It is of great practical significance to develop high-efficiency farmland water-saving irrigation technology. Existing irrigation systems lack decision-making guidance, can only achieve uniform irrigation in the field. In order to solve this problem, the ZigBee wireless network was used to do the real-time data acquisition of embedded moisture sensor in paper. And then put forward a variable irrigation data processing method based on the quadratic smoothing prediction algorithm. This method can adjust smooth weight according to the theory value to achieve the optimal prediction data, finally get the variable operation prescription map. The sprinkler machine variable control system was developed here. This system is mainly composed of Siemens s7-200 programmable logic controller, PC, soil water sensor, GNSS position information module, etc. It can adjust the sprinkler machine step by step by simulating the percentage timer, and real-time adjust the walk step length and speed to realize variable irrigation. In order to evaluate the performance of the system, the effect evaluation experiment of traditional irrigation and variable irrigation was carried out. Field contrast test results indicated that variable irrigation operation efficiency was higher than traditional irrigation about 23.19%. The practical application shows that the designed variable irrigation system has good applicability and operability, can significantly improve the irrigation effect, and achieve the purpose of saving water and increasing income.