Research On Distribution Network Reactive Power Optimization Based On Improved Cloud Particle Swarm Optimization

Reasonable distribution of reactive power optimization plays a vital role in improvingpower quality and reducing the network loss. With the constant improvement of the nationaleconomy and people’s living standard, whether Grid can provide users with reliable and safetransmission or not has become the hot topic in electric power industry, one of the effectivemeasure is reactive power optimization. By adjusting the control variables, the power griddistribution can be reasonable allocation, reduce network losses and improve voltage qualityconditions. Classical optimization algorithms must ensure that the accurate mathematicalmodel, such as linear programming method and nonlinear programming method. However,under the condition of the population size is larger, in the late search, the cloud particle swarmis losting species diversity, so the algorithm can’t converge to global optimal. Therefore, inorder to optimize power grids, improving the cloud particle swarm optimization algorithm isof great significance.Firstly, the research background and significance of the reactive power optimization ofthis thesis is introduced, the reactive power relations with voltage distribution network and thenetwork loss and common reactive power control equipment are analyzed. Reactive poweroptimization mathematical model with active network loss minimum is established, the fastdecoupled algorithm is chosen to flow analysis. What’s more, in the late search, the cloudparticle swarm is losting species diversity, so the algorithm can’t converge to global optimal.The improvement measures are put forward, using the cloud genetic crossover operator of thegenetic algorithm to let cloud particles cross and join the chaos immigration operator search.Using cloud genetic crossover operator of genetic algorithm to lose this part of the diversityof particles and has reached the convergence of particles were crossed, keep cross after theparticle has reached to the global optimal particle characteristics, after introducing chaosimmigration, let cloud particles with chaotic search after cross. Finally, through theoptimization of multi-peak value function instance, verify the validity and reliability of theimproved algorithm, the search precision and speed of the improved algorithm, increased thediversity of population. The simulation results show that the improved algorithm is effective.In order to verify the improved algorithms can achieve the desired effect in the simplerand more complex power distribution network, the IEEE-30nodes and the IEEE-57node arerespectively adopted to make reactive power optimization calculation, reactive poweroptimization program is written in MATLAB language. The simulation results are analyzed,the results shows that the improvement has been made in the search speed and the precisionand reach reactive power optimization of the expected requirements, which compared withbasic particle swarm and cloud particle swarm algorithms in reactive power optimization calculation.