为进一步提高船舶核动力二回路系统的安全性与可靠性，提出了一种子空间映射的船舶核动力二回路系统闭环故障诊断方法。针对船舶核动力二回路系统闭环控制结构导致控制器输入和测量输出存在的相关性问题，所提方法利用控制器输入和输出数据构造汉克尔矩阵，结合已知控制器的参数，构造了与控制器输入和测量输出无关的辅助矩阵。基于信号子空间方法，设计了一种数据驱动稳定核描述（Stable Kernel Representation, SKR）的闭环故障诊断方法，通过对未来和过去数据的辅助矩阵与输出矩阵进行LQ分解，获得二回路系统的数据驱动SKR，即残差产生器。然后，根据训练的输入和输出数据设计阈值、检验统计量与故障检测逻辑，实现对二回路系统有效的故障诊断。最后，实验主要在二回路系统的给水泵和冷凝器上进行算法测试，实验结果验证说明所提方法可以有效的检测二回路系统发生故障诊断。

To further improve the safety and reliability of ship nuclear power of secondary circuit systems, a subspace-aided closed-loop fault diagnosis approach is proposed for the ship nuclear power of the secondary circuit systems. Aiming at the correlation issues between controller input and measurement output caused by the closed-loop control structure of the ship of nuclear power of secondary circuit systems, An auxiliary matrix independent of the controller input and measurement output is constructed using known controller parameters, controller input, and measurement output data. Based on the subspace-aided method, a closed-loop fault diagnosis approach is designed for data-driven stable kernel representation (SKR). By performing LQ decomposition on the auxiliary matrix and output matrix of future and past data, the data-driven SKR of secondary circuit systems can be computed, namely the residual generator. Then, through the offline input and output as the training data, the threshold values, test statistics, and fault detection logic are designed for the implementation fault diagnosis of the secondary circuit system. Finally, the experiment mainly conducted algorithm testing on the feedwater pump and condenser of the secondary circuit system. The experimental results verified that the proposed method can effectively detect and diagnose faults in the secondary circuit system.

热能动力技术重点实验室开放基金资助项目（TPL2021C01）