Abstract: Low-field nuclear magnetic resonance(LF-NMR)was used to analyze the variation of the moisture content inside the broad bean and different water states. Based on the mathematical model of heat transfer and mass transfer of broad bean,the dehydration process of broad bean was numerically simulated by COMSOL Multiphysics software. The effects of temperature and pressure on the dehydration characteristics of broad bean were obtained under simulated environmental conditions of 0,4 and 8℃ and the vacuum degree of 95 kPa,97 and 99 kPa. In addition,the temperature and water distribution at different points of broad bean were analyzed. The results showed that there was a significant linear relationship between the moisture content and the NMR signal amplitude of broad bean with the determination coefficient R²=0.972 and the significance level P<0.01.During the vacuum dehydration process,the free water evaporated at the fastest rate,the semi-combined water content showed an increasing-decreasing trend and the final content decreased by 88.5%,the bound water content had only a small fluctuation and the final content decreased by 45.3%. The increasing of vacuum pressure would promote the evaporation of water,and the increase of temperature would shorten the dehydration time.