Multi-Step-Ahead Prediction of Chatter State in Cold Rolling Process based on FDA-GAM

Mill chatter seriously restricts the improvement of production efficiency and the development of new products. Thus, a method of cold rolling chatter monitoring and early warning based on the combination of Functional Data Analysis (FDA) and General Autoregression Model (GAM) is proposed. Firstly, the multi-source heterogeneous cold rolling data were preprocessed by FDA. Then, the sample space with 15 process parameters as input feature and vibration energy as output feature was constructed based on chatter mechanism. Finally, the multi-step-ahead predictions of cold rolling chatter were realized through five classical machine learning algorithms in GAM under different working conditions and the optimal selection of algorithms was completed based on the maximum prediction step. The results showed that the method of FDA-GAM provided an effective solution to delays, false alarm and distortion in chatter prediction and improved the rapid response ability when chatter occurs.