基于可见-近红外光谱和深度森林的蓝莓成熟度判别

doi:10.3964/j.issn.1000-0593(2024)11-3280-07

摘要
参考文献
相关文章 (15)

全文: PDF (8070 KB)
输出: BibTeX | EndNote (RIS)

摘要：为快速准确对蓝莓果实成熟程度进行分类，采用近红外光谱检测技术和深度森林算法，建立了蓝莓成熟度的判别模型。采用LabSpec 5000光谱仪采集了三种不同成熟程度的蓝莓标准样品，共获取了150组光谱样本。为确定最佳输入模型特征数目，对原始光谱数据进行SavitzkyGolay卷积平滑处理，采用主成分分析将平滑处理后的数据降至4个主成分，并采用多项式特征衍生方法对每个主成分进行2、3、4、5阶的特征衍生，最终在深度森林中确定最佳的特征衍生阶数为4。为检验深度森林的成熟度判别效果，将其与随机森林、极端梯度提升树算法(xgboost)及stacking融合模型进行了对比，对各模型确定了最佳超参数组合，深度森林和stacking融合模型采用了手动调参，随机森林和xgboost采用了贝叶斯优化算法进行了超参数寻优。模型评估指标采用准确率、混淆矩阵、受试者工作特征曲线（ROC）、AUC度量及抗噪能力。研究结果表明，在测试集上，深度森林和stacking融合模型的准确率均为95.56%，随机森林和xgboost的准确率为93.33%；深度森林的AUC值为1，随机森林、stacking融合模型、xgboost的AUC值分别为0.99、0.98、0.96，深度森林和stacking融合模型的抗噪能力优于随机森林和xgboost。该研究的深度森林模型整体上判别效果优于其他三种模型，为蓝莓成熟程度判别提供了技术支持。

关键词：可见-近红外光谱；深度森林；蓝莓；成熟度；无损检测

Abstract：To quickly and accurately classify the maturity of blueberries, this study established a discriminant model for blueberry maturity based on near-infrared spectroscopy detection technology and deep forest algorithms. A LabSpec 5000 spectrometer was used to collect three different maturity levels of blueberry standard samples, and a total of 150 spectral samples were obtained. To determine the optimal number of input model features, the original spectral data was subjected to SavitzkyGolay convolution smoothing, and then principal component analysis was used to reduce the smoothed data to 4 principal components. The polynomial feature derivation method derived 2nd, 3rd, 4th, and 5th order features for each principal component. The optimal feature derivation order in the deep forest was considered 4th order. To test the maturity discrimination effect of the deep forest, it was compared with random forest, extreme gradient boosting tree algorithm (xgboost), and stacking fusion model. In the comparison, the optimal hyperparameter combination for each model was determined. The deep forest and stacking fusion model used manual parameter tuning, while random forest and xgboost used a Bayesian optimization algorithm for hyperparameter optimization. The model evaluation indicators were accuracy, confusion matrix, receiver operating characteristic (ROC) curve, AUC measurement, and anti-noise ability. The results showed that on the test set, the accuracy of the deep forest and stacking fusion model was 95.56%, while that of random forest and xgboost was 93.33%. The AUC value of deep forest was 1, while that of random forest, stacking fusion model, and boost were 0.99, 0.98 and 0.96, respectively. The anti-noise ability of deep forest and stacking fusion model was better than that of random forest and xgboost. Overall, the deep forest model in this study had a better discrimination effect than the other three models and provided technical support for blueberry maturity discrimination.

Key words：Visible-near infrared spectrum;Deep forest;Blueberry;Maturity;Non-destructive testing

收稿日期: 2023-04-10 修订日期: 2024-03-17

中图分类号:

TS255

基金资助: 中央高校基本科研业务费专项资金项目（2572020BL01），黑龙江省自然科学基金项目(LH2020C050)资助

通讯作者: 冯国红 E-mail: fgh_1980@126.com

作者简介: 王宏恩，2000年生，东北林业大学机电工程学院工业工程系硕士研究生 e-mail: 1718442562@qq.com

引用本文:

