| It is acknowledged vastly that reactive power is an important variable related with the security and stability of power grid. The reactive power optimization aims at cutting down electrical power losses and improving voltage quality, through changing the position and capacity of reactive power compensation equipment, to ensure the system run in security, stability, high quality and economically. In general, Power distribution network is directly correlated with consumers who is concerned highly about the quality and reliability of electric energy. Hence, more and more studies on reactive power optimization come into our sight.Artificial fish-swarm algorithm is improved in this paper according to the characteristics of reactive power optimization of the distribution network. First, According to the shortcoming of artificial fish-swarm algorithm, and improve the algorithm. In order to guarantee the global optimal solution will be founded, take the influence of global optimal position to the artificial fish-swarm algorithm. Artificial fish-swarm algorithm is consisted of four fundamental conduct which is called as foraging behavior, cluster behavior, collision behavior and random behavior. When conduct the iterative calculation based on the behaviors above, we will find that the rate of convergence becomes slower in the later period. So we take eating behavior and feedback behavior into account.Next, considering the limitations of a single intelligent algorithm, two kinds of hybrid intelligent algorithm is presented, namely Chaotic Artificial Fish-Swarm Algorithm(CAFSA) and Genetic Artificial Fish-Swarm Algorithm(GAFSA). Chaotic Artificial Fish-Swarm Algorithm is known for the natures of ergodic by adding chaotic search in the basis of AFSA, in order to overcome the shortcomings that original version is inclined to involve local minimum value only. Results show that the rate of convergence and optimization are much better compared with before. Moreover, considering that the rate and precision of convergence may decline dramatically as the non-global extreme value point is distributed intensely, crossover and mutation mechanism of Genetic Algorithm is applied to Artificial Fish-Swarm Algorithm. It is marked that the artificial fish individuals turn into diversity. Consequently, the time spent for convergence get shorter, and global effect is improved at the same time.In the end, the three kinds of algorithm are applied in reactive power optimization of distribution network. Establish the objective function based on the smallest system network power loss and the exceed node voltage. At the same time, take the capacity of the reactive power compensation device and transformer adjusted to the mathematical model. The computer programs for reactive power optimization is completed according to these hybrid intelligent algorithm proposed. It is verified to be effective and practical when the programs are applied to IEEE33 network. Results show that hybrid algorithm in the performance, number of iterations and accuracy are better than the artificial fish-swarm algorithm. Hybrid intelligent algorithm can provide a more economical, high quality’s solution about reactive power optimization of distribution network. |
