Insect tea species identification method based on electronic nose
By combining electronic nose technology with PCA and DFA to construct an insect tea species identification model, the problem of rapid, non-destructive, and objective identification of insect tea species has been solved, achieving efficient and low-cost insect tea species identification, which is suitable for insect tea quality assurance and market supervision.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGXI SUBTROPICAL CROPS RESEARCH INSTITUTE(GUANGXI SUBTROPICAL AGRICULTURAL PRODUCTS PROCESSING RESEARCH INSTITUTE)
- Filing Date
- 2026-04-30
- Publication Date
- 2026-05-29
AI Technical Summary
Existing technologies cannot quickly, non-destructively, and objectively identify insect tea varieties. Sensory evaluation is highly subjective and unstable, while physicochemical analysis is costly and inefficient, and there is a lack of identification methods suitable for the market.
An electronic nose recognition method based on PCA and DFA was adopted. By collecting the aroma feature values of insect tea, a species discrimination model was constructed. Combined with principal component analysis and discriminant function analysis, species differentiation was achieved.
It enables rapid, non-destructive, and objective identification of insect tea varieties, with an identification accuracy rate of 91.84%. It is suitable for rapid batch judgment, reduces testing costs, and is applicable to insect tea quality assurance and market supervision.
Smart Images
Figure CN122109458A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of insect tea aroma detection technology, specifically relating to a method for identifying insect tea species based on an electronic nose. Background Technology
[0002] Insect tea is a traditional health drink unique to ethnic minority areas in my country, hailed as "tea that is not tea." Its formation mechanism is highly unique, involving the excrement (commonly known as "insect droppings") produced by specific insects such as the rice leaf borer and the fragrant armyworm after they feed on the leaves of specific plants like vine tea, tea trees, and jujube trees. This excrement is then collected, sifted, sterilized, and dried. Insect tea is rich in amino acids, proteins, sugars, flavonoids, polyphenols, minerals, and other nutrients and bioactive substances. Its medicinal and health-promoting value is long-standing and well-established: Li Shizhen's *Compendium of Materia Medica* from the Ming Dynasty records its effects of clearing heat and relieving summer heat, strengthening the stomach and aiding digestion, and stopping diarrhea and treating hemorrhoids. Modern pharmacological research further confirms that insect tea also possesses multiple functions such as anti-oxidation, antibacterial and anti-inflammatory effects, regulating blood sugar and lipids, and improving intestinal flora structure. Currently, it is mainly produced in ethnic minority areas such as Hunan, Guizhou, and Guangxi. As its health-promoting value is widely recognized, its market acceptance and scale are gradually increasing.
[0003] The quality characteristics of insect tea (such as aroma, taste, color, and efficacy) are closely related to the production environment, host plant species, insect species, and subsequent processing techniques. While insect teas from different production areas (such as Chengbu in Hunan, Chishui in Guizhou, and Guilin in Guangxi) all exhibit a compact, granular appearance with a rough surface and high porosity, their internal flavor compounds differ significantly. For example, tea insect tea contains caryophyllene, 3-methoxybenzaldehyde, 2-ethyl-1-hexanol, thiamethoxam, and geraniol; jujube leaf insect tea contains 5-methylfurfural, o-methylisopropylbenzene, mesitylene, and p-propenyltoluene; and Uncaria insect tea contains 1H-pyrrole-2-carboxaldehyde, 1-(4-methylphenyl)ethyl ketone, 1-phenyl-1-propanone, hexadecane, and nonanal. These differences directly determine the aroma, taste, health benefits, and market value of insect teas. They are also key indicators for distinguishing different types of insect teas and judging their quality grades, directly impacting the market value, brand reputation, and consumer perception of insect teas.
[0004] However, the current identification of insect tea varieties mainly suffers from the following problems and limitations: (1) Sensory evaluation is highly subjective and unstable: The current mainstream methods rely heavily on experienced professionals (such as evaluators and buyers) to make comprehensive judgments and distinctions through sensory means such as visual observation (color, particle shape), olfactory perception (dry aroma, wet aroma), and taste experience (broth color, flavor). This method is not only highly subjective and has large individual differences, but is also significantly affected by the experience, physical condition, and environmental factors of the evaluated personnel, resulting in low accuracy, poor repeatability, and difficulty in unifying classification standards. Especially for different insect teas with similar flavor characteristics, or for newcomers to the industry, the misjudgment rate is relatively high. (2) High cost, low efficiency and strong destructiveness of physicochemical analysis: Although modern instrumental analysis methods, such as high performance liquid chromatography (HPLC), gas chromatography-mass spectrometry (GC-MS), and inductively coupled plasma mass spectrometry (ICP-MS), can accurately determine specific chemical components (such as catechins, flavonoids, and mineral element fingerprints) in insect tea to help distinguish the types, these methods usually require complex sample pretreatment processes (extraction, purification, derivatization, etc.), expensive instruments and equipment, high operational skills, long detection cycles, and the detection process can cause irreversible damage to the sample. These limitations make it only suitable for precise laboratory analysis and cannot be adapted to the actual scenarios of insect tea circulation, rapid market supervision and batch sample screening. (3) Lack of rapid, non-destructive and objective identification methods: Considering the shortcomings of existing technologies, the market urgently needs a technology that can overcome the subjectivity of sensory evaluation and the limitations of physicochemical analysis to achieve rapid, non-destructive, high-throughput, low-cost and objective identification of insect tea types.
