Although externally bonding FRP with adhesive has been widely adopted for structural strengthening, the rapid and accurate prediction of early-age interfacial performance in CFRP-strengthened steel structures remained challenging. This study investigated the interfacial performance of CFRP-strengthened notched steel beams over curing periods ranging from 3 to 168 h, utilizing two types of adhesives with conventional curing (CC) and rapid curing (RC) rates. Furthermore, an electromechanical impedance (EMI)-integrated deep learning (DL) approach, based on a convolutional neural network-long short-term memory-sparrow search algorithm (CNN-LSTM-SSA) model, was developed to automatically predict bond-slip characteristic parameters at various curing stages using raw EMI responses. The results revealed that the interfacial performance varied significantly and generally improved with increasing curing time. The maximum shear stress was peaked at 72 h and 48 h for the CC and RC series specimens, respectively, with the maximum improvement reaching 238.32 % during the curing period. Moreover, the proposed model accurately predicted early-stage interfacial performance, achieving R2 values of 0.98, 0.94, and 0.97 for initial stiffness, fracture energy, and maximum shear stress, respectively. Additionally, the proposed network outperformed traditional machine learning and deep learning methods in terms of prediction accuracy, strong noise resistance (5dB), and robustness. These findings highlight the significant potential of the proposed method for the rapid and accurate estimation of early-age interfacial performance in FRPstrengthened structures.
