Method for quantitatively detecting eugenol in natural membrane based on electronic nose technology
阅读说明：本技术 一种基于电子鼻技术定量检测可得然膜中丁香酚的方法 (Method for quantitatively detecting eugenol in natural membrane based on electronic nose technology ) 是由 潘磊庆 韩璐 屠康 张充 朱静怡 于 2020-12-30 设计创作，主要内容包括：本发明涉及一种基于电子鼻技术定量检测可得然膜中丁香酚的方法,包括样品制备、采集气味信息、气质联用测定挥发性成分、模式识别及构建定量预测模型,属于食品包装检测技术领域。本发明通过电子鼻技术结合气质联用技术获取了生物膜气味变化值以及挥发性物质的成分,通过构建的判别模型快速、无损、准确地判断出丁香酚的含量。本发明可以用于生物基包装生产过程中的监测,防止因丁香酚含量过高导致其迁移至包装食品中危害消费者的生命安全,节省了人力和时间。(The invention relates to a method for quantitatively detecting eugenol in a natural membrane based on an electronic nose technology, which comprises the steps of sample preparation, odor information acquisition, volatile component determination by gas chromatography-mass spectrometry, mode identification and quantitative prediction model construction, and belongs to the technical field of food packaging detection. The invention obtains the odor change value of the biological membrane and the components of volatile substances by combining the electronic nose technology with the gas chromatography-mass spectrometry technology, and quickly, nondestructively and accurately judges the content of the eugenol by the established discrimination model. The method can be used for monitoring in the production process of bio-based packaging, prevents the eugenol from migrating to the packaged food to harm the life safety of consumers due to overhigh content of the eugenol, and saves labor and time.)
1. A method for quantitatively detecting eugenol in a natural membrane based on an electronic nose technology is characterized by comprising the following steps: comprises the following steps
S01 preparing natural membranes with different eugenol contents;
s02, selecting an electronic nose for detecting the natural membrane, wherein the electronic nose comprises a plurality of metal oxide type gas sensors, and the metal oxide type gas sensors are respectively used for identifying aromatic components, nitrogen oxides, ammonia compounds, aromatic compounds, hydrogen gases, alkanes compounds, aromatic compounds, hydrocarbons, hydrogen sulfides, alcohols, partial aromatic compounds, organic sulfides and alkanes;
s03 obtaining a response value of each metal oxide type gas sensor using each of the natural membranes obtained in the electronic nose detecting step S01;
s04, preliminarily judging the sensitivity of the sensor in the electronic nose to the soluble membrane with different eugenol concentrations by using a Principal Component Analysis (PCA) method and a Linear Discriminant Analysis (LDA), and if the sensitivity meets the requirement, carrying out the next step;
s05 screening characteristic sensors according to the contribution of the response values of the gas sensors to distinguishing different eugenol concentrations, specifically: optimizing a sensor array of the electronic nose by adopting Load Analysis (LA) and a continuous projection algorithm (SPA) according to the response value of each metal oxide type gas sensor, and comprehensively considering and screening out characteristic sensors;
s06, establishing a characteristic sensor prediction model according to the screened characteristic sensors, and quantitatively analyzing the concentration of eugenol in the natural membrane to be detected according to the prediction model.
2. The method for quantitatively detecting eugenol in the available membranes based on the electronic nose technology as claimed in claim 1, wherein the method comprises the following steps: the raw materials for preparing the natural membrane in the step S01 comprise glycerol, tween, natural polysaccharide and eugenol, the prepared natural membrane comprises 8 types of natural membranes with different eugenol contents, and the eugenol parameters of the 8 types of natural membranes are respectively 0mg/g, 5 mg/g, 10mg/g, 20mg/kg, 30mg/g, 40mg/g, 50mg/g and 60mg/g based on the natural membrane.
3. The method for quantitatively detecting eugenol in the available membranes based on the electronic nose technology as claimed in claim 1, wherein the method comprises the following steps: the method for acquiring the response value in step S03 is: placing a single available natural membrane sample with different eugenol concentrations in a beaker, covering tin foil paper on the cup mouth for sealing, fully emitting the smell of the available natural membrane, and starting detection after the smell of the available natural membrane is balanced;
taking 10 samples of the available membrane with each eugenol concentration;
extracting the 60 th response value of each sensor as a required response value;
the response value is the resistance G of the gas sensor after the gas sensor is contacted with the volatile matter of the natural film and the resistance G of the sensor after the sensor is contacted with the clean air0Ratio of (G/G)0）。
4. The method for quantitatively detecting eugenol in the available membranes based on the electronic nose technology as claimed in claim 1, wherein the method comprises the following steps: and (4) determining the volatile component content of the soluble film with different eugenol contents in the step S01 by adopting gas chromatography-mass spectrometry between the step S03 and the step S04, and verifying the detection accuracy of the electronic nose.
