核农学报2012，Vol.26Issue(8)：1154-1158,5.
基于最小二乘支持向量回归的鹅肉弹性的可见-近红外光谱测定
DETERMINATION OF ELASTICITY OF GOOSE MEAT USING VISIBLE-NEAR INFRARED SPECTROSCOPY AND LSSVR
摘要
Abstract
To improve and simplify the prediction model of elasticity of goose meat,the optimized characteristic spectral wavelengths were extracted from NIR spectra of goose meat combined with synergy interval PLS(siPLS) and genetic algorithm(GA),then prediction model of elasticity of goose meat was developed using least squares support vector regression(LSSVR).Recovery distances obtained by universal testing machine were used as actual value of elasticity of goose meat.Firstly,sym8 wavelet with two levels decomposition was used to complete the pretreatment of the original visible-near infrared spectroscopy.Secondly,4 subintervals,i.e.No.3,5,9 and 13 were selected by siPLS,and 74 characteristic wavelengths were selected in these spectral regions by GA.Finally,74 characteristic wavelengths were used to build prediction model based on LSSVR.The determination coefficient(R2) and the root mean squared error of prediction(RMSEP) for LSSVR prediction model were 0.9096 and 0.0588,respectively.This work proved that siPLS-GA could determine characteristic spectral wavelengths and improve the prediction accuracy of LSSVR model.关键词
可见-近红外光谱/弹性/最小二乘支持向量回归/联合区间偏最小二乘法/遗传算法Key words
visible-near infrared spectroscopy/elasticity/least squares support vector regression(LSSVR)/synergy interval PLS(siPLS)/genetic algorithm(GA)分类
化学化工引用本文复制引用
赵进辉,袁海超,刘木华,涂冬成,吁芳..基于最小二乘支持向量回归的鹅肉弹性的可见-近红外光谱测定[J].核农学报,2012,26(8):1154-1158,5.基金项目
国家高技术研究发展计划（863计划）项目 （）
国家自然科学基金项目 （）
