CN108195895A - A kind of tea leaf nitrogen content rapid detection method based on electronic nose and spectrophotometric color measurement instrument - Google Patents

Abstract

The invention relates to a rapid detection method of nitrogen content in tea tree leaves based on an electronic nose and a spectrophotometer. Through the electronic nose sensor's response difference and overload analysis to tea tree leaf odor substances, 5 sensors with large contribution rates are optimized and screened out. . Secondly, the three main factors of the electronic nose detection system (gas collection container volume, headspace time, and headspace temperature) were optimized. Then through the linear regression analysis, it is determined that the b ^* value in the spectrophotometer can be used as a sign to judge the nitrogen content of leaves. Based on the five main sensors of the electronic nose and the b ^* value of the spectrophotometer as the eigenvalues, a tea tree leaf nitrogen prediction model combining the electronic nose and the spectrophotometer was established. By calculation, the accuracy rate of the model building group can reach 95%, and the accuracy rate of the verification group can reach 92.5%.

Description

A rapid detection of nitrogen content in tea tree leaves based on electronic nose and spectrophotometer method

technical field

The invention relates to a rapid detection method for the nitrogen content of tea tree leaves based on an electronic nose and a spectrophotometer, which belongs to the field of rapid detection of plant nutrition, and in particular to a method for rapidly detecting the nitrogen content of tea tree leaves by combining an electronic nose and a spectrophotometer method.

Background technique

The electronic nose uses the characteristic that each gas sensor has a different response to complex component gases, and simulates the human olfactory system to identify a variety of odors with the help of data processing methods, so as to analyze and evaluate the odor quality. The electronic nose shows unique advantages in the quality monitoring, quality evaluation and safety inspection of meat, tea, wine, fruits and vegetables, etc. It can track the processing technology and inspection process online, and it is fast and sensitive without damage to the product. Spectrophotometer technology has the characteristics of fast detection speed, high precision, and is suitable for off-line operation in laboratory environment and production environment. It uses spectral analysis method to measure the color of objects, and can obtain accurate color data. Electronic nose combined with other analytical instruments for data fusion analysis is currently a popular analysis method. But there is no report on the combination of electronic nose and spectrophotometer to identify the nitrogen content of tea tree leaves.

Nitrogen is one of the most important nutrients of tea tree. Traditional tea tree nitrogen detection methods such as plant external morphology diagnosis, plant total nitrogen diagnosis, and plant nitrate diagnosis are complex in operation, cost a lot of money and lack timeliness. With the continuous emergence of new detection technologies, non-destructive diagnostic techniques such as chlorophyll meter method, chlorophyll fluorescence technology, and remote sensing technology have been gradually applied to the diagnosis of nitrogen in tea trees. In addition, remote sensing technology for identifying objects based on their spectral reflection characteristics has also become a possible means for real-time monitoring and rapid diagnosis of plant nitrogen.

Contents of the invention

The invention provides a method for quickly detecting the nitrogen content of tea tree leaves based on an electronic nose and a spectrophotometer.

The inventor first optimized and screened out 5 sensors (W5S, W1S, W1W, W2S, W2W) with a large contribution rate through the electronic nose (Germany PEN3 type) sensor’s response difference and overload analysis to the odor substances of tea tree leaves. As the main sensor for subsequent detection. Afterwards, the detection system of tea tree leaves by electronic nose was optimized. Secondly, through linear regression analysis, it was determined that the b ^* value in the spectrophotometer can be used as a sign to judge the nitrogen content of leaves. Based on the response values (G/G0) and b ^* values of W5S, W1S, W1W, W2S, W2W, a tea tree leaf nitrogen prediction model was established by combining electronic nose and spectrophotometer. At the same time, the model is verified and predicted. The accuracy rate of the model building group can reach 95%, and the accuracy rate of the verification group can reach 92.5%.

A method for quickly detecting the nitrogen content of tea tree leaves based on an electronic nose and a spectrophotometer, the steps are as follows:

(1) Material preparation: select 20 pieces of mature tea tree leaves without damage (3-4 pieces from the top bud down), clean them and dry them;

(2) Electronic nose detection: Put the leaf prepared in step 1) into a 50mL gas collection bottle, seal it with tinfoil; use the electronic nose to detect gas by headspace sampling method. Place the gas collection bottle in an environment of 30°C for headspace preheating, and the headspace time is 30 minutes. The detection parameters are: sensor cleaning time 100s, automatic zeroing time 10s, sample preparation time 5s, sample measurement interval time 1s, internal flow rate 300mL/min, sample flow rate 300mL/min, the signal at 80-83s is used as the sensor signal analysis W5S (sensitive to nitrogen oxides), W1S (sensitive to methyl compounds), W1W (sensitive to inorganic sulfides), W2S (sensitive to alcohols, aldehydes and ketones), W2W (sensitive to aromatic composition, sensitive to organic sulfide) the response value G/G0 of 5 sensors;

