Method for predicting content of glutamic acid in yellow wine based on electronic tongue data
阅读说明：本技术 一种基于电子舌数据预测黄酒中谷氨酸含量的方法 (Method for predicting content of glutamic acid in yellow wine based on electronic tongue data ) 是由 刘太昂 刘太行 刘振昌 周晶晶 周央 吴治富 朱峰 刘远 刘婷婷 朱鲁阳 于 2021-09-17 设计创作，主要内容包括：本发明的目的就是为了克服黄酒中谷氨酸检测方法的速度慢、成本高等缺陷而提供一种简便快捷、低成本、无污染的检测方法。通过对黄酒直接进行电子舌扫描,得到黄酒的电子舌数据,再通过映射转化得到基于电子舌中间数据,基于电子舌中间数据,利用支持向量机算法建立谷氨酸的定量预报模型。利用建立的定量预报模型可以实现对新的黄酒样品中的谷氨酸快速检测。(The invention aims to provide a simple, convenient, quick, low-cost and pollution-free detection method for overcoming the defects of low speed, high cost and the like of a detection method for glutamic acid in yellow wine. Electronic tongue data of the yellow wine are obtained by directly scanning the yellow wine, intermediate data based on the electronic tongue are obtained by mapping and converting, and a quantitative forecasting model of the glutamic acid is established by utilizing a support vector machine algorithm based on the intermediate data of the electronic tongue. The established quantitative forecasting model can be used for realizing the rapid detection of the glutamic acid in the new yellow wine sample.)
1. A method for predicting the content of glutamic acid in yellow wine based on electronic tongue data is characterized by comprising the following steps:
1) collecting a plurality of yellow wine samples of different brands, respectively taking a small amount of samples to directly perform electronic tongue scanning to obtain electronic tongue data of the yellow wine samples, wherein the electronic tongue data is composed of 847-dimensional data, detecting the content of glutamic acid in the yellow wine samples by using a high performance liquid chromatography, and the electronic tongue and glutamic acid content data of the yellow wine form basic data of a rapid detection model; 2) because the electronic tongue data is composed of 847 dimensions, high-dimensional data is mapped to a low-dimensional space by adopting a mapping function to obtain low-dimensional data of the electronic tongue data, and basic data is converted into intermediate data composed of the low-dimensional data and glutamic acid content data; 3) based on the intermediate data, a prediction model of glutamic acid in the yellow wine is established by adopting a support vector machine regression algorithm; 4) and collecting a new yellow wine sample, scanning the electronic tongue of the new yellow wine sample, substituting the obtained electronic tongue data into a mapping function to obtain low-dimensional data of the new yellow wine sample, and finally substituting the low-dimensional data of the new yellow wine sample into a regression model of a support vector machine to directly predict the glutamic acid content of the new yellow wine sample.
2. The method for predicting the content of glutamic acid in yellow wine based on electronic tongue data as claimed in claim 1, wherein the mapping function in the step 2) maps high-dimensional data to a low-dimensional space, and part of the mapping functions are as follows:
P1=+3.417E-4[SRS_0]+3.152E-4[SRS_1]+3.358E-4[SRS_2]+……+3.210E-5[BRS_119]+3.220E-5[BRS_120]-62.214；
P10=-8.511E-4[SRS_0]-6.174E-4[SRS_1]-3.195E-4[SRS_2]+……+1.590E-5[BRS_119]+1.250E-5[BRS_120]+2.014。
Technical Field
The invention relates to the technical field of yellow wine component analysis and detection, in particular to a method for predicting the content of glutamic acid in yellow wine based on electronic tongue data.
