Abstract: Large-scale wind power access may bring about potential safety risks to the load recovery process under extreme output scenarios. In order to cope with this problem, firstly, the predicted wind power outputs and the extreme output value are obtained by using the Copula theory, the Latin hypercube sampling and the value-at-risk (VAR) method. Secondly, based on the analysis of the active static security region (ASSR) of the load recovery time steps, the safety adjustment factors that characterizes the active power adjustment margins of different load nodes are proposed. Furthermore, by taking the load importance and safety adjustment factor as the weighting factors to measure the restoration values, the load restoration mixed integer non-linear programming (MILP) model with the optimization goal of maximizing the time step restoration value is established and solved with the CPLEX. Having the example of the IEEE-39 node system, the simulation results show that adjusting the restoration plan according to security region check results of the restoration plan in the extreme wind power output scenario, the load restoration effect and the safety of the restoration process are balanced.