A device for nondestructive detection of oxidation deterioration of edible vegetable oil

Abstract

The application provides a device for nondestructive testing of oxidation deterioration of edible vegetable oil, and belongs to the field of food quality testing, comprising a visual detection device, an olfactory detection device, a first 12V DC power supply, a first electromagnetic valve switch, a two-way electromagnetic valve, a first activated carbon filter, a light source controller, a computer, RS-485, DAM-3055N, a second 12V DC power supply, a three-way electromagnetic valve, a second activated carbon filter, a gas sampling pump, a 5V DC power supply, a third activated carbon filter and a second electromagnetic valve switch. The application can identify oxidation deterioration of edible vegetable oil by nondestructively testing the color and smell changes of the edible vegetable oil, and solves the problems of sample destruction, complex sample pretreatment, long time consumption and high professional and technical requirements for testers in the prior art.

Description

Technical Field

[0001] This invention relates to the field of food quality testing, and in particular to a non-destructive testing device for oxidative deterioration of edible vegetable oils. Background Technology

[0002] Edible vegetable oils are easily affected by factors such as light, enzymes, and heat during storage, leading to oxidation, off-flavors, and negatively impacting the consumer's eating experience. Severely oxidized edible vegetable oils produce harmful substances that can not only cause food poisoning but also accelerate aging, leading to cardiovascular, respiratory, and nervous system diseases, and even inducing cancer. More seriously, oil oxidation often occurs silently and is difficult to detect. By the time consumers consume the oil and realize it has oxidized and spoiled, harmful substances have already accumulated and exceeded the limits set forth in GB 2716-2018, "National Food Safety Standard for Vegetable Oils." Therefore, detecting the degree of oxidation and spoilage in edible vegetable oils is of great significance.

[0003] Methods for detecting the oxidation and deterioration of edible vegetable oils generally include the weight gain method, differential pressure scanning calorimetry, acid value method, iodometric method, thiobarbituric acid method, chromatography, mass spectrometry, and Fourier transform infrared spectroscopy. However, these methods have drawbacks such as being highly destructive to samples, time-consuming, costly, and using toxic reagents.

[0004] GB 2716-2018, the National Food Safety Standard for Vegetable Oils, stipulates that vegetable oils should have their proper color and odor. However, the oxidation of oils produces various oxidation products. These products dissolve in the vegetable oil, affecting not only its color but also its odor. Oxidative deterioration of oils generally manifests as a deepening of the oil's yellow or red color; and the emission of aromas such as gasoline, mint, rust, cucumber, roasted, caramel, chicken, and vinegar. Summary of the Invention

[0005] The purpose of this invention is to provide a non-destructive testing device for oxidative deterioration of edible vegetable oils. This device can identify oxidative deterioration of edible vegetable oils by non-destructively detecting changes in color and odor. It solves the problems of existing oil oxidation deterioration testing methods, such as being destructive to samples, having complex sample pretreatment, being time-consuming, and requiring high levels of professional skills from testing personnel.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] In a first aspect, a non-destructive testing device for oxidative deterioration of edible vegetable oils is characterized by comprising: a visual detection device, an olfactory detection device, a first 12V DC power supply, a first solenoid valve switch, a two-way solenoid valve, a first activated carbon filter, a light source controller, a computer, an RS-485, a DAM-3055N, a second 12V DC power supply, a three-way solenoid valve, a second activated carbon filter, a gas sampling pump, a 5V DC power supply, a third activated carbon filter, and a second solenoid valve switch.

[0008] A two-way solenoid valve is installed on one side of the visual inspection device. The outlet of the two-way solenoid valve is connected to the visual inspection inlet of the visual inspection device through a gas pipeline. The first activated carbon filter is connected to the inlet of the two-way solenoid valve through a gas pipeline. A three-way solenoid valve is installed on the other side of the visual inspection device. The normally closed inlet of the three-way solenoid valve is connected to the visual inspection outlet of the visual inspection device through a gas pipeline. The normally open inlet of the three-way solenoid valve is connected to the third activated carbon filter through a gas pipeline. The outlet of the three-way solenoid valve is connected to the olfactory inspection inlet of the olfactory inspection device through a gas pipeline. The olfactory inspection outlet of the olfactory inspection device is connected to the second activated carbon filter through a gas pipeline. The second activated carbon filter is connected to the inlet of the gas sampling pump through a gas pipeline. The outlet of the gas sampling pump is equipped with an exhaust gas filter.

