Contact Mistuning Identification in Bladed Disks Using Limited Experimental Data and Data-Driven Techniques

Contact-induced mistuning at the blade–disk interface is a major source of variability in the dynamic response of bladed disks, yet its identification remains challenging due to the intrinsic complexity of the contact phenomenon. Contact mistuning in this framework is defined as the variation of the contact topology at the blade–disk interface, i.e., the spatial distribution of contacting and non-contacting regions governing the mechanical interaction between components. This work proposes a data-driven framework for the identification of contact mistuning based on a reduced set of experimentally measured maps. A statistical representation of the contact space is constructed using Principal Component Analysis. This representation is subsequently exploited to generate physically consistent synthetic contact patterns. These are combined with blade intrinsic mistuning parameters to build a dataset linking contact conditions and blade variability to the natural frequencies of a disk–one-blade assembly. A k-nearest neighbors algorithm is employed for inverse identification using a subset of measured natural frequencies. The identification relies on modes selected through sensitivity analysis, and robustness is improved using a weighted distance metrics. The methodology is validated using both a reduced-order model based digital twin and experimental data. The results show accurate identification of contact patterns, confirmed through comparison of modal properties of the full mistuned assembly. Therefore, the proposed framework provides a practical tool for the experimental identification of contact mistuning in bladed disks.