5. The method for quantitatively detecting eugenol in the available membranes based on the electronic nose technology as claimed in claim 1, wherein the method comprises the following steps: the establishment of the prediction model adopts an automatic scaling algorithm for preprocessing, and establishes the prediction model of the eugenol concentration in the biological membrane based on a Partial Least Squares (PLS) algorithm and a Support Vector Machine (SVM) algorithm.
6. The method for quantitatively detecting eugenol in the available membranes based on the electronic nose technology as claimed in claim 1, wherein the method comprises the following steps: and establishing prediction models of all the sensors, comparing the prediction models with the prediction models of the characteristic sensors, and verifying whether the prediction models of the characteristic sensors are applicable or not.
Technical Field
The invention relates to a method for detecting eugenol, in particular to a method for quantitatively detecting eugenol in a natural membrane based on an electronic nose technology.
Background
The polysaccharide is a water-insoluble extracellular polysaccharide produced by bacteria, has thermal gelation and non-toxic properties, and is widely used in the food industry and related fields. In addition, the excellent film-forming properties and biological properties of the available polysaccharides have attracted attention, and are now available as raw materials for bio-based films. In view of safety, additives in bio-based food packaging have tended in recent years to select natural compounds instead of synthetic agents.
Eugenol is a natural oily liquid which, due to its abundant resources and low price, has been widely used in pharmaceuticals, cosmetics and foods, and moreover, eugenol has a broad-spectrum antibacterial property and a high antioxidant activity, and the incorporation of this natural compound in bioactive films helps to protect against certain deterioration reactions, and therefore it has attracted great interest in active packaging for the preservation of foods. The currently published articles mainly introduce the formula, performance, release rule and the like of the biological film added with eugenol, and have less research on the safety detection of the biological film in the biological basement membrane. Although recognized as a Generally Recognized As Safe (GRAS) substance by the U.S. Food and Drug Administration (FDA), eugenol is shown to have carcinogenic and mutagenic effects in mice in the chemical safety specification (MSDS), and additional literature studies indicate that eugenol is hepatotoxic and can cause aspiration pneumonia and coma, renal failure, disseminated intravascular coagulation, and the like. GB9685-2016 specifies that eugenol as an addition aid of a packaging material has no requirement for detecting the migration limit, but does not specify the addition limit and a detection method by a standard.
Various analytical methods for detecting eugenol have been developed, including High Performance Liquid Chromatography (HPLC), gas chromatography-mass spectrometry (GC-MS), and electrochemical methods. Although chromatography is widely used for simultaneous analysis of complex samples and multiple components, the instrument operation and maintenance costs involved are generally high, and the samples generally require more complex pre-treatment and longer detection time, which is not suitable for rapid detection. The electrochemical method has the advantages of simple operation, high response speed and high sensitivity, but the sensor has the advantages of complex manufacturing process, selectivity and short service life. The above analysis methods all cause damage to the sample to a certain extent, and are not suitable for quality detection of a large number of samples, so that the development of a nondestructive analysis method for measuring eugenol is of great significance.
In recent years, the nondestructive testing technology has been increasingly regarded as an emerging technology, and among them, the nondestructive testing technology for electronic noses has been widely researched and applied in a plurality of fields such as environmental monitoring, product quality detection, medical diagnosis, explosive detection and the like. The electronic nose is designed for simulating a mammal olfactory system, and the gas sensing array and the response pattern detect the odor condition of a specific position in real time, so that the electronic nose is concerned about the advantages of simple sample processing, short response time, good identification effect and real-time no damage. Eugenol has strong clove fragrance and slightly spicy fragrance, and the concentration of volatile fragrance of the eugenol is in positive correlation with the content of the eugenol.
Disclosure of Invention
The invention aims to capture the smell information of eugenol without pretreatment by utilizing an electronic nose technology according to the volatility of the eugenol, and realize the rapid nondestructive detection of the content. In order to achieve the purpose, the invention provides a method for quantitatively detecting eugenol in a natural film based on an electronic nose technology, which utilizes a gas sensor to obtain odor information of the natural film added with different eugenol concentrations, quickly and nondestructively judges the concentration of the eugenol, and prevents the eugenol from migrating to packaged food to harm the life safety of consumers due to overhigh content of the eugenol.
