The Prediction of Apple Pests Based on CEEMD-GWO-GRU

The study aimed to address the harm caused by frequent occurrence of Carposina sasakii by proposing a predictive model (GEEMD-GWO-GRU) and a warning mechanism. This model combined Complementary Ensemble Empirical Mode Decomposition (CEEMD) and Grey Wolf Optimization Algorithm (GWO) with Gated Recurrent Unit (GRU). The historical data on Carposina sasakii was first decomposed using CEEMD, then each eigenfunction modeled through GWO-GRU. Finally, the prediction of each eigenfunction was integrated to develop an apple and peach microcephalus prediction and early warning model based on GRU. Results indicated that the CEEMD-GWO-GRU model was more accurate in predicting apple Carposina sasakii disease compared to other methods, displaying an average absolute percentage error of 0.823% and a coefficient of determination of 0.961. This method has potential as a new strategy for agricultural pest and disease prediction.