Background
Yellow wine is a traditional Chinese national wine and also a wonderful flower in Huaxia treasure and wine. Along with the rapid development of economy, the improvement of life quality of people and the enhancement of health consciousness of China, yellow rice wine is increasingly popular with consumers due to the characteristics of low alcohol degree, nutrition, safety and health care. The ingredients in the yellow wine, particularly volatile flavor substances, determine the taste and quality of the yellow wine, but at present, the detection of the ingredients is mainly completed by high performance liquid chromatography, gas chromatography and the like, but the yellow wine is still limited to a laboratory stage due to low maturity, lack of relevant standards or high price. Therefore, in order to meet the requirements of consumers on the quality of yellow wine and improve the advantages of yellow wine enterprises in increasingly violent competition, thereby promoting the expansion of the yellow wine consumption market, the earning of product export and the healthy development of the yellow wine industry, a new method for rapidly detecting volatile flavor substances in yellow wine is increasingly urgent and important.
The electronic tongue technology is a novel detection means for analyzing and identifying the taste of liquid developed in the middle of the century. The electronic tongue is used for detecting the volatile flavor substances of the yellow wine, the advantages are prominent, firstly, the sample is directly detected without any pretreatment, secondly, the detection speed is high, and the electronic tongue only needs ten seconds to several minutes for detecting one sample, and is much faster than other instruments.
The patent aims at detecting glutamic acid in the yellow wine and tries to establish a quantitative forecasting model of the glutamic acid content based on electronic tongue data of the yellow wine. The rapid detection of glutamic acid in the yellow wine is realized through the established quantitative model, and the yellow wine enterprise is served.
Disclosure of Invention
The invention aims to provide a simple, convenient, quick, low-cost and pollution-free detection method for overcoming the defects of low speed, high cost and the like of a detection method for glutamic acid in yellow wine. Electronic tongue data of the yellow wine are obtained by directly scanning the yellow wine on the electronic tongue, intermediate data based on the electronic tongue are obtained by mapping and converting, and a quantitative forecasting model of the glutamic acid is established by utilizing a support vector machine regression algorithm based on the intermediate data of the electronic tongue.
The purpose of the invention can be realized by the following technical scheme:
a method for predicting the content of glutamic acid in yellow wine based on electronic tongue data comprises the following steps:
1) collecting a plurality of yellow wine samples of different brands, respectively taking a small amount of samples to directly perform electronic tongue scanning to obtain electronic tongue data of the yellow wine samples, wherein the electronic tongue data is composed of 847 dimensional data. Detecting the glutamic acid content in the yellow wine sample by using high performance liquid chromatography. The data of the electronic tongue and the glutamic acid content of the yellow wine form basic data of a rapid detection model.
2) Since the electronic tongue data is composed of 847 dimensions, the high-dimensional data is mapped to the low-dimensional space by using a mapping function, and the low-dimensional data is obtained. The basic data is converted into intermediate data consisting of low-dimensional data and glutamic acid content data;
3) based on the intermediate data, a prediction model of glutamic acid in the yellow wine is established by adopting a support vector machine regression algorithm;
4) and collecting a new yellow wine sample, scanning the electronic tongue of the new yellow wine sample, substituting the obtained electronic tongue data into a mapping function to obtain low-dimensional data of the new yellow wine sample, and finally substituting the low-dimensional data of the new yellow wine sample into a regression model of a support vector machine to directly predict the glutamic acid content of the new yellow wine sample.
Compared with the prior art, the invention has the following advantages:
firstly, low cost: although the detection of the glutamic acid in the yellow wine sample is finished by the high performance liquid chromatography in the previous work, after the forecasting model is established, the new yellow wine sample needing to detect the glutamic acid content is not detected by the high performance liquid chromatography, but the new yellow wine sample is directly scanned on the electronic tongue and is directly substituted into the model for forecasting, so that a lot of cost is saved in the later period.
Secondly, no pollution: after the glutamic acid quantitative model is established, the glutamic acid forecasting process is directly carried out on a new yellow wine sample by using an electronic tongue instrument, and no chemical is used in the whole process, so that the environment is not polluted.
Thirdly, the test is simple: the electronic tongue instrument can complete the detection of one sample within a few seconds, and the whole process can be completed by only one person, so the detection is quick and simple.
Drawings
FIG. 1 is a graph of electronic tongue data for yellow wine samples.
