Two-Stage Problem of Optimizing Smart Grid Energy Consumption at the Enterprise

The paper considers the energy consumption opti-mization problem for an enterprise exploiting not only the central power supply system but also local generators and energy storage devices. To solve this problem, the paper proposes optimizing the energy consumption of the enterprise in two stages, i.e., solving the problems of optimal energy consumption scheduling in the first stage and optimal control over an enterprise's Smart Grid operating modes in the second one. The problem of optimal energy consumption scheduling was solved using linear programming method. The study implements a model of the enterprise's simple Smart Grid based on the author-developed environment compatible with the OpenAI Gym interface. Popular Reinforcement Learning algorithms were used to control the Smart Grid operating modes. The algorithms showed good convergence and successfully coped with the task set.

Перевод пока недоступен