| Abstract |
In the realm of energy internet, the pivotal roles of community integrated multi-energy systems cannot be overstated, as it is instrumental in mitigating energy expenses, augmenting the utilization of renewable energy sources, and upholding equilibrium in supply and demand dynamics. A multi-energy scheduling strategy based on improved twin delayed deep deterministic policy gradient algorithm is proposed in this study, which can effectively decrease the energy cost of the community integrated multi-energy system energy. Firstly, the groundwork entails the establishment of a comprehensive model encompassing diverse energy conversion and storage devices deployed within the community. Moreover, the purpose of the scheduling strategy is to achieve the minimization of the energy cost. The cost minimization problem is mathematically formalized as a Markov decision process. Then the improved TD3 method is employed to address the optimization problem, where convolutional neural networks and attention mechanism are introduced to extract temporal features and improve the exploration efficiency of TD3. Finally, the simulation results are empirically validate the efficiency of the proposed community integrated multi-energy scheduling strategy, specifically in reducing the system's energy costs and ensuring equilibrium between energy supply and demand. © 2023 IEEE. |
