Abstract: Capacity-load ratio is an important reference index for power grid planning and construction. At present, the determination method of capacity-load ratio is single, which cannot accurately meet the needs of power grid in areas with unbalanced development of power loads and large differences in load characteristics, which is not conducive to the reliable operation of power grid and the improvement of investment benefits. The fine planning method of capacity-load ratio is studied, the factors affecting capacity-load ratio are analyzed, and the optimization model of theoretical calculation parameters is built. Backpropagation(BP) neural network is used to study the relationship between power load simultaneity rate and main influencing factors, and to optimize the load dispersion coefficient. Considering the influence of voltage deviation, network loss, line N−1 pass rate and so on, the optimization method of transformer safe running rate parameters considering the safety of power grid and power quality is proposed. The influence of load growth and load density on load development reserve coefficient is studied, and a calculation model of substation capacity reserve life based on load scale is proposed. The feasibility of the proposed refined capacity-load ratio parameter optimization method is verified by actual regional power grid data. The parameter optimization model constructed can fully consider the development level of regional power grid and realize the refined planning and differentiated allocation of regional capacity-load ratio, which has important guiding significance for the accurate planning of power grid.