A method for rapidly identifying different processing methods and host plants of epimedium based on electronic nose technology

Abstract

The present application relates to the technical field of identifying Cynomorium songaricum, in particular to a method for rapidly identifying Cynomorium songaricum with different processing methods and host plants based on electronic nose technology, comprising the following steps: sample processing; 1.0 g of sample is loaded into a 10 mL silica gel covered syringe bottle, and after 25 min of water bath balancing at 60 DEG C, the sample is detected by an electronic nose; and statistical analysis. The present application uses electronic nose technology to detect and analyze Cynomorium songaricum with different processing methods and from different host plants, to understand the influence of different host plants and processing methods on the flavor of Cynomorium songaricum, and to provide a strong theoretical basis for the rapid quality identification and characteristic flavor evaluation of Cynomorium songaricum. In data analysis, the Gaussian Naive Bayes (GNB) classification model statistical method is used to qualitatively identify the characteristic odor of Cynomorium songaricum through the retention value of the electronic nose, and a discrimination model is constructed to lay a solid theoretical foundation for future qualitative prediction.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of identifying Cynomorium songaricum, in particular to a method for rapidly identifying different processing methods and host plants of Cynomorium songaricum based on electronic nose technology. BACKGROUND

[0002] With the increasing pursuit of healthy life, many health care products with medicinal and edible properties have gradually entered daily life. Among them, Cynomorium songaricum, as a rare medicinal and edible plant, is particularly favored by the market. Products on the market often falsely label their effects, use inferior products to replace good ones, and consumers cannot obtain actual health care effects after use. Therefore, how to quickly and accurately identify the quality of Cynomorium songaricum has become an important problem in current research.

[0003] As a new type of detection technology, electronic nose has the characteristics of rapidness, accuracy and non-destructiveness. At present, electronic nose has been widely used in the detection of food flavor quality, but has not been applied to Cynomorium songaricum. At present, there is no related invention patent application.

[0004] In summary, the existing technology has the following problems: there is a lack of fast and effective methods and technical means for identifying the quality of Cynomorium songaricum. By introducing electronic nose technology, it is expected to solve this problem and provide strong technical support for the healthy development of the Cynomorium songaricum industry. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a method for rapidly identifying different processing methods and host plants of Cynomorium songaricum based on electronic nose technology, which can quickly and accurately identify different processing methods and host plants of Cynomorium songaricum, and is of great significance for protecting consumer rights and promoting the healthy development of the Cynomorium songaricum industry.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0007] A method for rapidly identifying different processing methods and host plants of Cynomorium songaricum based on electronic nose technology, characterized in that it comprises the following steps:

[0008] Step 1, sample processing; the different host plants of the sample are Cynomorium songaricum and Nitraria tangutorum Bobr.; the different processing methods of the sample include the following: (1) 3 kinds of slice thickness: 0.5 cm, 1 cm and 2 cm thick succulent stem flake; (2) 6 kinds of drying methods: oven 105℃ fixation 30 min, 65℃ drying to constant weight; oven 105℃ fixation 30 min, 85℃ drying to constant weight; oven 65℃ drying to constant weight; oven 85℃ drying to constant weight; vacuum freeze-drying: 0.096 hP, -100℃ drying to constant weight; natural drying: placed outdoors to dry naturally to constant weight; 3 kinds of slice thickness and 6 kinds of drying methods are combined into 18 kinds of processing methods; the sample is crushed with a high-speed pulverizer and sieved through an 80-mesh sieve for subsequent analysis;

[0009] Step two, 1.0 g of sample was loaded into a 10 mL silica gel cap vial, and after 25 min of water bath equilibrium at 60℃, the electronic nose was used for detection; the detection conditions and parameters were as follows: air inlet rate 1 L·min-1; sensor cleaning time 180 s; sensor cleaning flow rate 6 L·min-1; sample detection time 90 s;

[0010] Step three, statistical analysis; the electronic nose sensor response value data were used as the independent variable, and the processing method and host plant were used as the dependent variable, the data were shuffled, and 80% of them were randomly selected as the training set, and the remaining 20% were used as the test set, a Gaussian Naive Bayes (GNB) model was established, and the machine learning algorithm was further used to accurately identify different categories of horny goat weed samples; each determination index had three replicates, and the obtained data were analyzed and processed using SPSS PRO software and Origin 2022 software.

