Evolutionary Strategies For Solving Dynamic Optimization Problems

There are many Dynamic Optimization Problems (DOPs) in real-word applications. Studying and solving these DOPs have realistic significance. Recently, Evolutionary Algorithms (EAs) have been widely used to solve DOPs. Therefore, studying and devising efficient evolutionary strategies for DOPs are very important.Based on the existent efficient evolutionary strategies, this thesis does study from three aspects as follows.(1) Devising a novel updating strategy for the associative memory scheme. According to the relationship of the environmental information and the individual, this thesis proposes the environmental information based updating strategy. This strategy is tested in different environments on three dynamic test functions. The experimental results demonstrate that the proposed updating strategy can improve the performance of the associative memory based algorithm significantly.(2) Proposing two memory retrieving strategies, i.e. the survivability-based retrieving strategy and the diversity-based retrieving strategy, for the associative memory scheme. In these two strategies, the environmental information is retrived according to the survivability and the diversity, respectively. In experiments, these two strategies are compared with a state-of-the-art retrieving strategy. The experimental results demonstrate that these two strategies are both efficient for DOPs.(3) An adaptive hybrid immigrants scheme is proposed by combining the memory scheme, the elitism scheme and the random immigrants scheme together adaptively. In this scheme, three kinds of immigrants are created. Furthermore, the number of these three kinds immigrants is adjusted adaptively according to their contribution to the search ability. The experimental results demonstrate that this scheme can improve the environmental adaptability for EAs.With the studies in the dynamic optimization, the environmental information based updating strategy, the survivability-based retrieving strategy, the diversity-based retrieving strategy and the adaptive hybrid immigrants scheme are proposed in this paper. These works are very important for the research and the application of the evolutionary strategies to solve DOPs.