Research on Summer Cooling Load Prediction of Combined Cooling, Heating and Power System

Combined cooling, heating and power (CCHP) system is one of the effective means to improve energy efficiency, reduce carbon emission and environmental pollution. However, the traditional model-based CCHP scheduling algorithm is unable to adapt to the uncertainty of demand-side dynamic load and the imbalance of resource scheduling. Aiming at minimizing the daily operation cost and meeting the summer cooling and electricity demand of users, this paper proposes a deep reinforcement learning algorithm (DRL) Twin Delayed Deep Deterministic policy gradient algorithm - Gate Recurrent Unit (TD3-GRU) to optimize the resource scheduling of the CCHP system for summer cooling conditions. Characteristics and characteristic sets are carefully selected as load predict model inputs by using Extreme Gradient Boosting (XGBoost). The effectiveness of the algorithm is verified using the CCHP system in Shanghai World Expo Park as the research object. The results show that the DRL algorithm can effectively predict the summer cooling demand of users and match well with CCHP energy scheduling.