A model and method for identifying different varieties of celery based on characteristic odor differences of an electronic nose
The OPLS-DA model, established by combining electronic nose technology with clustering and discriminant analysis, solves the problem of low efficiency in celery variety identification, and achieves rapid and accurate variety differentiation. It is applicable to the identification of Western celery, native celery, and their hybrid offspring.
Patent Information
- Application Number
- CN202610346031.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-20
- Publication Date
- 2026-06-23
AI Technical Summary
Existing technologies are insufficient for quickly and accurately identifying different varieties of celery. Traditional methods are time-consuming, labor-intensive, or costly, and the application of electronic nose technology in celery variety identification has not yet been reported.
An OPLS-DA qualitative identification model based on an electronic nose was established by utilizing the characteristic response values of at least 10 metal oxide sensors and combining cluster analysis, principal component analysis, and orthogonal partial least squares discriminant analysis to achieve rapid identification of celery varieties.
It enables rapid and accurate identification of celery varieties, with a single sample testing time of only a few minutes. The model parameters R²Y and Q² are both greater than 0.9, and the results are objective and reliable, overcoming the subjectivity and complex preprocessing problems of traditional methods.
Smart Images
Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] This invention relates to the field of plant identification technology, specifically to a model and method for identifying different varieties of celery based on the characteristic odor differences of an electronic nose. Background Technology
[0002] Celery (Apium graveolens L.) is an important vegetable crop with a unique aroma and various health benefits. Based on morphological characteristics, celery is mainly divided into two types: Western celery (Apium graveolens var. dulce) and native celery (Apium graveolens var. secalinum). Western celery has broad, thick, and solid petioles, while native celery has slender, hollow petioles. Furthermore, breeders have developed many new varieties that combine characteristics of both through hybridization. Different varieties of celery exhibit significant differences in aroma characteristics, and aroma is a key indicator for evaluating celery quality and consumer acceptance.
[0003] Currently, celery variety identification mainly relies on morphological observation and molecular marker technology, but these methods are either time-consuming and labor-intensive or costly. Volatile component analysis typically employs gas chromatography-mass spectrometry (GC-MS), but this method involves complex pretreatment and a long detection cycle, making it difficult to meet the needs of rapid identification. Electronic nose technology, an analytical technique that simulates the human olfactory system, uses a sensor array to detect the overall information of volatile components in a sample, generating "fingerprint data." It offers advantages such as speed, convenience, and non-destructive testing, and has been widely used in food identification and agricultural product quality evaluation.
[0004] However, there are currently no reports on the use of electronic nose technology for the rapid identification and classification of different celery varieties (especially Western celery, native celery, and their hybrids). Therefore, developing a celery variety identification method based on electronic nose technology has significant application value. Summary of the Invention
[0005] Purpose of the invention: The purpose of this invention is to address the shortcomings of existing technologies by providing a model and method for identifying different varieties of celery based on the characteristic odor differences of an electronic nose, thereby solving the problems of difficulty, low efficiency, and strong subjectivity in celery variety identification in existing technologies.
[0006] Technical solution: The present invention describes a model for rapidly identifying different varieties of celery based on the characteristic odor differences of an electronic nose. The model is a qualitative identification model established based on electronic nose technology with the characteristic response values of at least 10 metal oxide sensors as indicators.
[0007] Furthermore, the characteristic response values of at least 10 sensors are derived from metal oxide sensors, including sensors sensitive to aromatic components, nitrogen oxides, sulfides, hydrides, alkanes, alcohols, and aldehydes.
[0008] A method for identifying celery based on the above model specifically includes the following steps:
[0009] Step 1: Using the stable response values of at least 10 metal oxide sensors of the electronic nose to the volatile components of celery samples as variables, perform cluster analysis and principal component analysis in sequence to obtain the preliminary classification and grouping of celery samples.
[0010] Step 2: The preliminary classification and grouping results described in Step 1 are verified using orthogonal partial least squares discriminant analysis to construct a qualitative identification model for celery varieties based on characteristic odor differences. The parameters R²Y and Q² of the OPLS-DA qualitative identification model are both greater than 0.9, indicating that the model is stable and reliable. The score map shows that the Western celery variety, the native celery variety, and the hybrid offspring form three independent clustering regions in space, with complete separation between groups and no overlap.
[0011] Step 3: Take a celery sample to be identified and use an electronic nose detection system to measure the response values of at least 10 sensors. Project the response values of the sample onto the OPLS-DA qualitative identification model established in Step 2. Based on the region to which the sample belongs in the model, determine its variety category. If the projection is located in the celery variety cluster area, it is identified as a celery variety; if the projection is located in the local celery variety cluster area, it is identified as a local celery variety; if the projection is located in the hybrid breeding offspring cluster area, it is identified as a hybrid breeding offspring, thereby achieving rapid identification of different celery varieties.
