Multi-algorithm fusion–based intelligent decision-making method for robotic belt grinding process parameters

Robotic belt grinding is a productive method frequently used to finish intricate parts. To address the challenges of ensuring grinding efficiency and quality due to the complexity of the robotic belt polishing process, a multi-algorithm fusion–based intelligent decision-making method is proposed in this study. The broad learning system (BLS) was utilized to construct prediction models for the process parameters of robotic belt grinding. This model then served as the objective function for subsequent process optimization models. Second, to achieve a more uniform Pareto front, several strategies were implemented to improve the multi-objective gray wolf optimizer (MOGWO). Then, the best process parameters were extracted from the Pareto front using the technique for order preference by similarity to the ideal solution (TOPSIS). The proposed approach achieved a 24.24% reduction in surface roughness (Ra) and a 3.82% increase in material removal rate (MRR) compared to those of existing processing methods. These results demonstrate that this intelligent approach significantly improves the determination of process parameters for robotic belt grinding, supporting the advancement of intelligent manufacturing.