Research On Fault Diagnosis Method Of Reciprocating Compressor Based On LCD And Multi-scale Fuzzy Entropy

Reciprocating compressors are one of the most important equipment in petrochemical and other fields.Due to the complex structure of reciprocating compressors and many parts,failures are prone to occur.If failures cannot be discovered and eliminated in time,this will not only cause a large amount of economic losses in petrochemical enterprises,but also may cause safety accidents,resulting in casualties.Diagnosis is very necessary.Aiming at the characteristics of the reciprocating compressor fault vibration signal with high noise and high complexity,in order to accurately achieve fault diagnosis,this paper proposes based on local characteristic-scale decomposition(LCD)and Multiscale Fuzzy Entropy(MFE)combined reciprocating compressor fault diagnosis,and with the research object of reciprocating compressors,a series of work on the most common gas valve failures among reciprocating compressor failures.First,the development status,basic structure,working principle,fault type and fault mechanism of the fault diagnosis of reciprocating compressors are described.Secondly,the algorithm flow of the local feature scale decomposition(LCD)method and three other adaptive time-frequency analysis methods are described,and the defects of the EMD,LMD and ITD methods are pointed out,and the four time-frequency analysis methods are simulated through simulation experiments The decomposition effect of the software is compared and analyzed in three aspects: relevance,number of iterations and running time.It shows that the LCD method can avoid the problems of envelope error,end-point effect,component distortion and complicated calculation process,and can be better applied to the data processing of non-stationary signals.Third,the basic theory and algorithm flow of multi-scale fuzzy entropy are described.Multi-scale fuzzy entropy can describe the complexity of time series under different scale factors,avoiding the problem of incomplete fault information obtained under a single scale factor.Using simulation analysis,the multi-scale fuzzy entropy is compared with the multi-scale entropy,which proves that the multi-scale fuzzy entropy has relative consistency in describing the complexity of time series.Finally,take the gas valve failure of 2d12 reciprocating compressor as an example,the fault diagnosis of reciprocating compressors based on LCD and multi-scale fuzzy entropy is proposed.In the case of a reciprocating compressor valve,And use Extreme Learning Machine(ELM)as the recognition classifier to classify and recognize the four states of thevalve to achieve fault diagnosis.In the case analysis,the method combining LCD and MFE is compared with other methods.Experiments show that this method can effectively extract the fault characteristics of the reciprocating compressor and diagnose the fault types with different fault characteristics.