A fault feature extraction method for reciprocating compressor based on optimized local mean decomposition

Reciprocating compressors are widely used in petroleum and chemical industries, excessive clearance caused by bearing wear is the main form of failure. The fault vibration signal of reciprocating compressor bearing has strong non-stationary characteristics, aiming at the deficiency of adaptive algorithm in smoothing process, this paper presents an optimize local mean decomposition (OLMD) method. In interpolation calculation, interpolation points such as extreme value and eigenvalue are added, the optimized local mean function and envelope estimation function were obtained by using monotone cubic Hermite interpolation method, thus the interpolation fitting accuracy is improved. Through analog signal analysis and field test signal analysis, different LMD methods were used to resolve the large fault signal of large axle bushing clearance of type 2D12 reciprocating compressor, quantitative comparative analysis with the help of relevant evaluation indicators, through the above research, it is proved that the mono-mite interpolation LMD algorithm can be used to accurately extract and diagnoses the bearing clearance fault characteristics of reciprocating compressor.