Optimization strategy based on robust model predictive control for RES-CCHP system under multiple uncertainties

The economic operation of renewable energy sources (RES) integrated with a combined cooling, heating, and power system significantly improves energy utilization and reduces environmental crises. However, multiple uncertainties in RES generation and load consumption predictions cause an imbalance between supply and demand, negatively impacting system efficiency and economics. To address this limitation, an integrated uncertainty rolling optimization framework for combining the probability distribution of prediction and uncertainty optimization is established. Under this framework, an optimization strategy based on a robust model predictive control is proposed for handling the multiple uncertainties of source load. Here, the probability distributions of RES generation and load consumption predictions are determined using the Gaussian process regression method, and a minimum-maximum rolling optimization model is developed. Under the uncertain scenarios of RES generation and load consumption, the optimization model is converted into a tractable form to obtain a robust schedule that minimizes operating costs. The over-conservatism of robust optimization can be mitigated by adjusting the uncertainty budget. Many case studies are further conducted to confirm the effectiveness of the proposed method. Results show that the operating costs decreased by 11.5% compared with the traditional model predictive control strategies in an uncertain scenario.

Automated summary

This study establishes an integrated uncertainty rolling optimization framework to address the supply-demand imbalance in the economic operation of renewable energy sources integrated with a combined cooling, heating, and power system. By combining the probability distribution of prediction and uncertainty optimization, the framework proposes an optimization strategy based on a robust model predictive control to handle the multiple uncertainties. The effectiveness of the proposed method is confirmed through case studies, which show a decrease in operating costs compared to traditional model predictive control strategies.