Fault pattern recognition of rolling bearings based on characteristic waveform sparse matching

A method of fault pattern recognition for rolling bearings was proposed on the basis of sparse matching of a characteristic waveform (CW).With a well-designed search algorithm,multi-section CWs were extracted from a vibration signal.A representative CW was obtained by learning from the extracted CWs.Then,the representative CW was acted as an atom model to construct a dictionary and a pattern matching dictionary.Pattern recognition was conducted through one-order matching analysis in the pattern matching dictionary.Employing the signals of a normal bearing,ball fault,inner race fault and outer race fault for pattern recognition,the result indicates that the method is valid and robust.