<p>Ensuring the reliability of the intelligent manufacturing system (IMS) is crucial to advancing the transformation of traditional manufacturing into intelligent manufacturing. However, the inherent complexity and dynamic nature of IMS create significant challenges for risk and uncertainty management. Failure mode and effects analysis (FMEA) serves as a valuable tool for identifying and evaluating potential system failures. This paper addresses traditional FMEA limitations in complex IMS by proposing an innovative framework. First, we extend the conventional risk factor set by incorporating repair and economic factors for a more comprehensive evaluation. The double hierarchy hesitant triangular fuzzy linguistic term set enables experts to express multi-dimensional emotional tendencies in hesitation. Moreover, the method based on the removal effects of criteria is employed to determine the weights of risk factors in IMS, effectively dealing with partial correlations among them. Furthermore, the ranking of failure modes utilizes the fuzzy exp-TODIM method based on weighted generalized hybrid Hausdorff-preference distance, which improves the measurement of distances between hesitant evaluations and enhances sorting precision. Finally, the framework is validated through an IMS case study, with sensitivity and comparative analyses confirming its robustness and superiority. Our approach contributes to IMS reliability analysis and establishes new directions for complex system reliability research.</p>