Accelerated strategies of evolutionary algorithms for optimization problem and their applications

Global optimization is important for solving practical optimization problems especially in engineering. Generally the problems encountered are very different in structure and have very different features. In trying to solve them a big variety of ideas and algorithms have been proposed. Among the rest, evolutionary algorithms (EAs) represent a powerful search and optimization paradigm. EAs have been applied in a vast number of application areas. In some cases, they have advantages over existing computerized techniques. However, the efficiencies and reliability of EAs for solving the complex and multimodal optimization problems cannot satisfy the practical requirements at present. Currently, it is an important research area for EAs.; This thesis has studied in depth such adaptations. Through analysis of the reasons leading to the low efficiencies of EAs for solving the complex and multimodal optimization problems, we have proposed a sequence of accelerated strategies for the different optimization problems. They are the exclusion-based selection operators for the global optimization problems, the Fourier series auxiliary function for the difficult optimization problems, and the adaptive elitist-population search techniques for the multimodal optimization problems. We believe these are all valuable approaches to improve EAs for the optimization problems.; The mainstream instances of EAs' models traditionally consist of genetic algorithms (GA), genetic programming (GP), evolution strategies (ES) and evolutionary programming (EP). The standard GA (SGA) evolves a population of the binary strings' chromosomes (candidate solutions). For reducing a redundancy of the standard binary coding and a resampling of the SGA's search process, a splicing/decomposable binary coding and exclusion-based selection operators have been proposed. Incorporating the strategies with SGA yields an accelerated genetic algorithm—the fast-GA. The empirical results have demonstrated that the fast-GA is more suitable for solving complex global optimization problems. (Abstract shortened by UMI.)...