(3) Spectrophotometer detection: fix the leaves prepared in step 1) with a sample holder; use the CM-5 spectrocolorimeter, and use the LAB colorimetric system to detect 20 leaves under the D65 light source (simulated sunlight) b ^* value, get the average value of b ^* value and substitute it in the formula of step 4);

(4) Substitute the W5S, W1S, W1W, W2S, W2W response values (G/G0) and b ^* values into the following formula, and the maximum Y value can be judged as the nitrogen content of the leaf.

In the formula: G/G0 ₂ , G/G0 ₆ , G/G0 ₇ , G/G0 ₈ , G/G0 ₉ respectively represent W5S, W1S, W1W, W2S, W2W sensors, and the signals at the detection time of 80-83s Response value G/G0.

The electronic nose is a portable electronic nose of the German AIRSENSE Company-PEN3 type.

The nitrogen content prediction model of the tea tree leaves established by the invention can more accurately predict the nitrogen content of the tea tree leaves. The accuracy rate of the test modeling group can reach 95%, and the accuracy rate of the verification group can reach 92.5%, which can better predict the nitrogen content of tea tree leaves.

Description of drawings

Figure 1 shows the response and overload analysis of different sensors

The figure shows that S2, S6, S7, S8, and S9 have high response values to gas substances and a large contribution rate. (There are 10 sensors in the German PEN3 portable electronic nose, respectively W1C, W5S, W3C, W6S, W5C, W1S, W1W, W2S, W2W, W3S, and the marks S1-S10 in the figure represent the above 10 sensors in turn)

Figure 2 shows the optimization of electronic nose detection conditions

The figure shows that the volume of the gas collector is 50mL, the headspace preheating temperature is 30°C, and the headspace time is 30min is the optimal detection system.

Figure 3 is the linear regression analysis of the nitrogen content of tea tree leaves and the L*, a*, b* values of the electronic nose

The figure shows that the linear relationship between L ^* value and leaf total nitrogen content is the lowest, although a ^* value shows a certain linearity, but the coefficient of determination is lower than b ^* value. The b ^* value can be used as a sign to judge the nitrogen content of leaves.

Detailed ways

The invention takes conventional green-leaf tea tree varieties (non-albino and purple varieties) as the background, and builds a model in combination with a colorimeter and an electronic nose. There are also albino and purple varieties in tea tree varieties, and albino varieties and purple varieties belong to leaf color variation varieties. The leaf color is white, yellow or purple, and the color difference has a large degree of variation. They do not belong to conventional tea tree varieties. Therefore, the leaf color variation of these two types of varieties is out of the scope of this model.

A method for rapid detection of nitrogen content in tea tree leaves based on electronic nose and spectrophotometer technology, the method is as follows:

(1) Detection of nitrogen content in tea tree leaves: take undamaged conventional green tea tree mature leaves (the 3-4th leaf with the top bud down), and after curing at 105°C for 5-10 minutes, dry at 80°C. Weigh 0.3000-0.5000g of dried and ground (0.25mm) sample, and place it in a digestion tube. Add 8mL of concentrated sulfuric acid, shake gently, and place overnight. Put the digestion tube on the digestion furnace to heat, remove it when the solution is uniform brown-black, add 10 drops of hydrogen peroxide, shake well, then heat to a slight boil, digest for about 5 minutes, remove and repeat the dropwise use 10 drops of hydrogen peroxide, then boil. Repeat this 3 to 5 times, cook until the solution is colorless or clear, then remove and filter. Draw 2-5 mL of the filtrate, and use the Kjeldahl nitrogen analyzer to measure the nitrogen content.

