基于深度信念网络与混合波长选择方法的蓝莓糖度近红外检测模型优化

doi:10.3964/j.issn.1000-0593(2022)12-3775-08

Abstract
References
Related Citation (15)

Download: PDF (4144 KB)
Export: BibTeX | EndNote (RIS)

Abstract Using near-infrared spectroscopy technology combined with synergy interval partial least square (SiPLS), competitive adaptive reweighted sampling (CARS), successive projection algorithm (SPA) and uninformative variable elimination (UVE) feature extraction methods, the universal detection models of blueberry sugar content were established by deep belief network (DBN) to achieve on-line non-destructive rapid detection. The near-infrared spectra of 280 blueberry samples of “Bluecrop” and “Reka” were collected, and the sugar content of blueberries was determined by a hand-held refractometer. Firstly, using the outlier samples detection based on joint X-Y distances (ODXY) method to detect abnormal samples, there were 2 and 4 abnormal samples from Bluecrop and Reka respectively. After eliminating 6 abnormal samples, the remaining 274 samples were divided into a training set and test set in a ratio of 3∶1 by the sample set partitioning based on the joint X-Y algorithm (SPXY). Secondly, Compared and analyzed the improvement effect of Savitzky-Golay smoothing (S-G smoothing), centralization, multiplicative scatter correction and other pretreatment methods on the original spectrum of blueberry. Using SiPLS to reduce the spectral dimension and filter the characteristic band and using SPA, UVE and CARS, choose characteristic wavelengths again. We established partial least square regression (PLSR) and DBN models with the optimal characteristic wavelengths. The results showed that the optimal pretreatment method of the blueberry sugar content near-infrared detection model was S-G smoothing, the optimal band of blueberry sugar screening by SiPLS method were 593~765 and 1 458~1 630 nm, and the UVE algorithm was used to select 159 optimal wavelengths from 346 variables screened by SiPLS. When establishing the DBN model of blueberry sugar content, we analyzed the influence of different hidden layer numbers on the detection model, and the root means square error of cross-validation (RMSECV) as a fitness function, the particle swarm optimization (PSO) was used to optimize the number of neurons in each hidden layer between [1, 100]. It is found that the RMSECV of the DBN model reached the minimum value of 0.397 7 when the hidden layer was 3 layers, and the hidden layer node number was 67-43-25. Whether in the full spectrum or modeling characteristic wavelengths, the near-infrared DBN models of blueberry sugar content were superior to the conventional PLSR method. In particular, the characteristic wavelengths selected by the UVE method can greatly reduce the modeling variables, and the model accuracy was high. The correlation coefficient (R_P) and root mean square error (RMSEP) of the optimal PLSR model were 0.887 5 and 0.395 9, respectively. TheR_P and RMSEP of the optimal DBN model were 0.954 2 and 0.310 5. The research shows that the detection model of blueberry sugar content based on the characteristic wavelength extracted by SiPLS-UVE combined with the deep belief network method can better complete the accurate online analysis of blueberry sugar content, and the method is expected to be applied to the internal quality detection of blueberries and other fruits and vegetables.

Key words：Near-infrared spectroscopy; Blueberry; Sugar content; Uninformative variable elimination; Deep belief network; Nondestructive testing

Received: 2021-10-13 Accepted: 2022-03-11

ZTFLH:

O657.3

Corresponding Authors: ZHU Yu-jie, FENG Guo-hong E-mail: zhuyujie004@126.com；fgh_1980@126.com

Cite this article:

ZHU Jin-yan,ZHU Yu-jie,FENG Guo-hong, et al. Optimization of Near-Infrared Detection Model of Blueberry Sugar Content Based on Deep Belief Network and Hybrid Wavelength Selection Method[J]. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2022, 42(12): 3775-3782.

URL:

https://www.gpxygpfx.com/EN/10.3964/j.issn.1000-0593(2022)12-3775-08 OR https://www.gpxygpfx.com/EN/Y2022/V42/I12/3775

[1] GUO Yang, GUO Jun-xian, SHI Yong, et al（郭阳，郭俊先，史勇，等）. Food & Machinery(食品与机械), 2021, 37(6): 81.
[2] JIE Deng-fei, CHEN Meng, XIE Li-juan, et al（介邓飞，陈猛，谢丽娟，等）. Transactions of the Chinese Sociely of Agricultural Engineering(农业工程学报), 2014, 30(9): 229.
[3] GAO Sheng, WANG Qiao-hua, LI Qing-xu, et al（高升，王巧华，李庆旭，等）. Chinese Journal of Analytical Chemistry(分析化学), 2019, 47(6): 941.
[4] ZHANG Juan（张娟）. Food and Fermentation Sciences & Technology(食品与发酵科技), 2020, 56(6): 136.
[5] Katarzyna W, Julia S, Igor K, et al. Journal of the Science of Food and Agriculture, 2019, 99(13): 5953.
[6] Masithoh R E, Pahlawan M F R, Wati R K. IOP Conference Series: Earth and Environmental Science, 2021, 752(1): 012047.
[7] WU Shuang, LI Guo-jian, JIE Deng-fei（吴爽，李国建，介邓飞）. Food & Machinery(食品与机械), 2020, 36(12): 132.
[8] ZHOU Zhao-yong, HE Dong-jian, ZHANG Hai-hui, et al（周兆永，何东健，张海辉，等）. Food Science(食品科学), 2017, 38(14): 297.
[9] ZHANG Meng, ZHAO Zhong-gai（张萌，赵忠盖）. Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2020, 40(8): 2512.
[10] WANG Gong-ming, QIAO Jun-fei, GUAN Li-na, et al（王功明，乔俊飞，关丽娜，等）. Acta Automatica Sinica(自动化学报), 2021, 47(1): 35.
[11] LI Zi-chen, SHI Xin-yu, YU Long, et al（李自臣，史新宇，禹龙，等）. Computer Applications and Software(计算机应用与软件), 2019, 36(2): 189.
[12] YIN Bao-quan, SHI Yin-xue, SUN Rui-zhi, et al（尹宝全，史银雪，孙瑞志，等）. Transactions of the Chinese Society for Agricultural Machinery(农业机械学报), 2015, 46(S1): 122.
[13] WANG Jun, ZHOU Cong, HUANG Jia-peng, et al（王钧，周聪，黄家鹏，等）. Journal of Food Safety & Quality(食品安全质量检测学报), 2021, 12(5): 1798.
[14] HU Xiao-nan, PENG Yun-fa, LUO Hua-ping, et al（胡晓男，彭云发，罗华平，等）. The Food Industry(食品工业), 2015, 36(5): 232.
[15] LIU Yun-hong, WANG Qing-qing, SHI Xiao-wei, et al（刘云宏，王庆庆，石晓微，等）. Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2019, 35(13): 291.
[16] FENG Qian, LI Qing, QUAN Wei, et al（冯茜，李擎，全威，等）. Chinese Journal of Engineering(工程科学学报), 2021, 43(6): 745.
[17] WU Lin-feng, MENG Yu-wei, YU Rong-dong, et al（吴林峰，孟瑜炜，俞荣栋，等）. Zhejiang Electric Power(浙江电力), 2020, 39(9): 60.