하구둑 주요 변수 기반 베이지안 모델을 활용한 종합성능등급 예측에 관한 연구 = A study on the prediction of comprehensive performance grade using a bayesian model based on key variables of estuary barrage

다국어 초록 (Multilingual Abstract)

Estuary barrages are critical water infrastructure facilities exposed to complex external loads such as upstream flooding and tidal backflow. Given that failures may result in large-scale inundation and the disruption of essential functions, the development of a quantitative performance prediction model is essential. This study proposes a Bayesian Network–based performance assessment model to quantitatively predict the comprehensive performance grade of estuary barrages. Six key variables representing structural, operational, and environmental factors were selected, and their causal relationships were represented in a DAG. A total of 148 performance assessment cases were collected and divided into a training set (118 cases) and a validation set (30 cases) to derive CPT based on empirical data. The validation results indicated that the proposed model shows satisfactory predictive performance, with an accuracy of 0.867 and a macro F1-score of 0.883. In particular, the model achieved a balanced and stable prediction for grades B and C, and misclassifications were limited mainly to adjacent grades in the confusion matrix analysis. Nevertheless, the proposed Bayesian model can serve as a practical decision-support tool for predicting performance degradation in estuary barrages, and its predictive capability could be further improved by incorporating time-series data and additional performance grade cases in future studies.