Fault Diagnosis Method For Compressor Valves In Terms Of Morphological Fractal

Reciprocating compressor due to its many vibration sources, complicated structure, lead to the valve vibration signal not only has the time-frequency characteristics, but also contain certain morphology fractal characteristics. This paper study on the valve fault diagnosis adaptive waveform decomposition and morphology fractal method, the main work is as follows:1. Carried out valve fault diagnosis based on empirical mode decomposition (EMD), local mean decomposition (LMD) and adaptive waveform decomposition (AWD) methods which are time-frequency domain decomposition algorithm, realize fault features of valves, analyzed and compared several methods of typical noise reduction effect. The results show that the noise reduction result of the AWD algorithm is better than that of the EMD, LMD algorithm.2. Carried out valve fault diagnosis based on the entropy value method (permutation entropy, sample entropy and approximate entropy) and traditional fractal dimension (box dimension and correlation dimension) method, which are nonlinear analysis methods. Finally, realized the typical failure mode identification of the air valve, and analysis the effect of failure mode identification. The results show that the traditional nonlinear algorithm can only identify valve normal and fault condition, can't identify several kind of typical failure mode.3. Carried out valve fault diagnosis based on the morphology of credits fractal dimension method and the higher order cumulates fractal dimension method. Combined with the piecewise fractal dimension and general improvement methods, such as cover area, then analyzed the environmental noise sensitivity and valve fault pattern recognition effect of two algorithms, finally realizes the valve typical fault mode.