可见近红外光谱检测河套蜜瓜糖度和硬度研究——基于LS-SVM  

Research of Sugar Content and Firmness of Hetao Muskmelon Using VIS-NIR Spectroscopy——Based on LS-SVM

作　　者：张德虎[1] 田海清[1] 刘超[1] 肖传晶[1] 武士钥 

机构地区：[1]内蒙古农业大学机电工程学院,呼和浩特010018

出　　处：《农机化研究》2014年第2期10-14,共5页Journal of Agricultural Mechanization Research

基　　金：国家自然科学基金项目(31160248);中国博士后科学基金项目(20110491551);教育部高等学校博士学科点专项科研基金项目(20111515120008)

摘　　要：最小二乘支持向量机是水果品质可见近红外光谱检测中新近发展起来的建模方法。为此,详细介绍了最小二乘支持向量机的工作原理以及基于Matlab环境的LS-SVM工具箱的使用;在此基础上,运用主成分分析和LS-SVM法建立了基于可见近红外光谱的河套蜜瓜糖度和硬度检测模型,并分析了样品糖度、硬度的真实值和预测值的相关性。预测结果表明:糖度和硬度预测值与真实值决定系数R2分别为0.865和0.852。LS-SVM法建模速度快、准确率高、易于实现,在可见近红外光谱数据分析中有很大的实用价值。Least square support vector machine （LS-SVM） tection in visible near infrared spectroscopy （VIS-NIR）. It is the newly developed modeling method of fruit quality de- was introduced thoroughly the working principle of and the application of LS-SVM toolbox based on Matlab environment in the paper. On this basis, principal component analysis and LS-SVM were used for establishing detection models of sugar content and firmness of Hetao muskmelon based on vis- ible near infrared spectroscopy. And the correlation between actual values and predicted values of sugar content and firm- ness of samples was analyzed. The predicted results showed ： the determination coefficients （ RE ） of the predicted and ac- tual values of sugar content and firmness were 0. 865 and 0. 852. LS-SVM has rapid speed of modeling, high accuracy characteristics, and it is easy to achieve. It had great practical value in the data analysis of VIS-NIR spectroscopy.

关 键 词：最小二乘支持向量机 河套蜜瓜 漫透射光谱 糖度 硬度 

分 类 号：S123[农业科学—农业基础科学]