引用本文:[点击复制]
[点击复制]
【打印本页】【下载PDF全文】 查看/发表评论下载PDF阅读器关闭

←前一篇|后一篇→

过刊浏览    高级检索

本文已被:浏览次   下载 本文二维码信息
码上扫一扫!
GEP优化的多输出RBF网络作物生理参数建模
闵文芳,江朝晖,李婷婷,祁钊,饶元
0
(安徽农业大学信息与计算机学院,合肥 230036)
摘要:
针对常用的回归和神经网络作物建模方法存在的输出单一、参数优化困难和预测精度不足等问题,利用基因表达式编程优异的全局搜索能力和RBF神经网络多输出任意非线性函数逼近特点,设计了1种GEP优化的RBF多输出模型算法GEP-RBF。以水稻和番茄的5个关键环境因子为输入、以叶片CO2交换率和蒸腾速率为输出,进行建模验证。结果显示,在预测的均方根误差指标上,GEP-RBF模型与GA-RBF和RBF相比,水稻的CO2交换率和蒸腾速率分别降低了约28.4%、38.0%和89.9%、62.8%,番茄的CO2交换率和蒸腾速率则分别降低了约56.9%、48.4%和75.3%、67.1%;在多输出结果的平衡性指标上,相比GA-RBF和RBF,GEP-RBF模型提高了约16.4%~ 77.4%。结果表明,GEP-RBF模型具有良好的预测精度和多输出平衡性,是一种有效的作物生长建模方法。
关键词:  作物模型  基因表达式编程  优化  RBF神经网络
DOI:10.13610/j.cnki.1672-352x.20170208.012
投稿时间:2016-04-29
基金项目:农业部国际科技合作项目(948计划, 2015-Z44和2016-X34), 安徽省自然科学基金(1508085MF110)和安徽省科技攻关项目(1501031102)共同资助。
Multi output RBF network based on GEP optimization of modeling for crop physiological parameters
MIN Wenfang,JIANG Zhaohui,LI Tingting,QI Zhao,RAO Yuan
(School of Information and Computer, Anhui Agricultural University, Hefei 230036)
Abstract:
In order to address such problems as single output, parameter optimization difficulties, and lack of prediction accuracy etc. in modeling and predicting for the conventional?plants based on regression and neural network, A multi output RBF network based on GEP optimization was designed with the help of strong global search ability of GEP and multi output arbitrary nonlinear function approximation of RBF network. Five key environmental factors of rice and tomato served as input, leaf CO2 exchange rate and transpiration rate as output, the proposed method was adopted in modeling and verifying. Experimental results showed: in view of the root mean square error, compared with GA-RBF and RBF, CO2 exchange rate and transpiration rate in rice using the GEP-RBF model were reduced by ~28.4%, 38.0% and 89.9%, 62.8%, respectively, while those in tomato were reduced by ~56.9%, 48.4% and 75.3%, 67.1%, respectively; on the balance of multiple output result, compared with GA-RBF and RBF, using the GEP-RBF model could improve it by ~16.4% - 77.4%. The study indicated that the GEP-RBF model has good prediction accuracy and multi output balance, and it is an effective method for crop growth modeling.
Key words:  crop model  gene expression programming  optimization  RBF neural network

用微信扫一扫

用微信扫一扫