Intelligent Diagnosis Technology Research For Reciprocating Compressors Based On Sensitive Feature Extraction

Reciprocating compressors play an important role in the petrochemical industry, and they almost compress the gas which is toxic or flammable, so fault diagnosis technology seems obviously important for the safety of reciprocating compressors. In order to improve the accuracy of fault diagnosis, in this paper a reciprocating compressor automatic fault diagnosis system is established, after feature extraction respectively for valve fault and piston fault. The main work is as follows:(1)Because bad valve’s temperature isn’t inconsistent with good valve’s, principal component analysis (PCA) is used to extract a feature named by "main feature vector". And it shows this feature can characterize valve fault through feature selection.(2) Through establishing mechanical model of piston parts, two features named by "frequency ratio" and "energy ratio" are extracted. And it shows these features can characterize piston fault through feature selection.(3)In order to improve the efficiency of fault diagnosis, and reduce the workload of fault diagnosis personnel, a reciprocating compressor automatic fault diagnosis system is established by using "main feature vector", "frequency ratio" and "energy ratio". And the experimental result shows this system can diagnose faults automatically, efficiently, and accurately.