Research Of The Method Of Fault Model Identification Based On The Neural Network

Gears is the most common transmission and easy to be damaged. The technology condition of gears has direct effects on the motion of the mechanize system. Therefore, the research of the gears fault has economic and society efficiency for decreasing the maintenance expresses of the plant and raising the quality of the product.ANN has strong parallel running, fault-tolerant and self-learning' capacity. ANN can finish the gain of knowledge by the model sampling and . memory the knowledge into the weights of topological structure. It is not like expert system, which needs repeat three periods-conceptualization, formalization and improvement.In this paper, research the fault gears' three models by the vibration signal. Pick up the character vector by using the wavelet packet and energy method as the ANN input neural vector. The ANN use BP algorithm with training by teacher. This method need not pick obviously up character from the system model structure. The second layer neural has the function of picking up character. The network is not sensitive to the uncompleted-features and has fault-tolerant. The experiment testified this approach could set effectively apart the fault models.In the process of analyzing data, combine the advantage of VB and MATLAB and exploit the interface program to convenience the users analyze the signal.