Research On Evolution Strategy With Algorithms And Applications

Recent years, the requirements for the high speed computation and the efficient of the artificial intelligence are increasing more and more. When the computer science crossover much with other science subjects, people focus on seeking one kind of algorithms with high artificial ability, such as self-organic, self-adaptive ,self-learning and so on . Evolutionary Algorithm is a kind of optimal algorithm based on searching, which simulates the process of life evolution. Due to its property of simulating evolution, it has high ability for self-organic, self-adaptive and self-learning. It can automatically gain and accumulate the knowledge about the searching space, reduce the searching space by using the connatural knowledge, control the searching process, dynamically debase the complication and get the final optimal results. Because the Evolutionary Algorithms need only few requirements about the original problem and it has high efficiency and easy usability, so it was used in many kinds of fields recently.As one of the branch of the Evolutionary Algorithms, Evolution Strategy has the representation with real value, the excellent ability to deal with complex non-linear problem, and the self-adaptive ability. It also can guarantee the multiformity of the searching domain. It is significant with the research on Evolution Strategy with the algorithms and applications, and it will help us to deal with the much more complicated problems in different fields, especially for the problems that can not be computing with the traditional optimal algorithms.In the first two parts of the thesis, it firstly gives the introduction of Evolutionary Algorithms with the basic theory, the way of sorting and the development history. Then it respectively states the theory of three branches, Genetic Algorithms, Evolution Programming, and Evolution Strategy to conclude the character of the Evolutionary Algorithms. At last, it compares the glossary of the Evolutionary Algorithms and the biology.In the third and fourth parts of the thesis, it expounds the basic theory and technology of the Evolution Strategy. They are Representation, Evaluation, Recombination, Mutation, Selection and Termination. It emphasizes the recombination part. Following the introduction of standard recombination operator, it gives tow new recombination technology—Building Block Crossover and global intermediate/weighted recombination, which are based on Building Block Hypothesis and Genetic Repair Hypothesis. At last it gives the conclusion of the characters and localizations of Evolution Strategy, by giving the comparison with Genetic Algorithms.In the last part, it states the outlook of the Evolution Strategy in both the mathematics theory and the applications ways, and analyses the exiting problems and the situation of applications. Because the Evolution Strategy is a rising subject, so the theory of it is very limited and the development mainly focus on realization, improvement and application of the algorithms. Though it still can not be used broadly now, due to its high quality of non-linear and self-adaptive, it has very excellent outlook in almost every science and engineering fields, such as complicated none-linear problems, computer science, biology, engineering, economy field, chemical plant design and society science. At last point out the direction for the development of the Evolution Strategy in future.