[0009] The first 12V DC power supply is connected to the first solenoid valve switch and the visual inspection device through a circuit. The first solenoid valve switch is connected to the two-way solenoid valve through a circuit. The light source controller is connected to the visual inspection device through a circuit. The second 12V DC power supply is connected to the second solenoid valve switch and the DAM-3055N through a circuit. The second solenoid valve switch is connected to the three-way solenoid valve through a circuit. The 5V DC power supply is connected to the olfactory detection device through a circuit.

[0010] The visual inspection device is connected to a computer via a signal line, the computer is connected to an RS-485 via a signal line, the RS-485 is connected to a DAM-3055N via a signal line, and the DAM-3055N is connected to an olfactory inspection device via a signal line.

[0011] Furthermore, the visual inspection device includes a CMOS camera, a visual inspection air inlet, a cover, a visual inspection air outlet, a buckle, a lens, a sample container, a chamber, and a light source.

[0012] The cover is located on top of the visual inspection device, and the chamber is located at the bottom of the cover. The interior of the chamber has a headspace for placing sample containers. A light source is embedded at the bottom of the chamber. The chamber has a square structure, and adjustable buckles are embedded on the four sides of the outer perimeter of the chamber.

[0013] The top of the cover has a hole for embedding a CMOS camera. The CMOS camera is embedded in the hole, and the lens of the CMOS camera extends into the headspace. The top two ends of the cover are respectively provided with a visual inspection air inlet and a visual inspection air outlet, which are symmetrically distributed on both sides of the CMOS camera. The sides of the cover are respectively provided with four slots, which are tightly connected with buckles. A sealing gasket is provided at the junction of the inner wall of the cover and the compartment.

[0014] Furthermore, the olfactory detection device is provided with an olfactory detection air inlet and an olfactory detection air outlet at both ends, and an air chamber is provided inside the olfactory detection device. The air chamber contains 14 different types of metal oxide semiconductor gas sensors, which are distributed on both sides of the air chamber.

[0015] Furthermore, the CMOS camera is a color camera with a resolution of not less than 1920×1080. The lens of the CMOS camera has low distortion and achromatic design. The CMOS camera is connected to the computer via a GigE cable.

[0016] Furthermore, the lid and the perimeter of the compartment are made of black, light-blocking organic plastic, while the bottom of the compartment is made of transparent organic plastic.

[0017] The headspace chamber has a diameter of 100mm and a height of 100mm. The chamber body is a square structure of 140mm×140mm. The diameter of the hole that can embed the CMOS camera is 27.8mm.

[0018] Furthermore, the light source is a surface light source, the light source is white, the color temperature of the light source is 6000-7500K, the maximum output current of the power supply of the light source is 1.2A, and the illuminance is fixed at 3000lx. The light source is connected to the light source controller through a circuit.

[0019] The sample container is a transparent cup with an outer dimension of Φ100×90mm. The sample container is used to hold the edible vegetable oil sample to be tested.

[0020] Furthermore, the 14 different types of metal oxide semiconductor gas sensors are model numbers TGS2619, WSP2110, MP905, MP503, MP801, MP702, MQ137, MP4, TGS2609, TGS2600, TGS2603, TGS2620, TGS2602 and MQ138.

[0021] The outer dimensions of the air chamber are 240mm×80mm×40mm, the inner dimensions of the air chamber are 170mm×30mm×40mm, and the wall thickness of the air chamber is 5mm.

[0022] Secondly, a non-destructive method for detecting the oxidative deterioration of edible vegetable oils is characterized by comprising the following steps:

[0023] Step S1. Turn on the power, connect the MVS software to the CMOS camera of the vision inspection device, and connect the DAM-3000M software to the DAM-3055N. Set the gas sampling pump flow rate to 600 mL / min, calibrate it with the LZB-3WB glass rotor flow meter, and wash and preheat the gas for at least 1 hour.

[0024] Step S2. Take 200 mL of the oxidized edible vegetable oil sample to be tested and place it in the headspace chamber sample container. Secure the lid and capture and save the image information of the sample in the MVS software.

[0025] Step S3. MVS performs sampling, and the white balance is fixed at 1682-1024-1526. After the image stabilizes, the sample is saved. The white balance value is the camera white balance result when no edible vegetable oil sample is added to this device.

[0026] Step S4. Wait 90 minutes, then simultaneously open the No. 1 solenoid valve switch and the No. 2 solenoid valve switch to change the gas path, and collect 240 seconds of odor data in the DAM-3000M software.

