• Nuclear Engineering and Technology
• /
• v.55 no.2
• /
• pp.506-515
• /
• 2023

This study is to find a method to learn engine sound after the start-up of a diesel generator installed in nuclear power plant with an unsupervised deep learning algorithm (CNN autoencoder) and a new method to predict the failure of a diesel generator using it. In order to learn the sound of a diesel generator with a deep learning algorithm, sound data recorded before and after the start-up of two diesel generators was used. The sound data of 20 min and 2 h were cut into 7 s, and the split sound was converted into a spectrogram image. 1200 and 7200 spectrogram images were created from sound data of 20 min and 2 h, respectively. Using two different deep learning algorithms (CNN autoencoder and binary classification), it was investigated whether the diesel generator post-start sounds were learned as normal. It was possible to accurately determine the post-start sounds as normal and the pre-start sounds as abnormal. It was also confirmed that the deep learning algorithm could detect the virtual abnormal sounds created by mixing the unusual sounds with the post-start sounds. This study showed that the unsupervised anomaly detection algorithm has a good accuracy increased about 3% with comparing to the binary classification algorithm.

• Nuclear Engineering and Technology
• /
• v.55 no.2
• /
• pp.603-622
• /
• 2023

Human error (HE) is an important concern in safety-critical systems such as nuclear power plants (NPPs). HE has played a role in many accidents and outage incidents in NPPs. Despite the increased automation in NPPs, HE remains unavoidable. Hence, the need for HE detection is as important as HE prevention efforts. In NPPs, HE is rather rare. Hence, anomaly detection, a widely used machine learning technique for detecting rare anomalous instances, can be repurposed to detect potential HE. In this study, we develop an unsupervised anomaly detection technique based on generative adversarial networks (GANs) to detect anomalies in manually collected surveillance data in NPPs. More specifically, our GAN is trained to detect mismatches between automatically recorded sensor data and manually collected surveillance data, and hence, identify anomalous instances that can be attributed to HE. We test our GAN on both a real-world dataset and an external dataset obtained from a testbed, and we benchmark our results against state-of-the-art unsupervised anomaly detection algorithms, including one-class support vector machine and isolation forest. Our results show that the proposed GAN provides improved anomaly detection performance. Our study is promising for the future development of artificial intelligence based HE detection systems.

• Nuclear Engineering and Technology
• /
• v.55 no.6
• /
• pp.2034-2046
• /
• 2023

Industrial control systems in nuclear facilities are facing increasing cyber threats due to the widespread use of information and communication equipment. To implement cyber security programs effectively through the RG 5.71, it is necessary to quantitatively assess cyber risks. However, this can be challenging due to limited historical data on threats and customized Critical Digital Assets (CDAs) in nuclear facilities. Previous works have focused on identifying data flows, the assets where the data is stored and processed, which means that the methods are heavily biased towards information security concerns. Additionally, in nuclear facilities, cyber threats need to be analyzed from a safety perspective. In this study, we use the system theoretic process analysis to identify system-level threat scenarios that could violate safety constraints. Instead of quantifying the likelihood of exploiting vulnerabilities, we quantify Security Control Measures (SCMs) against the identified threat scenarios. We classify the system and CDAs into four consequence-based classes, as presented in NEI 13-10, to analyze the adversary impact on CDAs. This allows for the ranking of identified threat scenarios according to the quantified SCMs. The proposed framework enables stakeholders to more effectively and accurately rank cyber risks, as well as establish security and response strategies.

• Nuclear Engineering and Technology
• /
• v.56 no.1
• /
• pp.233-240
• /
• 2024

The wall friction correlations of oscillatory natural circulation loops are highly loop-specific, making it difficult to perform 1-D system simulations before obtaining specific experimental data. To better predict the friction characteristics, the nonlinear dynamics of a toroidal single-phase natural circulation loop were numerically investigated, and the transition effect was considered. The k-k_L-ω transitional turbulence and k-ω SST turbulence models were used to compute the flow characteristics of the loop under different heating powers varying from 0.48 to 1.0 W/cm², and the results of both models were compared with previous experiments. The mass flow rates and friction factors predicted by the k-k_L-ω model showed a better agreement with the experimental data than the results of the k-ω SST model. The oscillation frequencies calculated using both models agreed well with the experimental data. The k-k_L-ω transitional turbulence model provided better friction-factor predictions in oscillatory natural circulation loops because it can reproduce the temporal and spatial variation of the wall shear stress more accurately by capturing the movement of laminar, transition turbulent zones inside unstable natural circulation loops. This study shows that transition effects are a possible explanation for the highly loop-specific friction correlations observed in various oscillatory natural circulation loops.

• Nuclear Engineering and Technology
• /
• v.56 no.9
• /
• pp.3626-3643
• /
• 2024

One of the issues in the development of the sodium-cooled fast reactor (SFR) using transuranic (TRU) metallic fuel is the absence of criticality benchmark experiment that faithfully mocks up the nuclear characteristics of the target design for validation of the reactor core design code and its uncertainty quantification (UQ). This study aims to quantify the criticality uncertainty of a typical TRU burner with metallic fuel by using the standard upper safety limit (USL) estimation framework based on the similarity analysis of existing benchmark experiments but elaborated in two aspects:1) application of two-sided rather than one-sided tolerance interval and 2) inclusion of additional uncertainty to account for fission products and minor actinides not included in the benchmark experiments. To conduct the similarity analysis and evaluate the nuclear-data induced uncertainty, existing, well-verified computing codes were integrated, including the nuclear data sampling code SANDY, the nuclear data processing code NJOY, and the continuous-energy Monte Carlo code McCARD. Finally, using the SFR benchmark database comprising both publicly available and proprietary benchmark experiments, the criticality uncertainty of the TRU core model with metallic fuel was evaluated.

• Proceedings of the Korean Nuclear Society Conference
• /
• 2004.10a
• /
• pp.678-678
• /
• 2004

• Proceedings of the Korean Nuclear Society Conference
• /
• 2004.10a
• /
• pp.676-677
• /
• 2004

• Proceedings of the Korean Nuclear Society Conference
• /
• 2004.10a
• /
• pp.87-88
• /
• 2004

• Proceedings of the Korean Nuclear Society Conference
• /
• 2001.05a
• /
• pp.69.1-69
• /
• 2001

• Proceedings of the Korean Nuclear Society Conference
• /
• 2001.05a
• /
• pp.487.2-487
• /
• 2001