To fully calculate the randomness, intermittency, and volatility of the clean energy output, improve the new energy consumption capacity of power system in the future time scale, and improve and optimize the new energy planning scheme is of great significance to promote the safety, reliability and economic operation of the power grid. This paper proposes to use the single station-partition-whole network progressive route to study the new energy limit consumption capacity of the power grid, and puts forward the grid operation intelligent adjustment method using the PSD-BPA software. The binary approximation method and improved particle swarm algorithm are used for single station evaluation; the optimization model under different constraints is established respectively for partition and whole network evaluation, and the polynomial model of new energy node injection power influence is used as equivalence to the complex constraints in the provincial network. In addition, for the new energy output probability modeling, this paper compares the advantages of non-parametric kernel density estimation and Gaussian mixture model, and conducts spatial correlation analysis on the performance of new energy plant output. Finally, the effectiveness of the proposed model and algorithm is verified by practical examples.