An Advanced Optimization Strategy Using PBPNN-MOPSO for Enhancing Hot Stamping Forming Performance of Automotive Aluminium Alloy Brackets

Aluminum alloy, as a lightweight material, is increasingly adopted in automotive manufacturing due to its substantial advantages in weight reduction and mechanical performance. However, accurately predicting mechanical behavior in the hot stamping process remains challenging, primarily because of complex physicochemical interactions and the high sensitivity of the material to process parameters. This study presents an innovative approach by integrating the Parallel Back Propagation Neural Network (PBPNN) with Multi-objective Particle Swarm Optimization (MOPSO) to construct a robust prediction and optimization framework. Specifically, the approach is employed to optimize key hot stamping parameters such as deformation temperature, deformation speed, and deformation degree for automotive aluminum alloy brackets. The neural network model’s predictive accuracy is enhanced significantly by utilizing the [Formula: see text]-means clustering algorithm to effectively partition input data and prevent convergence to local minima. Experimental validations indicate that the PBPNN-MOPSO optimized model reliably predicts yield strength under various process conditions, achieving a maximum prediction error of only 13.6%. Additionally, the optimized parameters notably improve mechanical performance consistency, illustrating the method’s applicability and efficiency in industrial practice. Overall, this research provides a reliable computational framework and effective methodological guidance, offering important theoretical insights and practical solutions for optimizing complex aluminum alloy forming processes in automotive applications.