[0011] In view of the unique odor characteristics of horny goat weed, the electronic nose technology is used to detect and analyze horny goat weed from different host plants and processed by different processing methods, understand the influence of different host plants and processing methods on the flavor of horny goat weed, and provide a strong theoretical basis for the rapid quality identification and characteristic flavor evaluation of horny goat weed. In data analysis, the Gaussian Naive Bayes (Gaussian Naive Bayes, GNB) classification model statistical method is used, the retention value of the electronic nose is used to qualitatively identify the characteristic odor of horny goat weed, and a discrimination model is established, which lays a solid theoretical foundation for future qualitative prediction.

[0012] The present application has the following beneficial effects:

[0013] 1. Rapidity: Electronic nose technology has the characteristics of rapid response, which can complete the detection and identification of odor in a short time, greatly shortening the time of horny goat weed quality identification and improving the detection efficiency.

[0014] 2. Accuracy: Electronic nose technology simulates the olfactory function of human nose and can perceive very small odor differences, accurately analyzes and judges the characteristic odor of horny goat weed, and realizes the accurate identification of horny goat weed quality.

[0015] 3. Non-destructive: Compared with traditional chemical analysis methods, electronic nose technology does not need to process or destroy the sample, can maintain the integrity of the sample, and avoids the errors that may be introduced in the sample processing process.

[0016] 4. Intelligence: Through intelligent sensory technology, the electronic nose can automatically detect and analyze horny goat weed from different host plants and processing methods, reduce the interference of human factors, and improve the objectivity and reliability of the detection.

[0017] 5. Wide applicability: Electronic nose technology is not only suitable for quality identification of Cynomorium songaricum, but also can be applied to quality detection of other traditional Chinese medicinal materials or foods with unique odor characteristics, and has wide applicability and application prospect.

[0018] In summary, the electronic nose technology is used for quality identification of Cynomorium songaricum in the present application, which has the advantages of rapidity, accuracy, non-destructiveness, intelligence and wide applicability, and provides strong technical support for quality control and market supervision of Cynomorium songaricum industry. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 Electronic nose radar fingerprint of odor characteristics of 18 kinds of processing methods of Cynomorium songaricum in the present application;

[0020] Figure 2 Confusion matrix heat map of GNB model training set (A) and test set (B) results of 18 kinds of processing methods in the present application;

[0021] Figure 3 Confusion matrix heat map of GNB model training set (A) and test set (B) results of 3 kinds of slicing thickness in the present application;

[0022] Figure 4 Confusion matrix heat map of GNB model training set (A) and test set (B) results of 6 kinds of drying methods in the present application;

[0023] Figure 5 Confusion matrix heat map of GNB model training set (A) and test set (B) results of 3 kinds of drying methods in the present application;

[0024] Figure 6 Electronic nose radar fingerprint of odor characteristics of Cynomorium songaricum of different host plants in the present application;

[0025] Figure 7 Confusion matrix heat map of GNB model training set (A) and test set (B) results of different host plants in the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0027] Embodiment 1

[0028] Sample processing

[0029] Different host plants: fresh Cynomorium songaricum and Nitraria sibirica.

[0030] Different processing methods: (1) 3 kinds of slice thickness: 0.5 cm, 1 cm, 2 cm thick succulent stem slices. (2) 6 kinds of drying methods: oven 105℃ fixation 30 min, 65℃ drying to constant weight; oven 105℃ fixation 30 min, 85℃ drying to constant weight; oven 65℃ drying to constant weight; oven 85℃ drying to constant weight; vacuum freeze-drying: 0.096 hP, -100℃ drying to constant weight; natural drying: placed outdoors to dry to constant weight. Each treatment is numbered, where letters A~C represent slice thickness, numbers 1~6 represent drying methods, and the combination of numbers and letters represents processing methods. The specific numbering is shown in Table 1. The above samples are pulverized with a high-speed pulverizer and sieved through an 80-mesh sieve for subsequent analysis.

