Improved Genetic Algorithm of Ａdaptive Real Range Search
Zhang Minghui,Wang Shangjin
(School of Energy and Power Engineering,Xi'an Jiaotong University, Xi'an 710049,China)

Abstract: In conventional real genetic algorithm, a minimum and a maximum value for each design variable must be set before genetic operators are given. However, no any information about the minimum and maximum values has been known, and the design space is set blindfoldly and stochastically. A new type of real genetic algorithm named adaptive real range search genetic algorithm is proposed, in which a range of real numbers will move adaptively in each generation by using the mean value and the standard deviation of the previous generations. In addition, an improved Gauss mutation operator of evolutionary strategy is used in order to speed up convergence. In order to verify algorithmic rationality and validity, the improved genetic algorithm is applied to compute a multimodal function and study shape optimization of a centrifugal impeller. The results show that this method excels the conventional real genetic algorithm in the convergence and robust.
Keywords: genetic algorithm;adaptive real search;Gauss mutation operator