随着船舶动力系统智能化和自动化水平的提升，其复杂性和故障概率显著增加，对安全性和可靠性提出了更高要求。针对传统容错控制方法依赖精确模型的局限性，提出了一种基于Youla参数化的数据驱动容错控制方法。该方法通过自适应观测器实时辨识冷凝器的系统参数，并利用残差驱动的Youla参数化控制器进行在线参数学习，实现了控制性能的实时优化与容错控制。仿真结果表明，所提方法能够有效诊断冷凝器故障并实现容错控制，显著提高了冷凝器系统的可靠性和稳定性。这项工作的创新点在于将数据驱动方法应用于船舶动力系统，为解决复杂难以建模的非线性系统的故障诊断与容错控制问题提供了新思路。

With the improvement of the intelligence and automation level of ship systems, the integration and complexity of their power systems have become increasingly higher, resulting in a greatly increased probability of failure. This puts higher demands on safety and reliability. Effective fault diagnosis and fault-tolerant control technology are essential to ensure the navigation safety of ship power systems. Most of the existing fault-tolerant control methods for power systems require known accurate models. As the complexity of the system increases, accurate models will be difficult to obtain, which makes it difficult for model-based methods to achieve satisfactory results. In this work, a data-driven fault-tolerant control method is proposed under the framework of Youla parameterization. This method uses an adaptive observer to identify the system parameters of the condenser in real time, and then uses the residual to drive the Youla parameterized controller for parameter learning, achieving the purpose of online optimization of control performance and fault-tolerant control. Finally, the effectiveness of the proposed method is verified using a simulation model of a ship power condenser system.

热能动力技术重点实验室开放基金资助项目（TPL2021C01）