优化神经网络模型在作物需水量预测中的应用  

The Application of Optimal Neutral Network Model in the Forecast of the Crop's Water Demand

作　　者：马淋军[1] 杨宁[2] 严雪萍[2] 

机构地区：[1]江苏大学汽车与交通工程学院,江苏镇江212013 [2]江苏大学电气信息工程学院,江苏镇江212013

出　　处：《农机化研究》2009年第12期169-171,共3页Journal of Agricultural Mechanization Research

基　　金：国家自然科学基金项目(50376021)

摘　　要：在MATLAB6.5中建立BP神经网络模型,用比较试验法对网络结构优化起关键作用的隐层节点数和优化训练算法进行了仿真试验,从而确定了合理优化的BP神经网络预测模型。采用一组样本数据来训练建立好的优化模型,并通过一组非样本数据来验证训练好的网络模型。误差结果证明,该优化模型能快速与准确地预测作物需水量,完全能够满足农业灌溉的精度要求。The BP optimal neutral network model was built in MATLAB6.5. The simulation experiment about the hidden layer nodes and the optimal training which play an important role in the optimization of network structure was carried out. So the rational BP optimal neutral network model was built. A set of sample data was adopt to train the optimal model which has been built, and a set of non - sample data was adopt to verify the network model which has trained. The error shows that the optimal model can forecast the water demand of the crop quickly and accurately, and meet the accuracy requirements of agricultural irrigation completely.

关 键 词：节水农业 作物需水量 预测 神经网络 优化模型 

分 类 号：S126[农业科学—农业基础科学] TP183[自动化与计算机技术—控制理论与控制工程]