Volatile organic pollutants produced during people’s living and production are the main source of stink pollution. The rapid and efficient detection and evaluation of odors in gas through effective methods is a key step for improving the level of supervision and ensuring people's living health. Because the chemical composition of the stink and odor gas is complex, the single odor intensity detection and evaluation method can hardly meet the requirement. Therefore, based on the odor research of domestic and foreign scholars, a multi-component odor mixture prediction model is established in this paper. The prediction model comprehensively considers and analyzes the interaction between odor and volatile organic compounds (VOCs), and the relationship between the concentration of chemical components and the intensity of odor; basing on the linear equation between the odor intensity and the logarithm of the activity to modify the U model and the vector model can better achieve detection and evaluation of the total odor intensity of the mixture gas, and can improve the prediction accuracy of the odor intensity prediction model. The research has important guiding significance for improving domestic evaluation and detection level of gas odor.