The Application Of Particle Swarm Optimization(PSO)algorithm In Power System VAR Optimization

Particle Swarm Optimization(PSO)algorithm is one of embranchments of swarm intelligence. In the middle of 1990th. PSO Algorithm was put forword by Doctor Kennedy and Doctor Eberhart for the first time,which is based on simulating intelligent colony action. PSO Algorithm has been used on scientific research and engineering,which has already settled some practical problems successfully.From then on, its importance is discovered increasingly.PSO Algorithm is a randomization search method. main characteristic of it is the colony search game information change of the individual in the colony. The main characteristic of it is the colony search game and the information change of the individual in the colony. the search independent of the grads, robust ,astringency,and its currency is better, hasn't the differentiability; the process of bed into the optimized problem is simple. It don't need the deep realize of the problem. So it is very applied to the complicated, nonlinear problem that the tradition search method difficult to resolve. It is very fit for the computer operation. However, the shortcoming of PSO Alforithm is easily plunging into the local minmmum.Aiming at the shortcoming above, this paper put forword an improved PSO,and its result is satisfactory. There are some improvements:First, in this paper ,Simplex method is used to initialize instead of the random initialization.The new method makes the prime particle swarm has better quality,that help to get the minmmum in the whole particle swarm.Second, when the particles severity gather, the aberrance is operated to increase particles'diversity and improve search ability.These methods can make the optimum result jump out of the area of the local minmmum.Third,some parameters of basic PSO Alforithm are changed including inertia coefficient W and acceleration coefficients C1 and C2 . The maximum of iterative time and the precision of optimum result are combined to form a new finishing criterion. After the parameters are changed,the optimum result has better precision.Fourth, the coefficients of active power losses and capacitor investment are evaluated dynamically.it can wash out results including variables slopping over to quicken constringency.According to the practice of VAR optimization in Power system,a more reasonable mathematical model is set up, which which is fit for practical application, the summation of active power losses is least, voltage beyond limits and capacitor investment is the least.The mended measures are used in IEEE-6, and IEEE-30 nodes distribution network and the results indicate that these improvements are propitious to optimization of reactive power compensation.