This work presents simulation results on passive structural health monitoring of a vibrating elastic structure for defect localization. We consider an object that consists of a shell with vibration sensors surrounding a complex, vibrating structure such that they are coupled at a finite number of points. The non-defective model of the external shell is called an acoustic digital twin (ADT). We use a combination of the concept of an acoustic digital twin (ADT) together with an adjoint-based high resolution array processing approach to detect a defect. The “dataset” in the presence of a defect was obtained from simulations and was “measured” on a linear array placed on the surface of a candidate ADT. A set of sensors placed at the connecting points provides the time histories of the excitation forces, which are used as inputs to the ADT in order to obtain the fields at the surface in the unperturbed case (no defect). The residual field (difference between defect and non-defect case) was used in the Matched Field Processing (MFP) algorithm in order to successfully locate the defect on the surface of the object.