Data-Driven Operation Cost Modeling of Power System Considering Wind Power Fluctuation

Abstract

Large scale random and intermittent wind power integration will significantly increase the additional operation cost of power system, and the variation characteristics are complex and difficult to accurately quantify. Different from the existing work, this paper first proposes a double-layer clustering method for wind power fluctuation scenarios. Secondly, a production simulation model of new energy power system is established to generate a comprehensive, real and representative sample set of power system operation cost. Finally, a data-driven mapping model between wind power fluctuation and the operation cost of power system and thermal power unit is established by using deep neural network algorithm. The proposed method can accurately simulate the variation trend of power system operation cost in different seasons and wind power fluctuation scenarios. The results show that the simulation error of thermal power unit and total system operation cost can reach 3%-13% and 4%-18%. This work provides insights and policy guidance for improving the wind power consumption rate and power system cost.