[Report] "Detecting Nuclear Power Plant Anomalies with AI and Holding Countermeasure Meetings in the Metaverse"

Operation of KHNP Central Research Institute Integrated Prediction and Diagnosis Center
Digital Twin for Nuclear Power Plant to be Established for the First Time in February Next Year
Structural Seismic Proof Test Center Also Completed in January This Year

On the 12th, when entering the Integrated Predictive Diagnosis (AIMD) Center located within the Korea Hydro & Nuclear Power (KHNP) Central Research Institute in Yuseong-gu, Daejeon, a large monitor installed at the front immediately caught my eye. The screen intuitively displayed the operational status of five nuclear power headquarters?Hanbit, Hanul, Saeul, Kori, and Wolseong?using numbers and visual data.

At around 3 p.m. on the day of the visit, the health index of the Hanbit headquarters among the five was shown as 98%. This was because a warning signal was detected from one piece of equipment. Yesonghae, head of the Digital Plant Group at KHNP, explained, "When such abnormal signs appear, we analyze the cause together with the equipment manager of the relevant nuclear power plant and find countermeasures to prevent accidents." Using this automatic predictive diagnosis technology at the AIMD Center, KHNP was able to prevent 14 major equipment failures last year.

AIMD Center Started Operation in 2022... Preemptive Diagnosis of Abnormal Signs at 26 Nuclear Power Plants Nationwide

The AIMD Center uses artificial intelligence (AI) to monitor and diagnose nuclear power plants and began operations in 2022. Six staff members, including vibration analysis experts, are stationed there. On average, the center automatically diagnoses the condition of over 100 pieces of equipment daily.

The Integrated Prediction and Diagnosis (AIMD) Center located within the Korea Hydro & Nuclear Power Central Research Institute. Photo by Korea Hydro & Nuclear Power

Here, sensors are attached to 12,387 pieces of equipment such as turbines, generators, coolers, pumps, and transformers at 26 nuclear power plants operated by KHNP nationwide, enabling real-time monitoring. The core mission is to detect abnormal signs in advance using AI and big data technologies to prevent accidents. KHNP extracts features from data accumulated over more than 10 years and uses machine learning technology to precisely diagnose equipment conditions.

For rotating and power equipment, deep learning technology is used to analyze infrared (IR) images and diagnose equipment abnormalities. Additionally, a homogeneous equipment comparative diagnosis technology is utilized to compare and analyze the same equipment installed at the 26 nuclear power plants. In hazardous areas where humans cannot enter, unmanned sensors are installed for monitoring.

Kim Daeyong, head of the Instrumentation Research Institute at KHNP Central Research Institute, explained, "By using AI, abnormal signs can be detected before humans recognize them, and large-scale equipment can be monitored simultaneously, which also leads to workforce reduction."

World’s First Development of Digital Twin... Enables Problem Prediction and Inspection in Virtual Space

Along with the predictive diagnosis system, KHNP is also developing the world’s first digital twin for nuclear power plants. A digital twin refers to a virtual space representation of an actual nuclear power plant. Using the digital twin, situations that may occur in reality can be simulated and the results predicted.

KHNP plans to implement avatars and metaverse systems in the digital twin using 3D and virtual reality (VR) technologies. For example, if an abnormal signal is detected in a reactor coolant pump, the site can be inspected using an avatar. Also, experts located apart can gather in the virtual space to discuss cause analysis and countermeasures. On the left side of the AIMD Center monitor, the Hanul Nuclear Power Plant Unit 5 coolant pump was implemented in digital twin form and monitored in real time.

KHNP is currently developing a digital twin for Saeul Nuclear Power Plant Units 1 and 2, which apply the APR1400 reactor type. The second phase of the project is underway and is scheduled for completion in February next year. In the future, digital twin technology will also be applied to innovative small modular reactors (SMRs). Except for the digital twin implemented by the U.S. GE at thermal power plants, there are no other cases of digital twin implementation at power plants.

Completion of Structural Seismic Test Center... Development of Equipment That Operates Safely Even During Earthquakes

On the same day, I also visited the Structural Seismic Test Center at KHNP Central Research Institute, which was completed in January. This center is equipped with a large vibration table measuring 5 meters by 5 meters, a small vibration table measuring 2 meters by 2 meters, and more than 20 types of nondestructive equipment to verify the structural integrity of nuclear power plants. These test devices will be used for seismic verification, extreme testing, and structural integrity evaluation of major nuclear power plant equipment and structures. Currently, certification as a performance verification institution in the seismic testing field by the Korea Nuclear Safety Foundation is being pursued.

A test being conducted on a large shaking table at the Structural Seismic Performance Test Center within the Korea Hydro & Nuclear Power Central Research Institute. Photo by Korea Hydro & Nuclear Power

Wearing a safety helmet, I observed a test where the large vibration table simulated an earthquake with a gravitational acceleration of 0.2g (approximately magnitude 6.5 on the Richter scale). The MCC (Motor Control Center) equipment equipped with a seismic isolation device (which reduces seismic loads) on the vibration table swayed stably from side to side. In contrast, the MCC equipment without any device experienced increasingly severe shaking toward the top. Kim Jincheol, deputy head of the Structural Seismic Group at the Central Research Institute, explained, "Through such experiments, we can develop equipment that can operate safely during actual earthquakes."

Panoramic view of the Seismic Structural Test Center at Korea Hydro & Nuclear Power Central Research Institute. Photo by Korea Hydro & Nuclear Power

The seismic design standards for nuclear power plants differ conceptually from those for general buildings. While the Building Act aims to prevent collapse and ensure human safety, the Nuclear Safety Act aims for "normal operation without impairment of safety functions." The OPR1000 type installed domestically (such as Shin-Kori Units 1 and 2) is designed to withstand 0.2g, and the APR1400 type (such as Shin-Hanul Units 1 and 2) is designed to withstand 0.3g. The strongest earthquake recorded domestically so far was the 2016 Gyeongju earthquake, with a magnitude of 5.8 on the Richter scale, which corresponds to 0.1g in gravitational acceleration.

KHNP Central Research Institute conducts research on domestic earthquakes and tsunamis and also investigates fault lines. Park Jonghee, head of the Structural Seismic Group, stated, "Nuclear power plants installed domestically are safe from earthquakes," and added, "Various experts are conducting research to ensure that nuclear power plants can operate safely under any circumstances."