[0012] Further, step 1 specifically involves: using the stable response values of at least 10 metal oxide sensors of the electronic nose to the volatile components of the celery sample as variables, and employing chemometrics software to perform cluster analysis on the normalized data. For example, Origin software is used to draw a cluster heatmap to observe the natural aggregation trend of the samples. Then, principal component analysis is performed using the dimensionality-reduced principal components as coordinates. Based on the spatial distribution of the samples on the score map, the preliminary classification and grouping of the celery samples are obtained.
[0013] Furthermore, the celery samples in step 1 include Western celery varieties, native celery varieties, and hybrid breeding offspring. The Western celery varieties include Ventura (VT), Queen (PC), and California King (CK); the native celery varieties include Yellow Heart Celery (HY), Red Celery (PY), Shanghai Celery No. 1 (SCC), Bao Celery (BQ), White Celery (SW), and Shen Celery No. 4 (S4); and the hybrid breeding offspring include Shangnong Jade Celery (SHY), Shen Celery No. 2 (S2), and Shen Celery No. 3 (S3).
[0014] Furthermore, the identification result of the region to which the sample to be identified belongs in the celery variety OPLS-DA qualitative identification model is based on the projection position of the model as the basis for determining the variety classification.
[0015] Further, in step 3, the celery sample to be identified needs to undergo pretreatment. The pretreatment includes cutting fresh celery petioles into small sections and homogenizing them, or freezing-drying them and then pulverizing them through a 10-20 mesh sieve, weighing them, and placing them in a headspace vial for headspace sampling. The parameters for the electronic nose detection are set as follows: clean air as the carrier gas, carrier gas flow rate of 280-320 mL / min, headspace incubation temperature of 35-45℃, headspace incubation time of 25-35 min, data acquisition time of 140-160 s, and taking the stable response value of the sensor as the characteristic response value.
[0016] This invention also provides a model for rapidly identifying different celery varieties based on the characteristic odor differences of an electronic nose, which can be applied to celery variety identification, celery flavor quality evaluation, or celery superiority breeding.
[0017] Beneficial effects: Compared with the prior art, the advantages of the present invention are as follows:
[0018] (1) The present invention uses electronic nose technology, and the detection time for a single sample is only a few minutes. It does not require a complicated pretreatment process, which greatly improves the efficiency of celery variety identification and solves the problem of time-consuming and laborious traditional morphological identification.
[0019] (2) The present invention establishes an OPLS-DA qualitative identification model by combining cluster analysis, principal component analysis and orthogonal partial least squares discriminant analysis. The model parameters R²Y and Q² are both greater than 0.9, indicating that the model is stable and reliable. The score map shows that celery, Chinese celery and their hybrid offspring are completely separated in space and have no overlap, which can realize the accurate identification of different varieties of celery.
[0020] (3) The present invention uses an electronic nose sensor array to quantitatively determine the volatile component response value of celery samples, obtains objective odor fingerprint data, overcomes the problem of strong subjectivity and individual variation in traditional sensory evaluation, and makes the identification results repeatable and objective.
[0021] (4) This invention can be used for the rapid identification of celery, native celery and their hybrid offspring. It can also be applied to the evaluation of celery flavor quality and the screening of superior breeding varieties, providing new technical means for celery variety identification and quality evaluation. Attached Figure Description
[0022] Figure 1 This is a clustering heatmap of different celery varieties in Example 1;
[0023] Figure 2 Principal component analysis (PCA) score graphs of different celery varieties in Example 1;
[0024] Figure 3 The OPLS-DA permutation test chart in Example 1
[0025] Figure 4 This is an OPLS-DA score chart of different groups of celery in Example 1. Detailed Implementation
[0026] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the embodiments described.
[0027] Example 1: Electronic nose odor information collection and analysis of different celery varieties
[0028] 1. Experimental Materials
[0029] Twelve celery varieties were selected, including three Western celery varieties: Ventura (VT), Queen (PC), and California King (CK); six native celery varieties: Yellow Heart Celery (HY), Red Celery (PY), Shanghai Celery No. 1 (SCC), Bao Celery (BQ), White Celery (SW), and Shen Celery No. 4 (S4); and three hybrid breeding progeny: Shangnong Jade Celery (SHY), Shen Celery No. 2 (S2), and Shen Celery No. 3 (S3). All varieties were planted under the same conditions, with unified water and fertilizer management. Celery was harvested uniformly on the 70th day after transplanting, and fresh celery petioles were collected for testing.
