Feature Extraction Of High-speed Machine Wheel Bearings Based On Resonance Demodulation Method

Rolling bearing is a most widely used part of industrial machinery, whose working state is related to the reliability and productivity of the entire machinery working level. Therefore, researching rolling bearing fault feature extraction methods is significant for the continued efficient operation of machinery and avoiding unexpected accidents, especially in the early weak fault feature extraction methods. Taking rolling bearing as the research object, the paper focuses on the fault feature extraction based on resonance demodulation principle and carry out related research work.First, analyze the mathematical model of the rolling bearing fault signal according to the high-speed locomotive rolling bearing structure, which reveals the nature of the shock pulse and non-Gaussian. Besides, the paper also gives theoretical typical fault characteristic frequency calculation formula and analyzes the principle and limitations of resonance demodulation.Second, in order to solve the problem of the determination of resonance frequency band and the lower signal to noise ratio in fault diagnosis of bearing, the paper propose two methods combing the empirical mode decomposition algorithm with the demodulated resonance technology according to the adaptive decomposition characteristics of EMD algorithm. The simulation analysis and experiment results show that these improved methods are effective. Experimental results show that the EMD algorithm combined with the resonance demodulation method not only can effectively extract the fault feature but also highlight the fault feature and improve the SNR of the fault signal.Third, as a generalization of the spectral correlation density, the paper introduces modulation intensity distribution(MID algorithm), in the given selectivity factor ?f conditions,which return a relationship between carrier signal frequency f and modulating signal frequency α exhibited in the three dimensional figure. Combination slice analysis of MID(C-SMID) has better anti-noise property and less amount of calculation according to the mathematical derivation, the simulation analysis and experimental demonstration. Above features satisfy the industrial real-time and provides a feasible idea for actual industrial production.Finally, to make real-time monitoring diagnosis of rolling bearings, the paper design voltage-mode resonance demodulation circuit, Current-mode resonance demodulation circuit, relative power supply circuit and Anti-jamming circuit respectively according to the resonance demodulation principle. Analysis transmission characteristics and device parameters’ affect for the voltage-mode circuit, the testing experiment results by the rolling bearing fault simulation experiment platform QPZZ-II show that the voltage-mode circuit has a good selectivity and can extract fault effectively. The paper propose a novel circuit--- current-mode resonance demodulation circuit, under microampere level, high-frequency signals, which can also extract shock pulse fault from the strong background noise effectively. The current-mode circuit provides a novel idea for real-time monitoring diagnosis of rolling bearings and a practical Signal conditioning circuit for sensor design.