Common security technology for liquid dangerous goods is unreliable and inefficient. This paper proposes a approach which using characteristics of scattering parameters curve to identify liquid dangerous goods. When centimeter waves penetrate liquids that have different dielectric constants and viscosities

the characteristics of the corresponding scattering parameter curves are different. The instrument uses Ultra-wideband technology to collect a large number of sampling points. These points describe the scattering parameter curve accurately. The perceptron model classifies the liquid by identifying the characteristics of the scattering parameter curve. Experiments show that the recognition rate is over 96%.