Abstract: In order to improve the accuracy and stability of the liquid fertilizer variable control system, the control system has better anti-interference and adaptive capabilities. Taking the liquid fertilizer variable control system as the research object, the mathematical model of the liquid fertilizer variable control system is established, and the conventional PID control algorithm, fuzzy adaptive PID control algorithm, BP neural network PID control algorithm are used, and the three control algorithms are simulated using MATLAB. The system quality and performance parameters are analyzed from the system simulation results. The control system uses BP neural network controller to perform actual system data collection to verify the actual operation effect of the system. By comparing the quality indicators of each algorithm, the rise time, transition time, static difference, and overshoot of the BP neural network are better than conventional PID and fuzzy adaptive PID algorithms. It is verified that the BP neural network PID control uses the liquid and fertilizer variable control system as the control object. The actual operation effect shows good robustness, response speed and operation accuracy within 5%, which can realize variable operation. The performance of BP neural network PID control algorithm applied to liquid fertilizer variable control system is better than PID and fuzzy adaptive algorithm.