Electronic nose

Overview

The electronic nose can also be called an odor analyzer or an odor fingerprint analyzer. It is a new type of detection instrument developed in the mid-1990s for analyzing and identifying the overall characteristics of odor substances. ThinkSenso electronic nose is an intelligent sensory analysis device based on the principle of bionic simulation, using a variety of non-specific sensors to obtain multi-level characteristic information of samples, and distinguishing the odors of different samples through computer pattern recognition and fuzzy judgment.

ThinkSenso electronic nose detection technology mainly uses the gas sensor array to respond to all the volatilized gas components of the measured sample as a whole. It can be divided into five parts, including sample collection, sensor array, signal processing and presentation, pattern recognition and data processing. An ideal electronic nose sensor can achieve a sensitivity comparable to that of a human nose, and can have excellent performance such as repeatability and stability. When the volatile gas of the measured sample passes through the sensor array, a series of physical and chemical reactions between all the sensors and the volatile gas will be collected and recorded by converting them into signals. The gas compound keeps a dynamic balance by constantly being adsorbed and eluted on the surface of the sensor. The electronic nose can reflect the overall information by detecting the volatile components of the sample. It can also roughly reflect the volatile gas of the measured sample containing a certain type of component by observing and comparing the different response signals of different sensors to different sample substances. gas.

Schematic:

The main detection objects of the electronic nose are volatile flavor substances. The 10~14 or more different metal oxide sensors equipped in the electronic nose detection system have good selectivity to organic substances such as alcohols, aldehydes, alkanes, carbon oxides, nitrogen oxides, aromatic compounds, etc. . When one or more flavor substances pass through the electronic nose, the "odor fingerprint" of the flavor substance can be sensed by the sensor and extracted by a special intelligent pattern recognition algorithm. The "odor fingerprint" information of different flavor substances is different, and some specific flavor substances (characteristic flavor of samples) can represent the multivariate influence of samples in different raw material origins, different receipt times, different processing conditions, and different storage environments. General Quality Information under . The sample odor can be learned and recognized by the electronic nose detection system. Users pass PCA, DFA, SIMCA, PLS, Euclidean Dis Euclidean distance, CORRELATION correlation coefficient, MAHALANOBIS Mahalanobis distance, DP dis individual recognition rate, DI dis distinction index, Scores Dd score value separation degree, Eigenvalue Dd feature value separation degree And so on, the built-in intelligent algorithm can intuitively get the test result graph. The electronic nose is mainly suitable for detecting liquid and solid samples containing volatile substances.

Application field：

Food: wine, beverages, tea, aquatic products, livestock products, poultry and egg products, bee products, edible oil, grain storage, fruits and vegetables and processed products, milk and dairy products, seasonings and fermented foods, various soups, flavors Spices, health food, etc.

Pharmaceutical: identification of raw materials of traditional Chinese medicine, quality control of prepared pieces of Chinese medicine, formula of western medicine, odor analysis, etc.

Tobacco: origin, grade, variety judgment of various medicinal herbs, smoke analysis of finished tobacco.

Chemical industry: cosmetics, chemical raw materials and other quality and process control.

Environment: Odor pollution, gas leakage, sewage odor monitoring and on-site inspection.

Medical diagnosis: Exhaled breath examination of patients responds to diseases, tumor screening, digestive function, etc.

Host technical parameters：

1. Sensors: sensor 14 and more high temperature metal oxide sensors

2. Sensor core working temperature: 200~500℃.

3. Sensor air chamber: The material of the sensor chamber is an aluminum alloy cavity.

4. Air source: external air generator or cylinder.

5. Pressure adjustment range: 0.05-0.15MP, 0.1MP is recommended.

6. Sample detection flow rate: 0.01~2.00L/min stepless adiustment.

7. Sensor cleaning flow: 0.01~6.00L/min stepless adiustment.

8. Sampling system: use a sampling needle, a quantitative loop (10ml) and an air compressor (or steel cylinder) for sampling. After sampling with the injection needle, inject it into the quantitative loop, and use the air pressure to bring the tested sample and the filtered carrier gas into the detection gas chamber.

9. Sampling system life: non-consumable, continuous working time (full load, 24 hours uninterrupted test) more than 160000 hours.

10. Data acquisition and analysis software: It can adjust the detection flow rate and set the acquisition time; it can arbitrarily select the sensor to be tested and display the corresponding curve and radar distribution map of the sample. Sensor contribution rate analysis and more than 20 kinds of analysis and pattern recognition methods. .

11. Continuous work: The main unit of the electronic nose can continuously maintain the state of uninterrupted power supply, and the test performance is stable without being affected.

