Akademik Veri Yönetim Sistemi
International Conference on Intelligent and Fuzzy Systems (INFUS 2025), İstanbul, Türkiye, 28 Temmuz 2025, cilt.1, ss.688-697, (Tam Metin Bildiri)
This study addresses the critical need for advanced risk assessment in unde ground mine sites by proposing a hybrid model of two machine learning algo- rithms: Random Forest (RF) and Support Vector Regression (SVM) and Fuzzy Inference System (FIS). Building on the success of these models, the given study aims to standardize the assessment of occupational health and safety risks in activities conducted in underground mines using an artificial expert system. Initially, datasets were created by leveraging real expert judgments for specific conditions. Subsequently, these datasets were employed in training the SVM and RF machine learning algorithms to create a decision support system. This decision support system linguistically expresses the probability of risk occu- rence by utilizing the severity of damage and combinations of potential hazards and processes based on the conditions of the conducted activity. Linguistic expressions were matched with triangular fuzzy numbers and the numerical repr sentation of risk situations in activities was determined employing the princples of fuzzy inference systems. Based on the data obtained, activities were priortized according to the risk levels they encompass. Risk mitigation planning was then organized based on prioritization, addressing more risky activities before those with lower risk levels. The results of this study offer a practical framework for assessing and addressing risks in real-world scenarios, providing valuable insights for effective risk management in underground mining operations.