Analysis of the influence of loads distribution of hydropower stations on the results of optimization algorithms

Abstract

Aiming to solve the problem: the conventional loads distribution of hydropower stations calculated by single model corresponding to multiple optimal algorithms, the way that various models corresponding to multiple optimal algorithms are proposed in the paper. According to characteristics and tasks of hydropower station, loads distribution models should be chosen reasonably. Genetic Algorithms, Simulated Annealing Algorithm, Particle Swarm Optimization Algorithm are introduced different loads distribution models respectively to calculate examples that undertaking different tasks of hydropower stations. When the Particle Swarm Optimization Algorithm is applied to calculate the minimum flow consumption model, because of its fast convergence and strong optimization ability and other characteristics, it is regarded as the best one among the three algorithms; Simulated Annealing Algorithm is in the calculation of the maximum power output model reflecting the greater advantage. BACKGROUNDThe hydropower station achieves economic scheduling, which will increase economic efficiency about 1%~3% [1] and decrease the energy consumption. The speed and accuracy of optimization algorithms calculating the loads distribution of the hydropower stations is the key to ensure economic operation of hydropower station. However, calculation speed and accuracy of optimization algorithms depends on the algorithms themselves, but also rely heavily on models. Therefore, studying the effects of loads distribution model of hydropower stations on the optimization results is significant for operating the hydropower station economically. In this respect, domestic and foreign researchers formed a series of theories gradually after decades of hard working and practicing [3][4][5][6][7][8][9][10][11] , such as nonlinear programming method, dynamic programming method. Although Nonlinear Programming and Dynamic Programming method have their own advantages, the problems should not be ignored including dimension of disaster, computation complex and slow. In addition, the type that a single model corresponds to various optimization algorithms and choosing a priority algorithm is used frequently in previous research in this area. In this paper, two basic models of loads distribution are introduced. One of the two models named minimum flow consumption model is used consuming the minimum water as the target function, and the other is the maximum power output model that generating maximum power is established for the objective function. Introducing Genetic algorithms, Simulated Annealing Algorithms and Particle Swarm Optimization in the two models respectively, and applying to examples. Analysis the results and explore loads distribution model effects on algorithms. LOADS DISTRIBUTION MODELS OF HYDROPOWER STATIONSActually minimum flow consumption model has the "electric definite water" criteria and maximum power output model is the "water definite electric" criteria. Minimum flow consumption model(MFCM) Objective functionwhere k=unit's number; Q ...