Rice, as one of the world’s major staple crops, is highly susceptible to various factors such as extreme weather conditions, differences in cultivation types, and the diversity of varieties. Rice crops are increasingly threatened by various pests and diseases, particularly by “double-migration pest” (Brown planthopper and Rice leaf roller), which have shown a tendency for severe infestations. The current rice industry faces widespread issues of excessive pest control, leading to pesticide residue exceeding safe limits, causing environmental pollution in farmlands, and posing a threat to food security to some extent. This paper focuses on Brown planthopper and Rice leaf roller in Hunan Province, proposing a pest prediction method based on GCN-AGRU using multidimensional data collected from multiple pest monitoring stations in Hunan. This method considers the mutual influence of meteorological conditions and pest occurrences in various counties and cities, constructing a graph structure that reflects the spatial relationships between monitoring stations. By calculating the distance weights between stations, the model effectively identifies the spatial dependencies of pest occurrences. Additionally, GRU is introduced to enhance the ability to extract temporal sequence features, and an attention mechanism is employed to identify important features. Experiments demonstrate that the proposed GCN-AGRU prediction method achieves high accuracy and reliability in predicting pest trends over multiple days.
本站僅提供期刊文獻檢索。
