Research on Multi-objective Planning Optimization Algorithm Based on Cluster Analysis

Abstract

AI has introduced new solutions to multi-objective decision-making in agricultural resource optimization. Still, the current methods struggle with handling complicated constraints and heterogeneous resource types effectively. This study proposes an integrated approach combining linear programming (LP) with K-means++ clustering to enhance the adaptability in diverse agricultural contexts. Specifically, the linear programming (LP) model firstly optimizes resource allocation according to real-world constraints, while K-means++ clustering groups crops with similar characteristics, selecting representative crops from each cluster, and allows simplification of crop structure and enhancing resilience. Compared to traditional K-means, K-means++ improves the stability of the initialization, ensuring more reliable crop clustering results, and largely avoids local minima. This approach has been applied to a case study in a village of Northern China, demonstrating its effectiveness. Using the LP model alone increases annual revenue from 5.66 to 6.58 million yuan annually, while combining with K-means++, revenue increases further to 7.89 million yuan, exceeding 2.23 million yuan compared to the original. This research also integrates risk modeling, simulating disruptions from the environment and market, demonstrating strong robustness in maintaining revenue stability in different scenarios. This hybrid model not only improves profitability but also provides a scalable and robust strategy for sustainable agricultural planning.