摘要
采用近红外光谱技术结合数据降维的方法,建立了哈密瓜可溶性固形物含量的预测模型,对原始光谱进行特征区间选择,共选取了6个子区间,432个光谱变量;将6个联合子区间的光谱数据分别结合特征选择竞争性自适应重加权采样算法、遗传算法、连续投影算法(successive projections algorithm, SPA)提取特征波长;再使用选取的特征波长以及特征区间波长作为模型的输入变量,利用极限学习机和偏最小二乘算法(partial least squares, PLS)建立哈密瓜可溶性固形物含量预测模型。结果显示,反向区间偏最小二乘算法+SPA+PLS建立的预测模型最优,模型的校正集相关系数为0.923 4,预测集相关系数为0.878 8,模型能够准确预测哈密瓜可溶性固形物含量。
The method of near-infrared spectroscopy and data dimension reduction method were used to establish the prediction model of soluble solids content in Hami melon. Six sub-ranges and 432 spectral variables were selected from the original spectrum. The spectral data of six joint sub-regions were combined with competitive adaptive reweighted sampling(CARS), genetic algorithm(GA) and successive projections algorithm(SPA) to extract the feature wavelength respectively. Then, the selected characteristic wavelength and interval wavelength was used as the input variables of the model. The extreme learning machine(ELM)and partial least squares(PLS)were used to establish the prediction model of soluble solids content in Hami melon. The results showed that the prediction model of BiPLS + SPA + PLS was the best. The correlation coefficient of the corrected set and the predicted set was 0.923 4 and 0.878 8 respectively. The model can accurately predict the content of soluble solids in Hami melon.
作者
郭阳
史勇
郭俊先
李雪莲
黄华
GUO Yang;SHI Yong;GUO Junxian;LI Xuelian;HUANG Hua(College of Electrical and Mechanical Engineering,Xinjiang Agricultural University,Urumqi 830052,China;College of Mathematics and Physics,Xinjiang Agricultural University,Urumqi 830052,China)
出处
《食品与发酵工业》
CAS
CSCD
北大核心
2022年第2期248-253,共6页
Food and Fermentation Industries
基金
新疆维吾尔自治区教育厅自然科学重点项目(XJEDU2020I009)
国家自然科学基金面上项目(61367001)。