[0005] Electronic nose technology, as an artificial olfactory analysis technique that simulates the biological olfactory system, has the core advantage of rapidly identifying and classifying complex odor substances by combining the overall response of volatile gases in a sample ("odor fingerprint") with pattern recognition algorithms through a specific sensor array. This technology boasts significant advantages such as simple sample pretreatment (usually requiring only simple crushing or headspace equilibration), fast analysis speed (minutes to tens of minutes), non-destructive or minimally invasive detection, and relatively simple operation, demonstrating enormous application potential in various fields such as food quality control, wine origin traceability, tea grading, and disease diagnosis. However, the formation mechanism and aroma component system of insect tea differ fundamentally from those of conventional tea: the aroma of insect tea originates from the synergistic transformation of insect metabolism and host plant components, resulting in a more complex composition and potentially more subtle differences between different species. This poses specific challenges to the selectivity and sensitivity of the electronic nose sensor array, as well as the anti-interference capability of subsequent data processing algorithms. Currently, there are no systematic studies or practical application reports on the use of electronic nose technology for species identification of this special product, insect tea, both domestically and internationally.
[0006] Based on this, we developed an electronic nose recognition method that is adapted to the unique aroma characteristics of insect tea and can effectively distinguish different types of insect tea, filling the gap in existing technology and possessing both important theoretical research value and broad industrial application prospects. Summary of the Invention
[0007] In view of the above, it is necessary to provide a method for identifying insect tea species based on electronic nose. This method constructs a species discrimination model based on PCA and DFA, which can objectively, accurately and quickly distinguish insect tea from different plant raw material sources, and solves the technical problem that conventional detection methods cannot be applied to insect tea.
[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows.
[0009] A method for identifying insect-fed tea species based on an electronic nose, comprising the following steps.
[0010] (1) Pretreatment of sample aroma: Select training insect-collecting tea and test insect-collecting tea, prepare dry tea samples, and immediately seal these samples in headspace bottles for aroma enrichment.
[0011] (2) Odor data acquisition: The headspace gas enriched in the headspace vial in step (1) is detected using an electronic nose to obtain the response curves and response values of each sensor over time.
[0012] (3) Aroma feature value extraction: Select 3 to 5 response values of each sensor response change curve that reach the steady stage as the aroma feature value of the sample. The steady stage refers to the period when the response value change rate is less than 0.1% per second.
[0013] (4) Construction of species distribution model: Based on the aroma feature values extracted in step (3) of the training set sample, principal component analysis (PCA) is applied to process them to generate PCA differentiation map reflecting the distribution area of different species of insect tea in the dimensionality reduction aroma space.
[0014] (5) Discriminant model construction: Based on the aroma feature values of the training set samples and their corresponding species labels, discriminant function analysis (DFA) is applied to construct a species discrimination model for insect tea.
[0015] (6) Test set sample discrimination: The aroma feature values obtained after processing the test set samples in steps (1)-(3) are input into the discriminant function analysis (DFA) model constructed in step (5) to obtain the discrimination results.
[0016] In this invention, the method further includes step (7): projecting the aroma feature values of the test set samples onto the PCA differentiation map obtained in step (4), and performing visual verification in combination with the classification results of step (6).
[0017] In this invention, further, in step (1), the training insect-collecting tea and the testing insect-collecting tea are made from jujube trees (Solanum nigrum). Choerospondias axillaris ), vine tea ( Ampelopsis grossedentata ), Uncaria ( Uncaria rhynchophylla ),tea tree( Camellia sinensis ), sweet potato ( Ipomoea batatas ), maple tree ( Liquidambar formosana It is made from the leaves or tender stems of plants.
[0018] In this invention, the aroma enrichment operation in step (1) is specifically as follows: weigh 3.0g of insect tea sample and place it in a headspace vial, then place the headspace vial in a 100°C water bath for 30min, take it out and let it stand at room temperature for 3min, and then transfer it to a 50°C water bath for 10min.
[0019] In this invention, the electronic nose is further described as a PEN-3 type electronic nose.
[0020] In this invention, the parameters of the electronic nose are further set as follows: sample preparation time 5s, automatic zero-point calibration time 10s, initial injection flow rate: 300mL / min, chamber flow rate: 300mL / min, data sampling interval time 1s, rinsing time 60s, and data acquisition time 80-120s.