王宏恩，冯国红，徐华东，张润泽. 基于可见-近红外光谱和深度森林的蓝莓成熟度判别[J]. 光谱学与光谱分析, 2024, 44(11): 3280-3286.
WANG Hong-en, FENG Guo-hong, XU Hua-dong, ZHANG Run-ze. Identification of Blueberry Ripeness Based on Visible-Near Infrared Spectroscopy and Deep Forest. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2024, 44(11): 3280-3286.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2024)11-3280-07 或 https://www.gpxygpfx.com/CN/Y2024/V44/I11/3280

[1] HU Yi-lei，JIANG Hong-zhe，ZHOU Hong-ping，et al(胡逸磊，姜洪喆, 周宏平，等). Food Industry Science and Technology(食品工业科技)，2021，42(20)：377.
[2] XIE Guo-fang，LIU Na，LU Dan，et al(谢国芳，刘娜，卢丹，等). Economic Forest Research(经济林研究)，2020，38(2)：209.
[3] CHEN Bei-wen，CHEN Gan(陈贝文，陈淦). Computer Generation(计算机时代)，2022，(7)：62.
[4] GUAN Ye-qing，WANG Dong，LI Nan，et al(关晔晴，王冬，李楠，等). Modern Food Science and Technology(现代食品科技)，2022，38(10)：290.
[5] ZHU Jin-yan，ZHU Yu-jie，FENG Guo-hong，et al(朱金艳，朱玉杰，冯国红，等). Spectroscopy and Spectral Analysis(光谱学与光谱分析)，2022，42(12)：3775.
[6] Pu H B, Liu D, Wang L, et al. Food Analytical Methods，2016，(1)：235.
[7] MA Xin，WANG Biao，LI Chun，et al. Advances in Laser and Optoelectronics，2022，59(19)：356.
[8] LIU Jin-xiu，HE Xiao-wei，LUO Hua-ping，et al(刘金秀，贺小伟，罗华平，等). Food Research and Development(食品研究与开发)，2022，43(15)：158.
[9] SHAO Yi-wei，CHEN Jia-yu，LIN Cui-ying，et al(邵怡伟，陈嘉宇，林翠颖，等). Aeronautical Journal(航空学报)，2022，43(08)，118.
[10] YANG Jing，GUO Wei-yu，YANG Wen-bin，et al(杨静，郭韦昱，杨文彬，等). Laser Journal(激光杂志)，2022，43(10)：150.
[11] National Health Commission of the People's Republic of China，State Administration for Market Regulation(中华人民共和国国家健康委员会，国家市场监督总局). Determination of Total Acid in Food According to National Food Safety Standards(GB 12456—2021 食品安全国家标准食品中总酸的测定). Beijing: Published by China Standard Press(北京：中国标准出版社)，2021.
[12] ZHANG Zhi-gang，LIU Yu-fang，LI Chang-cheng，et al(张志刚，刘玉芳，李长城，等). Zhejiang Agricultural Journal(浙江农业学报)，2021，33(8)：1402.
[13] HUANG Min-min，LI Wei，CHEN Mei-zhen，et al(黄敏敏，李巍，陈美珍，等). Anhui Agricultural Science Bulletin(安徽农学通报)，2021，27(4)：51.
[14] Mahayothee B，Rungpichayapichet P，Yuwanbun P，et al. Journal of Food Measurement and Characterization，2020，14(2)：992.
[15] FENG Guo-hong，ZHU Yu-jie，XU Hua-dong，et al(冯国红，朱玉杰，徐华东，等). Journal of Northeast Forestry University(东北林业大学学报)，2020，48(6)：56.
[16] HUANG Bao-feng，QI Ting-ting(黄宝凤，祁婷婷). Credit Reporting(征信)，2021，39(7)：51.
[17] YU Chen-hui，WANG Hao-ming，LI Yi，et al(于晨辉，王浩名，李一，等). High Voltage Appliances(高压电器)，2023，59(2)：37.
[18] YANG Xing-li(杨杏丽). Computer Science(计算机科学)，2021，48(8)：209.
[19] ZHU Rui-jin，DONG Ya-li，TANG Bo(朱瑞金，董亚丽，唐波). Grids and Clean Energy(电网与清洁能源)，2022，38(6)：24.