Composite materials have low quality and high strength, and are widely used in many industrial equipment such as ships, aircraft and automobiles. The cantilever beam structure is one of the main structures of equipment failure, and their main failure mode is crack. In this study, the crack of composite cantilever beam is identified by the constructed digital model and monitoring data. First, the forward modeling of the composite beam whose geometric shape and material properties have been determined is carried out to calculate the natural frequencies of the healthy beam. Secondly, the response data of composite beams are monitored in real time, and the characteristics of the monitoring data are extracted through signal processing methods to obtain the corresponding natural frequencies of the monitoring beams. The monitored natural frequency is compared with the natural frequency calculated by the forward model to obtain the relative change rate of the natural frequency. Finally, an artificial intelligence algorithm is constructed to determine the crack location according to the modal curvature, establish the potential relationship between the geometric size of the crack and the relative change rate of the natural frequency, so as to realize the crack identification of the composite cantilever driven by the digital model. The proposed method can not only accurately identify crack for the composite beam, but also provide an effective mean for predictive maintenance.