Intelligent Diagnosis and Monitoring for Abnormal Operation Event of Reactor Coolant System
-
摘要: 为解决传统基于深度学习（DL）的智能故障诊断模型无法监测系统未知异常运行事件的问题，本研究基于变分推断的概率深度神经网络（VI-PDNN）构建反应堆冷却剂系统智能诊断框架，对异常事件类别实现诊断同时量化评估输出结果的不确定性。该框架能够有效利用已知与未知运行事件的不确定性差异实现对未知异常运行事件的有效监测预警。最后，基于已建立反应堆模拟机仿真数据对本文提出方法进行验证。研究结果表明，提出的方法不仅能够针对系统已知事件获取较高的诊断精度，同时能有效监测预警未知异常事件，为实际环境下反应堆系统运行实时智能状态诊断与监测提供一种有效技术手段。Abstract: In order to solve the problem that the traditional deep learning (DL)-based intelligent fault diagnosis model cannot monitor the unknown abnormal operation events, this work constructs an intelligent diagnosis framework based on the probabilistic deep neural network (VI-PDNN) model of variational inference, which realizes the diagnosis of the abnormal event categories of the reactor coolant system and quantitatively evaluates the uncertainty of the output results. The framework can effectively utilize the uncertainty difference between known and unknown operating events to achieve effective monitoring and warning of unknown abnormal operating events. Finally, the proposed methodology is validated based on simulation data from an established reactor simulator. The results show that the proposed method not only obtains high diagnostic accuracy for known events, but also effectively monitors and warns against unknown abnormal events, providing an effective means for real-time intelligent state diagnosis and monitoring of reactor system operation in real environment.
-
图 1 基于VI-PDNN模型的智能诊断与监测流程
Figure 1. Intelligent Diagnosis and Monitoring Process based on VI-PDNN Model
图 4 已知事件与B-1异常事件混淆矩阵结果比较
Figure 4. Comparison of Confusion Matrix between Known Events and B-1 Abnormal Event
图 5 已知事件与B-2异常事件混淆矩阵结果比较
Figure 5. Comparison of Confusion Matrix between Known Events and B-2 Abnormal Event
图 6 不同方法针对B-1异常事件的不确定性评估结果
Figure 6. Uncertainty Evaluation Results of Different Methods for B-1 Abnormal Event
图 7 不同方法针对B-2异常事件的不确定性评估结果
Figure 7. Uncertainty Evaluation Results of Different Methods for B-2 Abnormal Event
图 8 不同方法模型训练与测试评估时间对比
Figure 8. Comparisons of Model Training and Testing Time with Different Methods
表 1 反应堆冷却剂系统运行已知与未知事件
Table 1. Known and Unknown Operation Events for
事件编号 事件详细信息 事件类型 N-1 稳态运行 已知 N-2 反应堆功率升高10% 已知 N-3 反应堆功率升高20% 已知 A-1 冷却剂丧失事故 5% 已知 A-2 一回路主泵转速下降50% 已知 A-3 一回路主泵卡轴 已知 A-4 给水泵转速下降50% 已知 A-5 给水泵卡轴 已知 B-1 意外的反应性引入事故 未知 B-2 冷却剂丧失事故 20% 未知 
下载: 导出CSV
表 2 不同方法针对已知事件的诊断性能对比
Table 2. Comparison of Diagnostic Performance for Known Events with Different Methods
模型 平均准确率 平均F1分数 传统模型 94% 89% 提出的方法 98% 99%
下载: 导出CSV
表 3 不同方法针对异常事件的检测率对比
Table 3. Comparisons of Abnormal Event Detection Rates with Different Methods
模型类别 异常事件检测率 B-1 B-2 深度集成方法 78% 74% MC Dropout 方法 92% 100% 提出的方法 100% 100%
下载: 导出CSV
-
[1] 刘才学,罗能,何攀,等. 反应堆关键设备健康监测与故障诊断技术研究进展[J]. 核动力工程,2023, 44(3): 8-20. [2] 张恒,吕雪,刘东,等. 核电人工智能应用: 现状、挑战和机遇[J]. 核动力工程,2023, 44(1): 1-8. [3] 朱少民,夏虹,彭彬森,等. 基于PCA的主泵传感器状态监测模型[J]. 核动力工程,2020, 41(3): 170-176. [4] YAO Y T, WANG J, XIE M, et al. A new approach for fault diagnosis with full-scope simulator based on state information imaging in nuclear power plant[J]. Annals of Nuclear Energy, 2020, 141: 107274. doi: 10.1016/j.anucene.2019.107274 [5] YU Y, PENG M J, WANG H, et al. Improved PCA model for multiple fault detection, isolation and reconstruction of sensors in nuclear power plant[J]. Annals of Nuclear Energy, 2020, 148: 107662. doi: 10.1016/j.anucene.2020.107662 [6] 曹桦松,孙培伟. 基于PCA-RBF神经网络的小型压水堆故障诊断方法研究[J]. 仪器仪表用户,2021, 28(1): 49-55. [7] LEE G, LEE S J, LEE C. A convolutional neural network model for abnormality diagnosis in a nuclear power plant[J]. Applied Soft Computing, 2021, 99: 106874. doi: 10.1016/j.asoc.2020.106874 [8] 邓强,王航,彭敏俊,等. 基于时间卷积胶囊网络的核动力装置事故诊断技术研究[J]. 原子能科学技术,2023, 57(2): 302-312. doi: 10.7538/yzk.2022.youxian.0218 [9] NGUYEN H P, LIU J, ZIO E. A long-term prediction approach based on long short-term memory neural networks with automatic parameter optimization by Tree-structured Parzen Estimator and applied to time-series data of NPP steam generators[J]. Applied Soft Computing, 2020, 89: 106116. doi: 10.1016/j.asoc.2020.106116 [10] 孙原理,宋志浩. 基于卷积长短期记忆网络和人工鲸鱼算法的核反应堆运行事件诊断方法研究[J]. 核动力工程,2022, 43(4): 185-190. [11] WILSON A G, IZMAILOV P. Bayesian deep learning and a probabilistic perspective of generalization[C]//Proceedings of the 34th International Conference on Neural Information Processing Systems. Vancouver: NeurIPS, 2020: 4697-4708. [12] MISHRA S, AYYUB B M. Shannon entropy for quantifying uncertainty and risk in economic disparity[J]. Risk Analysis, 2019, 39(10): 2160-2181. doi: 10.1111/risa.13313 [13] 姚源涛. 基于深度知识迁移的新型核系统诊断预测与容错技术研究[D]. 合肥: 中国科学技术大学,2021. [14] YAO Y T, WANG J Y, XIE M. Adaptive residual CNN-based fault detection and diagnosis system of small modular reactors[J]. Applied Soft Computing, 2022, 114: 108064. doi: 10.1016/j.asoc.2021.108064 [15] Rahaman R. Uncertainty quantification and deep ensembles[J]. Advances in neural information processing systems, 2021, 34: 20063-20075. [16] SADR M A M, GANTE J, CHAMPAGNE B, et al. Uncertainty estimation via Monte Carlo Dropout in CNN-based mmWave MIMO localization[J]. IEEE Signal Processing Letters, 2022, 29: 269-273. doi: 10.1109/LSP.2021.3130504 -
计量
- 文章访问数: 7
- HTML全文浏览量: 2
- PDF下载量: 0
- 被引次数: 0
