Research on Anomaly Detection and Fault Diagnosis Technology for Complex Systems in Nuclear Power Plants Based on State Estimation
-
摘要: 为实现对核电厂复杂系统微小故障的快速发现和精准定位，本文首先设计并搭建了预测性运维技术实验台架（后简称“实验台架”），然后利用非线性估计及不确定度分析算法构建异常检测模型，最后利用实验台架生成的运行数据对故障诊断技术方案（即先通过基于数据驱动的方式进行异常检测，再配合热工水力学分析进行故障定位）进行测试。结果显示，实验台架能够产生复杂系统的可控运行数据，异常检测模型能够及时发现系统出现的异常状况，针对异常信号配合热工水力学分析获得的故障诊断结果与实验台架注入的故障一致。因此，实验台架有效缓解了现有状态估计、异常检测算法应用于核电厂场景时主要采用模拟数据或其他工业场景运行数据进行测试验证的困难，同时证明了该故障诊断技术方案能够实现对于核电厂复杂系统微小故障的诊断及定位。
-
关键词:
- 预测性运维技术 /
- 复杂系统微小故障诊断 /
- 异常检测 /
- 状态估计
Abstract: To achieve rapid detection and precise localization of minor faults in complex nuclear power plant systems (NPPS), this study first designed and built a predictive operation and maintenance technology test facility (hereafter referred to as the PHM test facility). Then, using nonlinear estimation and uncertainty analysis algorithms, an anomaly detection model was constructed. Finally, the operational data generated by test facility were used to test the proposed fault diagnosis technical scheme, which firstly employs a data-driven anomaly detection model to confirm the presence of anomalies, and then utilizes thermal-hydraulic analysis to localize faults. Results demonstrate that the PHM test facility can successfully produce controllable operational data of complex systems. The anomaly detection model can effectively identifiy system anomalies in a timely manner, while the fault diagnosis results obtained in thermal-hydraulic analysis for anomaly signal is consistent with the pre-injected faults in the test facility. This test facility addresses the current challenges in nuclear power applications where state estimation and anomaly detection algorithms predominantly rely on simulated data or operational data from other industrial scenarios for validation. Furthermore, it confirms the capability of the proposed fault diagnosis technical scheme to detect and localize minor faults for complex systems in NPPS. -
图 1 RCV微小泄漏故障注入系统流程图
MP—压力仪表；MD—流量仪表；MT—温度仪表；HV—手动阀门；PU—循环水泵。
Figure 1. Flowchart of the RCV Minor Leak Fault Injection System
图 7 上充流量仪表018MD异常检测结果
Figure 7. Anomaly Detection Result of Charging Flow Instrument 018MD
图 8 上充调节阀开度异常检测结果
Figure 8. Opening Anomaly Detection Result of Charging Control Valve
图 9 再生式回热器冷端入口流量仪表异常检测结果
Figure 9. Anomaly Detection Result of Regenerative Heat Exchanger Cold End Inlet Flow
表 1 模型参数及说明
Table 1. Model Parameters and Description
参数名 参数值及说明 检测仪表数量/个 27 训练数据集点数 7000 测试数据集点数 600 核函数 $ k\left(\boldsymbol{x},\boldsymbol{y}\right)=1-\dfrac{\left|\left|\boldsymbol{x}-\boldsymbol{y}\right|\right|}{\left|\left|\boldsymbol{x}\right|\right|+\left|\left|\boldsymbol{y}\right|\right|} $ α值 0.9 
下载: 导出CSV
-
[1] World Nuclear Association. Nuclear Power Economics and Structuring - 2024 Edition[R]. England: WNA, 2024. [2] HINES J W, SEIBERT R. Technical review of on-line monitoring techniques for performance assessment, NUREG/CR-6895[R]. Washington: U. S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, 2006. [3] COBLE J B, RAMUHALLI P, BOND L J, et al. Prognostics and health management in nuclear power plants: a review of technologies and applications, PNNL-21515[R]. Richland, WA: U. S. Department of Energy, 2012. [4] AYO-IMORU R M, CILLIERS A C. A survey of the state of condition-based maintenance (CBM) in the nuclear power industry[J]. Annals of Nuclear Energy, 2018, 112: 177-188. doi: 10.1016/j.anucene.2017.10.010 [5] 陈豫龙,陈晓明,赵炳全. 威布尔分布在核电站可靠性数据库中的应用[J]. 原子能科学技术,2003, 37(4): 349-352. doi: 10.3969/j.issn.1000-6931.2003.04.013 [6] LIU J, SERAOUI R, VITELLI V, et al. Nuclear power plant components condition monitoring by probabilistic support vector machine[J]. Annals of Nuclear Energy, 2013, 56: 23-33. doi: 10.1016/j.anucene.2013.01.005 [7] AL RASHDAN A Y, GIRAUD K M K, GRIFFEL M, et al. An applied strategy for using empirical and hybrid models in online monitoring, INL/EXT-20-59688[R]. United States: U. S. Department of Energy, 2020. [8] QI Z F, LI W, ZHANG Y W, et al. Application of nonlinear principal component analysis technique to nuclear power plants[C]//The Proceedings of the International Conference on Nuclear Engineering (ICONE). Tokyo: The Japan Society of Mechanical Engineers, 2019: 1733. [9] PARK P, MARCO P D, SHIN H, et al. Fault detection and diagnosis using combined autoencoder and long short-term memory network[J]. Sensors, 2019, 19(21): 4612. doi: 10.3390/s19214612 [10] HINES J W, SEIBERT R, ARNDT S A. Technical review of on-line monitoring techniques for performance assessment, NUREG/CR-6895[R]. Washington, DC: Office of Nuclear Regulatory Research, 2006. [11] FANTONI P F. Experiences and applications of PEANO for online monitoring in power plants[J]. Progress in Nuclear Energy, 2005, 46(3-4): 206-225. doi: 10.1016/j.pnucene.2005.03.005 -
下载:
计量
- 文章访问数: 21
- HTML全文浏览量: 8
- PDF下载量: 0
- 被引次数: 0