The invention adopts the following specific scheme: a method for quantitatively detecting eugenol in natural membrane based on electronic nose technology comprises the following steps
S01 preparing natural membranes with different eugenol contents;
s02, selecting an electronic nose for detecting the natural membrane, wherein the electronic nose comprises a plurality of metal oxide type gas sensors, and the metal oxide type gas sensors are respectively used for identifying aromatic components, nitrogen oxides, ammonia compounds, aromatic compounds, hydrogen gases, alkanes compounds, aromatic compounds, hydrocarbons, hydrogen sulfides, alcohols, partial aromatic compounds, organic sulfides and alkanes;
s03 obtaining a response value of each metal oxide type gas sensor using each of the natural membranes obtained in the electronic nose detecting step S01;
s04, preliminarily judging the sensitivity of the sensor in the electronic nose to the soluble membrane with different eugenol concentrations by using a Principal Component Analysis (PCA) method and a Linear Discriminant Analysis (LDA), and if the sensitivity meets the requirement, carrying out the next step;
s05 screening characteristic sensors according to the contribution of the response values of the gas sensors to distinguishing different eugenol concentrations, specifically: optimizing a sensor array of the electronic nose by adopting Load Analysis (LA) and a continuous projection algorithm (SPA) according to the response value of each metal oxide type gas sensor, and comprehensively considering and screening out characteristic sensors;
s06, establishing a characteristic sensor prediction model according to the screened characteristic sensors, and quantitatively analyzing the concentration of eugenol in the natural membrane to be detected according to the prediction model.
Further, in step S01, the raw materials for preparing the natural membrane include glycerin, tween, natural polysaccharide and eugenol, the prepared natural membrane includes 8 types of natural membranes with different eugenol contents, and the eugenol parameters of the 8 types of natural membranes are respectively 0mg/g, 5 mg/g, 10mg/g, 20mg/kg, 30mg/g, 40mg/g, 50mg/g and 60mg/g based on the natural membrane.
Further, the method for acquiring the response value in step S03 is as follows: placing a single available natural membrane sample with different eugenol concentrations in a beaker, covering tin foil paper on the cup mouth for sealing, fully emitting the smell of the available natural membrane, and starting detection after the smell of the available natural membrane is balanced;
taking 10 samples of the available membrane with each eugenol concentration;
extracting the 60 th response value of each sensor as a required response value;
the response value is the resistance G of the gas sensor after the gas sensor is contacted with the volatile matter of the natural film and the resistance G of the sensor after the sensor is contacted with the clean air0Ratio of (G/G)0）。
Further, the volatile component content of the soluble membrane with different eugenol content in the step S01 is measured by adopting gas chromatography-mass spectrometry between the step S03 and the step S04, and the detection accuracy of the electronic nose is verified.
Further, the establishment of the prediction model adopts an automatic scaling algorithm for preprocessing, and the prediction model of the concentration of eugenol in the biological membrane based on a Partial Least Squares (PLS) algorithm and a Support Vector Machine (SVM) algorithm is established.
Furthermore, a prediction model of all the sensors is established and compared with the prediction model of the characteristic sensor, and whether the prediction model of the characteristic sensor is applicable or not is verified.
The beneficial effects produced by the invention comprise: (1) the invention obtains the change of the odor of the natural membrane by using the electronic nose technology and can more intuitively obtain the odor information of the natural membrane.
(2) The invention utilizes the gas chromatography-mass spectrometry technology to measure the components of the natural membrane volatile substances with different eugenol concentrations, detects 27 compounds and 4 types of volatile substances together, has effective and comprehensive information, and verifies that the natural membrane containing eugenol with different concentrations is sensitive to the response of an electronic nose.
(3) The invention establishes a qualitative discrimination model for the available membranes with different eugenol concentrations by utilizing PCA and LDA, can distinguish the available membranes with different eugenol concentrations, and provides a basis for the odor determination of other substances based on the electronic nose technology.
(4) The invention optimizes the sensors of the electronic nose by using LA and SPA, and screens out the sensors with characteristics.
