Farmland irrigation is an essential foundation for good crop growth, while traditional farmland irrigation techniques cannot fully consider the impact of factors such as natural precipitation and crop transpiration on crop growth, which can, to a certain extent, result in poor irrigation decisions and a complex farmland environment that cannot be monitored promptly, thereby reducing farmland production efficiency. This study designs a farmland irrigation control system based on a composite controller. Firstly, an irrigation control method is proposed to establish a prediction model for future rainfall and crop transpiration using historical meteorological data. The composite controller is designed based on the prediction model to realize an irrigation control operation with an irrigation value as the control quantity, a water and fertilizer machine, and a solenoid valve as the actuators. Secondly, an intelligent irrigation control cloud platform based on Java language is designed to monitor farm information and irrigation operation records in real-time to facilitate visual management. Finally, the prediction accuracy is high, based on the prediction model results, which can provide a specific reference basis. The superiority of the proposed controller is verified by simulation using MATLAB/Simulink. The results show that the proposed controller can be well suited for nonlinear control systems and has good control performance while ensuring high tracking accuracy, strong robustness, and fast convergence.