Intelligent auxiliary function method for multimodal global optimization problems

When solving multimodal global optimization problems, many auxiliary function methods are sensitive to their parameters and of great difficulty in dealing with high dimensional problems. Aiming to overcome these two disadvantages, a new intelligent auxiliary function method is proposed in this paper. Firstly, the smoothing function is employed to eliminate the solutions worse than the best one found so far. Based on the smoothing function, a novel auxiliary function is constructed, in which there is only one easily-adjusted parameter. This auxiliary function can avoid the unwilling “Mexican hat” effect caused by improper parameter settings. Then, properties of the auxiliary function are analyzed. In order to improve the searching ability, the proposed auxiliary function and intelligent optimization techniques are assembled in the designed method, which can help the method deal with high dimensional problems. Finally, 13 different benchmarks are used to test the influence of the parameter and the performance of the searching method. Experimental results indicate the effectiveness of the proposed method.