The sensory evaluation of these samples was conducted in tea research institute of Yunnan academy of agriculture science, and the analysis of inner components in Pu-erh tea college of Yunnan agricultural university.

Firstly, a large number of sample data of ammunition internal component is obtined by ultrasonic detector, by using the data fusion method, the neural network fusion modle is built to fusion and handle the feature of the first wave amplitude, frequency, gain and other characteristics of integration, and eventually the distinction is realized.