近些年，极端自然灾害引发的配电网大停电造成了十分可观的经济损失。为了提升配电网的灾害应对能力，本文设计了多阶段配电网韧性的量化评估指标，并提出了配电网灾害应对的完整优化决策方法。论文的主要工作如下： 第一，设计了多阶段配电网韧性的量化指标，以此评估配电网的灾害应对能力。考虑极端灾害中配电网的调控特性，划分了配电网灾害应对的决策阶段，设计了不同决策阶段的配电网韧性量化指标。随后，本文提出了提升配电网灾害应对能力的完整优化决策流程。 第二，提出了配电网应急供电资源的灾前调配优化决策方法。建立了配电网的资源调配优化模型，实现不同种类供电资源在微电网与关键负荷间的调配。提出了求解资源调配模型的启发式方法，将优化模型转化为混合整数二次规划问题，进而实现高效求解。 第三，提出了配电网关键负荷的灾后恢复优化决策方法。建立了利用储存有限供电资源的微电网恢复配电网关键负荷的机会约束规划模型。提出了求解优化模型的两阶段启发式方法，通过建立并更新配电网的恢复策略表，将优化模型转化为较小规模的线性整数规划问题，提高了恢复策略的求解效率。 第四，提出了配电网关键负荷恢复策略的暂态可行性校验方法。使用暂态仿真刻画恢复供电过程满足暂态安全约束的配电网初始运行状态集合，在执行策略前对其进行可行性校验。进而提出配电网运行状态的优化调整方法，增强恢复策略在配电网中的实用性。 第五，提出了应急供电模式下微电网的动态负荷切除优化决策方法。建立了微电网动态负荷切除的随机规划模型以及与其对应的马尔可夫决策模型。提出了最大化马尔可夫决策在有限时间内收益的求解方法，实现微电网根据当前的资源储备情况动态切除内部负荷。 综上，本文提出的优化决策方法充分利用了配电网中有限的供电资源和微电网的调控特性，提高了极端灾害下配电网关键负荷的应急供电水平。本文所提方法有助于提高配电网的防灾抗灾能力，减小大停电事故范围及其所造成的经济损失，具有重要的工程应用价值。

In recent years, major power outages caused by extreme natural disasters incurred considerable economic losses in distribution systems. In order to enhance the ability of a distribution system to cope with disasters, this thesis designs a multi-stage quantitative index set to evaluate its resilience, and proposes an optimization decision-making method for disaster relief in the distribution system. The main work of this thesis is as follows:Firstly, a multi-stage quantitative index set is designed to evaluate the ability of a distribution system to cope with disasters. Based on the operating features of a distribution system under extreme disasters, this thesis divides the disaster-relief decision-making process into several stages. Quantitative indices are designed to evaluate the resilience of a distribution system in these stages. Then, this thesis proposes an optimization decision-making process to enhance the ability to cope with extreme disasters in distribution systems.Secondly, an optimization decision-making method is proposed to allocate emergency generation resources in a distribution system ahead of an extreme natural disaster. A resource allocation optimization model is proposed to allocate different kinds of generation resources among microgrids and critical loads in the distribution system. A heuristic method is proposed to solve the model. By transforming the optimization model into a mixed integer quadratic program, the resource allocation plan can be obtained efficiently.Thirdly, an optimization decision-making method is proposed to restore critical load during the power outage of a distribution system after an extreme natural disaster. A chance-constrained program is proposed to restore critical load using microgrids with limited generation resources. A two-stage heuristic method is proposed to solve the model. By creating and updating a restoration strategy table, the optimization model is transformed into a linear integer program of a much smaller scale, and the efficiency of the restoration strategy solution is therefore enhanced.Fourthly, a verification method is proposed to verify the transient feasibility of critical load restoration strategies. Transient simulation is used to obtain the initial operating state set of a distribution system, which satisfies the transient operating constraints in the restoration procedure. The set obtained is used to verify the transient feasibility before a restoration strategy is actually executed in the distribution system. A method is proposed to adjust the operating state of the distribution system optimally, and enlarge the applicable scope of the candidate strategy.Fifthly, an optimization decision-making method is proposed to shed load dynamically in the emergency power supply mode of a microgrid. A stochastic program of the dynamic load shedding is proposed, as well as a Markov decision process model corresponding to it. A method is proposed to obtain the optimal load shedding strategy which maximizes the economic benefit of the microgrid in a limited time period. When adopting the strategy, the microgrid is able to shed load dynamically according to its available generation resources.In summary, in order to raise the level of emergency power supply to critical load, the optimization decision-making method proposed in this thesis makes full use of the limited generation resources within a distribution system and the operational features of microgrids. The proposed method is very helpful in improving the ability of a distribution system to cope with extreme natural disasters, and reducing the range of major power outages as well as the incurred economic losses, demonstrating the significant engineering value of the proposed method.