FIG. 2 is a diagram of the modeling result of a forecasting model of a glutamic acid support vector machine in yellow rice wine.
FIG. 3 is a graph of the results of the prediction model of the support vector machine for glutamic acid in yellow wine.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments.
Example (b):
a method for predicting glutamic acid content in yellow wine based on electronic tongue data is provided. The method comprises the following steps:
(1) and (4) collecting 18 yellow wine samples of different brands, and respectively taking a small amount of samples to directly perform electronic tongue scanning to obtain electronic tongue data of the yellow wine samples. The electronic tongue instrument consists of 7 sensors, each of which can collect 121 data, so that the total electronic tongue data consists of 847-dimensional data. And detecting the content of glutamic acid in 18 yellow wine samples by using high performance liquid chromatography. Thus, the data of the content of the electronic tongue and the glutamic acid of the yellow wine sample form the basic data of the rapid detection model, and the basic data is a data matrix of 18 x 847. The yellow wine electronic tongue data is shown in figure 1. Exemplary data for the yellow wine electronic tongue portion is shown in table 1.
Table 1 electronic nose part example data
SRS_0
SRS_1
GPS_0
GPS_1
STS_0
STS_1
UMS_0
UMS_1
SPS_0
SPS_1
SWS_0
SWS_1
BRS_0
BRS_1
1642.96
1641.13
1555.40
1565.77
1085.24
1077.92
1508.11
1496.21
1089.82
1087.68
1373.26
1350.98
1885.82
1883.38
1471.19
1467.53
1606.96
1600.86
1252.74
1244.20
1352.51
1342.14
1339.08
1321.39
1250.00
1210.64
1930.37
1930.37
1605.74
1618.86
1635.03
1621.61
1170.97
1172.50
1307.96
1302.47
1253.05
1242.06
1256.71
1216.43
1914.81
1915.72
1446.48
1441.60
1597.81
1595.98
1256.10
1257.01
1353.73
1371.43
1323.52
1317.12
1242.98
1210.33
1926.10
1921.22
1388.51
1387.90
1613.67
1597.81
935.74
939.40
1398.88
1392.17
1260.67
1247.55
1509.03
1496.52
2044.48
2043.26
1318.95
1400.10
1389.12
1411.09
965.64
1012.02
1484.31
1492.55
1356.78
1364.71
1897.42
1753.41
-525.99
-525.08
1613.67
1625.27
1536.79
1526.42
1013.54
1021.17
1428.78
1431.83
1235.66
1231.69
1392.78
1378.14
2030.75
2029.83
1618.25
1633.20
1545.94
1537.70
954.05
937.57
1416.58
1403.77
1210.94
1204.23
1433.06
1422.07
2046.92
2043.87
1538.62
1542.89
1578.28
1570.96
1008.36
1013.54
1458.68
1451.97
1282.34
1276.84
1402.55
1390.95
2028.61
2030.44
1563.64
1567.30
1592.93
1583.16
1016.29
1017.20
1442.82
1432.14
1251.52
1244.50
1389.73
1379.66
2027.39
2024.95
1547.77
1554.48
1589.88
1584.38
1016.59
1019.95
1449.53
1442.21
1255.49
1247.55
1433.36
1421.77
2033.49
2035.32
1623.74
1630.76
1589.27
1578.59
1033.37
1023.00
1480.96
1449.53
1261.59
1248.77
1406.82
1397.97
2031.36
2028.00
(2) And for the basic data of the data matrix of 18 x 847, mapping the high-dimensional data to the low-dimensional space by adopting a mapping function to obtain the low-dimensional data, and taking the first 10-dimensional low-dimensional data. These low-dimensional data are processes that also lose part of the information, but do not affect the modeling results. The base data was converted to a linear combination of low-dimensional 847-dimensional data, which thus contained a large amount of raw data information, of course constituting intermediate data between the mapped data and the glutamic acid content data, which was an 18 x 10 data matrix. The partial mapping function is as follows:
P1=+3.417E-4[SRS_0]+3.152E-4[SRS_1]+3.358E-4[SRS_2]+……+3.210E-5[BRS_119]+3.220E-5[BRS_120]-62.214
……
P10=-8.511E-4[SRS_0]-6.174E-4[SRS_1]-3.195E-4[SRS_2]+……+1.590E-5[BRS_119]+1.250E-5[BRS_120]+2.014
the mapping of the transformed partial electronic tongue data is shown in table 2.
