Maize Growth Simulation Analysis and Yield Prediction Based on Integrated Learning Algorithm and WOFOST Model

Maize growth status and yield prediction are the key issues in maize planting management. Traditional prediction methods rely on single crop growth models or machine learning algorithms, which have the problems of insufficient prediction accuracy and limited adaptability. The combination of ensemble learning and crop model provides a new idea for improving the prediction accuracy. In this study, an improved maize growth prediction model was proposed based on the WOFOST model, combined with the irrigation model and the ensemble learning algorithm. By introducing K-means clustering and data assimilation technology, the model parameters were optimized to improve the prediction performance. The experimental data are derived from NASA's meteorological data from 1990 to 2020, and the results show that the improved model performs well in the prediction of key indicators such as leaf area index, soil water content and total biomass, and its MAE and MSE are reduced to 2.836 and 7.581, respectively, with a coefficient of determination R—2 It is improved to 0.942, which is significantly better than the traditional WOFOST model and other machine learning algorithms, demonstrating the efficiency and reliability of the improved model in maize growth prediction.