Based On The Bearing Failure Of The Audio Signal Dhmm Diagnosis And Dsp

With the popularizment of lean production, mid- and small-scale enterprise adopt stricter requirements concerning the automation of the equipments. Therefore enhancing the research of the technique about the troubleshooting diagnostic , especially about the bear. And now applied diagnostic apparatus mostly base on the vibration signals. The apparatus need the acceleration transducer which is expensive but the technique applied to the system is simple and the precision is not high when the system is under unfavourable circumstances of complex noises. The article structure a system which adopts audio signals picked easily ,DSPs(Digital Signal Proccessers) that is highly cost-effective , DHMM(Discrete Hidden Markov Model) which is excellent for modeling.Principle Duties and conclusions are as follows:(1) structuring the off-line diagnostic system for bear malfunction which mainly complete information processing for audio signals and load the system files rapidly with GEL(General Extention Language) function;(2) analyzing the feactures and the work mechanisms basing on matlab environment of audio signals for bear malfunction and arguing the practicality;(3)improving the MFCC(Mel-Frequency Cepstrum Coefficients) , applying the Cepstrum-lift MFCC to malfunction diagnosis and proving the latter is better than the former by a comparative analysis;(4)improving the modeling method of DHMM,simpling the model parameters and decreasing its calculation quantity so much.Experiments prove that the average correctness of the system is 98 percent,and its performance index can satisfy realtime requirement. So the system provide a successful scheme of the application for DSP applying to bear malfunction diagnosis.