Research Gear's Fault Pattern Recognition Based On BP Neural Network

The gear is used in automobile transmission as the most important component. It is frequently responsible for transmission efficiency. Gear is the most common used and also one of the components which is the easiest to damage. Therefore, it's essential for gear carring on the fault diagnosis, and receiving more and more attention.Pattern recognition for the gear is the goal of the fault diagnosis. The characteristic extraction of signal is caning on the the foundation of pattern recognition. At present, traditional methods such as time region analysis and frequeney region analysis are mainly applied to feature extraction of gear. In order to improve the accuracy and rapidity of fault diagnosis, there are appear many advanced methods to gear's fault diagnosis, such as small wave transform method, enveloping analysis method etc.The vibration signal is the gear fault characteristic information carrier, in this thesis, it is systematically studied character extraction methods of gears' singal in this dissertation. It is used various methods such as time-domain analysis, frequency-domain analysis time and time-frequency analysis and enveloping analysis to comprehensively analysis gear fault signal, and extract features of representative fault. Simulation gear fault experiment, in extraction time-domain analysis and frequency-domain analysis obtain fault characteristic information. Pattern recognition fault is carried by using artificial neural network. The neural network have the unique structure and the process information in method, applies the gear fault diagnosis on it in analysis, to solving the gear's fault diagnosis technical question in a new method. On the Matlab platform, the PNN network and the BP neural network take as foundation, the gear's fault pattern can be recognition. Through to the datas training ample's analysis confirmation, making some improvements for the BP Neural network, through contrast, confirms that the new BP neural network in the automobile transmission gear fault diagnosis is more validity. The result indicates the effectiveness of this method since that it can improve the efficiency of diagnosis.