Research On Reconfiguration Technology For Optimal Operation Of Power Distribution Network

Power distribution network reconfiguration(PDNR)is a very important method of optimizing the operation of power distribution system.The power load can be transferred between the feeders or the substations through switching operation,so as to improve the operation index and realize the economic operation of the distribution system,thus it is widely used in the actual electricity dispatching.Moreover,the gradual emergence of smart distribution system with growing penetration of distributed generators(DGs)and modern FACTS devices,availability of new meta-heuristic optimization techniques and growing stochastic environment have brought challenges and opportunities to the power distribution network reconfiguration methods.The methods have evolved from the early single objective,computationally slow and mainly heuristic algorithms into the present day modern and multi-objective stochastic methods equipped with super fast simulators and latest visualizing tools.However,no matter how the structure and composition of distribution system changes,the basis of the PDNR is the traditional static reconfiguration problem.A real-time static reconfiguration method that can dynamically monitor the change of network structure in multi-scene is the foundation for solving dynamic reconfiguration and stochastic model reconfiguration.According to the analysis of current research status,two fundamental difficulties of PDNR methods are summarized as the extremely large combinatorial solution space and the requirement of a fast loss estimation technique for feasible solution.Considering the related studies in graph theory,the further research is made to find the better solution.The innovations and contributions of this dissertation are listed as follows.1.The connection between "mesh" planning mode and "tree" operation mode of distribution system is explored by applying graph theory.Besides,three kinds of mutual conversion of "mesh" and "tree" are put forward,which provide the support of mathematical theory for the follow-up study.2.The topological constraints of distribution system generating spanning trees from graph are analyzed,a new coding method of spanning tree is proposed based on Dandelion coding,for Dandelion coding can generate solution space without infeasible solution quickly and accurately.Then the size of solution space is futher reduced by the way of prejudging the certainly closed swtiches and looking for optimal solution in a subset.The optimal model of parallel PDNR is established,setting the minimum power loss as objective and taking the power flow equation,node voltage constraint,branch flow constraint and topology constraint into account.Finally,parallel computing is used to find the global optimal solution for the power flow calculation and the network loss calculation in feasible solution space in the simulation of the IEEE 33-bus case,the efficiency of the reconfiguration model is improved under the premise of determining the global optimal solution.3.In order to compress the extremely large combinatorial solution space of PDNR,adapt to power grid structure changes in actual conditions and satisfy the needs of dynamic reconfiguration,an algorithm of dynamic shortest loop-partitioning combining the advantages of feasible tree-bases mode and loop mode is put forward.The distribution network is firstly transmuted into weighted graph,then,the optimal feasible tree-base is found through minimum spanning tree of weighted graph,and the shortest loop is further determined with applying the shortest path algorithm.A fast method to judge the topology radial property for a reconfiguration solution based on shortest loop encoding mode and connected graph theory is proposed.A new Meta-heuristics Improved Fireworks Algorithm(IFWA)is presented to optimize the radial distribution network while satisfying the operating constraints.The shortest loop encoding,the improved explosion operator and the mutation operation are introduced to adapt to discrete combination problem and improve the search success ratio and efficiency.The proposed method has shown advantages over other methods in the simulation of the PG&E 69-bus case.