China’s agricultural water resources are scarce, irrigation water utilization is low, and waste is serious. In water-saving irrigation systems, the precise measurement and prediction of root soil moisture is of great significance. The BAS-Beetle antennae search algorithm performed well in numerical prediction in recent years, but it is greatly affected by the initial value. A beetle in each iteration of the traditional beetle antennae search (BAS) was improved into a beetle population. The IBAS-BP prediction model was established and optimized by the improved beetle antennae search algorithm, and the measured shallow root soil moisture content data was used to predict the soil moisture content of winter wheat root system at a depth of 50cm. The results showed that compared with the PSO-BP prediction model, GA-BP prediction model and the original BAS-BP model, the IBAS-BP model not only effectively avoided the possibility of the network falling into a local minimum, but also had higher prediction accuracy and better robustness. Based on this, accurate predictions can be made on the soil moisture content of the winter wheat root system, which provided a basis for the rational use of agricultural water resources and the construction of water-saving agriculture.