[0031] Table 1 Numbering of slice thickness, drying method and processing method

[0032] 2. Instruments

[0033] Table 2 Test instruments

[0034] Instrument name Model Instrument manufacturer Electric heating constant temperature blast drying oven DHG-9420A Shanghai Hengsheng Scientific Instrument Co., Ltd. Vacuum freeze dryer CoolSafe 100-4 Genebio International Trading (Shanghai) Co., Ltd. High-speed universal pulverizer FW100 Test Instruments Co., Ltd. One ten-thousandth analytical balance BSA224S Sartorius Scientific Instruments (Beijing) Co., Ltd. Standard test sieve 80 mesh Shaoxing Shangyu Huafeng Hardware Instrument Co., Ltd. Constant temperature water bath HH-S26 Jintan City Datu Automation Instrument Factory Electronic nose system cNose-18 Shanghai Baosheng Industry Development Co., Ltd.

[0035] 3. Experimental methods

[0036] 1.0 g of the sample to be tested was placed in a 10 mL silica gel cap vial, and after equilibration in a 60℃ water bath for 25 min, the detection was performed. The detection conditions and parameters were as follows: inlet air flow rate 1 L·min-1; sensor cleaning time 180 s; sensor cleaning flow rate 6 L·min-1; sample detection time 90 s. The classification of substances corresponding to each sensor of the electronic nose is shown in Table 3.

[0037] Table 3 Classification of substances corresponding to electronic nose sensors

[0038] 4. Statistical analysis

[0039] The electronic nose sensor response value data were used as the independent variable, and the processing method and host plant were used as the dependent variable. The data were shuffled, and 80% of them were randomly selected as the training set, and the remaining 20% were used as the test set. A Gaussian Naive Bayes (GNB) model was established, and machine learning algorithms were used to further accurately identify different categories of Cynomorium songaricum samples.

[0040] Each determination index had 3 replicates, and the obtained data were analyzed and processed using SPSS PRO software and Origin 2022 software.

[0041] 4.1 Identification of Cynomorium songaricum with different processing methods based on electronic nose technology

[0042] 4.1.1 Analysis of electronic nose response values

[0043] As can be seen from Figure 1 , 18 electronic nose sensors respond to the volatile components of different processing methods of Cynomorium songaricum, but there are differences between the response characteristics of different sensors, and the contribution to the test data is also different. Under different slice thickness, the electronic nose response values of the same drying method are basically consistent; while under the same slice thickness, the electronic nose response values between different drying methods have a large gap, which is consistent with the conclusion in the previous section that the drying method has a greater impact on nutrients and antioxidant activity. Overall, the response value of the oven-dried sample is the highest, followed by the sun-dried sample, and the freeze-dried sample is the lowest. The four sensors with the greatest impact on the odor response value of Cynomorium songaricum are S1, S2, S10, and S15. Among them, S1 sensor is mainly sensitive to short-chain alkanes, S2 sensor is mainly sensitive to carbon-containing substances, S10 sensor is mainly sensitive to hydrogen-containing gases, and S15 sensor is mainly sensitive to carbon-containing substances, alcohols and aldehydes. This indicates that the differences in the odor of Cynomorium songaricum processed by different methods are mainly concentrated in short-chain alkanes, carbon-containing substances, hydrogen-containing gases, and alcohols and aldehydes. In summary, the odors of Cynomorium songaricum processed by different methods are different.

[0044] 4.1.2 Establishment of GNB classification model based on electronic nose

[0045] (1) Establishment of GNB classification model based on electronic nose for 18 processing methods

[0046] By observing the confusion matrix heat map of the training set of 18 processing methods ( Figure 2 A), that is, the cross set of model prediction and actual situation, it is found that all the samples in the prediction set are predicted correctly. Looking at the confusion matrix of the test data ( Figure 2 B), it is found that there are 5 samples predicted correctly, namely A2, A3, A5, C5 and C6; 6 samples predicted incorrectly, namely 1 B3, B4, B5, C1 and 2 A1. As shown in Table 4-1, the accuracy, precision and recall of this model on the training set are all 100.0%, and the F1-Score is 1.000, which performs very well. However, the accuracy and recall on the test set are both 45.5%, the precision is 40.9%, and the F1-Score is 0.424, which is generally evaluated. This shows that the GNB discrimination model established based on electronic nose data cannot accurately predict Cynomorium songaricum samples processed by different methods.

