As a method in the evaluation of smart grid, anti-Shannon-entropy weight can compute index weight which is in even distribution or normal distribution, and a reasonable evaluation result can be obtained, but it is not fit for the weight computation when the index includes a distinctive event in non-unique small probability through the research. For the above problem, a novel index weight algorithmanti-nonextension-entropy weight algorithm is proposed and applied to evaluating microgrid. After constructing the typical microgrid evaluation indexes, some microgrids whose indexes include a distinctive event in non-unique small probability is taken as the research objects to find the difference of two evaluation methods based on the different anti-entropy weight algorithm, and the evaluation testing of microgrid is also performed using two methods. Theoretical analysis and simulation results show that in comparison with the traditional evaluation method the novel evaluation method can not only inherit the merits of the traditional evaluation method but also obtain the reasonable weight of index including a distinctive event in non-unique small probability, which improves the precision of microgrid evaluation.