Application research of pattern recognition for failure detection

Considering the characteristics of the mechanical equipment and the shortage of the existed diagnosis method, a failure detection method based on pattern recognition is defined. Short time signal technology and linear distinguish function are adopted, alone feature is combined for deciding. The method of sensitive and efficient is provided for detecting bearing fault and is applied on bearing diagnosis and frication diagnosis. From the analysis result of the actual vibration signals, the proposed method is proved to be efficient in detecting and diagnosing the rolling element bearing faults. The results indicate it has abilities of early stage forecast fault.