您好，欢迎访问广东省农业科学院 机构知识库！
广东省农业科学院机构知识库

全部

  • 全部
  • 标题
  • 学者
  • 机构
  • 关键词

Development of comprehensive prediction models for pumpkin fruit sensory quality using physicochemical analysis, near-infrared spectroscopy, and machine learning

文献类型： 外文期刊

作者： Xu, Yingchao 1 ; Luo, Jiayu 1 ; Xue, Shudan 1 ; Jin, Qingmin 1 ; Zhu, Jitong 1 ; Lu, Sen 1 ; Meng, Qitao 1 ; Du, Hu 1 ; Fu, Manqin 2 ; Zhong, Yujuan 1 ;

作者机构： 1.Guangdong Acad Agr Sci, Vegetable Res Inst, Guangdong Key Lab New Technol Res Vegetables, Guangzhou 510640, Peoples R China

2.Guangdong Acad Agr Sci, Sericultural & Agrifood Res Inst, Key Lab Funct Foods, Guangdong Key Lab Agr Prod Proc,Minist Agr & Rural, Guangzhou 510610, Peoples R China

关键词： Pumpkin; Quality; Spectroscopy; Machine learning; Prediction model

期刊名称：JOURNAL OF FOOD COMPOSITION AND ANALYSIS （ 影响因子：4.0； 五年影响因子：4.1 ）

ISSN： 0889-1575

年卷期： 2024 年 134 卷

页码：

收录情况： SCI

摘要： Pumpkins are vegetables rich in nutrients. Despite the significant impact of sensory attributes on market success, guidance on standardizing quality factors related to sensory quality is lacking. Also, the development of sensory quality prediction models has not been explored in detail. This study aimed to examine 17 quality traits, and proposed evaluation standards incorporating analysis of sensory and key attributes based on 140 cultivars. The correlation analyses revealed significant correlations between taste scores and dry matter, soluble solids, starch, and sucrose contents. A linear equation was established to predict the sensory quality: Y1 (taste scores) = 31.2806 + 0.2011 x X1 (dry matter) + 2.3834 x X2 (soluble solids content) + 0.0917 x X3 (starch) + 0.0307 x X4 (sucrose) - 0.0127 x X5 (malic acid). The credibility of this model, referring to the proportion of variance in the taste scores, was 95%, and the R2 value was 0.841. Meanwhile, regression prediction models for taste score were developed based on near-infrared spectra data using partial least squares and support vector regression, with coefficient of determination (R2) of 1 and 0.895, root mean square errors (RMSEs) of 0.083 and 2.36, R2P of 0.757 and 0.773, and RMSE of prediction (RMSEp) of 3.17 and 3.06 scores, respectively. This study provided a material basis and standard system for evaluating pumpkin quality and highlighted the effective combination of near-infrared spectroscopy with machine learning techniques.

  • 相关文献
作者其他论文 更多>>
置顶
购物车
反馈

© 京ICP备09089781号-29 主办单位：广东省农业科学院

学术资源： 中国农业科学院 农业专业知识服务系统 农科联盟资源共建共享平台 中国农业科技文献与信息服务平台 中国农业期刊集成服务平台