12. Application field: suitable for detecting gas, liquid and solid samples containing volatile substances. In wine, beverages, tea, aquatic products, livestock products, poultry, egg and meat products, bee products, edible oil, grain, fruits and vegetables and processed products, dairy products, condiments and fermented foods, various soups, flavors and fragrances, health care The quality and quality control of food and other foods, authenticity identification, shelf life and freshness evaluation, protection of origin, discrimination of different brands, different varieties and samples of different processing methods, qualitative and quantitative analysis of sensory attributes of samples, etc. have been obtained. full application.

14. Host size: 495mm*245mm*410mm

15. Host weight: 15KG

16. Host power: 80W

Autosampler Technical Parameters：

1. Sampling method: automatic headspace sampling (software preset sampling program)

2. Injector station: 30 positions

3. Headspace bottle specification: 40ml

4. Injector size: 565mm*460mm*455mm

5. Sampler weight: 35KG

6. Injector power: 100W

Application examples:：

With its advantages of stability, objectivity, universality and repeatability, the intelligent electronic nose can be used as an effective supplement and extension of the classic artificial sensory evaluation. The following are some of the cases, and I hope they can inspire your application development.

1. Application of electronic nose-SmartNose in judging product similarity

In the discriminant function analysis (LDA) result graph, each sample in the graph is distributed in different areas in the graph, and there is no overlap between them, and the repeatability is good; the DI value is 100.00%, indicating that each sample can be distinguished. It can be seen that samples 3 (lemon distilled wine) and 4 (mango distilled wine) are clustered together, samples 8 mango rice wine (stainless steel aging) and 9 mango wine (oak barrel aging) are clustered together, and sample 5 (planted hawthorn wine) ) 6 (kiwi fruit wine) 7 (Osmanthus wine) were divided into the same area, and sample 1 (Silver Osmanthus wine) 2 (lemon fermented wine) 10 (Osmanthus osmanthus wine) was in another area. Based on the above results, it can be seen that the taste of each type of fruit wine is different; the taste of wine made by distillation is similar, indicating that the production method of wine (distillation and non-distillation) has a certain impact on the taste of wine, different types The differences between distilled spirits were smaller than samples of the same raw material made in different ways.

2.Application of electronic nose-SuperNose in the investigation of abnormal samples of rice wine

From the sensor response radar images of different rice wine samples, it can be seen that the radar images of A1/A2/A3 samples are relatively close, and the odor response signals of A4 and these three samples are quite different. Their smell differences are mainly reflected in sensors 1, 2, 4, 6, 7, 9, 10, and 11. For the four samples, sensor optimization is performed, and the optimized sensor result is: S11_P_S7_P_S6_P. In the PCA graph, the abscissa represents the percentage of the first principal component of the overall information, and the ordinate represents the percentage of the second principal component of the overall information. In the figure, the DI value is 94.63%, which shows that the overall discrimination effect of the principal component analysis method is very good. The total contribution rate of principal component 1 and principal component 2 is 99.95%, which can basically represent all the information of the sample. In the figure, the repeatability of each sample is good. In addition, samples 1-3 are clustered together, and 4 is alone in its own area, indicating that the odor information of samples 1-3 is relatively similar, and 4 is quite different from them. It shows that in the production process, the 1/2/3 stages are all in normal production, and there is a production abnormality in the 4th stage, and the key problem needs to be checked.

3. The application of electronic nose-iNose in the prediction of star anise content

Taking the actual mass fraction of the sample as the abscissa, and the output result of the electronic nose as the ordinate, the PLS linear fitting was performed, and the correlation coefficient reached 0.9896. The average relative error of the predicted mass fraction of the star anise liquid was 6.8%, which was less than 10%. For spices, a plant tissue with very complex composition and structure, it can be considered that the model can better predict the mass fraction of star anise cooking liquid.

4. Application of electronic nose-iNose in chicken detection

In the figure, the DI value is 94.2%, which shows that the overall discrimination effect of the principal component analysis method is very good. The smells of samples 2 and 3 are relatively similar, and the smells of samples 1 and 4 are quite different from these two samples. Combined with the results of human sensory perception, sample No. 4 has a good smell, and samples No. 2 and 3 have poor smell, which is consistent with the results of the electronic nose that samples 2 and 3 are relatively close to each other, while other samples are far apart . According to the experimental results, it can be speculated that if the Maillard reaction with sugar is to obtain a product with a better smell, the enzymatic hydrolysis of chicken protein needs to be controlled at a certain molecular weight. Excessive (sample 3) or insufficient enzymatic hydrolysis (sample 2) may cause the reaction The product smells bad.