[0047] Table 4-1 Evaluation results of GNB model for 18 processing methods

[0048] Item Accuracy Recall rate Precision F1-Score Training set 100.0% 100.0% 100.0% 1.000 Test set 45.5% 45.5% 40.9% 0.424

[0049] (2) Establishment of GNB classification models for three slice thicknesses

[0050] Depend on Figure 3 As shown in Figure A, the GNB model correctly predicted 28 samples in the training set for the three slice thicknesses, including 11 0.5 cm slice samples, 13 1 cm slice samples, and 4 2 cm slice samples; it incorrectly predicted 15 samples, including 3 0.5 cm slice samples, 3 1 cm slice samples, and 9 2 cm slice samples. Figure 3 As shown in B, a total of 7 samples were correctly predicted in the test set, including 4 0.5 cm slice samples, 2 1 cm slice samples, and 1 2 cm slice sample; only 2 samples were incorrectly predicted, both of which were 2 cm slice samples. Table 4-2 shows that the model achieved a precision and recall of 65.1% and a accuracy of 68.3% on the training set, with an F1-Score of 0.627, demonstrating good performance. On the test set, the model achieved a precision and recall of 63.6% and a accuracy of 81.8%, with an F1-Score of 0.579, also showing good evaluation results. This indicates that the GNB discrimination model for the three slice thicknesses established based on electronic nose data can accurately predict *Cistanche deserticola* samples with different slice thicknesses.

[0051] Table 4-2 Evaluation results of GNB model for three slice thicknesses

[0052] (3) Establishment of GNB classification model for six drying methods based on electronic nose

[0053] Depend on Figure 4 As shown in Figure A, the GNB model correctly predicted 36 samples across the six drying methods in the training set. These included 5 samples that underwent blanching at 105℃ for 30 min and drying at 65℃, 6 samples that underwent blanching at 105℃ for 30 min and drying at 85℃, 7 samples that were dried at 65℃, 5 samples that were dried at 85℃, 6 sun-dried samples, and 7 freeze-dried samples. Seven samples were incorrectly predicted: 1 sample that underwent blanching at 105℃ for 30 min and drying at 65℃, 3 samples that underwent blanching at 105℃ for 30 min and drying at 85℃, and 3 samples that were dried at 85℃. Figure 4It can be seen from Table 4-3 that the accuracy and recall rate of the model obtained on the training set are both 83.7%, the precision is 86.9%, the F1-Score is 0.836, and the performance is good. The accuracy and recall rate obtained on the test set are 72.7%, the precision is 81.8%, the F1-Score is 0.764, and the evaluation effect is also good. It shows that the GNB discrimination model of 6 kinds of drying methods established according to the electronic nose data can accurately predict the different slice thickness of the samples of Cynomorium songaricum.

[0054] Table 4-3 Evaluation results of GNB model of 6 kinds of drying methods

[0055] (4) Establishment of GNB classification model of 3 kinds of drying methods based on electronic nose

[0056] From Figure 5 It can be seen from Table 4-3 that the accuracy and recall rate of the model obtained on the training set are both 83.7%, the precision is 86.9%, the F1-Score is 0.836, and the performance is good. The accuracy and recall rate obtained on the test set are 72.7%, the precision is 81.8%, the F1-Score is 0.764, and the evaluation effect is also good. It shows that the GNB discrimination model of 6 kinds of drying methods established according to the electronic nose data can accurately predict the different slice thickness of the samples of Cynomorium songaricum. Figure 5 It can be seen from Table 4-3 that the accuracy and recall rate of the model obtained on the training set are both 83.7%, the precision is 86.9%, the F1-Score is 0.836, and the performance is good. The accuracy and recall rate obtained on the test set are 72.7%, the precision is 81.8%, the F1-Score is 0.764, and the evaluation effect is also good. It shows that the GNB discrimination model of 6 kinds of drying methods established according to the electronic nose data can accurately predict the different slice thickness of the samples of Cynomorium songaricum.

