Fourier infrared spectroscopy (FTIR) technology is an important method for material characterization. However, it is limited by the diffraction limit. The spatial resolution limit of traditional Fourier infrared spectrometer is on the order of micrometers and cannot be applied to the characterization of nanomaterials. Nano Fourier transforms infrared spectroscopy (Nano-FTIR) is an emerging super-resolution spectroscopic surface analysis technology, which exhibits tremendous performance in nanomaterial characterization research with the characteristics of nano-scale spatial resolution, wide spectral range and high chemical sensitivity. Qualitative and quantitative research on the source of the high spatial resolution of Nano-FTIR signals and the extraction process of spectral signals in the system can provide the important basis for the design and development of Nano-FTIR instruments and the interpretation of sample spectral characterization results. Based on the typical instrument structure and basic working principles, an equivalent research model is established in the multi-physics finite element analysis software COMSOL, and the important details of the model and the numerical calculation process are explained separately. In the simulation study, the model first calculated the electromagnetic field enhancement in the model space based on Maxwell’s electromagnetic wave theory, and then simulated the “line sweep” process of the probe at the interface between two materials with large differences in dielectric constants, discussed the near-field enhancement of the tip-sample and the spatial resolution of the signal. Subsequently, a numerical model was proposed with the scattered power of the probe and sample as the research object, and the process of the modulation and demodulation extraction of the signal by the probe tapping, different incident angle of light and demodulation frequencies were also discussed. Finally, in order to verify the rationality of the model, the spectral responses of SiO2 thin film samples of three thicknesses of 20, 100 and 300 nm in the wavenumber range of 900～1 250 cm-1 were simulated, and the simulated spectra were compared with the measured results. The results show that as the thickness of the sample increases, the spectral signal is correspondingly enhanced and predicted spectra in agreement with the experimental spectra. Spectra predicted by our model are more consistent in peak shape compare with spectra simulated by the tip-sample electric field strength used to some previous studies. The proposed numerical model can be used for the prediction of Nano-FTIR spectra, in addition, the model also has certain generality, and can be used for Tip-enhanced Raman spectra and Terahertz spectra based on scattering near-field optical microscopy (s-SNOM) technology.