Data-model fusing driven robust configuration optimization model and decision-making method for renewable energy generation base considering electric-hydrogen conversion

In order to guarantee the long-distance transmission of renewable energy from the large renewable plants (LRP)to the regions with tight electricity supply to meet high load demand, it is necessary to rationally configure various types of power sources in the plant. For the LRP configuration problem, this paper first proposes a data-model fusing driven inversion method to predict renewable energy output. Then, combining stochastic optimization and information gap theory, a multi-scenario confidence gap configuration model for a renewable energy base at the sending side considering electric-hydrogen conversion is proposed. Finally, a 4E evaluation theory based on energy, economy, elasticity, and environment is proposed to provide a comprehensive evaluation of the configuration results of different energy storage models. The results show that 1) the MAE, RMSE, and R2 of the proposed data-model fusing driven inversion method are improved by 20.1 %, 24.8 %, and 9.5 %, respectively, relative to the single data-driven model; 2) The unidirectional conversion model of electro-hydrogen has a better performance compared to bi-directional conversion: 4.7 % higher than bi-directional conversion in terms of NPV, 2.8 % lower cost per unit kWh, and 26 % lower cost of environmental pollution; 3) On a long time scale, the configuration priority of hydrogen, electric and pumped storage will change over the next 15 years. By 2030, the electric-hydrogen unidirectional conversion model is optimal due to the decreasing unit cost of electrolyzers and increasing unit cost of pumped storage. Considering three kinds of energy storage mixed configuration, by 2030, "electric hydrogen unidirectional conversion + pumped storage" will exceed "pumped storage + electric energy storage", become the configuration optimal choice.