[0021] Compared with the prior art, the present invention has at least the following beneficial effects.
[0022] 1. This invention employs electronic nose technology to simulate a biological olfactory system. It detects the overall response of volatile aroma components in insect tea using a specific sensor array, and combines this with pattern recognition algorithms to achieve species identification. Compared to existing methods that rely on experienced professionals to make comprehensive judgments using visual, olfactory, and gustatory senses, this invention is unaffected by subjective factors such as the reviewer's experience, physical condition, and environmental factors. The detection results are objective, stable, and repeatable, effectively solving the technical challenges of strong subjectivity, large individual differences, and difficulty in unifying classification standards in sensory evaluation.
[0023] 2. This invention, through a single-factor experimental system, screened key detection parameters such as enrichment temperature, detection temperature, enrichment time, sample weight, and injection flow rate, and for the first time established the optimal combination of detection parameters for the specific object of insect tea (3.0g sample, enrichment at 100℃ for 30min, detection at 50℃, and flow rate of 300mL / min). It is well known to those skilled in the art that different types of insect tea contain their own characteristic aroma components (e.g., tea insect tea contains caryophyllene, 3-methoxybenzaldehyde, 2-ethyl-1-hexanol, and thiamethoxam; jujube leaf insect tea contains 5-methylfurfural, o-methylisopropylbenzene, etc.; and Uncaria insect tea contains 1H-pyrrole-2-carboxaldehyde, 1-(4-methylphenyl)ethyl ketone, 1-phenyl-1-propanone, hexadecane, and nonanal, etc.), while the aroma components of ordinary tea are mainly linalool, geraniol, and ionone. It is precisely because of this unique characteristic of the insect tea aroma system that conventional detection parameters are difficult to achieve ideal differentiation. The control group experiment in the examples confirmed that, using conventional detection parameters (room temperature, 400 mL / min flow rate), there was overlap between *Tea Leaf Insect Tea* and *Uncaria Rhizoma Insect Tea*, and between *Liquidambar formosana* leaf Insect Tea and *Sweet Potato Leaf Insect Tea* in the PCA diagram, making effective differentiation impossible. However, after using the optimized parameters of this invention, the six insect-infested teas achieved clear and non-overlapping separation in the PCA visualization space, laying a solid foundation for building a high-precision discrimination model. This indicates that the present invention has creatively optimized the aroma characteristics of insect-infested teas, solving the technical problem that conventional tea detection methods cannot be directly applied to insect-infested teas.
[0024] 3. Currently, there is a lack of research and public reports on the systematic application of electronic nose technology for the identification of insect tea species. This invention is the first to apply electronic nose technology to the identification of insect tea species, establishing a complete technical chain from sample pretreatment, odor data collection, feature extraction, PCA distribution map construction, DFA discriminant model construction to visualization and verification of the discrimination results. This method can accurately identify insect tea from different plant sources (jujube leaves, vine tea, maple leaves, uncaria, tea leaves, and sweet potato leaves). Verification shows that the method of this invention achieves an accuracy rate of 91.84% in identifying insect tea from six different plant sources, and the identification results are objective and reliable. This provides effective technical support for ensuring the quality of insect tea, maintaining brand reputation, combating counterfeit and substandard products, and building a product traceability system, and has significant practical significance and industrial application value for promoting the high-quality development of the insect tea industry.
[0025] In summary, the method for identifying insect tea species based on electronic nose provided by this invention is fast, non-destructive, requires no complex sample pretreatment, is simple, environmentally friendly, has low detection costs, and provides intuitive and reliable results. It is suitable for batch, rapid, objective, and accurate identification of insect tea species. Attached Figure Description
[0026] Figure 1 The graph shows the response curve of the electronic nose after 30 minutes of enrichment at room temperature.
[0027] Figure 2 The response curve of the electronic nose after enrichment at 100℃ for 30 minutes.
[0028] Figure 3 The electronic nose response curve is shown at a detection temperature of 20℃.
[0029] Figure 4 The electronic nose response curve is shown at a detection temperature of 35℃.
[0030] Figure 5 The electronic nose response curve is shown at a detection temperature of 50℃.
[0031] Figure 6 The electronic nose response curve is shown at 100℃ for 10 min enrichment time.
[0032] Figure 7 The electronic nose response curve is shown at 100℃ for 20 min enrichment time.
[0033] Figure 8 The electronic nose response curve is shown at 100℃ for 30 min enrichment time.
[0034] Figure 9 The electronic nose response curve is shown at 100℃ for 40 min enrichment time.
[0035] Figure 10 The image shows the response curve of the electronic nose when the sample weight is 1.0g.
[0036] Figure 11 The electronic nose response curve is shown when the sample weight is 2.0g.
[0037] Figure 12 The electronic nose response curve is shown when the sample weight is 3.0g.
[0038] Figure 13 The image shows the response curve of the electronic nose when the sample weight is 4.0g.