[0027] Step S5. DAM-3000M performs sampling. Select the COM port connected to the acquisition module, search for the acquisition module at the preset baud rate, click the search result, enable saving when starting sampling, and click Start Sampling;

[0028] Step S6. Simultaneously close the No. 1 solenoid valve switch and the No. 2 solenoid valve switch, start gas washing, remove the sample container from the headspace chamber, replace the sample to be tested, and perform parallel measurements 5 times for each edible vegetable oil sample.

[0029] Step S7. Perform feature-level fusion on the color and odor data collected by this device, and divide this data into a test set and a training set. Use machine learning methods to obtain a predictive model of oil oxidation for subsequent use.

[0030] Furthermore, in step S2, the images obtained by the MVS software each day are segmented using Matlab software to uniformly divide the images into small images of 25×25 pixels. The color information of each small image is read and preserved as normalized results of red (R), green (G), blue (B), hue (H), saturation (S), a (green to red), and b (blue to yellow).

[0031] In step S4, the daily clean air signal baseline is subtracted from the daily electrical signal change data recorded by 14 different types of metal oxide semiconductor gas sensors, and data for 100-104 seconds of each signal is extracted.

[0032] Further, in step S7, the color and gas data are fused into 21-dimensional feature data: R, G, B, H, S, a, b, TGS2619, WSP2110, MP905, MP503, MP801, MP702, MQ137, MP4, TGS2609, TGS2600, TGS2603, TGS2620, TGS2602, and MQ138.

[0033] In step S7, machine learning uses PLSR, MLP, and SVR as base learners and LR as a meta learner to stack the data.

[0034] Advantages of this invention:

[0035] This invention uses a camera and gas sensors to detect the oxidative deterioration of oils in edible vegetable oils. The entire process requires no chemical reagents and does not damage the edible vegetable oil. The prediction error for the peroxide value of edible vegetable oil does not exceed 0.01g / 100g, which is less than 4% of the maximum control limit for POV (0.25g / 100g), meeting the precision requirement of 10% in GB5009.227-2023 "National Food Safety Standard - Determination of Peroxide Value in Food". This invention is inexpensive and easy to produce, assemble, and replace. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the structure of the present invention;

[0037] Figure 2 This is a cross-sectional view of the visual inspection device in this invention;

[0038] Figure 3 This is a cross-sectional view of the olfactory detection device in this invention;

[0039] Figure 4 This is a schematic diagram of a six-pin metal-oxide-semiconductor gas sensor in the prior art;

[0040] Figure 5 This is a schematic diagram of a three-pin metal-oxide-semiconductor gas sensor in the prior art;

[0041] Figure 6 This is a schematic diagram of a four-pin metal-oxide-semiconductor gas sensor in the prior art;

[0042] Figure 7This is a trend graph showing the effect of headspace placement time on load voltage in this invention;

[0043] Figure 8 This is a trend graph showing the change of load voltage with sampling time in this invention;

[0044] Figure 9 This is a schematic diagram of the sampling workflow in this invention;

[0045] In the figure: 1. Visual inspection device, 101. CMOS camera, 102. Visual inspection air inlet, 103. Cover, 104. Visual inspection air outlet, 105. Buckle, 106. Lens, 107. Sample container, 108. Chamber, 109. Light source;

[0046] 2. Odor detection device, 201. Odor detection air inlet; 202. Odor detection air outlet; 203. Metal oxide semiconductor gas sensor;

[0047] 3. 12V DC power supply (No. 1);

[0048] 4. Solenoid valve switch No. 1;

[0049] 5. Two-way solenoid valve;

[0050] 6. Activated carbon filter No. 1;

[0051] 7. Light source controller;

[0052] 8. Computer;

[0053] 9. RS-485;

[0054] 10. DAM-3055N;

[0055] 11. No. 2 12V DC power supply;

[0056] 12. Three-way solenoid valve;

[0057] 13. Activated carbon filter No. 2;

[0058] 14. Gas sampling pump;

[0059] 15.5V DC power supply;

[0060] 16. Activated carbon filter No. 3;

[0061] 17. Solenoid valve switch No. 2. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention.

[0063] The detection objects of this invention are edible vegetable oil oxidized at room temperature and edible vegetable oil recovered after high-temperature frying. The former represents the storage of daily edible vegetable oil, and the latter represents the recycling and use of frying oil, basically covering the usage process of daily edible vegetable oil.

