Absfruci-A distributed reactive power optimization method for area power system is presented in this paper. Considering the difference of network structure and the characteristics for reactive power optimization, area power system is divided into a subtransmission sub-system and distribution sub-systems according to layers and regions. So the optimization problem of large-scale area power system is decomposed into several optimization sub-problems of small-scale systems, which can reduce the difficulty of reactive power optimization problem significantly. The subtransmission sub-system and distribution sub-systems are independent relatively in power flow algorithms, objective functions and optimization algorithms. The optimization of the whole area power system is solved synthetically in the distributed computing way. Based on the proposed approach, a Client/Server software package is also introduced. The simulation results, greatIy improved voltage profiles and reduced power losses, show that this method is feasible and effective.Index Terms-reactive power optimization; distributed computing; power systems; programming I. INTRODUCTION ECENTLY, reactive power WAR) optimization has R received an ever-increasing interest because of its significant influence on secure and economic operation of power systems. Reactive power optimization aims to improve voltage profiles and reduce active power losses by regulating reactive power flow distribution. The problem has been formulated as 8 desired objective fiction, such as the total network power losses function, which is minimized while load constraints and operational constraints are met. Therefore, reactive power optimization is a mixed non-hear programming problem viewed as a NP-hard problem, which is awkward to solve.A wide variety of approaches based on traditional mathematical optimization algorithms, such as linear, nonlinear, quadratic, dynamic programmiog, Newton and interior point method [ 1 ]- [3] have been developed to solving the VAR optimization problem. One main disadvantage of these techniques is that they often trap in a local optimum rather Zhonpu Li is with the School of EIectrical EngineeriOg, Shandong
