Research On Optimization Of Controllers Using Distribution Population Based Genetic Algorithms

As a new optimization method, GA was widely used in the optimizations of many fields owing to the features of simplicity, easily handing and parallel processing. However GA theory is not prefect,such as there exist the problems of easily creating earliness and bad ability in local optima,etc. Enlightened by distribution of creature living in natural ecology environment, the Distribution Population based Genetic Algorithm (DPGA) is proposed in this paper. DPGA is applied to optimize the controller parameters for single variable systems and multi-variable systems. The simulation tests are made and the results demonstrate the efficiency of the proposed method. The main content of this thesis includes the following:1. A survey of the origin and the development status of genetic algorithm is summarized and the status of the parameter optimization of controllers is introduced. Also the principles of GA are introduced and the problems for further study on GA are given.2. Enlightened by distribution of creature living in natural ecology environment, the Distribution Population based Genetic Algorithm is proposed. Then DPGA is applied to obtain the optimal parameters of PID controllers.3. DPGA is used to get the optimal parameters of the neuron PID controller and the neuro-fuzzy controller. Simulation experiments results of controlling the hydraulic turbine generator and the cutting process show that the better performances of the controllers are reached.4. The multivariable neuron-PID controller based on DPGA is designed for the plants of the unit power plant and the distillation towers with multiple side-streams. The simulation tests of controlling a unit power plant and a distillation process are made and the results demonstrate the efficiency of the optimal multivariable controller.