[0039] Figure 14 The image shows the response curves of the electronic nose when the initial injection flow rate and the chamber flow rate are both 100 mL / min, where A represents the detection time of 100 s and B represents the detection time of 400 s.
[0040] Figure 15 The image shows the response curves of the electronic nose when the initial injection flow rate and chamber flow rate are both 200 mL / min, where A represents the detection time of 100 s and B represents the detection time of 200 s.
[0041] Figure 16 The electronic nose response curve is shown when the initial injection flow rate and chamber flow rate are both 300 mL / min.
[0042] Figure 17 The electronic nose response curve is shown when the initial injection flow rate and chamber flow rate are both 400 mL / min.
[0043] Figure 18 The image shows the electronic nose response curve when the initial injection flow rate and chamber flow rate are both 500 mL / min.
[0044] Figure 19 The electronic nose response curve of the insect tea sample is shown.
[0045] Figure 20-25 The electronic nose response curve is shown for the control group insect tea sample.
[0046] in, Figure 20 The image shows the electronic nose response curve for the Uncaria rhynchophylla tea sample. Figure 21 The image shows the electronic nose response curve for the Liquidambar formosana leaf insect tea sample. Figure 22 The image shows the electronic nose response curve for the sweet potato leaf insect tea sample. Figure 23 The electronic nose response curve of the jujube leaf insect tea sample is shown. Figure 24 This is a graph showing the electronic nose response curve for tea insect samples. Figure 25 The electronic nose response curve of the vine tea insect tea sample is shown.
[0047] Figure 26 PCA differentiation diagram of the control group insect tea samples.
[0048] Figure 27 PCA differentiation diagram of the insect tea samples in the experimental group (this application).
[0049] Figure 28-31 This is a PCA projection plot for predicting the sample. in, Figure 28 PCA projection diagram of tea leaves and insect-infested tea. Figure 29 PCA projection diagram of Uncaria rhynchophylla tea. Figure 30 This is a PCA projection diagram of Liquidambar formosana leaf and insect tea. Figure 31 PCA projection diagram of the predicted sample that does not belong to the above six types of insect tea. Figure 32-35 The image shows the DFA discrimination results for the predicted sample. in, Figure 32 The image shows the DFA discrimination results for tea leaves and insect-infested tea. Figure 33 The image shows the DFA discrimination results for Uncaria rhynchophylla tea. Figure 34 The image shows the DFA discrimination results for Liquidambar formosana leaf and insect tea. Figure 35 The image shows the DFA discrimination results for predicted samples that do not belong to the above six types of insect tea. Detailed Implementation
[0050] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described in detail below. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0051] 1. Experimental materials and instruments.
[0052] 1.1 Sample source.
[0053] Collect data from the entire Guangxi Zhuang Autonomous Region using *Ziziphus jujuba* (sweet jujube) trees. Choerospondias axillaris ), vine tea ( Ampelopsis grossedentata ), Uncaria ( Uncaria rhynchophylla ),tea tree( Camellia sinensis ), sweet potato ( Ipomoea batatas ), maple tree ( Liquidambar formosana Insect tea is prepared using the leaves or tender stems of insects as plant materials. After sensory evaluation to confirm the species attributes of each insect tea sample, it is randomly divided into training insect tea for model construction and test insect tea for model validation.
[0054] 1.2 Test instruments.
[0055] The experimental instrument used was the PEN3 portable electronic nose odor analysis system manufactured by AIRSENSE GmbH, Germany. This electronic nose consists of a gas sampling system, a sensor control system, and a signal acquisition system, and is equipped with 10 sets of highly sensitive metal oxide gas sensors. The performance descriptions of each sensor are shown in Table 1.
[0056]
[0057] When volatile substances from the sample enter the sensor chamber and come into contact with the heated metal oxide sensor array, the sensor resistivity G changes. The ratio of this change to the initial resistivity G0, G / G0, is the response value. The magnitude of the response value reflects the change in the volatile substance content: as the gas concentration increases, G / G0 deviates further from 1 (greater than 1 or less than 1); when the gas concentration is below the detection limit or there is no sensing gas, the response value is close to or equal to 1.
[0058] 2. Optimization and screening of electronic nose detection parameters.
[0059] To obtain the optimal detection parameters for the aroma characteristics of insect tea, this invention systematically screened the enrichment temperature, detection temperature, enrichment time, sample weight, and injection flow rate through single-factor experiments.
[0060] 2.1 Enrichment temperature screening.
[0061] 3.0g of dry tea sample was weighed and placed into a headspace vial. The samples were enriched for 30 minutes at room temperature and in a 100℃ water bath. After cooling to room temperature, odor data were collected by simultaneously inserting the injection needle and the gas replenishment needle. The gas replenishment needle was inserted deeper than the sampling needle to ensure gas circulation. The acquisition time was set to 100s, the initial injection flow rate was 400mL / min, and the chamber flow rate was 400mL / min. Each sample was repeated three times.
