Study on the Agricultural System in a Certain Region Based on the LSTM Model and Genetic Algorithms

In response to the two core demands in agricultural production, namely precision irrigation and sustainable transformation, this study conducts two key research tasks. First, it predicts the crop irrigation demand based on meteorological data. Initially, the 3σ principle is adopted to identify and label outliers isolated outliers are filled using linear interpolation, while consecutive outliers are filled with average value. Subsequently, two types of models are constructed: on the one hand, three evapotranspiration formulas (Hargreaves, Priestley-Taylor, and Makkink) are integrated, and the evapotranspiration (ET) amount is calculated through weighted fusion to establish the ET formula; on the other hand, a Long Short-Term Memory (LSTM) model is built via a two-layer architecture. These two models are then compared. Second, it carries out research on the transformation of organic agriculture based on farm economic and environmental data. First, farms are classified into geographical types (Plain, hilly, and mountainous). Then, a multi-objective optimization model is constructed with the goals of maximizing economic benefits and maximizing environmental benefits. Combined with constraints on transformation ratio and annual transformation, the genetic algorithm is used to solve the model.