(2) Optimization of the electronic nose detection system: put 20 pieces of tea samples into a triangular flask filled with absorbent paper and Na ₂ CO ₃ (the absorbent paper and Na ₂ CO ₃ were both dried before use), and sealed with tin foil. The PEN3 portable electronic nose was used for gas detection by the headspace sampling method, and the response value G/G0 of each sensor was recorded (indicating the resistance value G of the nth sensor after being exposed to the sample volatiles and the sensor after being filtered by standard activated carbon The ratio of the resistivity G0 of the gas). Three factors were considered for system optimization: volume of gas collection bottle, headspace preheating temperature, and headspace time. Among them, the gas collection bottle volume test is set to 3 treatments of 50, 100, and 150mL; the headspace preheating temperature test is set to 6 treatments of 30, 40, 50, 60, 70, and 80°C; the headspace time test is set to 5, 10, 15, and 20 , 25, 30min 6 treatments. Optimization experiments were carried out for each treatment. Electronic nose detection parameters are sensor cleaning time 100s, automatic zeroing time 10s, sample preparation time 5s, sample measurement interval time 1s, internal flow rate 300mL/min, injection flow rate 300mL/min, the signal at 80-83s is used as the sensor signal time point of analysis. Before and after each measurement, the sensor is cleaned and standardized. Principal component analysis (PCA), linear discriminant analysis (LDA) and overload analysis were performed using WinMuster software, and the optimal gas collection bottle volume, headspace preheating temperature and headspace time were screened out for subsequent experiments.

Preliminary tests have found that there are too few leaves, the concentration of gas substances volatilized by the leaves is too low, and the sensor of the electronic nose is not enough to effectively identify the gas; With water vapor, the resolution of the sensor will be reduced. So choose 20 blades to measure among the present invention.

(3) Screening of color difference eigenvalues: Clean the tea tree leaves, dry them, and fix them with sample holders (included with the equipment). Use the CM-5 spectrocolorimeter to detect the L*, a*, and b* values of the leaves under the D65 light source (simulated sunlight) using the LAB colorimetric system (L* means lightness; a* positive value is red, and negative value is red). green; b* is positive for yellow and negative for blue). Combine the L*, a*, b* values with the nitrogen content of tea tree leaves in (1) for linear regression analysis to screen out the optimal color difference feature value.

(4) The nitrogen content model of tea tree leaves is established: the response value (G/G0) of W5S, W1S, W1W, W2S, W2W (the optimal sensor screened in step 2) and b ^* value (the optimal sensor screened in step 3) Color difference feature value) combined with the nitrogen content of tea tree leaves was used to establish a prediction model using Matlab software.

Extract the response value (G/G0) and b* value of the above five optimal sensors at 80-83s to establish a prediction model. The prediction model of leaf nitrogen content based on odor combined with color is:

In the formula, G/G0 ₂ , G/G0 ₆ , G/G0 ₇ , G/G0 ₈ , and G/G0 ₉ respectively represent the signal responses of W5S, W1S, W1W, W2S, and W2W sensors at the detection time of 80-83s Value G/G0.

Put G/G0 ₂ , G/G0 ₆ , G/G0 ₈ , G/G0 ₇ , G/G0 ₉ , and b* into the above formula, and the group with the largest Y value can be judged as the nitrogen content of the leaf.

Example 1. Taking the optimization of the electronic nose detection system as an example

There are 10 PEN3 general-purpose sensors, namely W1C, W5S, W3C, W6S, W5C, W1S, W1W, W2S, W2W, W3S. S1-S10 marked in Fig. 1 represent the above-mentioned 10 sensors in turn. Due to the different gas components detected by the 10 sensors, the responses of the electronic nose sensors to the odorants of tea tree leaves were different. As shown in Figure 1, the three sensors S2, S7, and S9 all have high response values to gas substances, and the sensors S6 and S8 also have a certain response, but the response values are lower than those of S2, S7, and S9. The response values of the five sensors S1, S3, S4, S5, and S10 are all around 1.0, indicating that these sensors are not sensitive to the response of gaseous substances. As shown by the overload analysis, the sensor that contributes the most to leaf odor recognition is S6, followed by S9, S7, S2, and S8, and the contribution of other sensors is small. Therefore, S2 (W5S), S6 (W1S), S7 (W1W), S8 (W2S), and S9 (W2W) were selected as the optimal sensors for subsequent experiments.

As shown in Figure 2, the PCA data of 50, 100, and 150mL treatments overlap in pairs. Through linear angle LDA analysis (Fig. 2a-2), it was found that the 100mL and 150mL treatments overlapped, while the 50mL treatment samples could be clearly distinguished. Therefore, 50mL was selected as the final gas collector volume for the test. As shown in Figure 2b-1, the sample points at 30°C are relatively concentrated, and the sample points of other treatments have a large degree of dispersion within the group, and the treatments at 50, 60, 70, and 80°C overlap. Through 2b-2LDA data analysis, it can be seen that the sample points at 30°C and 40°C can be clearly distinguished. Based on PCA and LDA data analysis, the headspace preheating temperature was finally selected as 30°C. According to the data analysis in Figure 3c, the 30-minute treatment sample point data has a small degree of dispersion, and there is no overlap with other treatments, and the other groups of treatments have overlapping areas. The final headspace time was selected as 30 min.

