Bearing Fault Detection Based On Embedded System Design

As one of important parts in rotating machinery and equipment, the operation of rolling bearing affects the performance of mechanical equipment directly. Therefore, the real-time monitoring for rolling bearing which can identify the germination and trend of fault, is important in ensuring the safe operation of the equipment s and reducing or even avoiding major accidents. High frequency noise signals usually emerge in the early state of fault of rolling bearing. The research object of this topic is the high frequency noise signal. A kind of bearing fault detection system based on embedded is designed to detect running state of bearing.Two parts with hardware circuit design and software programming are included in this detecting system. Hardware circuit module is mainly to complete the high frequency noise signal acquisition and signal pretreatment process, including differential amplifier circuit, band-pass filter circuit, carrier modulation circuit, envelope detection circuit to complete signal processing, and then the control chip STM32F103VET6 built-in ADC conversion module converts analog signals to digital signals. The software programming completes the selection of signal display and storage, and other functions, where Keil MDK compiler environment is used. The embedded bearing detecting equipment displays real-time waveform chart and the histogram on the LCD, and calculates the corresponding time-domain indexes including the RMS, skewness, kurtosis, the failure characteristic value by selecting an appropriate sampling frequency. Once one or several indexes of time domain is beyond the preset threshold, the system saves the data automatically and begins to analyse the spectrum. The frequency spectrum peak value will be as the slice frequency to use in cyclic bispectrum analysis. Extracting the inherent characteristic frequency of rolling bearing is to make corresponding preparations for equipment maintenance.Finally, data of the Case Western Reserve University bearing experiment center and data from the field experiments are used to verify the feasibility and effectiveness of the detecting equipment in this paper. The experimental results show that the cyclic double spectral peak frequency of the slice method can effectively suppress gaussian noise andextract the characteristic frequency of rolling bearing. In this paper, the failure of characteristic value based on variable weight histogram cumulative error is related to the characteristic value of bearing failure status. It is more accurate and stable to state the characterization of bearing failure than the commonly used effective value, variance,skewness and kurtosis characteristics.