Huang Xiaoyu, Lin Yubin, Ruan Xiaofei, Li Jiyu, Cheng Nuo
Department of Evaluation Center, Economic and Technological Research Institute of State Grid Fujian Electric Power Co., Ltd, Fuzhou, Fujian, China.
PeerJ Comput Sci. 2023 Jul 25;9:e1482. doi: 10.7717/peerj-cs.1482. eCollection 2023.
The optimal scheduling of energy in a smart grid is crucial to the energy consumption of the entire grid. In fact, for larger grids, intelligent scheduling may result in substantial energy savings. Herein, we introduce an enhanced dynamic programming algorithm (DPA) that utilizes two state variables to derive the optimal power supply schedule. The algorithm accounts for the dynamic states of both batteries and supercapacitors in the power supply system to augment the performance of the dynamic programming model. Additionally, this study incorporates a long short-term memory (LSTM) deep learning model, which integrates various environmental factors such as temperature, humidity, wind, and precipitation to predict grid power consumption. This serves as a mid-point pre-processing step for smart grid energy consumption scheduling. Our simulation experiments confirm that the proposed method significantly reduces energy consumption, surpassing similar grid energy consumption scheduling algorithms. This is critical for the establishment of smart grids and the reduction of energy consumption and emissions.
智能电网中能源的优化调度对于整个电网的能源消耗至关重要。事实上，对于规模较大的电网，智能调度可能会带来可观的能源节约。在此，我们引入一种增强型动态规划算法（DPA），该算法利用两个状态变量来推导最优供电调度。该算法考虑了供电系统中电池和超级电容器的动态状态，以提升动态规划模型的性能。此外，本研究纳入了长短期记忆（LSTM）深度学习模型，该模型整合了温度、湿度、风、降水等各种环境因素来预测电网功耗。这作为智能电网能源消耗调度的中间预处理步骤。我们的模拟实验证实，所提出的方法显著降低了能源消耗，超过了类似的电网能源消耗调度算法。这对于智能电网的建立以及能源消耗和排放的减少至关重要。
