Integration of Topology Optimization and ANN-Based Shape Optimization for Weight Reduction of a Control Arm

This study employs an integrated approach using topology optimization, artificial neural networks (ANN) based models, and meta-heuristic algorithms to reduce the weight of a lower control arm, a component of automotive suspension. First, topology optimization of the part was performed using the finite element method, resulting in a 7.83% reduction in the weight of the lower control arm compared to the main model. Following topology optimization, the lower control arm was redesigned, six geometric design variables were created, and ANN-based proxy models were developed to estimate mass-stress values. These proxy models achieved faster and more efficient optimization than the finite element method. During the optimization process, comparisons were made with Genetic Algorithm (GA), Grey Wolf Algorithm (GWO), and Starfish Optimization Algorithm (SFOA). The results showed that meta-heuristic optimization reduced the weight of the lower control arm by approximately 11.4%, while the maximum stress increased by approximately 6%. Despite this increase in stress, the component remained within safe operating limits. Among the evaluated algorithms, SFOA produced a marginally lower weight estimate than GA and GWO under the adopted optimization settings. As a result of the study, it was shown that meta-heuristic algorithms supported by ANN-based surrogate models can be used for weight reduction in automotive sub-control arms.