To effectively identify various cavitiesin tunnel lining structures, the finite difference time domain method was used to forward simulation. The influence of cavity shape, filling medium and burial depth on electromagnetic reflection imaging effect and time-frequency characteristics of GPR was analyzed respectively. And those analyses were then validated through laboratory model tests.Results indicate that the reflection response characteristics of cavities are manifested by a downward-opening parabola. As the cavity top-boundary transitions gradually from straight line to arc or sharp angle, the number of reflection wave in the forward simulation decreases, the reflected signal is abate. Moreover, the energy in the continuous wavelet transform time-frequency diagram decreases gradually. The shallower the buried depth of the cavity and the larger the dielectric constant of the cavity filling medium, the stronger the simulated image signal. The results can provide reference for the prediction and judgment of disease development around cavities in tunnel lining structures.