[0057] Table 4-4 Evaluation results of GNB model of 3 kinds of drying methods

[0058] 4.2 Identification of different host plants of Cynomorium songaricum based on electronic nose technology

[0059] 4.2.1 Analysis of electronic nose response value

[0060] From Figure 6It can be seen that 18 sensors respond differently to volatile components of different host plants of Cynomorium songaricum. Overall, the electronic nose response value of Cynomorium songaricum is greater than that of Cynomorium coccineum. Among the 18 sensors, S4, S6, S9, S14, S15 and S16 have a greater impact on the response value of Cynomorium songaricum odor. Among them, S4 and S16 are mainly sensitive to sulfides, S6 is mainly sensitive to aldehydes and ketones, S9 is mainly sensitive to alkanes, alcohols, ketones and other substances, S14 is mainly sensitive to short-chain alkanes, S15 is mainly sensitive to carbon-containing substances, alcohols and aldehydes, indicating that the differences in the odor of different host plants of Cynomorium songaricum are mainly concentrated in sulfides, aldehydes and ketones, alkanes, short-chain alkanes, alcohols, ketones and aldehydes. The above shows that there are differences in odor characteristics between different host plants of Cynomorium songaricum.

[0061] 4.2.2 Establishment of Gaussian Naive Bayes (GNB) classification model based on electronic nose

[0062] From Figure 7 A can be seen that all samples in the GNB model training set of different host plants are predicted correctly, and from Figure 7 B can be seen that all samples in the test set model are also predicted correctly. As can be seen from Tables 4-5, the accuracy, precision and recall of this model on the training set and test set are all 100.0%, and the F1-Score is 1.000, which means that the evaluation effect on the training set and test set is very good, indicating that the GNB discrimination model of two host plants established according to the electronic nose sensor data can accurately predict Cynomorium coccineum (BW) and Cynomorium songaricum (BC) samples.

[0063] Table 4-5 Evaluation results of GNB model of different host plants

[0064] The embodiments of the present application are described above in combination with the drawings, but the present application is not limited to the above specific embodiments, and the above specific embodiments are only illustrative, not limiting, and those skilled in the art can make many forms under the inspiration of the present application without departing from the scope of the present application and the scope protected by the claims.

Claims

1. A method for rapidly identifying different processing methods and host plants, *Cistanche deserticola*, based on electronic nose technology, characterized in that... Includes the following steps: Step 1, Sample processing; The different host plants of the sample are Cynomorium songaricum and Cynomorium nidus; The different processing methods of the sample include the following: (1) Three slice thicknesses: 0.5 cm, 1 cm, and 2 cm thick fleshy stem slices; (2) Six drying methods: oven blanching at 105℃ for 30 min, drying at 65℃ to constant weight; oven blanching at 105℃ for 30 min, drying at 85℃ to constant weight; oven drying at 65℃ to constant weight; oven drying at 85℃ to constant weight; vacuum freeze drying: drying at 0.096 hP and -100℃ to constant weight; natural sun drying: placing outdoors for natural sun drying to constant weight; The three slice thicknesses and six drying methods are combined to form 18 processing methods; Step 2: Place 1.0 g of sample into a 10 mL silicone-capped vial, equilibrate in a 60℃ water bath for 25 min, and then detect using an electronic nose. The electronic nose can identify a variety of volatile gases, including: short-chain alkanes, carbonaceous substances, sulfides, hydrogen, nitrogenous substances, aldehydes, ketones, liquefied petroleum gas, and alcohols; to obtain the electronic nose response values ​​corresponding to 18 different processing methods of Cistanche deserticola. Step 3: Statistical Analysis; Using the electronic nose sensor response data as the independent variable and the processing method and host plant as the dependent variable, the data was shuffled, and 80% was randomly selected as the training set, with the remaining 20% ​​as the test set. A Gaussian Naive Bayes (GNB) model was established, and machine learning algorithms were used to further accurately identify different categories of Cynomorium songaricum samples. Each measurement index had three replicates, and the obtained data were analyzed using SPSSPRO and Origin 2022 software. The specific GNB models were established as follows: GNB classification models based on three slice thicknesses using the electronic nose; GNB classification models based on six drying methods using the electronic nose; GNB classification models based on three drying methods using the electronic nose; and GNB models based on two host plants using the electronic nose.

2. The method for rapidly identifying different processing methods and host plants *Cistanche deserticola* based on electronic nose technology according to claim 1, characterized in that: In step one, the sample is pulverized using a high-speed pulverizer and passed through an 80-mesh sieve for subsequent analysis.

3. The method for rapidly identifying different processing methods and host plants *Cistanche deserticola* based on electronic nose technology according to claim 1, characterized in that: In step two, the detection conditions and parameters are: intake rate 1 L·min -1 The sensor cleaning time is 180 seconds; the sensor cleaning flow rate is 6 L / min. -1 Sample testing time: 90 seconds.

Images