The growth of field crops needs appropriate soil nutrients. As a basic fertilizer, liquid manure provides biological nutrients for crop growth and increases the content of organic matter in crops. However, improper spraying not only reduces soil fertility but also destroys soil structure. Therefore, the precise control of the amount of liquid manure is of great significance for agricultural production and weight loss. In this study, we first built the model of spraying control, then optimized the BP neural network algorithm through a genetic algorithm. The stability and efficiency of the optimized controller were compared with PID, fuzzy PID and BPNN-PID control. The simulation results show that the optimized algorithm has the shortest response time and lowest relative error. Finally, platform experiments were designed to verify the four control algorithms at four different vehicle speeds. The results show that, compared with other control algorithms, the control algorithm described here has good stability, short response time, small overshoot, and can achieve an accurate fertilizer application effect, providing an optimization scheme for research on the precise application of liquid manure.