Improvement Of Partial Intelligent Optimization Algorithm And Analsis Of Its Mathmatical Thoery

Optimization theory with its algorithm is an important branch of mathematics, it is concerned with the problem of which project is the best optima in many projects and how to find the optima solution. With the ubiquity of these problems, the importance of optimization technique is valued and thus extended to be applied to many engineering fields. Owing to the feature of complexity, constraint, nonlinearity, multi-minimum, and difficulty of modeling, to find the intelligent algorithm which suits collateral large scale problems is a main goal and eye-catching direction of research. The study of optimization theory is of great importance to improve the properties of algorithm, broaden the applied fields of algorithm, and perfect the system of algorithm as well. Therefore, the study of optimization theory and its algorithm is an important issue which bears both theoretical meaning and practical significance.This theory involves genetic algorithm, simulated annealing algorithm, ant colony algorithm, artificial neural network, particle swarm optimization, etc. and compound optimization policy, which are developed by simulating or revealing some phenomenen of nature and course. Its principle and content relate to such areas as mathematics, physics, biological evolution, artificial intelligence, neural science and statistical mechanics. And it provides innovative thought and means to solve some complicated problems.The optimum result of traditional single algorithm is often not satisfactory enough to solve a complicated problem, and the shortcomings in itself also restricts a further improvement of algorithm. Whereas, directive searching method can be used in more areas and needn't use the specific information of the problems. So constructing new algorithm by combining their merits of them reasonably and filling the gap has strong interest to the people in engineering field that attaches equal importance to real-time and optimization.This paper studies the convergence by Markov chain to describe and analyze the convergence of PSO, simplifies the model of PSO, and analyzes the convergence by the study of coefficient. The theory of ant colony algorithm is studied by the method of dynamic system. The study anastomoses to the theory by simulation, and analyzes the effect of coefficient. The paper studies simulated annealing algorithm and itsmathematical theory, uses the strong jump property in high temperature and chemiotaxis search in hypothermia to improve the PSO, and avoids falling into local optima, so the result of searching is more precise. The mutation of 'PSO' makes the mutation of GA more reasonable, and the result is fine by means of the simulation of some functions.