arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2304.00644eess.SYcs.SY
收藏

基于核条件均值嵌入的随机系统分布鲁棒最优与安全控制

Distributionally Robust Optimal and Safe Control of Stochastic Systems via Kernel Conditional Mean Embedding

Licio Romao, Ashish R. Hota, Alessandro Abate

AI总结:

本研究提出一种基于核方法的分布鲁棒动态规划框架,通过核条件均值嵌入与最大均值差异构建模糊集设计鲁棒控制策略,证明了最优策略的马尔可夫确定性,并给出可计算近似方案,在TCL安全控制实验中验证了理论结果。

摘要:

我们提出了一种用于动态规划的新型分布鲁棒框架,该框架采用核方法设计反馈控制策略。具体而言,我们利用核均值嵌入将支配状态演化的转移概率映射到对应的再生核希尔伯特空间中。我们的核心思路是将条件均值嵌入与最大均值差异距离相结合构建模糊集,再借助分布鲁棒优化相关技术设计鲁棒控制策略。本文的主要理论贡献是利用泛函分析工具证明了该无限维极小极大问题的最优策略具有马尔可夫性且是确定性的。此外,我们还探讨了基于状态和输入离散化的近似方案,以提升该方法的计算可处理性。为验证理论结果,我们在恒温控制负载(TCL)的安全控制场景下开展了实验。

英文摘要:

We present a novel distributionally robust framework for dynamic programming that uses kernel methods to design feedback control policies. Specifically, we leverage kernel mean embedding to map the transition probabilities governing the state evolution into an associated repreducing kernel Hilbert space. Our key idea lies in combining conditional mean embedding with the maximum mean discrepancy distance to construct an ambiguity set, and then design a robust control policy using techniques from distributionally robust optimization. The main theoretical contribution of this paper is to leverage functional analytic tools to prove that optimal policies for this infinite-dimensional min-max problem are Markovian and deterministic. Additionally, we discuss approximation schemes based on state and input discretization to make the approach computationally tractable. To validate the theoretical findings, we conduct an experiment on safe control for thermostatically controlled loads (TCL).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