冬小麦水肥生产函数的PSO-SVM模型

PSO-SVM Model for Water-Fertilizer Production Function of Winter Wheat

作物的水肥生产函数研究是非充分灌溉理论的重要内容,也是通过灌溉制度优化提高农田水、肥利用效率的基础。本文建立了施肥-灌水-产量的支持向量机(SVM)预测模型,并通过粒子群优化算法(PSO)对支持向量机(SVM)的参数进行优化选择,得出最优的水-肥-产量支持向量机(SVM)预测模型,基于潇河试验站的灌水试验数据,分别建立了水肥生产函数的BP神经网络模型和PSO-SVM模型,并结合文献[12]中的作物生长模型进行了对比分析,结果表明PSO-SVM模型精度优于其他模型。

The research on crop water-fertilizer production function is an important part of deficit irrigation theory and the basis of ir rigation optimization for increasing utilizing efficiency of soil water and fertilizer. The Support Vector Machine （SVM） prediction model for fertilization-irrigation-yield is established in this paper, and the parameters of SVM are optimized by means of Particle Swarm Optimization （PSO）. Finally, the optimal SVM prediction model is obtained. Based on the irrigation experimental data of Xi- aohe experimental station, the BP neural network model, the crop growth model and PSO - SVM model of water-fertilizer production function are built, respectively. The results show that the model of PSO-SVM has a higher precision than other models.