Multi-objective comprehensive optimization of riveting quality of aircraft thin-walled parts based on response surface method and entropy weight method

Riveting assembly of aircraft thin-walled components inevitably induces coupled deformation and residual stress fields. Most existing studies only optimize deformation or residual stress independently, lacking a collaborative evaluation framework for the two indicators. Meanwhile, conventional weight assignment suffers from strong subjectivity, and the interaction mechanism of process parameters is insufficiently analyzed. In this paper, rivet length, hole diameter and upset head height are selected as process variables. A geometric-physical comprehensive evaluation system consisting of maximum Z -direction deformation, average circumferential compressive residual stress around holes and uniformity of compressive residual stress is established. A novel multi-objective collaborative optimization framework integrating response surface methodology (RSM), entropy weight method (EWM) and NSGA-II algorithm is proposed. Finite element (FE) simulations are carried out based on Box-Behnken Design (BBD), and second-order response surface models for each quality indicator are constructed. Analysis of variance (ANOVA) and multi-agent model comparison are implemented to verify model accuracy. The entropy weight method is adopted to objectively calculate indicator weights, and sensitivity analysis under multiple weight schemes is conducted to eliminate deviations caused by subjective weighting. The NSGA-II algorithm is employed to solve the optimal process parameters, with convergence and robustness of the algorithm validated. Three-dimensional response surfaces are plotted to reveal the coupled mechanical laws of interactive parameters. Further validation of the optimized results is performed via repeated FE calculation and single-rivet riveting experiments. The results show that within the given ranges of material, structural dimensions and process parameters in this work, the optimized parameter combination can reduce riveting deformation while improving the distribution of compressive residual stress around rivet holes. In terms of comprehensive evaluation index, the comprehensive performance of the optimized scheme is improved by 58.1% and 23.4% compared with the single-factor optimal scheme and PSO-based scheme, respectively. It should be emphasized that the above improvement ratios reflect the relative promotion of the comprehensive evaluation index, rather than the direct enhancement of structural fatigue life or individual physical performance indicators.