Complex profile optimization of marine diesel engine piston pin bore using hybrid GA-BP neural network and NSGA-II algorithm

Abstract

To address deformation mismatch and stress concentration in the pin holes of a steel-topped aluminum-skirted combined piston under service conditions, this study proposes a surface optimization methodology integrating axial and circumferential bore profiles. By constructing a genetic algorithm-optimized backpropagation neural network surrogate model combined with the NSGA-II multi-objective optimization algorithm and CRITIC weighting decision mechanism, this approach achieves multi-parameter collaborative optimization for the pin hole's intricate geometric configuration. Results demonstrate that compared to the original design, the optimized complex surface reduces peak contact pressure by 66.7 %, decreases equivalent stress by 52.0 %, and lowers equivalent stress at bolt counterbores by 44.1 %. Relative to axial profile-only optimization, the contact pressure is further reduced by 12.4 %. The proposed method effectively resolves stress inhomogeneity induced by elliptical deformation, with finite element simulations verifying that axial-circumferential collaborative optimization significantly enhances load distribution uniformity and fatigue resistance. This work provides a systematic algorithmic approach for high-reliability piston design, advancing the application of intelligent optimization techniques in engine component engineering.