Optimization of Production Process Parameters for Polishing Machine Tools in Crankshaft Abrasive Belt based on BP Neural Network and NSGA-II

Abstract

Abstract In order to improve the polishing quality of the connecting rod journals of crankshafts and reduce polishing time for abrasive belt polishing machines, a method for optimizing the polishing process parameters for connecting rod journals is proposed, combining BP neural network and NSGA-II algorithm. Firstly, the factors affecting the polishing quality are screened, and in consideration of practical production requirements, a five-factor four-level orthogonal experiment is designed. Then, a BP neural network is used to establish a nonlinear mapping relationship between the polishing process parameters and the polishing quality of the connecting rod journals. The predicted results from the BP neural network are used as fitness values, and the NSGA-II algorithm is employed to obtain the Pareto frontier optimal solution set and the corresponding combination of polishing process parameters. Based on the optimization results, two sets of recommended process parameter schemes are provided. Compared to the initial process parameters of the polishing machine, one group can improve the polishing quality by 0.101 \(\mu\) m and reduce the polishing time by 5 seconds, while the other group can achieve an additional polishing time of 10 seconds in exchange for an improvement in polishing quality of 0.151$\mu$m. Finally, experimental validation has been conducted for the two sets of polishing process parameters, and the results align with the optimization expectations.