Author(s)

Jintao Yang, Zhuo Liu, Shengkai Wan

Corresponding Author

Jintao Yang

Abstract

This paper focuses on determining the optimal crop planting scheme. After comparing traditional models such as linear programming and stochastic optimization, a modeling approach employing a hybrid optimization algorithm and multi-objective optimization strategy is adopted. Under the assumption of stable crop yields, planting costs, and sales prices, the model utilizes a genetic algorithm (GA) to optimize the allocation of crop planting areas. Additionally, reinforcement learning via Q-learning dynamically adjusts crossover and mutation operations to enhance the global search capability. By introducing two scenarios—unsalable crops and price reductions—and employing penalty functions to manage risks, a multi-objective programming model is established. The model is solved using the non-dominated sorting genetic algorithm (NSGA) to obtain Pareto optimal solutions, leading to the determination of optimal planting schemes for both scenarios. The results demonstrate the model's high profitability and stability across different conditions.

Keywords

Linear Programming, Non-Dominated Sorting Genetic Algorithm, Q-Learning