摘要：In view of the cow extensive farming, which lacks modern management tools, the cow estrus determine are often laid off. In this paper, a cow behavioral characteristic is designed. The system collects X, Y, Z axis acceleration data through triaxial accelerometer ADXL345, by ZigBee wireless network and RS485 sent to the information center, it's based on the support vector of binary tree to classify their behavior. The result shows, the algorithm for cow's stationary and moving classification accuracy was 93.26%, a slight and sharp motion accuracy was 84.81%, It provides an effective basis in judging cows in heat or in physical abnormalities. That system contributes to the refinement of cows feeding and health culture,and has great significance for the stable and healthy development of the dairy industry