[0062] The results are as follows Figure 1-2 As shown. By Figure 1 It can be seen that after 30 minutes of enrichment at room temperature, the response values of all 10 sensors were below 2, indicating weak signal intensity, which is not conducive to subsequent analysis. Figure 2 It can be seen that after enrichment at 100℃, the sensor response value increases to a maximum of 14.46, and the sensor response value at equilibrium is also higher than 3, indicating a significant enhancement in signal strength. Therefore, the enrichment temperature is determined to be 100℃.
[0063] 2.2 Screening based on temperature detection.
[0064] Weigh 3.0g of dry tea sample into a headspace vial, enrich it in a 100℃ water bath for 30 min, cool it to room temperature for 2-4 min, and then place it in a water bath at 20℃, 35℃, and 50℃ for analysis. The sampling time was set to 100s, the initial injection flow rate was 400mL / min, the chamber flow rate was 400mL / min, and each sample was repeated 3 times.
[0065] The results are as follows Figure 3-5 As shown. By Figure 3 It can be seen that when the detection temperature is 20℃, the response curve rises slowly, and the response value at equilibrium is low. From Figure 4 It can be seen that when the detection temperature is 35℃, the rate of ascent of the response curve increases, but the response value at equilibrium is still relatively low. Figure 5 It can be seen that when the detection temperature is 50℃, the response curve rises rapidly, and the response value is high and stable at equilibrium. Therefore, the detection temperature is determined to be 50℃.
[0066] 2.3 Enrichment time screening.
[0067] Weigh 3.0g of dry tea sample into a headspace vial and enrich it in a 100℃ water bath for 10 min, 20 min, 30 min, and 40 min respectively. After cooling to room temperature for about 3 min, place the headspace vial on a 50℃ water bath for analysis. The acquisition time is set to 100 s, the initial injection flow rate is 400 mL / min, the chamber flow rate is 400 mL / min, and each sample is repeated 3 times.
[0068] The results are as follows Figure 6-9 As shown. By Figure 6 It can be seen that when the enrichment time is 10 minutes, the overall response value in the steady-state phase is relatively low; from Figure 7 It can be seen that when the enrichment time is 20 min, the response value improves, but the equilibrium stage is still below 3; Figure 8 It can be seen that when the enrichment time is 30 minutes, the response value reaches a high level and the curve is stable; from Figure 9 It can be seen that when the enrichment time is 40 min, the response value does not show a significant improvement compared to 30 min. Considering both detection efficiency and signal strength, the enrichment time is determined to be 30 min.
[0069] 2.4 Sample weight screening.
[0070] Weigh out 1.0g, 2.0g, 3.0g, and 4.0g of dried tea samples and place them into headspace vials. Enrich the samples in a 100℃ water bath for 30 min, cool to room temperature for about 3 min, and then place the headspace vials in a 50℃ water bath for analysis. The sampling time was set to 100s, the initial injection flow rate was 400mL / min, the chamber flow rate was 400mL / min, and each sample was repeated 3 times.
[0071] The results are as follows Figure 10-13 As shown. By Figure 10 It can be seen that when the sample weight is 1.0g, the response value in the equilibrium phase is relatively low; from Figure 11 It can be seen that the response value is improved when the sample weight is 2.0g; from Figure 12 It can be seen that when the sample weight is 3.0g, the response value reaches a suitable level and the curve is stable; from Figure 13 It can be seen that when the sample weight is 4.0g, the response value is not significantly improved compared to 3.0g, and an excessively large sample weight may result in insufficient space inside the headspace vial. Therefore, the sample weight is determined to be 3.0g.
[0072] 2.5 Sample flow rate screening.
[0073] Weigh 3.0g of dry tea sample into a headspace vial, enrich it in a 100℃ water bath for 30 min, cool it to room temperature for about 3 min, and then place the headspace vial on a 50℃ water bath for analysis. The sampling time was set to 100 s, and the initial injection flow rate and chamber flow rate were set to 100, 200, 300, 400, and 500 mL / min for comparative experiments. Each sample was repeated 3 times.
[0074] The results are as follows Figure 14-18 As shown. When the injection flow rate is ≤200 mL / min ( Figure 14 , Figure 15 When the gas flow rate is too slow, volatile aroma components accumulate in the chamber, resulting in a longer time for the response curve to reach equilibrium, thus affecting detection efficiency. When the flow rate is ≥400 mL / min ( Figure 17 , Figure 18 Excessive airflow can physically disturb the sensor, accelerate the formation of turbulence in the chamber, disrupt the stability of the airflow field, and directly cause random fluctuations in the sensor's response signal, resulting in poor response stability. However, when the flow rate is 300 mL / min ( Figure 16 The response curve exhibited a moderate equilibrium speed, suitable response value intensity, and optimal stability. Therefore, the initial injection flow rate and chamber flow rate were both determined to be 300 mL / min.
