Optimization of complex surface milling parameters based on HSS-MFM and OBL-NSGA-II

Complex curved parts are widely used in some critical engineering equipment with high service performance, but machining such parts in computer numerical control (CNC) often has a series of problems such as unsatisfactory surface quality, high energy consumption, etc. Therefore, it is very important to optimize the milling parameters to improve the surface quality and reduce the cost. In this study, three main milling parameters were first extracted by physical tests and ABAQUS simulation tests, and two sets of sample data with different accuracies were obtained. A milling process parameter model was developed using Hybrid Stacked Scaling Function with Multi-Fidelity Metamodeling (HSS-MFM), which fitted the sample data obtained from both physical and simulation tests, with the maximum relative error of the fit not exceeding 10%. Secondly, the traditional optimization algorithm non dominated sorting genetic algorithm –II (NSGA-II) was improved, and the opposition-based learning (OBL) was introduced into the population update process to improve the search performance of the algorithm. The proposed OBL-NSGA-II is compared with three algorithms in 13 standard test cases, and the results show that the improved algorithm enhances its optimization performance. Finally, OBL-NSGA-II is used to optimize the milling model, and the optimal combination of milling parameters is finally obtained. The results not only improved the machining quality effectively, but also reduced the energy consumption, and improved the manufacturing level of complex curved surface parts, which provided a certain theoretical basis for the machining personnel.