NETWORK PATTERN RECOGNITION OF BALL BEARING FAULTS

SUMMARY: The task of condition monitoring and fault diagnosis of rotating machinery faults is significant but is often cumbersome and labour intensive. Automating the procedure of feature extraction, fault detection and identification of rotating machinery has the advantage of reducing the reliance on experienced personnel with expert knowledge. Rolling element bearing failure is one of the foremost causes of breakdown in rotating machinery. This paper proposes a schema to automate the diagnostic procedure based on a newly developed time frequency analysis technique-Matching Pursuit. Features are extracted based on Matching Pursuit analysis and subsequently input to a Feedforward Neural Network (FFNN) to classify bearing conditions including Healthy, Inner Race Fault (IRF), Outer Race Fault (ORF), and Rolling Element Fault (REF). The performance of the automatic diagnostic procedure was then evaluated using classification rate and Mean Square Error (MSE). The proposed procedure successfully classified the four conditions of rolling bearings