Multi-population genetic algorithm and improved accelerated particle swarm optimization-based robust optimization design method for the steering trapezoidal mechanism of vehicles

For the design of the steering trapezoidal mechanism of vehicles, various uncertainties must be considered to mitigate potential risks. These uncertainties pose challenges for metaheuristic algorithms to find the global optimum. To enhance the efficiency of robust design, we propose a novel hybrid method called Multi-population Genetic Algorithm and Improved Accelerated Particle Swarm Optimization (MPGA-IAPSO), which is a robust optimization method combined with deterministic design. Firstly, kinematic analysis of the steering trapezoidal mechanism is conducted and an optimization model is constructed. Then global sensitivity analysis using the Sobol method is introduced. An Improved Accelerated Particle Swarm Optimization (IAPSO) incorporating an adaptive stochastic strategy is proposed, which can address complex nonlinear constrained optimization problems. Subsequently, the MPGA-IAPSO method is developed to achieve an accurate solution. This method utilizes the “few individuals, many populations” strategy to search the solution space, introduces an elite initialization particle swarm and migration strategy to provide high-quality initial particle swarm and candidate solutions for robust optimization. Simulation results show that the optimized steering trapezoid is closer to the target steering relationship. Via this method, the standard deviation of the weighted absolute error decreases by 22.95% compared to deterministic optimization, and the quality of the maximum rack travel achieves the 8 sigma level. Therefore, the proposed method enhances the efficiency of robust design optimization, reduces computational cost, and holds practical significance in engineering applications.