Identification of soybean producing area in North China based on multi-element and fatty acid fingerprint characteristics

doi:10.19802/j.issn.1007-9084.2021111

Abstract

Abstract:

In order to explore the feasibility of data fusion technology in the origin identification of neighboring soybean producing area, 216 soybean samples were collected from Heilongjiang Agricultural Reclamation Jiusan Administration and Suihua area, and the contents of 13 mineral elements such as magnesium (Mg), aluminum (Al) and phosphorus (P) and 5 fatty acids such as palmitic acid, stearic acid and oleic acid were determined. Support vector machine (SVM) with 4 kernel functions were established by mineral elements, fatty acids, data-level fusion and feature-level fusion data. After parameter optimization by grid search algorithm combined with 50% cross-validation, the recognition accuracy of the model increased to 90.77%, 92.31%, 89.23% and 95.38% respectively. The results showed that the feature-level data fusion technology had a significant effect on the origin identification of neighboring soybean producing area, which was superior to the other three data identification technologies. It could accurately and effectively distinguish soybean origin in neighboring areas by feature-level data fusion technology to establish support vector machine origin identification model, which provided a new research direction for geographical indication product protection technology.

Key words: multi-element analysis, fatty acid analysis, data fusion, origin identification, northern soybean

CLC Number:

TP181

Jing-hui WANG, Yang LIU, Qi-you ZHENG, Xiao-tang CHENG, Chao-hui WANG. Identification of soybean producing area in North China based on multi-element and fatty acid fingerprint characteristics[J]. CHINESE JOURNAL OF OIL CROP SCIENCES, 2022, 44(3): 532-538.

Figures/Tables 11

Table 1

Common kernel functions

名称 Name	表达式 Expression	参数 Parameter	R中参数 Parameters in R
线性核 Linear kernel	$K (x i, x j) = x i T x j$		Cost
高斯核 Gaussian kernel	$K (x i x j) = e x p (- x i - x j 2 ÷ 2 δ 2)$	$δ > 0$	Cost，gamma
Sigmoid核 Sigmoid core	$K (x i, x j) = t a n h (β x i T x j + θ)$	$β > 0, θ < 0$	Cost，gamma，degree
多项式核 Polynomial kernel	$K (x i, x j) = (x i T x j) d$	$d ≥ 1$ 为多项式次数	Cost，gamma，degree，coef

Table 1

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Table 2

Table 3

Table 4

Table 5

Fig. 5

Table 6

References 10

1	刘澜. 大豆的营养成分及其综合利用前景［J］. 内蒙古民族大学学报（自然科学版）， 2014， 29（2）： 175-178. DOI：10.3969/j.issn.1671-0185.2014.02.016 . doi: 10.3969/j.issn.1671-0185.2014.02.016
2	鹿保鑫，马楠，王霞，等. 基于电感耦合等离子体质谱仪分析矿物元素含量的大豆产地溯源［J］. 食品科学， 2018， 39（8）： 288-294. DOI：10.7506/spkx1002-6630-201808045 . doi: 10.7506/spkx1002-6630-201808045
3	张勇，李雪，汪雪芳，等. 基于脂肪酸组成的进口大豆鉴别技术研究［J］. 食品安全质量检测学报， 2020， 11（8）： 2375-2379.
4	李吉光，周艳芬，王蓉. GC-M指纹图谱结合主成分分析法评价不同产地黑果枸杞脂肪酸的质量［J］. 宁夏大学学报（自然科学版）， 2017， 38（4）： 365-370.
5	韩永翔. 核桃不同品种营养品质差异与产地溯源［D］. 北京：中国林业科学研究院， 2019.
6	赵海燕，郭波莉，张波，等. 小麦产地矿物元素指纹溯源技术研究［J］. 中国农业科学， 2010， 43（18）： 3817-3823. DOI：10.3864/j.issn.0578-1752.2010.18.016 . doi: 10.3864/j.issn.0578-1752.2010.18.016
7	范胜栩，李斌，孙君明，等. 气相色谱方法定量检测大豆5种脂肪酸［J］. 中国油料作物学报， 2015， 37（4）： 548-553. DOI：10.7505/j.issn.1007-9084.2015.04.017 . doi: 10.7505/j.issn.1007-9084.2015.04.017
8	Borràs E， Ferré J， Boqué R， et al. Data fusion methodologies for food and beverage authentication and quality assessment-a review［J］. Anal Chim Acta， 2015， 891： 1-14. DOI：10.1016/j.aca.2015.04.042 . doi: 10.1016/j.aca.2015.04.042
9	Yao S， Li T， Liu H， et al. Traceability of Boletaceae mushrooms using data fusion of UV-visible and FTIR combined with chemometrics methods［J］. J Sci Food Agric， 2018， 98（6）： 2215-2222. DOI：10.1002/jsfa.8707 . doi: 10.1002/jsfa.8707
10	Singh D， Singh B. Investigating the impact of data normalization on classification performance［J］. Appl Soft Comp， 2019， 97（B）：105524. DOI：10.1016/j.asoc.2019.105524 . doi: 10.1016/j.asoc.2019.105524

成分 Ingredient	初始特征值 Initial eigenvalue			提取平方和载入 Extract the sum of squares and load
成分 Ingredient	特征值 Eigenvalue	方差百分比 Variance /%	累积 Accumulation /%	总计 Total	方差百分比 Variance /%	累积 Accumulation /%
1	5.88	45.26	45.26	5.88	45.26	45.26
2	1.38	10.59	55.85	1.38	10.59	55.85
3	1.06	8.17	64.02	1.06	8.17	64.02
4	1.04	6.46	70.48	1.04	6.46	70.48
5	0.78	6.02	76.50
6	0.69	5.28	81.78
7	0.53	4.14	85.92
8	0.47	3.63	89.55
9	0.43	3.34	92.89
10	0.34	2.58	95.47
11	0.25	1.88	97.35
12	0.24	1.87	99.22
13	0.10	0.78	100.00

成分 Ingredient	初始特征值 Initial eigenvalue			提取平方和载入 Extract the sum of squares and load
成分 Ingredient	特征值 Eigenvalues	方差百分比 Variance /%	累积 Accumulation /%	总计 Total	方差百分比 Variance /%	累积 Accumulation /%
1	3.02	60.40	60.40	3.02	60.40	60.40
2	1.01	16.25	76.65	1.01	16.25	76.65
3	0.60	11.93	88.58
4	0.39	7.72	96.30
5	0.19	3.70	100.00

数据集 Data set	线性核 Linear kernel	高斯核 Gaussian kernel	Sigmoid核 Sigmoid core	多项式核 Polynomial kernel
矿物质元素 Mineral elements	87.69	83.08	83.08	80.00
脂肪酸 Fatty acid	89.23	84.62	84.62	81.54
数据级融合 Data-level fusion	86.15	81.54	81.54	78.46
特征级融合 Feature-level fusion	92.31	90.77	90.77	84.62