Reactive Power Optimization Of Power System Based On Improved Particle Swarm Algorithm

The distribution of reactive power not only determines the voltage profile, but alsoaffects the security and economy of power systems. Reactive power optimization of powersystem is an effective means of improving voltage quality and reducing network losses.Reactive power optimization is a complex non-linear problem with characteristics ofmulti-variable, multi-constrain and discreteness. Artificial intelligence algorithms based onrandom search has good adaptability in solving this problem and a greater probability ofconvergence to global optimal solution.This paper studies the reactive power optimization model and related algorithms, andreviews the advantages and disadvantages of all kinds of optimization algorithm. The ParticleSwarm Optimization Based on Bacterial Chemotaxis is proposed, in allusion to suchinsufficients as tending to fall into the local optimal and slow convergence speed. Theimproved algorithm keeps the diversity of population in the whole calculation process;convergence and the calculation accuracy have been improved. Using the improved particleswarm algorithm in power system reactive power optimization problem, simulation results onthe IEEE standard system demonstrate the feasibility and effectiveness of the algorithm.With the expansion of power networks, it is becoming more and more difficult to get asatisfying global solution for the optimization of reactive power system using centralizedoptimization approaches. Cooperative evolutionary theory is introduced in this paper to solvethis problem. The difference between coevolutionary algorithm and the traditionalevolutionary algorithm is that coevolutionary algorithm considers synergism betweenpopulation and population, and population and environment.. According to thedecomposition-coordination principle, the centralized optimization of the whole system isdecomposed into a number of interactive sub-problems. Each sub-problem is mapped into asingle evolution sub-population. The populations interact with each other and co-evolvethrough a common system model. These means promote the continuous evolution of thewhole system. The simulation results indicate that this algorithm has a lot of advantages inlarge scale system. computational complexity of the problem is reduced, and the calculationtime is shortened.