Multistage Model Predictive Control with Simplified Scenario Ensembles for Robust Control of Hydropower Station

Abstract

This paper proposes simplification of the scenario ensembles that describe the uncertainty present in a hydropower plant. The simplified scenario tree is further used with a multistage model predictive control for optimal operation of the hydropower station. The proposed method reduces the number of considered scenario ensembles of water inflow forecast into the reservoir in the Dalsfoss hydropower plant, which leads to less computational demand of the multistage MPC. The method takes two steps: the creation of three synthesis scenario ensembles and the estimation of the probability of occurrence of the three synthesis scenario ensembles. The simulation results of multistage MPC with 4 different types of scenario ensembles demonstrate that the proposed simplified method reduces the computation demand of the multistage MPC by 15 times approximately, without degrading its performance.

Keywords

• multistage model predictive control
• uncertainty
• simplified method
• renewable energy