THE MODELS AND METHODS FOR OPTIMIZATION OF BASIC NETWORKS POWER SYSTEMS DEVELOPMENT IN THE CONDITIONS OF MARKET RELATIONS

Abstract

The paper analyzes existing technique feasibility of design decisions on long-term development of electric power systems. Showing the potential threats to reduce the quality of the design decisions, among them – the lack of objective economic criteria of efficiency of technical solutions, the artificial restriction of the search space at the preliminary stage of the selection of possible solutions. To solve the problem of formation and selection of the optimal project plan of development of power system is proposed to use the mathematical apparatus of genetic algorithms. The main idea is to encode information about characteristics of possible design solutions in the form of binary vector – genome. Next, the application of genetic operations leads to the generation of new design solutions, the quality of which is estimated by target function. The objective function is designed to assess the quality of design solutions, such as by means of economic efficiency criteria. The basic genetic operations - crossover is performed with genotypes of parental pairs of the current set of design decisions. Priority in the formation of parental pairs are design solutions that are characterized by the highest quality assessments. To provide reliable variety of design decisions carried in the evolution of additional genetic operations – mutation and inversion. Performing these operations provide a way out of local optima, which may fall evoltsionny algorithm. The paper presents the mathematical model and method of formation and feasibility making design decisions concerning future development of electric power systems based on mathematical apparatus of genetic algorithms. The functioning of the developed mathematical tools demonstrated for solving the transportation problem nutrition consumers from nodal power plant. It is shown that the proposed approach provides a stable convergence of the search process in a reasonable amount of steps without artificial restrictions of the search space and without use of additional expert information on the feasibility of possible design solutions. The problem of setting a genetic algorithm, that is the choice of the initial population, life expectancy of possible solutions, the probability characteristics of the genetic operations, etc. is considered. To set up a genetic algorithm is proposed to use the mathematical apparatus of fuzzy logic, which provides a definition of the parameters of the algorithm on the basis of verbal estimates.