Research on Resource Allocation and Water Saving Strategies in Deep Bayesian Network Driven Farmland Irrigation Systems

Aiming at the uncertainty of the solving algorithm of multi-objective optimal allocation model of water resources in irrigation area and the problem of optimal scheme selection, a multi-objective optimal allocation of water resources in irrigation area and scheme preference model based on Deep Bayesian Networks (BDNNs) driven by the optimized allocation of resources in the irrigation area is established with the agricultural irrigation system as the object of the study, and optimal allocation of resources in the irrigation area and water saving as the goal. Finally, this model is used to optimize resource allocation in the farmland irrigation system and save water resources. The results showed that in this paper, based on two irrigation schemes, the annual watering amount (335, 243, 384, 220 and 266 cubic meters) and the ratio of canal-well water use (0.867:0.699) in the flat water year for five crops, including barley and spring wheat, were determined for the final scheme. In addition, the irrigation water for winter, spring and summer canal-well consolidation was obtained as 0.882, 1.611 and 1.962 (10⁸·m³), respectively. Under the conditions of the final sub-optimized combined scheme, the irrigation ratios of the flat water year and the extra dry water year in the winter, spring and summer flat water years were determined to be 0.441:0.8055:0.9810 and 0.630:0.735:0.858, respectively. Obviously, reasonable regulation of the amount of canal irrigation diversion and the amount of agricultural irrigation extraction in different periods of time of each farmland irrigation area can make the intra-annual fluctuation of the groundwater level slower, and make the irrigation area’s groundwater level remain within a reasonable range.