Application of Multilayer Perceptron Optimized by Wavelet Packet and Laplace Feature in Early Warning of Hob Failure

Aiming at the problem that it is difficult to accurately predict the remaining life of the hob in the hob process, a hob failure warning method is proposed, which is combined wavelet packet and Laplace features to optimize the multi-layer perceptron. Firstly, the spindle signal in the process of hob machining is collected and the wavelet packet is used to reduce noise. Secondly, the time domain, frequency domain and wavelet energy characteristics of the signal are extracted after noise reduction. Then, the Laplace matrix is used to optimize the feature set. Finally, the optimized feature vector is input into the 3-layer perceptron for training, and the trained model is used for hob failure warning. Compared with the traditional method, this method is time-consuming and has a high warning accuracy rate. The experimental results show that the traditional method takes 6.703s and the prediction accuracy rate is 93.2
^%
; this method takes 1.101s and the prediction accuracy rate is 98.0%, which is 4.8% higher than the traditional prediction method; the time-consuming is the time-consuming traditional method 16.4% of the total.