Abstract
本文提出以遺傳演算法發展控制策略以解決最佳化控制問題。遺傳演算法是一種以目標為導向的平行搜尋技術，可用在許多最佳化問題上以尋找全域或近似全域之極值。本文提出一具有粗調及微調功能之新方式以修正遺傳演算法之運算而可模擬人類調整(tuning)方式及節省搜尋時間。於最佳化控制問題上，其控制量可被視為一個染色體。在本方法中染色體於演化過程中逐漸被切割成更細之段落以及更細的基因解析度。本文藉由一個二維最佳化控制問題的電腦模擬結果驗證此新的遺傳演算法之可行性及有效性。A strategy based on genetic algorithm is presented in thisstudy to solve optimal control problems. Genetic algorithm isa parallel goal-oriented search technique for optimization andcan be used to easily find out the global or nearly globaloptima for optimization problems. A new genetic algorithmembedding the ideas of coarse and fine tuning is presented inthis study to save the search time. These ideas areimplemented by a chromosome with variable numbers of sectionsand variable resolutions of gene. The control strategy can bedetermined through the new genetic algorithm with the controlaction being considered as the chromosome. Numerical results ofoptimal control problems for a second order nonlinear plantshow the feasibility and effectiveness of this proposed methodfor optimal control problems.