With the gradually exhaustion of traditional fossil energy, the increasing global concerns of environmental protection and the rapid development of the electric power science and engineering, the distributed generation (DG) technology based on renewable energy sources such as photovoltaic power and wind power has become a hot research topic in recent years. The penetration of DG in power systems is also increasing continuously. The active distribution network (ADN) is formed by integrating DG to traditional distribution network, supplemented with active, flexible and intelligent control strategies. The integration of DG changes the characteristics of power flow and voltage profiles in traditional distribution networks, and the stochastic and fluctuation characteristics of renewable energy bring uncertainties to distribution network operation and control. With large scale of renewable DG (RDG) integration, it has become a challenge of ADN to consume as more renewable energy as possible and reduce the power curtailment on the premise of safety operation. Meanwhile, the integration of RDG and energy storage make the loads no longer have only a single power supply path, which brings new opportunity in ADN to improve the power supply reliability and decrease the customer interruption cost due to the network failure. On the other hand, with the construction of smart grid and the development of distribution automation, the control strategies of ADN are diversified and the network are more flexible, which provides the means for ADN to seizing the opportunity and facing the challenge due to the RDG integration.
Based on the above background, this dissertation studies the operation optimization of ADN from the point of topology adjustment. By optimizing the network structure under normal operation condition, i.e., network reconfiguration, the power flow and voltage profiles can be optimized and the capability of ADN to consume the RDG can be improved. By optimizing the network structure under failure condition, i.e., island partition, the capability of RDG and energy storage to maintain an uninterrupted power supply can be fully utilized, and the impact of network failure can be alleviated. The contents of this dissertation are as follows.
1) A network reconfiguration method to increase the ADN’s capability to consume RDG is proposed. Aiming to maximize the RDG consumption in the ADN, a static network reconfiguration model for a single time section is built. The stochastic characteristics of RDG output power and load are considered, and the chance-constrained programming (CCP) is adopted to transform the stochastic optimization model to be a deterministic one. A reconfiguraiton strategy based on an improved intelligent algorithm is introduced. The intelligent algorithm is improved from the point of statistical analysis: the Pearson correlation coefficient is introduced to modify the iteration process.
2) A dynamic network reconfiguration based on second-order cone programming is proposed. Aiming to maximize the economic benefit of AND during a day, an optimal dynamic network reconfiguration model is built. An expanded DistFlow model which is suitable for flexible configuration of distribution network is proposed, based on which the operation constriants of the optimal model is given. The definition of “line voltage” is introduced to linearize the voltage magnitude drop constraint, and “active power flow” is introduced to linearize the radiality constraint. Togther with variable substitutions and constraint relaxation, the integrated model is transformed as a second-order cone programming problem.
3) An optimal static island partition algorithm based on energy indexes is proposed. A two-stage approach that integrates optimal island partition and power dispatch is proposed. Considering the power fluctuation characteristics of distributed energy resources (DERs) such as RDG, energy storage, electric vehicles (EVs), the energy indexes of the DERs and loads during the island partition period are defined, and a 2 stage optimal island partition method is proposed. The model in first stage maintains in a mathematical for of tree knapsack problem (TKP). Power dispatch and sequential power flow tests are given in the second stage, and the model modification method and island adjustment strategies are proposed based on the test result. By building a constraint set containing energy constraint, power constraint and energy transfer constraint, the island status is finally ensured to be feasible and optimized
4) A dynamic island partition method based on dynamic load weight is proposed. When a failure occurs in active distribution system, the network might be divided into several connected components, and topology of outage components is optimized. The total customer interruption cost is decreased by the cooperation of DER, network structure and loads considering the different interruption cost characteristics of multiple kinds of customers. Firstly, the interruption cost functions of customers are fitted, and the graph theory is introduced to build a theoretically optimal island partition model. Secondly, a simplified but practical model is derived based on the assumptions of time discretization, energy storage performance and customer interruption frequency. Fluctuation and stochastic characteristics of the renewable distributed generations and loads are considered and the stochastic model is transformed into a deterministic one by analytical method. A heuristic search algorithm based on the hill climbing method is proposed to solve the practical model.