[0064] The design and selection of the detection device were determined based on the detection results of headspace solid phase microextraction-full two-dimensional gas chromatography-time-of-flight mass spectrometry (HS-SPME-GC×GC-TOFMS).

[0065] Nine samples of Grade A soybean oil, Grade A corn oil, and a commercially available Grade A rapeseed oil from a large grain and oil company, from opening to becoming inedible (peroxide value > 0.25 mg / kg), were used as testing subjects. A total of 165 identical substances were found during the oxidation process of these three types of edible vegetable oils. These included 21 hydrocarbon components; 22 alkene components; 23 amine and alcohol components; 26 aromatic and furan components; 27 aldehyde components; 26 ketone components; and 20 acid and ester components. The content of these components varies depending on the composition, oxidation path, and oxidation time of different edible vegetable oils, macroscopically causing changes in the color and odor of the edible vegetable oils. These complex color and odor changes contain information about the degree of oxidation of the oils in the edible vegetable oils.

[0066] Based on the results of HS-SPME-GC×GC-TOFMS measurements, a non-destructive testing device for oxidative deterioration of edible vegetable oils was designed, such as... Figure 1 As shown, it includes a visual detection device 1, an olfactory detection device 2, a 12V DC power supply 3, a solenoid valve switch 4, a two-way solenoid valve 5, an activated carbon filter 6, a light source controller 7, a computer 8, an RS-485 9, a DAM-3055N 10, a 12V DC power supply 11, a three-way solenoid valve 12, an activated carbon filter 13, a gas sampling pump 14, a 5V DC power supply 15, an activated carbon filter 16, and a solenoid valve switch 17.

[0067] like Figures 1-3As shown, a two-way solenoid valve 5 is provided on one side of the visual inspection device 1. The outlet of the two-way solenoid valve 5 is connected to the visual inspection inlet 102 of the visual inspection device 1 via a gas pipeline. The first activated carbon filter 6 is connected to the inlet of the two-way solenoid valve 5 via a gas pipeline. A three-way solenoid valve 12 is provided on the other side of the visual inspection device 1. The normally closed inlet of the three-way solenoid valve 12 is connected to the visual inspection outlet 104 of the visual inspection device 1 via a gas pipeline. The normally open inlet of the three-way solenoid valve 12 is connected to the third activated carbon filter 16 via a gas pipeline. The outlet of the three-way solenoid valve 12 is connected to the olfactory inspection inlet 201 of the olfactory inspection device 2 via a gas pipeline. The olfactory inspection outlet 202 of the olfactory inspection device 2 is connected to the second activated carbon filter 13 via a gas pipeline. The second activated carbon filter 13 is connected to the inlet of the gas sampling pump 14 via a gas pipeline. The outlet of the gas sampling pump 14 is provided with an exhaust gas filter.

[0068] The first 12V DC power supply 3 is connected to the first solenoid valve switch 4 and the visual inspection device 1 through a circuit. The first solenoid valve switch 4 is connected to the two-way solenoid valve 5 through a circuit. The light source controller 7 is connected to the visual inspection device 1 through a circuit. The second 12V DC power supply 11 is connected to the second solenoid valve switch 17 and the DAM-3055N10 through a circuit. The second solenoid valve switch 17 is connected to the three-way solenoid valve 12 through a circuit. The 5V DC power supply 15 is connected to the olfactory detection device 2 through a circuit.

[0069] The visual detection device 1 is connected to the computer 8 via a signal line. The computer 8 is connected to the RS-4859 via a signal line. The RS-4859 is connected to the DAM-3055N10 via a signal line. The DAM-3055N10 is connected to the olfactory detection device 2 via a signal line.

[0070] As a preferred embodiment of the present invention, such as Figure 2 As shown, the visual inspection device 1 includes a CMOS camera 101, a visual inspection air inlet 102, a cover 103, a visual inspection air outlet 104, a buckle 105, a lens 106, a sample container 107, a chamber 108, and a light source 109.

[0071] The cover 103 is located on the top of the visual inspection device 1, and the chamber 108 is located at the bottom of the cover 103. The chamber 108 has a headspace for placing the sample container 107. A light source 109 is embedded in the bottom of the chamber 108. The chamber 108 has a square structure, and adjustable buckles 105 are embedded on the four outer sides of the chamber 108.

