Research on high accuracy gearbox fault classification method

摘要

In order to improve the intelligence of the gearbox fault diagnosis system, this paper uses BP neural network as the core algorithm to study gearbox fault diagnosis. First, the wavelet packet decomposition algorithm is used to decompose the gearbox vibration signal into 8 different frequency bands. Use the energy values of 8 frequency bands as the characteristic vector of the gearbox vibration signal, train the BP neural network, and determine the structural parameters of the BP neural network. The network is used to classify 4 types of gearbox fault signals. Each type of fault signal has 65 groups, a total of 260 groups of signals. The classification results show that the accuracy rate is above 95%.It shows that the wavelet packet algorithm is used to decompose the gearbox vibration signal, and the BP network is trained with the feature vector composed of energy values of different frequency bands. The network has good fault recognition performance.