(5) The odor change of the natural membrane with different eugenol concentrations is measured by an electronic nose technology, and the natural membrane is modeled by PLS and SVM based on all sensors and characteristic sensors respectively, so that the content prediction of eugenol in the natural membrane is finally realized. Wherein all models have good prediction performance (R)2 p> 0.89，RPD>3) While the best prediction model is a PLS prediction model based on all sensors, R2 pThe RMSEP is 4.612 mg/g, and the method provides a basis for the establishment of a quantitative prediction model based on the electronic nose technology for other substances. The modeling effect of the PLS model established based on the feature sensor is also good（R2 p= 0.948), provides basis for developing special electronic nose equipment for detecting eugenol, and greatly reduces the number of sensors and production cost.
Drawings
FIG. 1 is a flow chart of the present invention for detecting the content of eugenol in a natural membrane.
FIG. 2 is a graph showing the relative content change of the main volatile substances in the gas analysis of the consumable film of the present invention.
FIG. 3 is a radar chart of the signal response of the available membrane electronic nose of the present invention.
FIG. 4 is a graph of principal component analysis of sensor response values for a natural membrane of the present invention.
FIG. 5 is a graph of linear discriminant analysis of sensor response values for the natural membrane of the present invention.
FIG. 6 is a load analysis graph of sensor response values for a natural membrane of the present invention.
FIG. 7 is a RMSEP plot of SPA variable screening for a sensor of a ran membrane of the present invention.
Fig. 8 is an optimal prediction model curve of eugenol content in the available membrane based on PEN3 electronic nose of the present invention.
Table 1 shows the relative amounts of the major volatile substances in the available membranes of different eugenol concentrations according to the invention.
Table 2 is a model for predicting eugenol content in available membranes based on PEN3 e-nose.
Detailed Description
The present invention is explained in further detail below with reference to the drawings and the specific embodiments, but it should be understood that the scope of the present invention is not limited to the specific embodiments.
A method for measuring the content of eugenol in a natural membrane based on an electronic nose technology comprises the following specific implementation modes:
1. test materials
The film was prepared using a tape casting method. Weighing 4g of curdlan, adding 100ml of water, 1.2g of glycerol, 0.8g of Tween 80 and a series of eugenol (based on curdlan) with different concentrations to obtain different curdlan membranes (5 mg/g (E-5), 10mg/g (E-10), 20mg/kg (E-20), 30mg/g (E-30), 40mg/g (E-40), 50mg/g (E-50) and 60mg/g (E-60), magnetically stirring the membrane solution at room temperature for 30min, adjusting the pH to 4 by lactic acid, taking out the rotor, homogenizing for 2min, pouring the membrane solution into a polytetrafluoroethylene plate, and placing in a constant temperature and humidity box (25 ℃, 50% RH) for about 24h to take off the membranes for later use.
2. Testing instrument
PEN3 model electronic nose, available from Airsense, germany, comprising 10 metal oxide sensors, each: W1C (recognizing aromatic components), W5S (recognizing nitrogen oxides), W3C (recognizing ammonia-based and aromatic-based compounds), W6S (recognizing hydrogen gas), W5C (recognizing alkane-based and aromatic-based compounds), W1S (hydrocarbon-based substances), W1W (recognizing hydrogen sulfide), W2S (recognizing alcohol-based and partially aromatic-based compounds), W2W (recognizing aromatic compounds and organic sulfides), and W3S (recognizing alkane).
Gas chromatography-Mass Spectrometry (GC-MS) instrument (7890A/5975C, Agilent technologies, Inc. USA)
3. Collecting electronic nose information
Placing single natural membrane samples with different eugenol concentrations in a 150 ml beaker, covering the mouth of the beaker with tinfoil paper, placing the beaker in a headspace at 40 ℃ and sealing for 10 min to ensure that the smell of the natural membrane is fully emitted and the detection is started after the smell reaches the balance. 10 samples of native membrane were taken for each eugenol concentration. The sample interval gas washing time is 60 s, the zero setting time is 5s, the sample preparation time is 5s, the sample detection time is 60 s, and the flow rate is 300 mL/min. The response value S of the gas sensor is determined by the resistance G of the sensor after contacting the sample volatile matter and the resistance G of the sensor after contacting the clean air0Ratio of (G/G)0) The 60 th response value of each sensor is extracted as a feature value.
