This paper outlines main issues and key points of post-grid-project evaluation,and proposes a GA-PSO-BP neu-ral network algorithm.It establishes a predictive model for electricity in post-grid-project evaluation based on the proposed algorithm according to operation data of grid projects with different voltage levels in different regions of the province.The proposed method has been proved by a case experiment on an actual 500 kV transmission project to have a smaller devia-tion and a higher accuracy compared to conventional methods,and hence a certain applicative value.