Multi-objective Evolutionary Algorithm And Its Application In Power System

Multi-objective evolutionary algorithms (MOEA) are newly developed powerful toolsfor solving the multi-objective optimization problem, which widely exists in the practicalengineering. Application of the multi-objective evolutionary algorithms in power systemcan provide feasible schemes to power system operation and its control, therefore, the finaldecision can be made according to the different design requirements. Firstly, the basic concept of the multi-objective optimization is introduced. Anoverview of multi-objective evolutionary algorithms and their application in engineeringfields are given. Then, with the introduction of the multi-objective optimization problemexisted in the economic power dispatch and in the design of the controller, especially thedesign of the controller used for a FACTS-based stabilizer in power system, the applicationof multi-objective evolutionary algorithms in power system is summarized. Also, thepotential prospect of applying the multi-objective evolutionary algorithms in power systemis given. The dissertation consists of six main parts as follows: The first part of the dissertation presents a novel fuzzy multi-objective evolutionaryalgorithm (FMOEA). This algorithm helps in preventing the premature convergence of thetraditional evolutionary algorithm and finding the Pareto front effectively. The mainobjective of the multi-objective optimization is to find out the Pareto optimal solution.However, the problem of the premature in the rank-based MOEAs often makes it difficultto get the optimal solution. Also, the currently used weighted sum approach is difficult tofind out the Pareto optimal set when magnitude diversity for difference objectives exists. Inthe proposed FMOEA, a fuzzy evaluation factor is employed to modify the fitness functionof the traditional MOEA, also a niche and a novel elitism strategy are adopted, whichenables the Pareto optimal set to generate in the FMOEA. A number of multi-objectivefunctions are used to test the performance of the FMOEA. Test results show that theFMOEA has the capacity of effectively finding out the distributed Pareto optimal solution.In addition, the algorithm gives a very high efficiency of convergency. In the second part of the dissertation, the proposed FMOEA is used to handle theeconomic power dispatch problem as a multi-objective problem, in which*Project No.50007002; 50228707 supported by NSFC IIInon-commensurable objectives, such as the fuel cost, the pollution emission and the systemactive power losses are all taken into account. This problem is treated in the conventionalmethods in the following way. The pollution emission and the system active power lossesare taken into account as the constraint conditions. As a result, the multi-objectiveoptimization problem is converted to a single objective optimization problem with multipleconstraints. The FMOEA is used to solve this problem in this dissertation. The Pareto setfor this multi-objective economic power dispatch problem is obtained. The results ofapplying the FMOEA in such a multi-objective optimization show the effectiveness of theFMOEA in dealing with the variety of objective space. The third part of this dissertation deals with the coordination of power systemstabilizers in multi-machine power system using the FMOEA. Different indexes are used toevaluate the performance of a controller. Though simultaneously optimizing all the indexeswill provide the best performance, it is difficult even impossible in the real world. . Thispart of the dissertation demonstrates the essential issues and the selection method of theobjective function in the multi-objective controller design, through the controlleroptimization for a system with time delay. A coordinate design example of power systemstabilizers is presented to show the design procedure and the validity of the proposedalgorithms. Simulation results show that the PSSs...