Scholarlink

<p>We propose a stochastic control problem to manage cooperatively heterogeneous Thermostatically Controlled Loads (TCLs) to promote power balance in electricity networks. We develop a method to solve this problem with a decentralized architecture, in order to respect the privacy of individual users and to reduce both the telecommunications and the computational burden compared to the setting of an omniscient central planner. This paradigm is called federated learning in the machine learning community, hence the name <i>federated stochastic control problem</i>. The optimality conditions are expressed in the form of a high-dimensional Forward-Backward Stochastic Differential Equation (FBSDE), which is decomposed into smaller FBSDEs, which fully characterize the Nash equilibrium of a stochastic Stackelberg differential game. In this game, a coordinator (the leader) aims at controlling the aggregate behavior of the population, by sending appropriate signals, and agents (the followers) respond to this signal by optimizing their storage system locally. A mean-field-type approximation is proposed to circumvent telecommunication constraints and privacy issues. Convergence results and error bounds depending on the size of the population are obtained for this approximation. A numerical illustration is provided to show the interest of the control scheme and to exhibit the convergence of the approximation. An implementation is presented that answers the practical industrial challenges to deploy such a scheme.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Federated Stochastic Control of Numerous Heterogeneous Energy Storage Systems

  • 期刊论文

  • 发表日期：

    2026年2月18日

查看全文
  • Emmanuel Gobet,
  • Maxime Grangereau

摘要

We propose a stochastic control problem to manage cooperatively heterogeneous Thermostatically Controlled Loads (TCLs) to promote power balance in electricity networks. We develop a method to solve this problem with a decentralized architecture, in order to respect the privacy of individual users and to reduce both the telecommunications and the computational burden compared to the setting of an omniscient central planner. This paradigm is called federated learning in the machine learning community, hence the name federated stochastic control problem. The optimality conditions are expressed in the form of a high-dimensional Forward-Backward Stochastic Differential Equation (FBSDE), which is decomposed into smaller FBSDEs, which fully characterize the Nash equilibrium of a stochastic Stackelberg differential game. In this game, a coordinator (the leader) aims at controlling the aggregate behavior of the population, by sending appropriate signals, and agents (the followers) respond to this signal by optimizing their storage system locally. A mean-field-type approximation is proposed to circumvent telecommunication constraints and privacy issues. Convergence results and error bounds depending on the size of the population are obtained for this approximation. A numerical illustration is provided to show the interest of the control scheme and to exhibit the convergence of the approximation. An implementation is presented that answers the practical industrial challenges to deploy such a scheme.