基于多传感器融合的刀具剩余寿命预测

Here, to improve the prediction accuracy of tool wear state, a tool residual life prediction method based on multi-sensor fusion was proposed. In training phase, firstly, combining vibration, current and PLC controller information, data were preprocessed, and the time series analysis method was used to do feature extraction; then, aiming at the problem of single-frame sample lacking context information and being not able to cover the whole life cycle data, multi-frame combination and the mix-up method were used to enhance data; finally, a deep neural network was designed to learn complex nonlinear functions among multi-modal input features and tool residual life. In test phase, median filtering was used to remove effects of noise and obtain the final predicted value. The experimental results showed that the effectiveness of multi-sensor fusion is verified; using multi-modal data and introducing data enhancement can significantly improve the prediction accuracy of tool wear.