Establishment of Wheat Yield Prediction Model in Dry Farming Area Based on Neural Network

Abstract

The accurate prediction of wheat yield based on the wheat yield data of the previous years can help to balance the relationship between market supply and demand and formulate appropriate purchase price and make allocation plans, which has important practical significance for the study of national economic industrial structure. This study proposes an artificial neural network model for wheat yield prediction by simulating the information processing mode of brain neural network and combining the characteristics of wheat yield data in dry farming areas, uses IOWA operator to establish a prediction model combining BP, RBF, and GRNN neural network models, and conducts an empirical study on the models. The results show that the average error rate of BP, RBF, GRNN and IOWA-based combined neural network model is less than 10%, and the error rate of the combined prediction model is the smallest, suggesting that these four kinds of neural network models have a satisfactory effect on wheat yield prediction and the combined neural network prediction model has a better accuracy and effect for wheat yield prediction in the dry farming areas, providing an important theoretical basis for the accurate prediction of wheat yield in dry farming areas in the future.