Study on Quantifying Small RCS Leaks Using Deep Learning Sang Hyun Lee Advisor : Prof. Man Gyun Na, Ph.D. Department of Nuclear Engineering, Graduate School of Chosun University In nuclear power plants reactor coolant leakage can occur due to a variety of factors. Coolant leakages can lead to casual...
Study on Quantifying Small RCS Leaks Using Deep Learning Sang Hyun Lee Advisor : Prof. Man Gyun Na, Ph.D. Department of Nuclear Engineering, Graduate School of Chosun University In nuclear power plants reactor coolant leakage can occur due to a variety of factors. Coolant leakages can lead to casualties and economic losses. To prevent this, early detection of leakages is crucial to ensuring the safety of nuclear power plants. Currently, a detection system is being developed in Korea Atomic Energy Research Institute to identify reactor coolant system (RCS) leakages of less than 0.5 gpm. Typically, RCS leakage is detected by monitoring the temperature, humidity, radiation, and sump water level in the containment. However, detecting small leakages proves difficult because the resulting changes in containment humidity and temperature and sump water level are small. To address these issues and enhance the speed of leak detection, it is necessary to quantify the leakages and develop an artificial intelligence-based leakage detection system. In this study, temperature, relative humidity of the measured area, and distance were used as input variables among the variables obtained using the CUPID code. And, long short-term memory, bidirectional long short-term memory, and gated recurrent unit, were employed to predict the relative humidity in the leakage area for leakage quantification. Additionally, an optimization technique was implemented to reduce the learning time and improve the prediction performance. Through evaluation of the developed artificial intelligence model's prediction accuracy, the initial relative humidity of the leakage area was accurately predicted. And the amount of leakage was quantified, which is expected to be used in a future leakage detection system to detect small-scale coolant leakage by applying artificial intelligence even in the event of instrument failure or sensor failure. In addition, it is expected to secure time margin for operators to take actions and contribute to reducing secondary human errors.
Study on Quantifying Small RCS Leaks Using Deep Learning Sang Hyun Lee Advisor : Prof. Man Gyun Na, Ph.D. Department of Nuclear Engineering, Graduate School of Chosun University In nuclear power plants reactor coolant leakage can occur due to a variety of factors. Coolant leakages can lead to casualties and economic losses. To prevent this, early detection of leakages is crucial to ensuring the safety of nuclear power plants. Currently, a detection system is being developed in Korea Atomic Energy Research Institute to identify reactor coolant system (RCS) leakages of less than 0.5 gpm. Typically, RCS leakage is detected by monitoring the temperature, humidity, radiation, and sump water level in the containment. However, detecting small leakages proves difficult because the resulting changes in containment humidity and temperature and sump water level are small. To address these issues and enhance the speed of leak detection, it is necessary to quantify the leakages and develop an artificial intelligence-based leakage detection system. In this study, temperature, relative humidity of the measured area, and distance were used as input variables among the variables obtained using the CUPID code. And, long short-term memory, bidirectional long short-term memory, and gated recurrent unit, were employed to predict the relative humidity in the leakage area for leakage quantification. Additionally, an optimization technique was implemented to reduce the learning time and improve the prediction performance. Through evaluation of the developed artificial intelligence model's prediction accuracy, the initial relative humidity of the leakage area was accurately predicted. And the amount of leakage was quantified, which is expected to be used in a future leakage detection system to detect small-scale coolant leakage by applying artificial intelligence even in the event of instrument failure or sensor failure. In addition, it is expected to secure time margin for operators to take actions and contribute to reducing secondary human errors.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.