Study on the process and multi-objective optimization of laser cleaning anodic oxide film on 2024 aluminum alloy

Abstract

The process parameters of laser cleaning the anodic oxide film on 2024 aluminum alloy surface are numerous and have a wide adjustable range. The selection of its optimal parameters to achieve a dual improvement in cleaning quality and efficiency has become a critical technical bottleneck restricting engineering applications. To address this problem, this study designed a Box-Behnken design (BBD) response surface experiment with laser power, travel speed, and scanning speed as input variables, and post-cleaning surface roughness, oxygen content, and self-corrosion potential as response indicators. Response surface models and a BP neural network model optimized based on the MOEA/D-DE hybrid algorithm were established, and the prediction accuracy and generalization ability of the two models were compared and analyzed. In addition, this study proposed an improved multi-objective firefly algorithm (IMO-FA) integrating Latin hypercube initialization, adaptive variable step-size detection, and a strict crowding distance selection mechanism to solve the problems of traditional algorithms being prone to local clustering, premature convergence, uneven Pareto front distribution, and boundary shrinkage. Subsequently, the entropy weight-TOPSIS method was used to determine the optimal process parameter combination from the optimal Pareto solution set: average power 184.28 W, scanning speed 2447.10 mm/s, and travel speed 2.86 mm/s. The results show that compared with the optimal specimen in the BBD experiment, the surface roughness of the optimized specimen was reduced by 5.63%, the oxygen content was reduced by 21.43%, and the self-corrosion potential was increased by 2.12%. The optimized process parameters are beneficial to the improvement of cleaning surface quality.