TABLE 2 mapping of transformed partial electronic tongue data
P1
P2
P3
P4
P5
P6
P7
P8
P9
P10
39.74
7.99
-15.30
1.99
9.65
-0.27
-0.59
0.23
-0.03
0.25
-5.02
-0.24
7.12
-3.98
16.07
-1.63
-0.41
-0.30
0.46
0.87
4.81
8.28
13.26
-3.22
8.81
-6.02
-0.66
1.95
1.03
-0.17
-7.10
-1.32
3.96
-4.24
18.65
1.97
-0.51
-0.53
-1.72
-0.38
-3.38
-17.11
-3.73
-2.00
-10.37
-5.29
-0.37
4.52
0.02
0.95
-10.84
8.62
-7.03
3.01
-0.30
-15.71
2.16
-2.74
0.55
0.19
3.20
-3.07
8.63
7.76
-1.04
4.30
0.26
0.97
1.57
0.09
3.92
-3.32
9.15
6.10
-8.07
0.12
-0.87
-1.37
-2.69
1.25
0.53
-6.38
3.02
1.31
-3.80
1.93
0.13
-0.68
1.12
0.00
2.97
-6.71
3.93
2.92
-2.40
0.57
0.18
-0.31
0.28
-1.25
3.18
-5.98
2.95
1.83
-3.15
0.71
0.89
-0.31
-0.60
-0.48
4.22
-0.73
2.61
-1.63
-6.66
1.42
0.54
0.20
1.03
-0.52
2.14
-4.80
3.19
1.86
-0.23
3.06
1.05
-1.63
-0.33
-0.36
(3) And (4) establishing a prediction model of glutamic acid in the yellow wine by adopting a support vector machine regression algorithm according to the intermediate data. For parameters of a support vector machine algorithm, a linear kernel function is selected, a penalty factor is 55, and an insensitive function is 0.03.
(4) Collecting 2 new yellow wine samples, scanning the electronic tongues of the yellow wine samples, substituting the obtained electronic tongue data into a mapping function to obtain low-dimensional data of the new yellow wine samples, and finally substituting the low-dimensional data of the new yellow wine samples into a support vector machine regression model to directly predict the glutamic acid content of the 2 new yellow wine samples.
Example 1: modeling result of 18 yellow rice wine sample amino acid support vector machine forecasting models
And establishing glutamic acid quantitative prediction modeling for the intermediate data based on the 18 yellow wine electronic tongue data by using a support vector machine regression algorithm. And the correlation coefficient of the glutamic acid calculated value of the modeling result and the experimental detection value is 0.99. The results are shown in FIG. 1.
Example 2: leave-one-out test result of prediction model of 18 yellow rice wine samples based on amino acid support vector machine
An amino acid support vector machine forecasting model is established based on the electronic tongue data of 18 yellow wine samples, and the leave-one-out test result is shown in figure 2. The correlation coefficient of the glutamic acid calculated value and the experimental detection value of the leave-one-out test result is 0.89.
Example 3: forecasting glutamic acid of 2 new yellow wine samples
Collecting 2 new yellow wine samples, scanning the electronic tongues of the yellow wine samples, substituting the obtained electronic tongue data into a mapping function to obtain low-dimensional data of the new yellow wine samples, and finally substituting the low-dimensional data of the new yellow wine samples into a support vector machine regression model to directly predict glutamic acid data of the 2 new yellow wine samples. The results are shown in Table 3.
TABLE 3 prediction results
Sample number
Glutamic acid prediction value (mg/l)
1
90.15
2
118.93
- 上一篇：一种医用注射器针头装配设备
- 下一篇：一种烟草中香气成分的分析方法