Research on the end mill life prediction method based on feature fusion

In modern manufacturing, end mill wear monitoring is essential to ensure machining quality and enhance production efficiency. Traditional monitoring methods usually rely on a single source of information. However, they may face the problem of incomplete information in complex environment. Therefore, effective integration of multi-source information to accurately predict the life of end mills has become the focus of research. In this paper, we propose a method of multi-source information acquisition and processing based on the time window. Aiming at the limitations of traditional image processing methods. The image preprocessing technology based on CLAHE histogram equalization and Sobel operator is used to accurately extract the tool tip wear zone. Based on Pearson correlation analysis, the sensor signal features are fused by constructing polygonal radar chart. The fusion feature space polyhedron is introduced to fuse the multi-source wear information features with equal weight. The tool wear process is characterized by volume change of the polyhedron. In view of the temporal relationship of the tool wear process, considering the hyperparameter optimization of the model, the SCSBiLSTMSE neural network model is constructed. Experiments show that the model performs superiorly in predicting the remaining useful life of end mills. On this basis, the model is corrected through experimental data to further increase the accuracy of the prediction. This research provides a new solution for the acquisition, processing, fusion and life prediction of multi-source wear information of end mills, which has important theoretical significance and practical application value.