Intelligent bionic genetic algorithm (IB-GA) and its convergence
Fachao Li,Li Da Xu,Chenxia Jin,Hong Wang
DOI: https://doi.org/10.1016/j.eswa.2011.01.091
IF: 8.5
2011-01-01
Expert Systems with Applications
Abstract:As a new kind of intelligence optimization method, genetic algorithms, with the features of simple structure and strong adaptability, achieves great success in many real applications. However, it has many shortcomings such as a greater computation complexity and more chance of being trapped in local states. In this paper, through analyzing the deficiency of the existing genetic operation and the essential characteristics of creature evolution from the angle of improving evolution efficiency, we propose a compound mutation strategy based on mutation criteria function, a multi-reserved strategy based on intelligence evolution, and a weak arithmetic crossover strategy reflecting different evolution modes. Furthermore, we establish an intelligent bionic genetic algorithm with structural features (denoted by IB-GA, for short). Finally, we analyze the performances of IB-GA with the theory of Markov chains and simulation technology. The results indicate that IB-GA is essentially an extension of ordinary GA and obviously better than ordinary GA in terms of computation efficiency and convergence performance.