Reactive Power Optimization In Power System Based On CAGPSO And ISCO Algorithms

The reactive power optimization in power system refers to considering theparameters and loads of the network structure, makes the full use of the power systemreactive power, improves the voltage quality and reduces network losses by adjustingthe control variables such as generator’s terminal voltage, OLTC’s (On-Load TapChanger) transformer ratio and reactive power compensation device’s stalls. Thetraditional reactive power optimization algorithms have significant limitations and theyare improper in dealing with the discrete variables, so there are many intelligentalgorithms in recent years. The PSO algorithm has high convergence speed and is easyto implement, but it also exists ‘precocious phenomenon’, and in the later stage ofoptimization, the improvement is not good and is easy to be trapped in local minima.This paper researches the model and relative solutions of reactive poweroptimization and Newton-Laphson iteration is used as the calculation method of powerflow. Seeing minimum network loss as the objective function, We establish the singleobjective function, and the mathematical model of multi-objective reactive poweroptimization in this paper considers three aspects, which are net loss, voltage deviationand static voltage stable margin. Cloud model is firstly introduced into PSO in thisarticle and the particles are divided into two parts, close to or away from the bestparticle, the former particles’ weight of inertia will be adaptively adjusted byX-condition generator of cloud model. Considering the gradient theory, the articleproposes the cloud adaptive gradient particle swarm optimization (CAGPSO).We applyit into single objective reactive power optimization for power system with dynamicallyadjusting the penalty function. This paper improves the SCO algorithm with shrinkingsearch in the Simulating Fisher fishing Optimization algorithm (SFOA) and proposesthe Improved Social Cognitive Optimization (ISCO) algorithm, then applies it intomulti-objective reactive power optimization of power system. To meet the users’demand as much as possible, weight distribution according to individual preference is applied to deal with the weight processing problem in multi-objective reactive poweroptimization.We make the simulations in standard IEEE14and IEEE30node system byMATLAB7.0. The results show that the mathematical model and algorithms in thispaper are feasible, effective and economical.