The artificial intelligence-assisted short-term optimal scheduling of a cascade hydro-photovoltaic complementary system with hybrid time steps

Abstract

The coordinated scheduling of the cascade hydro-photovoltaic hybrid system can significantly alleviate the adverse impact from the intermittence of solar energy resources and hence improve the overall stability of bundled power output. The traditional short-term scheduling strategy of cascade hydro-photovoltaic systems is generally based on a fixed time step, which cannot efficiently adapt to the ever-changing photovoltaic output. In this paper, a novel artificial intelligence-assisted short-term scheduling model with hybrid time steps is proposed, where the time-frequency characteristics of photovoltaic output are taken into consideration. The photovoltaic output is estimated by a data-driven long short-term memory network. The Hilbert-Huang transform is used to analyze the time-frequency characteristics of the photovoltaic power. The hybrid scheduling time steps are determined according to the analyzed instantaneous frequency of photovoltaic power. A short-term optimal scheduling model based on data-driven photovoltaic output evaluation and hybrid time intervals is established to minimize the operational fluctuation of hydropower units and to track the planned output curve simultaneously. Finally, a comprehensive index involving the safety, economy, and operational efficiency is established to compare the overall performance of the hybrid time step model and that of the fixed time step models. The results demonstrate that the proposed hybrid time step model yields the lowest comprehensive ranking score of 1.00, obviously better than the figures for the fixed 60-minuite step model (2.86) and the fixed 15-min step model (2.20). It indicates that the proposed model can achieve the optimal compromise of overall performance via adaptive scheduling time steps.