Lightweight design of electric box-type trucks based on deep learning models

Recently, in order to save energy, there is increasing concern on reducing the weight of electric box-type trucks.Traditional design methods implement finite element analysis (FEA) repeatedly for large truck structures, which is extremely time-consuming.To solve this, we propose a framework that combines a ResNet152-based deep learning model with particle swarm optimisation.The deep learning model predicts key structural metrics, such as weight, maximum stress, maximum displacement, and energy absorption, while particle swarm optimisation searches the design space and tries to find the optimal weight under specific mechanical constraints.We evaluate our proposed method on a generated truck frame dataset, achieving an average R 2 of 0.993.Compared with the baseline models, our method achieves the most reduction in weights and the least computational cost, proving its better reliability and efficiency than other optimisation methods.