[0075] 2.6 Determination of optimal detection parameters.
[0076] In summary, through single-factor experimental system screening, the optimal detection parameters for electronic nose identification of insect tea species were determined as follows: accurately weigh 3.0g of dried insect tea sample, place it in a 100℃ water bath for 30min for enrichment, cool it to room temperature for 3min after enrichment, and then place the headspace vial in a 50℃ water bath for 10min before detection; the acquisition time was set to 100s, and the initial injection flow rate and chamber flow rate were both adjusted to 300mL / min. This parameter combination ensures that the volatile aroma signal of insect tea captured by the electronic nose is clear and stable, providing reliable technical support for the accurate identification of insect tea species.
[0077] 3. Example
[0078] 3.1 Method and process for identifying insect tea varieties.
[0079] The insect tea species identification method in this embodiment includes the following steps.
[0080] (1) Sample pretreatment: Weigh 3.0g of insect tea sample and place it in a headspace bottle. Enrich in a 100℃ water bath for 30min, cool at room temperature for 3min, and transfer to a 50℃ water bath for 10min to obtain the sample to be tested.
[0081] (2) Odor data acquisition: The PEN-3 electronic nose was used for detection to obtain the response curves and response values of each sensor over time.
[0082] (3) Aroma feature value extraction: The response value of each sensor response curve reaching the stable stage is selected as the feature value.
[0083] (4) Construction of PCA species distribution model: Based on the feature values of the training set samples, principal component analysis (PCA) is applied to generate PCA differentiation map reflecting the distribution areas of different species of insect tea in the dimensionality-reduced odor space.
[0084] (5) Construction of DFA discriminant model: Based on the feature values of the training set samples and their corresponding category labels, discriminant function analysis (DFA) is applied to construct a discriminant model for insect tea.
[0085] (6) Test set sample discrimination: Input the feature values of the test set samples into the DFA discrimination model to obtain the discrimination results.
[0086] (7) Visual verification: Project the aroma feature values of the test set samples onto the PCA differentiation map obtained in step (4), and perform visual verification in combination with the classification results in step (6) (see Section 3.5 for details).
[0087] In this embodiment, the specific parameters of steps (1) and (2) adopt the optimal combination of detection parameters determined in Section 2.6, as detailed in the experimental group in Section 3.2.2.
[0088] 3.2 Test methods.
[0089] 3.2.1 Control group.
[0090] Take 3.0g of each sample (jujube leaf insect tea, sweet potato leaf insect tea, vine tea insect tea, tea leaf insect tea, and hook vine insect tea) and place them in a headspace vial. Tighten the cap and let stand at room temperature for 15-20 minutes before testing. Insert the electronic nose sampling needle and the gas injection needle into the headspace vial simultaneously, ensuring the gas injection needle is inserted deeper than the sampling needle to guarantee gas circulation. Set the electronic nose sampling parameters as follows: sample preparation time 5s, automatic zero-point calibration time 10s, initial injection flow rate: 400mL / min, chamber flow rate: 400mL / min, data sampling interval 1s, rinsing time 60s, and data acquisition time 100s. Perform triple replicates for each sample.
[0091] 3.2.2 Experimental group of this application.
[0092] Weigh 3.0g of each of the following samples (jujube leaf insect tea, sweet potato leaf insect tea, vine tea insect tea, tea leaf insect tea, and hook vine insect tea) and place them in headspace vials. Tighten the caps and place the vials in a 100℃ water bath for 30 min to enrich. After cooling to room temperature for 3 min, transfer them to a 50℃ water bath for 10 min before testing. This set of parameters represents the optimal combination of detection parameters determined through single-factor experiments in sections 2.1-2.5. Insert the electronic nose sampling needle and the gas injection needle into the headspace vial simultaneously, with the gas injection needle inserted deeper than the sampling needle. The electronic nose sampling parameters are set as follows: sample preparation time 5s, automatic zero-point calibration time 10s, initial injection flow rate: 300mL / min, chamber flow rate: 300mL / min, data sampling interval 1s, rinsing time 60s, and data acquisition time 100s. Perform triple replicates for each sample.
[0093] 3.2.3 Statistical analysis.
[0094] The Airsense-WinMuster software配套的电子鼻was used for data processing and analysis. Principal Component Analysis (PCA) was used for sample discrimination analysis, and Discriminant Function Analysis (DFA) was used for qualitative determination of the prediction set samples.
[0095] 3.3 Collection results of electronic nose response values and extraction of characteristic values.
[0096] As Figure 19 shown, the response curves of all samples showed a trend of first rising sharply, reaching a peak, then gradually decreasing, and finally leveling off. Combining the response curves of all samples, the response values at 94 - 96 s when the response curves of each sensor reached the stable stage were selected as the characteristic values for sample identification. After calculation, within this 94 - 96 s period, the change rate of the response value of each sensor was less than 0.1% per second. For example, for the W1W sensor with a relatively significant change in response value, its average change rate during this period was about 0.05% per second, meeting the definition criteria of the "stable stage" described in this application.
