Abstract:The weight of dairy cows is an important data in the process of healthy breeding. Aiming at the current problems of dynamic weighing, a dynamic weighing algorithm was proposed based on empirical mode decomposition (EMD). Firstly, the nonlinear and nonstationary oscillation signals collected by the weighing equipment were preprocessed to obtain the effective part of the signal. Secondly, the effective part was initially judged, and if it met the preset condition, it was the walkingstop state, the arithmetic average method was used to obtain weighing value. Finally, if it did not accord with the walkingstop state, the EMD algorithm can be used in distinguishing the slow walking, fast walking and strenuous moving states of the animal, and calculating the dynamic weighing value. In the algorithm design, it was found that the data acquired under severe motion fluctuated greatly and the weight value needed to be calculated after filtering. The experimental results showed that the dynamic weighing algorithm proposed can judge the motion state of dairy cows. The calculated weight value was less than 0.16% in the walkingstop state compared with the static weight. The error rate was less than 1% in slow walking and fast walking. The error in the state of motion was within 1.35% under strenuous motion. The research method can provide technical support for dairy cow dynamic weighing technology.