Reactive Power Flow Based On Improved Clone Selection Algorithm With Uniform Design

Reactive power optimization is an effective measure to enhance voltage quality,reduce power loss and improve system stability. Therefore, reactive power optimizationis studied in this paper.Reactive power optimization problem is a typical non-linear optimization problemwith characteristics of multi-objective, multi-constrained, multi-variable, and mixture ofcontinuous variable and discrete variable. Intelligent Optimization Algorithms havedrawn extensive attention due to its good performance in convergence, such as GeneticAlgorithm, Particle Swarm Optimization and Immune Algorithm. As a class ofstochastic optimization methods based on population, the information included ininitinal population has an important influence on its performance. Initial population israndomly generated, the information included in the initial population have a greatrandomness. Unsufficient information will make algorithm converge to the localoptimum with high probability. It may converge to global optimum at last, but manyiterations are required, so the convergence rate is slow. In order to have enoughinformation, the larger population size is adopted. However, the number of evaluationsof the antibody is proportional to the population size in each generation, which willconsume long time, thus the optimization has poor efficiency. In this paper, the uniformdesign principle is used to construct initial population, which distributes initialpopulation in solution space uniformly, at the same population size, more informationincluded in initial population than random distribution. Therefore, the optimizationefficiency is improved.Clone Selection Algorithm (CSA) is a new type of intelligent optimizationalgorithm, mutation operator plays a very important role on its performance. In thispaper, the mutation operator has been studied. The hierarchic mutation strategy isproposed, in which antibodies go into the appropriate levels according to the affinity.Superior antibodies in lower-level have smaller mutation range and mutation rate whileinferior antibodies in topper-level have larger, which can improve the local and globalsearch capabilities simultaneously.Vaccine inoculation can effectively improve the algorithm performance. However,the positions for vaccine inoculation are determined on probability, which has a chanceto cause waste inoculation. Thus, a vaccine inoculation strategy based on the distance is put forward, namely the positions for vaccine inoculation are determined according tothe ordering from long distance to short distance by judging the distances between thepositions of inoculated gene individuals and the positions of the vaccine, so the waste ofinoculation can be avoided.Multi-objective optimization problem has the optimal solution set, namely Pareto-optimal solution set. The conventional method aggregates multi-objective intosingle-objective, which has great limitations. In this paper, the CSA is used for solvingmulti-objective optimization problem and the following improved strategies areproposed:①Multi-population parallel search, which can optimize multiple objectivessimultaneously, so the convergence performance is improved;②Construct non-inferiorsolution set, which can reserve many non-dominated solutions searched in evolutionaryprocess, and the solutions will distribute uniformly by removing the smaller crowdeddistance solution;③n on-dominated option, which avails to search real Pareto-frontwith the better convergence rate.④Optimal compromise solution selected, the fuzzy settheory is used to calculate the solution in the final non-dominated option, which offerthe flexible options for user.Combined uniform design initial population, hierarchic mutation strategy andvaccine inoculation with distance with clone selection algorithm, the improved cloneselection algorithm based on uniform design (UICSA) is proposed. Multi-objectiveoptimization problem is sloved with UICSA, which combined with multi-populationparallel search, construct non-inferior solution set, non-dominated option and optimalcompromise solution selected, the improved multi-objective clone selection algorithmbased on uniform design (UIMCSA) is proposed. UICSA applied for optimal reactivepower flow is evaluated on IEEE14-bus system and IEEE30-bus system comparedwith CSA and GA. It is show that UICSA has better in efficiency, accuracy and stability.UIMCSA applied for multi-objective optimal reactive power flow is evaluated on IEEE14-bus system and IEEE30-bus system compared with NSGA-II. It is show thatUIMCSA has better in convergence rate, Pareto-front and uniform distribution.