In order to detect the content of benzoic acid in corn flour, control the amount of benzoic acid used in corn flour to improve food safety grade, the content of benzoic acid in corn flour was studied using terahertz spectroscopy technology, and the terahertz benzoic acid spectral data of different mass fractions in corn flour was obtained. In order to improve the accuracy of the model, pre-processing methods such as moving average smoothing algorithm, standard normal transformation, multi-scattering correction, baseline correction, and normalization were used to eliminate the original spectral noise and useless information. The least squares support vector machine(LS-SVM), partial least squares(PLS), and multiple linear regression(MLR) spectroscopy models were constructed, and the model samples were not used for model evaluation. The model was evaluated based on the prediction set correlation coefficient eRMSEP and the prediction set root mean square error Rp. The results show that the least squares support vector machine model with the original terahertz time-domain spectral preprocessing has the strongest evaluation ability. The correlation coefficient of the prediction set Rp=0.9958, and the root mean square error of the prediction(RMSEP) set eRMSEP=0.0057, indicating the metrology method of THz-TDS binding chemistry can be used to detect the quantitative determination of benzoic acid in corn flour.