4. Gas chromatography-mass spectrometry for measuring volatile components
Detecting volatile components in the natural membrane sample by adopting an HS-SPME-GC-MS method, placing 0.5 g of the natural membrane sample in a glass bottle, heating the natural membrane sample in water bath at 90 ℃ for 10 min, adsorbing the natural membrane sample in a headspace at 45 ℃ for 30min, and analyzing an extraction head at a sample inlet at 250 ℃ for 5 min. Column HP-5 (30 m 0.25 μm), temperature program: maintaining at 40 deg.C for 1 min, heating to 150 deg.C at 6 deg.C/min, heating to 240 deg.C at 7 deg.C/min, and maintaining for 3 min. No flow splitting and sample introduction are carried out, the carrier gas is He, and the flow rate is 1 mL/min. Mass spectrum conditions: the ion source temperature was 230 ℃ and the quadrupole temperature was 150 ℃. The electron energy is 70 eV, and the mass scan range is full scan. Matching the mass spectrogram of the unknown compound with an NIST mass spectrum library (2008), selecting components with the matching degree of more than 80%, calculating the relative content of each chemical component (the peak area of each compound accounts for the percentage of the total peak area of the sample) by adopting a peak area normalization method, obtaining the variation trend of each volatile substance component in the natural membrane with different eugenol contents, and providing reference for the sensor response analysis of the electronic nose.
5. GC-MS volatile component analysis
As shown in table 1, 27 compounds were identified by GC-MS analysis, mainly comprising 9 alkanes, 8 alkenes, 5 aldehydes and 5 aromatics. With the increase of the concentration of eugenol in the natural film, the types of volatile components in the natural film are not obviously changed. Aromatic compounds based on eugenol have a remarkable tendency to increase in low-concentration films, but since the relative concentration of eugenol has reached a higher level from E-20 onwards, the relative concentration of eugenol tends to stabilize in higher-concentration films. While an increase in the relative content of eugenol results in a decrease in the relative concentrations of alkanes, alkenes and aldehydes. Figure 2 shows the trend of alkane, alkene, aldehyde, aromatic compound and eugenol in the available membrane of eugenol at different concentrations.
6. Response analysis of gas sensor to available membrane odors with different eugenol contents
Fig. 3 shows radar graphs of the detection response of gas sensors with different eugenol content natural membranes, and the more deviation of the radar values from 1 represents the higher gas concentration detected by the sensors. The radar plots visually show the difference in electronic nose response values for the available membrane samples at different eugenol concentrations. As the concentration of eugenol increases, the degree of differentiation of the available membranes increases for different eugenol concentrations. The response values of W5S, W1W and W2W among different samples change more obviously, the changes of W6S, W1S, W2S and W3S are not obvious, and W1C, W3C and W5C have almost no response. Among them, W2W showed an upward trend with increasing eugenol content, while W1S and W3S decreased, which is consistent with the results of GC-MS.
7. Qualitative analysis of gas sensor detection signal
Pattern recognition can display direct and easily understood qualitative and semi-quantitative analysis results. PCA is a projection method used to reduce the dimensionality of the data, compute variables that best describe the differences between samples, and rank according to the contribution ratio (called Principal Component (PC)). To investigate whether the electronic nose can distinguish between the available membranes of different eugenol concentrations, the data of 10 sensors were processed using PCA to reduce the complexity of the data. As shown in fig. 4, the contribution rate of PC1 was 67.1%, and the contribution rate of PC2 was 26%. The results show that natural membranes with large differences in eugenol concentration can be effectively distinguished, but that there is overlap between adjacent samples, especially that E-60 cannot be separated from other natural membranes. Since PCA is suitable for only a few samples at a time, when the number of the biofilm samples to be tested is increased, the biofilm samples with different eugenol concentrations become crowded, and particularly, the crowded samples in adjacent samples in the PCA chart are difficult to distinguish. The distribution trend of the sample films at different concentrations is indicated by red arrows.
To further investigate the electronic nose data, a qualitative discriminant model was established using LDA to rapidly distinguish between natural membranes with different eugenol concentrations. LDA is a statistical method that can determine which group a sample belongs to obtain the best discrimination result by maximizing the variance between classes and minimizing the variance within a class. As shown in fig. 5, the total LDA contribution of the achievable films was 92.81%, slightly lower than PCA analysis (93.1%). However, in LDA analysis, the dispersion between samples of native films was greater than PCA analysis, which can achieve the effect of effectively distinguishing native films of different eugenol concentrations. As the eugenol concentration in the available film increased, the samples distributed from large to small along the LD1 side, as indicated by the red arrows. The above results provide reference for further differentiation.
