Machine learning and bayesian network based on fuzzy AHP framework for risk assessment in process units

Risk assessment plays a crucial role in ensuring the safety of process units. Artificial intelligence has become increasingly prevalent in risk assessment and prediction, offering the potential for more precise outcomes when integrated with other techniques. This study is both descriptive and analytical in nature. The dataset utilized comprises 160 deviations identified through the HAZOP technique. A variety of evaluation algorithms were employed in this study, ranging from ensemble methods like Random Forest, Hist Gradient Boosting, XGBoost, and CatBoost, to traditional methods such as Logistic Regression, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Convolutional Neural Network (CNN). This broad array of algorithms enabled a comprehensive comparison of diverse modeling approaches, encompassing conventional statistical methods and cutting-edge machine-learning techniques. Among the algorithms tested, Random Forest, XGBoost, and CatBoost exhibited exceptional performance on the training and test datasets, achieving near-perfect AUC scores and accuracy values of 1.0000. In the fusion of Bayesian networks and Multi-Criteria Decision Making (MCDM), the options "Corrosion in Electrolysis Cells" and "Damage and Explosion of Cells" were given higher priority over other options. The findings from this study suggest that machine learning techniques, along with the amalgamation of Bayesian networks and MCDM, can serve as effective tools for risk assessment and the prioritization of risk options. By leveraging these methodologies, suitable control and preventive measures can be implemented to mitigate risks effectively.