Train Rolling Bearing Fault Diagnosis And Monitoring Systems Research

In modernized production, fault diagnosis technology for mechanical equipment is increasingly valued, and the rolling bearing is the most frequently used assembly of the mechanical equipment. Therefore, the fault diagnosis technology for rolling bearing is of great significance. By working on this subject, a variety of abnormal states or malfunction can be timely and accurately diagnosed in order to prevent or shoot the faults and consequently its dependability, safety and effectiveness can be achieved.In the research on this subject, hardware design is done in combination with the controlling capacity of SCM and the digital processing ability of DSP. The scheme of DSP+SCM makes the system capable of performing such major functions as signal acquisition, data processing, signal analysis, fault diagnosis, etc.The system in the thesis generally adapts to the collection and analysis of vibration signal from the rolling bearing and fault diagnosis. Data collection by high-speed A/D ensures both the amount of data required by data analysis and the analysis of the magnitude domain, time-domain and frequency domain of the data collected. In software design, the idea of modular design makes system maintenance, system improvement, and its function expansion easy and convenient. It can also be extended to other vibration signal collection and analysis.By utilizing ANN's favorable adaptometer, self-organization and powerful learning function, the network structure, based on Wavelet analysis, withit and Wavelet analysis combined, can be effectively determined by training sample set. The self-adaptation perpendicular minimum variance algorithm(srosl)basically eliminates the influence of relativity among samples and thus the training effects are guaranteed and the on-line training algorithm for RBF ANN is exercisable. The emulation reveals that the algorithm proposed in the thesis is effective and excellent results are achieved in nonlinear system control.