Consideration of streamflow forecast uncertainty in the development of short-term hydropower station optimal operation schemes: A novel approach based on mean-variance theory

Abstract

The uncertainty of streamflow forecast will cause the ineffectiveness on the short-term hydropower station optimal operation (SHSOO) and lead to the risk of producing less hydropower than planned. Traditionally, stochastic optimal theory represented by the stochastic dynamic programming model (SDPM) is commonly used for SHSOO, which is focus on the uncertainty during a single lead time and lack of an overall consideration of multiple independent lead times of differing length. Thus, this paper proposes a short-term hydropower station optimal operation model based on mean-variance theory (MV-SHSOOM) for reducing operational risk. The proposed multi-step model considers the forecasting uncertainty of different lead times as an integral component to guide a station’s optimal operation. Specifically, the forecast uncertainty was integrated by the joint probability distribution of forecast error using a multivariate hydrologic uncertainty processor (HUP) at first. Then, possible observed streamflow sequences were obtained using the joint probability distribution of forecast error and taken as input to the short-term optimal operation model of the hydropower station. Finally, the mean-variance theory was applied to construct the optimal operation model for balancing the power generation and operational risk. The proposed model was applied to the Jinxi hydropower station, China, to determine its performance. The results show that: 1) the dependent structure among the forecast errors at different lead times is significant in terms of Spearman correlation, and the joint probability distribution of forecast error can be effectively described by a multivariate HUP; 2) the proposed MV-SHSOOM is superior to the traditional deterministic dynamic programming model (DDPM) and SDPM, and can more effectively reduce operational risk while increasing power generation.