Publication

Abstract : Reliability modeling is a pivotal aspect of evaluating and enhancing the performance and trust worthiness of intricate systems. Conventional reliability models frequently encounter difficulties when dealing with the inherent uncertainties and ambiguities present in real-world systems. In recent years, Rough set theory has emerged as a promising approach to reliability modeling, particularly for handling imprecise and incomplete data. This work sheds light on a holistic approach to reliability modeling that leverages Rough set theory. Our method seamlessly integrates Rough set principles with established reliability analysis techniques, effectively mitigating the constraints associated with conventional models. By leveraging the strengths of Rough set theory, the proposed approach enables more accurate and robust modeling of complex systems. The key steps of the proposed approach include data pre-processing, attribute reduction, rule generation, and reliability evaluation. By exploiting the power of Rough set, a set of rules is generated to describe the relationships between system attributes and reliability outcomes. These rules provide valuable insights into the factors that significantly impact system performance and give optimal decision-making for reliability improvement. Finally, the reliability of the system is evaluated based on generated rules, enabling the estimation of failure probabilities and reliability metrics. We put our approach to the test through a case study involving a sophisticated industrial system. The results reveal its proficiency in comprehending and evaluating the complex interconnections among system attributes and reliability. Reliability modeling with the Rough set approach outperforms fuzzy reliability models, primarily because fuzzy sets fail to effectively capture type 1 uncertainty. When it comes to estimating type 2 uncertainty in the context of industrial reliability problems, both type 2 fuzzy sets and rough sets are commonly utilized.