Owing to the various uncertain factors during power system restoration

there are always deviations between the actual recovered load amounts and the predicted values. In order to overcome the shortcomings of existing deterministic load restoration model

trapezoidal fuzzy parameters are used to express the injected recovered load amounts. And the deterministic safety constraints are substituted by the chance constraints based on fuzzy parameters. By taking into account the influences of load restoration benefit and overload risk on load optimization

a load restoration optimization model for the last stage of network reconfiguration is proposed. To calculate the load weights

the fuzzy entropy is introduced to quantify the load uncertainty

and the load importance and load uncertainty are used as load evaluation indices. Finally

clear equivalent forms are used to change the uncertain load restoration model into a deterministic 0-1 programming problem which can be solved by the mixed integer programming method. Case studies show that the proposed model is able to balance the restoration benefit and the overload risk

and the decisions obtained are better adapted to load fuzzy uncertainty. This work is supported by National Basic Research Program of China(973 Program)(No. 2012CB215101)and National Natural Science Foundation of China(No. 51177092).