中文摘要：

采用定量结构-活性（QSAR）方法,分别建立了用32个有机挥发性化合物的结构描述符来预测味阈值（logNPT）的线性和非线性模型,所建模型的预测值与实验值都能很好的吻合.本文为预测有机挥发性化合物的味阈值提供了一个快速、简单、有效的途径.

英文摘要：

Nasal Pungency Thresholds（NPT） are important properties of compounds,and quantitative predic-tion modles of NPT are established in the present study.Quantitative structure-activity relationship（QSAR） approaches have been performed to study the NPT of 32 volatile organic compounds（VOCs） by multiple linear regression（MLR） and non-linear radial basis function neural network（RBFNN） methods.All the predicting results are in agreement with the experimental values well.The QSAR models provide a rapid,simple and valid way to predict the NPT of VOCs.