AI-CPS-Benchmark

AI-CPS-Benchmark是由阿尔伯塔大学等机构创建的公开数据集，包含9个工业级别的CPS，涉及汽车系统、化学反应器、航空航天等多个领域。数据集通过深度强化学习方法构建AI控制器，旨在评估AI增强的CPS与传统CPS的性能差异。创建过程中，研究团队根据系统需求设计了适当的奖励函数，并通过实验选择了最佳的AI控制器。该数据集主要用于研究AI控制器在CPS中的应用，以及探索如何通过结合传统与AI控制器来提升系统性能和可靠性。

AI-CPS-Benchmark is a public dataset developed by the University of Alberta and other research institutions. It includes 9 industrial-grade Cyber-Physical Systems (CPS) spanning multiple domains such as automotive systems, chemical reactors, and aerospace. The AI controllers incorporated in this dataset are constructed via deep reinforcement learning approaches, with the primary objective of evaluating the performance differences between AI-augmented CPS and conventional CPS. During the dataset development process, the research team designed appropriate reward functions based on system requirements, and selected the optimal AI controller through rigorous experiments. This dataset is primarily utilized for researching the application of AI controllers in CPS, as well as exploring methods to enhance system performance and reliability by integrating traditional and AI controllers.