Embodiment 2, taking tea tree leaf chromaticity detection as example

Remove the mature leaves from the non-etiolated tea tree, wash them quickly with deionized water, and dry them. The chromaticity detection was carried out with a CM-5 spectrocolorimeter. The surface color of tea tree leaves gradually changed from green to yellow with the aggravation of nitrogen deficiency. Figure 3 shows that the linear relationship between the L* value and the total nitrogen content of leaves is the lowest. Although the a* value shows a certain linearity, the coefficient of determination is lower than the b* value. In the LAB colorimetric system, +b* means yellow, and the larger the value of +b*, the heavier the yellowness of the tested sample. In this experiment, the b* values were all greater than 0, showing a linear correlation with the total nitrogen content, and the coefficient of determination was 0.9204, indicating that the b* value had the best correlation with leaf yellowing compared with other values. Therefore, the b* value can be used as a sign to judge the nitrogen content of leaves.

Example 3, taking the leaf nitrogen content prediction model based on odor combined with color as an example

Extract the response value (G/G0) and b* value of the above five odor sensors at 80-83s to establish a prediction model. The prediction model of leaf nitrogen content based on odor combined with color is:

In the formula, G/G0 ₂ , G/G0 ₆ , G/G0 ₇ , G/G0 ₈ , and G/G0 ₉ respectively represent the signal responses of W5S, W1S, W1W, W2S, and W2W sensors at the detection time of 80-83s Value G/G0.

Put G/G0 ₂ , G/G0 ₆ , G/G0 ₈ , G/G0 ₇ , G/G0 ₉ , and b* into the above formula, and the group with the largest Y value can be judged as the nitrogen content of the leaf.

Test example 1. Taking the verification and evaluation of the leaf nitrogen content prediction model as an example

The modeling set and validation set of the model each have 80 samples. The sample selection criteria of the modeling group and the verification group are the same, they belong to the same kind of samples and have the same treatment, some are used for experiments, and some are used for verification. The following table is mainly used to verify the accuracy of the model's analysis of nitrogen content. As shown in Table 1, the accuracy rate of the modeling group is 95%, and the accuracy rate of the verification group is 92.5%, indicating that the model can accurately predict the nitrogen content of leaves.

Table 1 The back judgment and verification results of the leaf nitrogen content prediction model based on odor and color

Claims (3)

1. a kind of tea leaf nitrogen content rapid detection method based on electronic nose and spectrophotometric color measurement instrument, it is characterised in that：Step It is as follows：

(1) 20 not damaged tea tree mature leafs are selected, cleans up, dries；

(2) the ready blade of step 1) is put into 50mL gas collection bottles, is sealed and sealed with masking foil；Using headspace sampling Method carries out gas detection using electronic nose；Gas collection bottle is placed in progress head space preheating, head space time in 30 DEG C of environment 30min；Detection parameters are：Sensor scavenging period 100s, automatic zero set time 10s, sample preparation time 5s, between sample measures Every time 1s, inner stream flow 300mL/min, sample introduction flow 300mL/min, using the signal at 80-83s as sensor signal point The time point of analysis；The response G/G0 of five sensors of W5S, W1S, W1W, W2S, W2W is extracted respectively；

(3) spectrophotometric color measurement instrument detects：The ready blade of step 1) is fixed with specimen holder；Color difference meter is divided using CM-5, is adopted With LAB colour systems under d 65 illuminant, the b of 20 blades is detected respectively^*Value, takes b^*It is worth average value to substitute into step 4) formula；

(4) by W5S, W1S, W1W, W2S, W2W response G/G0 and b^*Value substitutes into following equation, and the Y value maximum class value obtained is sentenced For the LTN content；

In formula：G/G0₂、G/G0₆、G/G0₇、G/G0₈、G/G0₉W5S, W1S, W1W, W2S, W2W sensor are represented respectively, are being detected Time is the signal response G/G0 at 80-83s.

2. a kind of quick side of detection of tea leaf nitrogen content based on electronic nose and spectrophotometric color measurement instrument as described in claim 1 Method, it is characterised in that：The electronic nose is PEN3 type portable electric noses.

3. a kind of quick side of detection of tea leaf nitrogen content based on electronic nose and spectrophotometric color measurement instrument as described in claim 1 Method, it is characterised in that：The tea tree is conventional greenery tea tree breed.