[0030] 2. Sample pretreatment
[0031] Cut fresh celery stalks into 2 cm long pieces, mix them thoroughly, and accurately weigh 3.0 g of the sample into a 20 mL headspace vial. Immediately seal the vial with a PTFE septum and let it stand at room temperature for 5 min before testing. Set up 3 biological replicates for each variety.
[0032] 3. Electronic nose detection
[0033] An electronic nose system equipped with an array of 10 metal oxide sensors was used for detection. The sensor array includes sensors sensitive to aromatic components, nitrogen oxides, sulfides, hydrides, alkanes, alcohols, and aldehydes.
[0034] The detection parameters are set as follows:
[0035] Carrier gas: Clean air
[0036] Carrier gas flow rate: 300 mL / min
[0037] Data acquisition time: 150 s
[0038] Cleaning time: 95 seconds
[0039] Data acquisition frequency: 1 time / s
[0040] Headspace incubation temperature: 40℃
[0041] Headspace incubation time: 30 min
[0042] Each sample was tested in parallel three times, and the average response value of the sensor within 100-120 seconds was taken as the stable response value to obtain the raw data.
[0043] Table 1. Models and detection limits of 10 electronic nose sensors
[0044] Sensor model Sensitive gas types Detection limit W1C Sensitive to aromatic compounds 10ppm W5S Sensitive to nitrogen oxides 1ppm W3C Sensitive to ammonia and aromatic compounds 100ppm W6S Sensitive to hydrogen 100ppb W5C Sensitive to alkanes and aromatic compounds 1ppm W1S Sensitive to methane 100ppm W1W Sensitive to sulfides and terpenes 1ppm W2S Sensitive to ethanol and some aromatic compounds 100ppm W2W Sensitive to organosulfur compounds 1ppm W3S Sensitive to alkanes 10ppm
[0045] 4. Data Analysis
[0046] 4.1 Cluster Heatmap Analysis
[0047] After normalizing the response values of the 10 sensors, a cluster heatmap was generated using chemometrics software. The results are as follows: Figure 1 As shown, the 12 celery varieties can be grouped into three categories: Category I is Western celery varieties (VT, PC, CK), Category II is native celery varieties (HY, PY, SCC, BQ, SW, S4), and Category III is hybrid offspring (SHY, S2, S3).
[0048] 4.2 Principal Component Analysis (PCA)
[0049] Principal component analysis was performed using the response values of 10 sensors as variables in SIMCA software. The results are as follows: Figure 2 As shown, the Western celery variety and the native celery variety are clearly distributed on both sides of the PC1 axis, and the hybrid offspring cluster together, achieving good segregation. The PCA score plot and the clustering heatmap results are consistent, indicating that the electronic nose can effectively distinguish different celery varieties.
[0050] 4.3 Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA)
[0051] To further verify the grouping effect, orthogonal partial least squares discriminant analysis was performed. The model parameters R²Y and Q² are both greater than 0.9, and the permutation test results are as follows: Figure 3 As shown, after 200 cross-validations, all permutation tests showed R² and Q² values lower than the original values, and the intercepts were negative, indicating that the model is stable, reliable, and free from overfitting. (Score chart) Figure 4 The results showed that there was complete separation and no overlap between the groups of celery varieties, native celery varieties and hybrid breeding progeny, and a qualitative identification model of celery varieties based on characteristic odor differences was successfully established.
[0052] Example 2: Application of Variety Identification in Unknown Celery Samples
[0053] 1. Experimental Materials
[0054] Ten batches of commercially available celery samples awaiting identification were collected, numbered S1 to S10, and originated from different production areas and farmers' markets.
[0055] 2. Sample pretreatment and electronic nose detection
[0056] Celery samples to be identified were taken and pretreated and tested with an electronic nose according to the method described in Example 1 to obtain stable response values of 10 sensors for each sample.
[0057] 3. Variety identification
[0058] The sensor response values of 10 batches of samples to be tested were projected into the OPLS-DA qualitative identification model established in Example 1, and the variety category of the sample was determined according to the projection position of the sample in the model.
[0059] The results showed that samples from batches S1, S3, S5, S7, and S9 were projected into the celery variety cluster area and identified as celery varieties; samples from batches S2, S4, and S6 were projected into the native celery variety cluster area and identified as native celery varieties; and samples from batches S8 and S10 were projected into the hybrid breeding offspring cluster area and identified as hybrid breeding offspring.
