APP下载

基于近红外的松子蛋白质品质分类处理

2018-12-10蒋大鹏张冬妍李丹丹曹军

计算技术与自动化 2018年3期

蒋大鹏 张冬妍 李丹丹 曹军

摘 要:为了探索松子基于近红外光谱的无损品质分类。建立松子蛋白质品质的分类数学模型。采用近红外测量获取松子光谱数据,运用SMO-SVM、Pegasos-SVM与LS-SVM方法建立松子蛋白质分类相关性模型,并对相应验证集上的数据进行预测验证。实验结果表明支持向量机精准率略高,但耗费时间比LS-SVM与Pegasos-SVM多。研究中所建模型均能达到一定程度上的良好分類,精准度均达到80%以上,可有效实现依据近红外光谱数据预测松子蛋白质含量等级的目的。此模型对于其他干果类食品的等级品质分类具有一定的实践指导意义与应用价值。

关键词:松子;近红外;支持向量机;蛋白质

中图分类号:TP181 文献标识码:A

Abstract: In order to explore the non-destructive quality classification of pine nuts based on near-infrared.Establishment of classification mathematical model of protein quality.Measuring the near infrared spectrum data acquired pine nuts.The support vector machine and the least squares support vector machine were used to establish the correlation model of pine nut protein classification,and the data on the corresponding verification set were predicted and verified.The experimental results of SMO-SVM show that the precision of support vector machine was slightly higher,but it takes more than one third of the least squares support vector machine.The accuracy of the model was 80% or more,which can effectively achieve the purpose of predicting the protein content of pine nuts based on near infrared spectroscopy data.This model has certain practical significance and application value for the grade quality classification of other dried fruit.

Key words: pine nuts;near infrared;support vector machine;protein

1 引 言

松子(Pinus koraiensis Sieb.et Zucc)是一种营养成分很高的食品,有极高的食用价值与药用价值,是东北三省对外重要特产之一。但在松子的无损处理、产品分级方面,一直采用原始低效的分级方法。传统的松子外部品质等级划分多采用人工分选或机械振动筛选方法实现。人工分级劳动强度大,工作者主观经验影响较大:震动筛选噪声大,功耗大,并且分级精准度不高,且会产生磕碰损伤。因此建立一种快速,准确,安全,非破坏性的松子品质分级检测系统,是十分重要的。而近红外光谱技术恰好是快速,准确非破坏性的品质分级检测系统的选择之一。虽然运用近红外光谱进行松子无损品质分类的相关文献较少,但有相关学者研究过运用近红外光谱进行干果类鉴定与品质比较[1]。……

登录APP查看全文