[0097] 3.4 Establishment of the model.
[0098] In this experiment, a total of six kinds of insect tea (wild jujube leaf insect tea, rattan tea insect tea, liquidambar formosana leaf insect tea, uncaria rhynchophylla insect tea, sweet potato leaf insect tea, tea leaf insect tea) were collected. All samples of each kind of insect tea were divided into a training set (used to construct PCA and DFA models) and a test set (49 samples randomly selected from each kind that did not overlap with the training set in terms of collection time, origin, and processing batch) according to a certain proportion. There was no overlap in the raw materials of insect tea for the training set and the test set in terms of harvest time, origin, and processing batch.
[0099] As Figure 26 shown in the PCA discrimination graph, the scatter plots of the six kinds of insect tea showed clear and non-overlapping separation in the dimensionality reduction space, intuitively demonstrating obvious differences in the aroma characteristics of different kinds of insect tea. After verification with 49 test set samples, the overall recognition accuracy of this method for the six kinds of insect tea reached 91.84%, indicating that the model has good discrimination ability.
[0100] 3.4.1 Construction of the PCA species distribution model.
[0101] The PCA analysis method in the electronic nose software was used to perform data conversion and dimensionality reduction on the 94 - 96 s signal response values of 10 sensors, and linear classification was performed on the eigenvectors. Finally, the main two-dimensional graph was displayed on the PCA graph. The greater the contribution rate, the better it can reflect the sample information.
[0102] Control group: The electronic nose was used to collect aroma data of the dry tea of the control group insect tea, and a response curve graph was obtained ( Figure 20-25 The response characteristic values of all samples were very similar, all below 3. Among them, the response characteristic values of sweet potato leaf insect tea, maple leaf insect tea, and tea insect tea were close to 1 on most sensors, indicating that the concentration of these aroma substances was below the detection limit or that the instrument had low sensitivity to the volatile aroma components in the samples. Based on the response curves of all samples, the response values at 94-96 seconds when the response curves of each sensor reached a stable phase were selected as the sample identification characteristic values. PCA analysis was then performed on the aromatic substance response values of all samples. Figure 26 As shown in the figure. The results show that the first principal component distinguishes 96.60% of the samples, the second principal component distinguishes 2.94%, and the cumulative contribution of the two principal components distinguishes 99.54%. However, the scatter plots of tea insect tea and Uncaria rhynchophylla insect tea, and Sinica leaf insect tea and sweet potato leaf insect tea overlap, and cannot be completely separated. It should be noted that the parameter settings of this control group (room temperature, 400 mL / min flow rate) are commonly used by those skilled in the art when detecting the aroma of conventional tea. However, when these conventional tea detection parameters are directly applied to insect tea, they cannot effectively distinguish different types of insect tea samples. This indicates that the control group method (room temperature, 400 mL / min flow rate) cannot distinguish tea insect tea from Uncaria rhynchophylla insect tea, and Sinica leaf insect tea from sweet potato leaf insect tea.
[0103] Experimental group: The response values of aromatic substances in the dried insect tea of the experimental group were analyzed by PCA, such as... Figure 27 As shown in the figure. The results show that the first principal component contributed 83.52% to the differentiation, the second principal component contributed 7.07%, and the cumulative contribution of the two principal components was 90.59%, indicating that the first two principal components can effectively characterize the aroma features of the samples and can be used for subsequent analysis. Figure 27 It can be seen that by using the optimized parameters of this application, the six insect tea samples in the training set can be completely distinguished in PCA analysis, and the scatter plots of the six insect teas have no overlapping parts.
[0104] This indicates that, compared to the control group parameters, the optimized pretreatment and detection parameters of this invention (enrichment at 100℃ for 30 min, detection at 50℃, and a flow rate of 300 mL / min) can effectively enhance the differences in response characteristics of different types of insect tea on the electronic nose sensor array, thereby achieving clearer category separation in the PCA projection space and laying the foundation for the construction of a subsequent high-precision discrimination model.
[0105] 3.4.2 Construction of DFA discriminant model.
[0106] Based on the aroma characteristics of the training set samples and their corresponding species labels (sour jujube leaf insect tea, sweet potato leaf insect tea, vine tea insect tea, tea leaf insect tea, and hook vine insect tea), a species discrimination model for insect tea was constructed using discriminant function analysis (DFA). This model projects the multidimensional feature space to a low-dimensional discriminant space through the discriminant function, maximizing the differences between various types of samples, thereby achieving species discrimination of new samples.
[0107] 3.5 Test set sample discrimination and verification.
[0108] Forty-nine predicted samples were randomly selected and tested using the electronic nose method of the experimental group (i.e., using the optimized parameters of this application: 3.0g, enrichment at 100℃ for 30min, detection at 50℃, flow rate of 300mL / min, and acquisition time of 100s). The response values of 94-96s were extracted as feature values and input into the DFA discrimination model constructed in Section 3.4.2 to obtain the discrimination results of each predicted sample.
