Acoustic emission (AE) technology is suitable for monitoring the status of high-speed train bearings owing to its high sensitivity and real-time dynamic monitoring capabilities. However, a complete theoretical model of the AE sensor, which is the core component of AE signal sensing equipment, has not yet been reported. The existing matching layer models do not account for wave attenuation in the matching layer. To address these shortcomings, we established a novel piezoelectric AE sensor design and modeling method. First, a rough contact model is established for piezoelectric ceramics, and the influence of roughness on size selection of piezoelectric ceramics is analyzed. Second, the sound intensity transmission coefficient (SITC) model of the matching layer is established considering the attenuation of AE waves, and the corresponding relationship between the attenuation coefficient and the optimum thickness of the matching layer is derived. Then a complete finite element (FE) model of an AE sensor is established, and the electroacoustic properties of the AE sensor are numerically simulated based on acoustic and piezoelectric coupling. Furthermore, AE sensors with different thicknesses of matching layers are constructed, and the validity of the mathematical model of the matching layer is verified through a lead-breaking experiment. Thereafter, a novel comprehensive performance evaluation index (CPEI) is designed through principal component analysis (PCA) based on hit parameters of AE sensors. Finally, the effectiveness and environmental adaptability of the AE sensor is verified by performance testing under complex conditions near an actual high-speed train line. The proposed method can provide a valuable theoretical framework for AE sensor design and status monitoring of high-speed train bearings.