Development Of State Detection System For Mechanical Equipment And Research Of Signal Processing Arithmetic

Mechanical condition monitoring and fault diagnosis occupies a crucial position in industrial production. In this paper, the significance of mechanical condition monitoring and the problem of non-stationary in fault diagnosis were introduced. Then a kind of new time-frequency analysis theory, Hilbert-Huang Transformation (HHT), was researched in detail. In view of the HHT being sensitive to noise interference, an improved method based on wavelet transform was mentioned, and subsequently the modified method was applied in fault diagnosis. From the two aspects of simulation and application, the advantages of HHT in the vibration signal analysis and the improvement of the anti-interference ability were proved.A system based on LabVIEW was developed for the purpose of mechanical condition monitoring and fault diagnosis. This system mainly consists of two parts, hardware and software. In the part of hardware, a new vibration measurement method based on ultrasonic was mentioned, in view of some limitations of the traditional methods and instruments. The vibration transducer based on ultrasonic was also developed. From the experiment comparing with eddy current displacement sensor, the feasibility and effective of vibration transducer designed are proved.The software part, which having the function modules of acquisition and storage, signal preprocessing (digital filter, wavelet noise eliminator), feature extraction of fault (wave analysis, correlation analysis, power spectrum, cepstrum, time-frequency analyzer of HHT), and the display of results, was designed in LabVIEW. Especially in the module of signal processing, the method of HHT was successfully designed in LabVIEW, witch expends the functions of LabVIEW in signal processing, and provides a new way to feature extraction of fault.The system was tested in the applications of actual vibration analysis. The function modules of time-domain analysis, spectrum analysis, autocorrelation, cepstrum and the time-frequency analysis of HHT were all used in the feature extraction. The results show that the system works well, and the system has its value in industry.