Performance Evaluation Method of Day-Ahead Load Prediction Models in a District Heating and Cooling System: A Case Study

Many researchers are devoted to improving the prediction accuracy of daily load profiles, so as optimize day-ahead operation strategies achieve most efficient district heating and cooling (DHC) systems; however, studies on strategy optimization generally isolated, which leaves following question: what day-head performance should be paid attention in DHC systems? In order explain this issue, taking an actual system a case study, paper proposes evaluation method for profiles by considering impact inaccurate system. The results show following: (1) When models developed, mean emphasized, there is no need painstakingly increase profile shapes. (2) CV RMSE suitable deviation measures (compared others, e.g., MAPE, MAE, etc.) models. A model with 27.8% (CV) only causes 3.74% costs; thus, enough meet engineering requirements paper.