A novel intuitionistic fuzzy OPARA-based multi-criteria decision-making framework for the evaluation of AI-enabled intelligent fault diagnosis systems

In intelligent fault diagnosis, an intelligent decision-making framework that is suitable to cope with uncertainty, vagueness, and conflict in the criterion should be adopted in AI-based diagnostic systems. These complications cannot be best captured through the traditional multi-criteria decision-making (MCDM) models, which compromise the reliability and accuracy of the models. In order to solve this issue, the paper will present a novel method, the intuitionistic fuzzy optimized pairwise ratio analysis (IF-OPARA), an effective method to reflect the contribution of expert opinions and performance ratings in terms of intuitionistic fuzzy numbers (IFNs). The integration allows a better depiction of uncertainty and gives a better view of the expert judgments during the process of the system-level assessment. Represented with a hypothetical case study centered on the evaluation and selection of AI-enabled fault diagnosis systems, three parameters of diagnostic accuracy, implementation cost and real-time adaptability, and three solutions, one of them being a model-based diagnosis, the other is a data-driven diagnosis, and the hybrid diagnosis, the proposed IF-OPARA framework gives the ranking score of A₂ = 0.452, A₁ = 0.312, and A₃ = 0.236, and the data-driven diagnosis is the most appropriate solution. The introduction of IF-OPARA indicates that the data-driven diagnosis takes first place on the list and, to a greater extent, can be characterized as a good compromise with the level of accuracy, cost, and flexibility. In addition, a comparison of IF-OPARA with other existing MCDM methods reveals that IF-OPARA is the strongest ranking method, the ranking order does not change with the changes of the main parameters, which is the insensitivity of the ranking method to the changes in the key parameters and sensitivity to the changes of the criteria. The results prove that IF-OPARA can produce a computationally efficient, reliable, and adaptable decision-support framework to estimate AI-based intelligent fault diagnosis frameworks with good potential to be applied to real-life settings where uncertainty is one of the most important parameters. • Proposes IF-OPARA, merging intuitionistic fuzzy sets with OPARA • Avoids normalization, preserving original decision-making data • Introduces range and nonlinearity adjustment for fair evaluations • Outperforms TOPSIS, WASPAS, CODAS, EDAS, and MARCOS in robustness • Offers reliable, efficient tool for AI-enabled fault diagnosis systems