8. Feature sensor screening
Radar plots show that some sensors contribute little to response discrimination. Therefore, the optimization of the sensor array can not only effectively eliminate useless and abnormal sensors, reduce the data volume, improve the accuracy and speed of operation, but also reduce the production cost of the system. The sensor array was optimized using LA and SPA to determine the characteristic sensor for measuring eugenol in the native membrane.
Fig. 6 shows a load analysis plot of sensor response values for samples of available membranes with different eugenol concentrations to evaluate the contribution of the sensor array to distinguish odor changes of available membranes with different eugenol concentrations. According to this figure, the sensors other than the W1C, W3C, and W5C sensors all scored higher (greater than 0.5) on the first principal component. The contribution of the W5S, W1S and W2S sensors to the first principal component is greater than the second principal component (both above 0.9), while the contribution of the W6S, W1W and W2W sensors to the second principal component is greater. Although the contribution ratio of the first principal component is larger than that of the second principal component, the contribution ratio of the second principal component is not low, and therefore the contributions of the first principal component and the second principal component should be considered in combination. According to the load graph of the sensor response values of the natural membranes with different eugenol concentrations, the contribution of the sensors W1C, W3C and W5C to the first main component and the second main component is small, so that the effects of the sensors W1C, W3C and W5C in identifying the odor change of the natural membranes with different eugenol concentrations are considered to be small and can be ignored. For the remaining sensors, the load factor scores of W6S, W1W, W2W and W5S, W1S, W2S are relatively close, indicating that there is a strong correlation between the sensors and their recognition effects are similar. The data may overlap, thus requiring further optimization and screening.
The main objective of SPA is to select the combination of variables with the least and most representative collinearity. The SPA approach can solve the co-linearity problem present in sensor arrays described above and eliminate redundant sensors. Figure 7 shows the root mean square error of the prediction set of variables. When the number of variables is 4, the value of RMSEP is relatively small and model performance is considered to be best, so the variables set when the number of variables is 4 are selected and the corresponding sensors are W5S, W1W, W2W and W3S. In view of load analysis of the integrated sensors, the W5S, W1W, W2W and W3S sensors were determined as the final optimized array (feature sensor). The results again demonstrate the consistency of the electronic nose signature and the GC-MS analysis of the available membranes.
9. The effect of the discrimination model of the content of eugenol in the natural membrane can be obtained
And (4) taking 80 samples (a modeling set: a prediction set =7: 3) as samples, and adopting an automatic scaling algorithm to carry out pretreatment, so as to establish a prediction model of the concentration of eugenol in the biological membrane based on PLS and SVM algorithms.
Where the Latent Variables (LVs) of the PLS model are 10. The SVM model adopted in the research mainly uses a root-base function as a kernel function, and parameters of the SVM model are a loss function epsilon (0.01), a penalty coefficient C (100) and a kernel function coefficient gamma (0.0316). The model prediction results are shown in table 2, and it can be seen that all models have good prediction performance (R)2 p> 0.89，RPD>3). From a modeling approach point of view, the PLS model outperforms SVM in eugenol concentration prediction based on all sensors and feature sensors. The prediction effect of the SVM prediction model based on the feature sensor modeling is slightly better than that of all sensors, but the prediction effect of the PLS model is opposite, which shows that the SPA algorithm can eliminate a small amount of effective information while reducing redundant information, thereby influencing the accuracy of the model. The number of electronic nose sensors is much smaller than the number of spectral wavelengths, and the speed of all even predictive models based on all sensors is very fast. Therefore, the establishment of a prediction model of the concentration of eugenol in the natural membrane based on all sensors is more beneficial to the retention of original information and the improvement of prediction accuracy. In summary, the optimal prediction model of eugenol concentration is based on the PLS prediction model of all sensors, and the prediction set correlation coefficient R2 pAt 0.952, the predicted root mean square error, RMSEP, was 4.612 mg/g (FIG. 8). R of the optimal prediction model of the concentration of eugenol in the natural membrane established by the experiment2 pClose to or even higher than previous findings, the quantitative and predictive model of eugenol in available membranes based on electronic nose proved to be effective. Meanwhile, characteristic sensors (W5S, W1W, W2W and W3S) are provided for developing the special electronic nose for detecting the eugenol so as to reduce the number of the sensors and the production cost.
TABLE 1
TABLE 1 continuation
TABLE 2
The above description is only a preferred embodiment of the present invention, and the present invention is not limited to the content of the embodiment. It will be apparent to those skilled in the art that various changes and modifications can be made within the technical scope of the present invention, and any changes and modifications made are within the protective scope of the present invention.