The Research And Development Of Rolling Bearing Fault Diagnosis System Based On DSP

Rolling bearing fault diagnosis is to identify bearing state through observe, analyze and process the signal which can reflect the state of the bearing, to ensure the bearings can effective and reliable operation under a certain load, to a certain operation speed and during a certain period of work. integrity of the rolling bearing fault diagnose process includes three parts: observe signal,feature extraction,fault diagnosis.Observe signal: is the key and difficult part of the whole research process, directly related to the accuracy of rolling bearing fault diagnosis. Through theoretical analysis and comparison, Adopt condition signal of rolling bearing with the sound level meter, under the Nyquist sampling theorem conditions, through A/D on the WT6701PA board to sample the signal, as internal storage space of DSP is limited, so use real-time data track of CCS, through C language program DSP external interrupt, then active DMA, real-time transmit sampling data to the PC, store as a data file.Feature extraction: is the process to analyze the sampling signal. With develop of wavelet analysis and spectrum analysis technology, creating the new field of using time-frequency analysis method to detect and diagnose the bearing fault. Wavelet transformation is a time-scale analysis, the scale is corresponded with frequency, so it is a kind of time-frequency analysis; wavelet transformation has the following characteristics: has high time resolution in the high frequency range, has high frequency resolution in the low-frequency range; Fast Algorithm-Mallat algorithm; using discrete wavelet transformation can decompose signal into various scales. The key is to know the changing of the sampling frequency in Mallat algorithm. based on the unique structure and programmability of DSP internal, making a high-speed real-time processing system with DSP core not only handle large quantities of data quickly, but also process signal on real-time, so select programmable TMS320C6701 DSP chip to realize the wavelet transformation. Then do spectrum analysis of the sampling signal through Matlab. By contrary, complete feature extraction.Fault diagnosis: contrary numbers of rolling bearing signal analysis, and calculate the fault characteristics frequency of the rolling bearing, to diagnose the fault of rolling bearings.