Genetic algorithm and gradient-based algorithm optimization of vehicle turning mechanism

The main function of vehicle turning mechanism is to realize the ideal relations of turn angle of the internal and external wheels when vehicles turning. At present the main methods on design computing and verifying turning mechanism have still been the planar graphing and analysis method, thus the design cycle is long and the computation precision is low, therefore it is very great significance to optimal design on the turning mechanism. Considered the boundary and performance constraints, the objective function is specified to minimize the sum of squares of relative error between actual angle and expectant angle of internal steering wheel, thus optimization model of vehicle turning mechanism is created. Global algorithms are known for their slower convergence to the true global optimum once the optimum region is found. This drawback of the genetic algorithm can be overcome by combining it with local gradient-based algorithms, which are known for their faster convergence. This hybrid approach improves the efficiency of the algorithm and also avoids the need to specify a good initial point for the derivative-based methods. The hybrid genetic algorithm and neural network method are developed in this paper, and the nonsmooth problems are solved effectively.