Improved Genetic Algorithm Based Niches For Power Network Planning

As an important guarantee for power industry to fulfill rapid, stable and sustainable development, power network planning is a strategic research. Transregional power network interconnection and the reforming of power market is the background for power network development in China. With the realization of an open market and the network interconnection in the whole country, the frequent exchange of regional power gives birth to a lot of uncertain factors. Power network planning is a huge and complicated matter concerning optimization of combination. The traditional methods of optimization can not meet the development of power network any more because of the enlarging scale of network and the getting more complicated structure. Therefore, confronted with constantly emerging new problems, It, the effort in exploring a new algorithm methods applied to power network planning through absorbing the latest achievement from the development of optimization methods, and by utilizing its characteristics, has its significance both theoretically and realistically.Genetic algorithm is a random search and optimization methods based on natural selection and genetic mechanism of the living beings. Recently, it has attracted a wide range of attention from abroad and at home for its huge potential in solving the complex optimization and its successful application in Industrial engineering field. At the same time, the genetic algorithm, itself, is developing continuously and many new improved genetic algorithms have been proposed. However, the early convergence of the genetic algorithm is still the main obstacle that confines its application. The basic ideas of niche technique derived from the biologic evolution phenomenon that the livings always tend to live together with the same species. That, applied to genetic algorithm, makes individuals evolve in a specific circumstance. By adopting the niche technique the improved genetic algorithm can avoid the reproducing of individuals with high adaptability in the late stage of evolution, therefore, can effectively prevent them from early convergence.This paper briefly reviews the general situation of research on power network planning; introduces background that power network are now facing; expatiates the significance of network structurepoints in power network planning; points out the characteristics of power network planning and its existing problems. Then, the paper proposes a new algorithmic method in power network planning i.e. genetic algorithm, by illustrating its main features and principles, and generalizing its application and existent shortages in power network planning.On the basis of introducing the fundamental theory of direct current model, the paper explains the certain safety analysis methods in power network planning and introduces the basic conception and methods of economy analysis in power network planning. Through analyzing the factors that influence power network planning, this paper builds up unitary mathematic optimization model, which also taking a lot of restrictions in to consideration.The paper analyzes in depth the reasons for the early convergence existing in the standard genetic algorithm. Based on this analysis, this paper introduces the conception of niche genetic algorithm, and states briefly the implement of this niche genetic algorithm. The paper designsalgorithm structure based on fitness sharing function of niche genetic algorithm. Regarding the characteristics of power network planning, the paper makes a discussion on defining the fitness sharing function. In order to improve the convergent capability of the algorithm, the paper introduced the duality mutation and the mechanism of competition between parent and filial generation. The example of Garve-6 is simulated in the paper. The parameters called Population diversity and Population sharing factor is introduced to describe the variation of the selection probability of and individual and the sum of the selection probability of the individuals of a niche under fitness proportionate selection scheme. By applying this algorithm, the power network planning in a certain region of central China is calculated.The result of this example shows that the algorithm proposed in paper enjoys a high superiority in the performance of the global convergence and rapid searching. The introduction of the niches explores a new broad view for the application of genetic algorithm in power network planning.