| Summary: | 碩士 === 國立中興大學 === 資訊科學與工程學系所 === 101 === Since Genetic Algorithms (GAs) have become very important in recent decade, there raises an interesting research area to enhance the performance of GAs when executing evolutionary process. Among various enhancements of GAs, there is a lack of enhancing initialized population and mutation operation. To this end, this thesis tends to provide a solution to improve GAs via enhancing both initialized population and mutation operation. Moreover, such solution is applied to train TSK-type neural fuzzy networks. The proposed solution named Two-Strategy performance based Search base Genetic Algorithm (TSPS-GA) is mainly used to generate well-performed population and execute mutation efficiently by searching and keeping well-performed chromosomes. Moreover, it can also consider both depth and breadth solution to prevent converge quickly. In other words, the TSPS-GA can benefit to not only provide good initial population but also provide efficiency mutation to improve evolutionary process. As shown in the experimental results, the proposed TSPS-GA can obtain excellent results than other well-know enhancement Genetic Algorithms.
|
|---|
