Application of EMD and SampEn to the fault diagnosis of reciprocating compressor valves

Aiming at the characteristics of fault signals of for a reciprocating compressor valve: shock and discontinuity, the EMD method was taken to decompose and extract the fault signal of every frequency, which was reconstructed through screening the data and finding the main vibration information. The decomposition results and reconstruction data of the reciprocating compressor fault signals were analyzed, and the fault information was extracted. For the complexity of decomposition results of the normal and fault signal, the sample entropy(SampEn) method was used to quantitatively analyze the fault information of every main IMF. EMD combined with reconstruction, and SampEn methods were applied to analyze the vibration signals of three kinds of conditions: normal state, valve gap, and valve less spring. The practical application abstracted the fault information exactly and proved the feasibility of the proposed method.