The model of optimal restoration control considering the income from restored loads and control cost， as well as the control risk is presented. A two-layer framework for optimizing is proposed， which uses a procedure of“parallel optimization for subareas and dynamic coordination among subareas”， to overcome the drawback of offline made restoration schemes based on fixed scenarios and fixed priority for countermeasures. Moreover， restoration strategy is extended from bottom-up one， which is widely discussed nowadays， to top-down one as well as hybrid one. Studies include the inner optimization of each strategy and coordination between different strategies. Based on a viewpoint of adaptive optimization and risk decision-making， related modules and algorithms are designed. A framework of the online decision support system is presented， the three parts of which are information collecting， analysis and control. The validity and adaptiveness of the decision support system are verified by simulations.