An interval scheduling method for the CCHP system containing renewable energy sources based on model predictive control.

Scheduling strategies are crucial for combined cooling, heating, and power-containing renewable energy source (RES-CCHP) systems to realize cost-effective operation. However, the prediction error generated by renewable energy sources (RES) and load leads to mismatch between the scheduling plan and actual demand. To address this problem, an interval scheduling method is proposed for the RES-CCHP system based on the model predictive control method. A genetic algorithm to optimize a back propagation (GA-BP) neural network combined with historical and weather data is used to predict the multistep RES and load to improve prediction accuracy. An interval optimization model (cost optimization) is then constructed based on RES and load prediction data to achieve an output consistent with the actual situation. Finally, feedback correction is used to compensate the RES and load prediction data online. The prediction error and interval are reduced, thereby improving the performance of the interval optimization method in achieving an optimal solution. This research is applicable to living spaces such as residential areas, schools, and hospitals. In this study, a hospital was considered as a case study to verify the effectiveness of the proposed method. A comparison of the results of the proposed method with those obtained in the following electrical load and following thermal load modes in a typical summer day revealed a cost reduction of 16.21% and 16.92%, respectively. The proposed method can improve the prediction accuracy of the RES and load of the RES-CCHP system and help decision makers to develop reasonable operation schemes and maximize profits. • Interval scheduling method for RES combined cooling, heating, and power systems based on model predictive control. • The weather data and genetic algorithm–BP neural network method accurately predict multistep source and load. • Interval optimization used upon considering source and load error of prediction caused by randomness. • Feedback correction to compensate prediction results and reduce optimization interval. [ABSTRACT FROM AUTHOR]