Innovative Risk Analysis in Petrochemical Processes: A Text Data Mining Approach for HAZOP Studies

This research presents a comprehensive methodology leveraging text data mining techniques on Hazard and Operability (HAZOP) data to intelligently analyze emerging petrochemical processes. The approach involves extracting subject nodes information related to deviations, causes, and consequences, enabling the intelligent forecasting of overall risk levels for potential abnormal events. Utilizing Natural Language Processing (NLP) with the TF-IDF approach, features determining outcomes are transformed into feature vectors from sentences in the HAZOP study worksheet. Various machine learning models, including KNN, Random Forest, Decision Tree, Support Vector Machine, Gradient Boosting, Neural Networks, and an Ensemble model, are applied to predict overall risk for each node. The results showcase the efficacy of the proposed methodology, demonstrating improved accuracy through KNN classifiers compared to other models. Hyperparameter tuning further enhances the performance of the KNN classifier. This research contributes to prioritizing nodes, examining accident patterns, and supporting HAZOP analysis for enhanced safety in petrochemical processes.