[0072] The top of the cover 103 has a hole for embedding a CMOS camera 101. The CMOS camera 101 is embedded in the hole, and the lens 106 of the CMOS camera 101 extends into the headroom. The top two ends of the cover 103 are respectively provided with a visual detection air inlet 102 and a visual detection air outlet 104, which are symmetrically distributed on both sides of the CMOS camera 101. The sides of the cover 103 are respectively provided with four slots, which are tightly connected to the buckles 105. A sealing gasket is provided at the junction of the inner wall of the cover 103 and the chamber.

[0073] As a preferred embodiment of the present invention, such as Figure 3 As shown, the olfactory detection device 2 has an olfactory detection air inlet 201 and an olfactory detection air outlet 202 at its two ends, and an air chamber is provided inside the olfactory detection device 2. The air chamber contains 14 different types of metal oxide semiconductor gas sensors 203, which are distributed on both sides of the air chamber.

[0074] Compared to infrared, electrochemical, and other gas sensors, the metal oxide semiconductor gas sensor embedded in the gas chamber has higher sensitivity, longer lifespan, and similar response to a class of substances, making it more suitable for complex gases.

[0075] In a preferred embodiment of the present invention, the CMOS camera 101 is a color camera, the sensor model of the CMOS camera 101 is IMX183, and the resolution of the CMOS camera 101 is not less than 1920×1080. In this embodiment, the resolution of the CMOS camera 101 is 5472×3648. The lens 106 of the CMOS camera 101 is designed for low distortion and achromatic aberration. The aperture of the lens 106 is F2.4-F16, the distortion parameter is 0.39%, the image plane parameter is 1.2", and the resolution is 25 million pixels.

[0076] The CMOS camera 101 acquires signals using MVS software. The CMOS camera 101 is connected to the computer 8 via a GigE cable to enable communication between the two.

[0077] In a preferred embodiment of the present invention, the lid 103 and the compartment 108 are made of black light-blocking organic plastic, and the bottom surface of the compartment 108 is made of transparent organic plastic.

[0078] The headspace chamber has a diameter of 100mm and a height of 100mm. The chamber body 108 is a square structure of 140mm×140mm. The diameter of the hole for embedding the CMOS camera 101 is 27.8mm.

[0079] In a preferred embodiment of the present invention, the light source 109 is a surface light source, the light source is white, the color temperature of the light source is 6000-7500K, the power supply of the light source has a maximum output current of 1.2A and a fixed illuminance of 3000lx, and the light source 109 is connected to the light source controller 7 through a circuit.

[0080] The sample container 107 is a transparent cup with an outer dimension of Φ100×90mm. The sample container 107 is used to hold the edible vegetable oil sample to be tested.

[0081] In a preferred embodiment of the present invention, the 14 different types of metal oxide semiconductor gas sensors 203 are model TGS2619, WSP2110, MP905, MP503, MP801, MP702, MQ137, MP4, TGS2609, TGS2600, TGS2603, TGS2620, TGS2602 and MQ138.

[0082] The models, accuracies, and detection objects of the embedded metal-oxide-semiconductor sensors are shown in Table 1:

[0083] Table 1. Model and parameter information of metal oxide sensors

[0084]

[0085]

[0086] The signal acquisition of the metal oxide semiconductor sensor 203 is achieved by a signal acquisition circuit. The signal acquisition circuits of gas sensors, categorized by the number of pins, are as follows: Figures 4-6 As shown.

[0087] The DAM-3055N was used as the load voltage VRL acquisition unit in the test circuit. The DAM-3055N has 16 differential analog inputs with an acquisition accuracy of 1‰, and is connected to the DAM-3000M computer software via RS-485. The sensor and channel correspondence is as follows: 0-TGS2619, 1-WSP2110, 2-MP905, 3-MP503, 4-MP801, 5-MP702, 6-MQ137, 9-MP4, 10-TGS2609, 11-TGS2600, 12-TGS2603, 13-TGS2620, 14-TGS2602, 15-MQ138. The load voltage change can then be detected in the DAM-3000M software.

[0088] The outer dimensions of the air chamber are 240mm×80mm×40mm, the inner dimensions of the air chamber are 170mm×30mm×40mm, and the wall thickness of the air chamber is 5mm.

[0089] The gas flow within the visual detection device 1 and the olfactory detection device 2 has been verified to meet the requirements through Fluent fluid simulation.

