Research on Crop Planting Based on K-Means Algorithm and Bi-Level Optimisation Model

The aim of this paper is to develop a reasonable crop planting plan by constructing a research framework based on K-means algorithm and two-layer optimisation model. To address the crop cultivation problem, data are first analysed from multiple perspectives and a two-layer planning model is developed. In the upper-layer model, multiple constraints of various economic indicators are considered, aiming at maximising the total return; while the lower-layer model is based on the results of the upper-layer and optimised for the annual return under the condition of fixed resources. In order to avoid local optimal solutions, the empire competition algorithm is used for the solution in combination with the adaptive Cauchy variation mechanism. In addition, in the face of some uncertainties, this paper constructs a robust optimisation model to achieve the balance between return and risk by statically adjusting the variable parameters. Based on the multi-objective robust optimisation model, the simulated annealing algorithm was used to obtain a better planting scheme. Finally, crop characteristics were classified by Kendall's tau-b correlation analysis and K-means algorithm, and alternative and complementary coefficients were further introduced to optimise the planting decision. The research results not only provide practical planting strategies for agriculture, but also provide effective management tools to cope with risk changes.