摘要
以蚁群优化算法(Ant Colony Optimization,ACO)结合支持向量机回归(Support Vector Regression,SVR)为新的改进方法进行特征波段的选择,并应用到对麦粒硬度的预测研究方面,来探测预测能力的可行性。研究结果表明:模型的相关系数R_(cv)=0.981 0,均方根误差RMSE_(cv)=0.038 2;预测结果的相关系数R_P=0.954 4,均方根误差RMSE_P=0.059 0。相比全光谱PLS、IPLS算法,减少一定的建模所用的变量数的同时又在预测能力和精度方面均有所提高,能够更好地体现波段的优选模型。证实了ACO–SVR法应用到粮粒硬度的预测研究方面是可行的。
The new improved method, which was combined the ant colony optimization （ACO） with the support vector regression （ SVR ）, was used for band selection, which was applied to the prediction research of wheat grain hardness. The feasibility of the forecasting ability was detected. The results showed that Rcv and RMSEcv of the model were 0.981 0 and 0.038 2, respectively. Rp and RMSEp were 0.954 4 and 0.059 0, respectively. Compared with the full spectrum of PLS and IPLS algorithm, it simultaneously reduced the number of certain variables used in the model ,and the prediction ability and the precision increased. It could better reflect optimization model of the wave band. It was confirmed that applying ACO-SVR method to the hardness prediction research of the grain kernels was feasible.
作者
张红涛
母建茹
阮朋举
李德伟
ZHANG Hong-tao MU Jian-ru RUAN Peng-ju LI De-wei(Institute of Electric Power, North China University of Water Resources and Electric Power, Zhengzhou 450045, Henan, Chin)
出处
《粮食与油脂》
北大核心
2016年第10期58-63,共6页
Cereals & Oils
基金
国家自然科学基金项目(31101085)
河南省科技攻关项目(162102110112)
华北水利水电大学教学名师培育项目(2014108)
关键词
近红外高光谱
蚁群优化
波段优选
硬度预测
支持向量机回归
the near-infrared （ NIR ） hyperspectral
ant colony optimization （ ACO ）
optimized selection of wave band
hardness prediction
support vector regression （ SVR ）