[0060] Experimental Example 1: Model Stability and Reproducibility Assessment
[0061] 1. Experimental Design
[0062] Three representative varieties—Ventura (VT), Yellow Celery (HY), and Shangnong Jade Celery (SHY)—were selected to represent Western celery, native celery, and hybrid offspring, respectively. Six parallel samples were prepared for each variety, and electronic nose detection was performed according to the method in Example 1, with continuous measurements for 3 days, twice a day.
[0063] 2. Results
[0064] The sensor response values of 18 samples were projected into the OPLS-DA model. The results showed that all Ventura samples were projected into the celery variety cluster region, all yellow celery samples were projected into the native celery variety cluster region, and all Shangnong Yuqin samples were projected into the hybrid breeding progeny cluster region, with no misclassifications. This indicates that the model established in this invention has good stability and reproducibility.
[0065] As described above, although the invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the invention itself. Various changes in form and detail may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims
1. A model for rapid identification of different celery varieties based on characteristic odor differences using an electronic nose, characterized by: The model is a qualitative identification model based on electronic nose technology, using the characteristic response values of at least 10 metal oxide sensors as indicators.
2. The model according to claim 1, characterized in that: The characteristic response values of the at least 10 sensors are derived from metal oxide sensors, including sensors sensitive to aromatic components, nitrogen oxides, sulfides, hydrides, alkanes, alcohols, and aldehydes.
3. A method for identifying celery based on the model described in claims 1-2, characterized in that... Specifically, the steps include the following: Step 1: Using the stable response values of at least 10 metal oxide sensors of the electronic nose to the volatile components of celery samples as variables, perform cluster analysis and principal component analysis in sequence to obtain the preliminary classification and grouping of celery samples. Step 2: The preliminary classification and grouping results described in Step 1 are verified using orthogonal partial least squares discriminant analysis to construct a qualitative identification model for celery varieties based on characteristic odor differences. The parameters R²Y and Q² of the OPLS-DA qualitative identification model are both greater than 0.9, indicating that the model is stable and reliable. The score map shows that the Western celery variety, the native celery variety, and the hybrid offspring form three independent clustering regions in space, with complete separation between groups and no overlap. Step 3: Take a celery sample to be identified and use an electronic nose detection system to measure the response values of at least 10 sensors. Project the response values of the sample onto the OPLS-DA qualitative identification model established in Step 2. Based on the region to which the sample belongs in the model, determine its variety category. If the projection is located in the celery variety cluster area, it is identified as a celery variety; if the projection is located in the local celery variety cluster area, it is identified as a local celery variety; if the projection is located in the hybrid breeding offspring cluster area, it is identified as a hybrid breeding offspring, thereby achieving rapid identification of different celery varieties.
4. The method according to claim 3, characterized in that: Step 1 is as follows: using the stable response values of at least 10 metal oxide sensors of the electronic nose to the volatile components of celery samples as variables, the normalized data are clustered using chemometrics software to draw a cluster heatmap to observe the natural aggregation trend of the samples; then, principal component analysis is performed using the dimensionality-reduced principal components as coordinates, and the preliminary classification and grouping of celery samples are obtained based on the spatial distribution of the samples on the score map.
5. The method according to claim 3, characterized in that: The celery samples in step 1 include Western celery varieties, native celery varieties, and hybrid offspring. The Western celery varieties include Ventura (VT), Queen (PC), and California King (CK); the native celery varieties include Yellow Heart Celery (HY), Red Celery (PY), Shanghai Celery No. 1 (SCC), Bao Celery (BQ), White Celery (SW), and Shen Celery No. 4 (S4); and the hybrid offspring include Shangnong Jade Celery (SHY), Shen Celery No. 2 (S2), and Shen Celery No. 3 (S3).
6. The method according to claim 5, characterized in that: The identification result of the region to which the sample to be identified belongs in the celery variety OPLS-DA qualitative identification model is based on the projection position of the model as the basis for determining the variety classification.
7. The method according to claim 3, characterized in that: In step 3, the celery sample to be identified needs to be pretreated. The pretreatment includes cutting the fresh celery petioles into small sections and homogenizing them, or freezing-drying them and then pulverizing them through a 10-20 mesh sieve, weighing them, and placing them in a headspace vial for headspace injection. The parameters for the electronic nose detection are set as follows: clean air as the carrier gas, carrier gas flow rate of 280-320 mL / min, headspace incubation temperature of 35-45℃, headspace incubation time of 25-35 min, data acquisition time of 140-160 s, and taking the stable response value of the sensor as the characteristic response value.
8. The application of the model for rapid identification of different celery varieties based on characteristic odor differences of electronic nose as described in claims 1-2 in celery variety identification, celery flavor quality evaluation, or celery superiority breeding.