Modified PSO Computational Intelligent Algorithm And The Application Of Multi-objectives Optimization

Optimization is a kind of application technology which is based on math and solves all kinds of practical problems. With the development of technology and penetration of subjects, new cross subjects come into being and new thinking style, new computation methods, especially the development of computer science and technology energize the study and development of optimization technology and supply more space of study. With the expansion of people understand and rebuild the world, people have brought forward to newer and higher request about science and technology, especially about high efficient optimization technology and computation methods. At the same time, for practical system, such as engineering field, especially artificial intelligence and control field, multi-objective, nonlinear, nondifferentiable and intermix systems come forth. Some problems which can't be solved by classic optimization methods must be solved by computation intelligence technology.The main contributions of this thesis are listed as following:1. Look back the theory and technology development history of computational intelligence and introduce its study background. Summarize traditional methods for multi-objectives optimization and methods based on evolutional computation, especially emphasize the current research progress of Genetic and PSO methods. Summarize mixed integer programming and uncertainty system optimization respectively.2. Synthesis particle swarm optimization (PSO) method. PSO is a new evolution method. It is attached importance because it has general convergence similar to Genetic method and faster convergence velocity. After introduce basal PSO method, modified PSO methods and their application fields are provided. Synthesis several modified PSO methods, compare the results with Genetic algorithm, a sample verified the efficiency of PSO. Discuss the development orientation of PSO.3. A Hybrid Particle Swarm Optimization (HPSO) Method is introduced and a Modified HPSO (MHPSO) method is proposed to cope with multi-objectivesoptimization problems. A fitness function method is also studied to solve constraints in optimization problems. It is proved to be efficiency by testing benchmark problems.4. Summarize solving methods for mixed integer programming. Introduce minimize warp method and GAMS software. A new method is proposed to solve mixed integer programming through combining minimize warp and GAMS. Prospect the development orientation of mixed integer programming.5. Synthesis current methods of ranking interval coefficients. An exampleverifies a new ranking method + rule is efficient to linear inequalityconstraints. For those problems whose objective functions have interval parameters, we choose interval coefficients and transfer objective functions from single objective to multi-objectives by decision maker's favorite high expectation and low uncertainty. An example is solved by hybrid PSO algorithm and the results verify the efficiency of the method.