A Study On Distribution Network Planning And Optimal Operation Based On Intelligence Optimization Algorithms

With the rapid development of national economy and the improvement ofpeople's living conditions, reliability and power quality play a more important role inpower supply in distribution system. How to promote current power distributionnetworks potentiality by configuration optimization and optimal operation whileguaranteeing its security and reliability and how to pursue scientific and long termdistribution network planning (DNP) are urgent problems for distribution powerplanners and operators. The complexity of DNP and distribution network optimaloperation (DNOO) lies in its large-scale, uncertain, multi-factor and NP hardcharacteristics. Mere experience of planners and operators cannot meet therequirements of modern distribution networks.In this paper, current research of DNP and DNOO is introduced;the searchefficiency and convergence property of several emerging intelligent optimizationmethods are analyzed and several hybrid intelligent optimization algorithms are putforward to compensate the deficiency of single optimization algorithms. Moreover,those proposed algorithms are employed to solve such problems as the substationlocating and sizing (SLS) and general tie-line planning (GTP), as well as distributionnetwork configuration (DNC) and reactive power optimization (RPO) in networkoperation. These algorithms can significantly improve work efficiency of planners andoperators. In sum, the work in this paper is as follows:1. The algorithm of Adaptive Mutation Particle Swarm Optimization (AMPSO)is presented to overcome the prematurity of Particle Swarm Optimization (PSO).According to Genetic Algorithm (GA), AMPSO observes the "clustering" of particleswarms by the population fitness's variance and conducts mutation operation onclustering particles. AMPSO indicates its global optimization searching ability byseveral classic mathematic function optimization problems. The method is used todeal with SLS problems in DNP and the results demonstrate its feasibility andreasonableness. Besides, in the SLS model, not only the influence of geographicinformation on substation construction costs is considered to avoid unacceptable areassuch as lakes and buildings;but also the effect of substation equipments costs is takeninto account. The model incorporates geographic and electric information, and SLSresults are very helpful for distribution network planners.2. Ant Colony Optimization (ACO) is employed to solve the optimization oftie-lines between substations in different connection modes under certain load densityand load transfer ratio of substations, with such heuristic factors as distances betweensubstations, surplus capacity and information along the routes. At the same time, fastand efficient PSO is used to tackle load transfer ratio of substations, an optimizationproblem to balance the economical efficiency and reliability, with the minimummarginal power supply cost as the objective function. Along with the tie-line planningwith ACO, the general planning result of substations in power distribution network isfinally formed.3. Refined Tabu Search (RTS) and Hybrid GA and PSO (HGAPSO) are proposedrespectively to solve power distribution network reconfiguration and timely dynamicreconfiguration. HGAPSO incorporates the merits of GA and PSO, whichcompensates GA's time consuming and PSO's local optimum problems. Thus itssearching speed and convergence property are improved greatly compared with eithersingle algorithm. Tabu Search itself has strong local searching ability;however, howto find the optimum solution always appeals to researchers' attention. According tothe specific characteristics of power distribution network, convergence conditions ofTabu Search are analyzed and efficient genetic coding strategy is employed to verifytheoretically that the method can certainly find the global optimum solution in DNC.The above two methods are proved to be correct and effective by results analysis onthree typical IEEE testing examples.4. Chaotic Optimization Algorithm (COA) and Chaotic Particle SwarmOptimization (CPSO) are addressed respectively to solve reactive power optimization.In consideration of the ergodicity and randomness of the chaotic system, the optimalsolution is figured out directly by transforming the chaotic variables into optimizingvariables in COA and the proposed method shows good performance to avoid localoptimum. PSO is good at dealing with large-scale integer programming problems,while it is apt to fall into local optimum. Overlapping particles which gradually losetheir searching ability in PSO are imposed on distinct ergodicity in COA, therefore,the presented CPSO algorithm combines the ergodicity in COA and the rapidity ofPSO in optima finding. CPSO is tested by two IEEE distribution network examplesand the results demonstrate its high efficiency and promising practical applications.