gases such as ethylene and natural gas due its advantages of high thermal efficiency and wide working medium.There might be irreparable losses when it fails to work.Therefore,the research on fault diagnosis methods of reciprocating compressor has become one of the research hotspots at home and abroad.The reciprocating compressor has complex structure,harsh working environment,numerous internal excitation sources,and complex and diverse fault types.The vibration signal exhibits strong characteristics of non-stationary,nonlinear,multi-component coupling,and often accompanied by strong noise.How to filter out useful fault information and judge the type of fault exactly are important and difficult issue of the reciprocating compressor fault diagnosis work.In view of the above characteristics of reciprocating compressor,based on the related research results at home and abroad and according to the resonance characteristics of vibration signal of reciprocating compressor,sparse decomposition method based on the particle swarm optimization resonance was utilized to decompose vibration signal,the high and low resonance components and residual components were obtained.Then the particle swarm optimization algorithm was used to optimize the multi-scale entropy parameter,and the method of multi-scale permutation entropy was performed on the low-resonance components to get the data of fault feature according to the optimal parameter combination,and the support vector machine was used for pattern recognition and classification.The results showed that the method could extract the fault information accurately and effectively improve the fault diagnosis accuracy of the reciprocating compressor.First of all,the literature was searched to acquire the current status and development trend of reciprocating compressors fault diagnosis at home and abroad,and the signal decomposition method was proposed.The common feature extraction method and intelligent identification technology of reciprocating compressor were studied and summarized.Secondly,the signal resonance property and resonance sparse decomposition algorithm flow were studied.Aiming at the problem of poor performance for the reason of artificial selection of high and low quality factors in the traditional resonance sparse decomposition method,the particle swarm optimization algorithm was applied to the high-low quality factor selection of the resonance sparse decomposition.The quality factor was optimized as the objective function was the kurtosis of low resonance component.The experimental results of the simulated signal and the reciprocating compressor measured vibration signal indicated that the method could separate the fault information effectively.Furthermore,based of the research into multi-scale permutation entropy algorithm theory,the necessity of multi-scale entropy parameter optimization was analyzed,the square functionof entropy skewness of the multi-scale permutation was taken as the objective function,and the particle swarm optimization algorithm was used to optimize the multi-scale entropy parameters.In the range of parameters,the particle swarm optimization was utilized to obtain the optimal parameter combination of each state and then extract the fault features.Finally,the structure,working principle,common faults and mechanisms,measuring point arrangement of the bearing and valve of 2D12 reciprocating compressor were introduced,and the fault diagnosis method and diagnosis process were determined.The experimental results showed that the method has better recognition accuracy and can distinguish the main fault types of reciprocating compressor bearings and valves accurately.