Abstract
Bat algorithm is a new intelligent optimization algorithm that is simple and easy to implement. But bat algorithm is easy to fall into local optimum and will appear premature convergence to lead to poor convergence precision. The elite multi-parent hybrid optimization algorithm is better than other optimization algorithms when solving complex function optimization problems. However, the algorithm consists of hybrid operation without mutation so as not to keep the diversity of population in the search process. Combining bat algorithm with elite multi-parent evolutionary optimization algorithm, the improved elite multi-parent hybrid optimization algorithm optimizing hybrid discrete variables was proposed. In this algorithm, firstly the rough optimization is carried out by bat algorithm, and then the accurate optimization is implemented by the elite multi-parent hybrid optimization algorithm. This kind of algorithm takes advantage of two algorithms and overcomes their shortcomings. The procedure as DIEMPCOA1.0 is to optimum design for three-shaft four-speed automobile gearbox with 20 design variables, 50 inequality constraints and eight equations. Optimization example shows that this algorithm has characteristics of no special requirements for the optimization design problems, better universality, reliable operation, higher calculation efficiency and stronger global convergence ability, so as to shorten the design cycle, reduce quality, reduce cost and improve quality.