[0109] To further verify the accuracy of the discrimination results, the sensor feature values of the predicted samples were projected onto the PCA discrimination map constructed in Section 3.4.1, and then combined with the DFA discrimination results for visual verification. For example... Figure 28-31 As shown, the predicted sample landing points in the PCA diagram fall into the corresponding category regions, which is consistent with the DFA discriminant model ( Figure 32-35 The results are basically consistent: Figure 32 The image shows the DFA discrimination results for tea leaves and insect-infested tea. Figure 33 The image shows the DFA discrimination results for Uncaria rhynchophylla tea. Figure 34 The image shows the DFA discrimination results for Liquidambar formosana leaf and insect tea. Figure 35 The image shows the DFA discrimination results for predicted samples that do not belong to the above six types of insect tea.
[0110] The discrimination results of the 49 predicted sample identification models are shown in Table 2: The prediction accuracy of the dry tea identification model for unknown samples is 91.84%, indicating that the model established in this experiment can accurately identify insect tea made from different plant materials.
[0111]
[0112] 3.6 Range of characteristic response values of electronic nose sensors for different insect teas.
[0113] The response value ranges of different types of insect tea on various sensors are shown in Table 3. As can be seen from Table 3, there are significant differences in the response values of different types of insect tea on different sensors. These differences constitute the characteristic fingerprint of electronic nose for identifying the types of insect tea.
[0114]
[0115] In summary, this invention screened the optimal parameter combination for electronic nose detection of insect tea (3.0g sample, enrichment at 100℃ for 30min, detection at 50℃, and flow rate of 300mL / min) through a single-factor experimental system, and established a method for identifying insect tea species based on the DFA discriminant model. This method allows for visual verification and analysis of the discrimination results using PCA distribution maps. The method is simple to operate, rapid, non-destructive, and low-cost, achieving an accuracy rate of 91.84% in identifying six different plant-based insect tea materials. It can effectively solve the problems of inferior products being passed off as superior ones and confusing labeling in the insect tea market, and has broad application prospects.
[0116] The above embodiments only illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention.
Claims
1. A method for identifying insect-fed tea species based on an electronic nose, characterized in that, The method includes the following steps: (1) Pretreatment of sample aroma: Select training insect-collecting tea and test insect-collecting tea, prepare dry tea samples, and immediately seal these samples in headspace bottles for aroma enrichment; (2) Odor data acquisition: The headspace gas enriched in the headspace vial in step (1) is detected using an electronic nose to obtain the response curves and response values of each sensor over time. (3) Aroma feature value extraction: Select 3 to 5 response values of each sensor response change curve that have reached the steady stage as the aroma feature values of the sample. The steady stage refers to the period when the response value change rate is less than 0.1% per second. (4) Construction of species distribution model: Based on the aroma feature values extracted in step (3) of the training set sample, principal component analysis (PCA) is applied to process them to generate PCA differentiation map reflecting the distribution area of different species of insect tea in the dimensionality-reduced aroma space; (5) Discriminant model construction: Based on the aroma feature values of the training set samples and their corresponding species labels, discriminant function analysis (DFA) is applied to construct a species discrimination model for insect tea; (6) Test set sample discrimination: The aroma feature values obtained after processing the test set samples in steps (1)-(3) are input into the discriminant function analysis (DFA) model constructed in step (5) to obtain the discrimination results.
2. The method as described in claim 1, characterized in that, It also includes step (7): projecting the aroma feature values of the test set samples onto the PCA differentiation map obtained in step (4), and performing visual verification in combination with the classification results of step (6).
3. The method as described in claim 1, characterized in that, In step (1), the training and testing insect-collecting teas are made from jujube trees (Solanum nigrum). Choerospondias axillaris ), vine tea ( Ampelopsis grossedentata ), Uncaria ( Uncaria rhynchophylla ),tea tree( Camellia sinensis ), sweet potato ( Ipomoea batatas ), maple tree ( Liquidambar formosana It is made from the leaves or tender stems of plants.
4. The method as described in claim 1, characterized in that, The aroma enrichment operation in step (1) is as follows: weigh 3.0g of insect tea sample and place it in a headspace bottle, then place the headspace bottle in a 100℃ water bath for 30min, take it out and let it stand at room temperature for 3min, and then transfer it to a 50℃ water bath for 10min.
5. The method as described in claim 1, characterized in that, The electronic nose mentioned is the PEN-3 type electronic nose.
6. The method as described in claim 1, characterized in that, The parameters of the electronic nose are set as follows: sample preparation time 5s, automatic zero-point calibration time 10s, initial injection flow rate: 300mL / min, chamber flow rate: 300mL / min, data sampling interval time 1s, rinsing time 60s, and data acquisition time 80-120s.