Three artificial sweeteners, sucralose, erythritol and xylitol, are qualitatively and quantitatively studied based on Terahertz time-domain spectroscopy combined with machine learning algorithms and optimization algorithms. The results show that the Sparrow Search Algorithm-Support Vector Machines/Support Vector Regression(SSA-SVM/SVR) model is optimal for qualitative and quantitative analysis of the mixture. The accuracy of classification prediction is up to 95.56%, and the optimal regression coefficient for quantitative regression prediction is 0.999 8, so that a high-precision classification and quantitative analysis of three sweetener-flour mixtures is achieved. This provides an effective and reliable method for the rapid detection of artificial sweeteners.