Abstract: Based on Fourier infrared spectroscopy, the compressive strength and spectral data of the Zenia insignis samples were measured. After eliminating the outliers, the spectral data were preprocessed by multiple scattering correction, and the SPA algorithm was used to select the spectral wave number. The compressive strength prediction model was established by using PLS, and the thickness and width of the samples were added as independent variables for comparison. Results show that after pretreatment and the SPA algorithm is adopted to the spectral data of band selection can strengthen the prediction ability of the spectrum, the prediction model of standard deviation determination coefficient R² and forecasting SEP 0.9000 and 1.5366, respectively, can satisfy the demand of Z. insignis nondestructive testing, if add the width and thickness of the sample as the independent variable, establish the forecast model of R² and SEP were 0.9046, 0.9325 and 1.7109, 1.3685.