[0090] The working process of the device of the present invention:

[0091] Air, driven by the gas sampling pump 14, passes through an air filter, enters the headspace chamber through the visual inspection device inlet 102, and flows out through the visual inspection device outlet 104, carrying the gas components in the sample headspace into the olfactory inspection device inlet 201. The gas is detected by 14 metal oxide semiconductor gas sensors 203 distributed within the chamber, and exits through the olfactory inspection outlet 202, passing through the activated carbon filter 13 to the inlet of the gas sampling pump 14. The gas flow path switch is controlled by two 12V DC power supplies via two corresponding solenoid valve switches, which control the two-way and three-way solenoid valves respectively. A computer acquires visual inspection signals via a GigE line and obtains olfactory inspection signals acquired by the DAM-3055N via RS-485.

[0092] Sample addition optimization: The sample container 107 has dimensions of Φ100×100mm, the lens 106 of the CMOS camera 101 has a length of 60mm, and the usable volume is 200mL. Edible corn oil of 50mL, 100mL, 150mL, and 200mL was added to the sample container 107, and images were acquired using MVS software. The results showed that the optimal sample addition volume was 200mL.

[0093] Headspace time optimization: The inner wall dimensions of the visual inspection device 1 are Φ100×100mm, the wall thickness of the sample container 107 is negligible, and the volume of the lens 106 is approximately 100cm³. 3 The actual usable volume of the headspace chamber is approximately 700 mL. After placing a 200 mL sample of edible corn oil in the headspace of the visual detection device 1 for a period of time, the difference between the load voltage value VRL of the 14 metal oxide semiconductor gas sensors 203 of the olfactory detection device 2 at the 30th second and VRL0 under pure air was recorded. The results are as follows: Figure 7 As shown, placing it in the headspace for 90 minutes is sufficient to meet the requirements of the olfactory detection system 2.

[0094] Sampling time optimization: After placing 200 mL of edible corn oil under headspace for 90 min, the load voltage was collected for 270 s using a DAM-3000M sampling pump at a flow rate of 600 mL / min. The results... Figure 8As shown, when solenoid valve switch 4 (number one) and solenoid valve switch 17 (number two) are pressed simultaneously, the data acquisition module begins recording data. The load voltage rises significantly within a short period, indicating that the gas collected in the headspace chamber is rapidly flowing into the gas chamber, triggering a response from the metal oxide semiconductor gas sensor. After reaching its maximum value, the load voltage continuously decreases as the pure air gradually dilutes the headspace gas, eventually reaching a stable state.

[0095] When the sampling time is within 0-50s, the load voltage rises sharply, resulting in the largest error in the sampling data. When the sampling time is greater than 150s, the load voltage is in a relatively stable state, at which point the headspace gas has completely flowed out of the gas chamber, and the change in the metal oxide semiconductor gas sensor is small, leading to a decrease in data resolution. When the sampling time is between 75-125s, the load voltage remains at a high response level, exhibiting strong resolution; and the voltage decreases slowly, resulting in a smaller error. Data from 100-104s is suitable for characterizing the odor information of edible vegetable oils in an oxidized state.

[0096] This invention also provides a non-destructive method for detecting the oxidative deterioration of edible vegetable oils, comprising the following steps:

[0097] Step S1. Turn on the power, connect the MVS software to the CMOS camera 101 of the vision inspection device 1, and connect the DAM-3000M software to the DAM-3055N10. The flow rate of the gas sampling pump 14 is 600 mL / min. It is calibrated with an LZB-3WB glass rotor flow meter. The gas is washed and preheated for at least 1 hour.

[0098] Step S2. Take 200 mL of the oxidized edible vegetable oil sample to be tested and place it in the headspace chamber sample container 107. Secure the lid 103 and capture and save the image information of the sample in the MVS software.

[0099] Step S3. MVS performs sampling, and the white balance is fixed at 1682-1024-1526. After the image stabilizes, the sample is saved. The white balance value is the camera white balance result when no edible vegetable oil sample is added to this device.

[0100] Step S4. Wait 90 minutes, then simultaneously open solenoid valve 4 (number 1) and solenoid valve 17 (number 2) to change the gas path, and collect 240 seconds of odor data in the DAM-3000M software.

[0101] Step S5. DAM-3000M performs sampling. Select the COM port connected to the acquisition module, search for the acquisition module at the preset baud rate, click the search result, enable saving when starting sampling, and click Start Sampling;

[0102] Step S6. Simultaneously close the first solenoid valve switch 4 and the second solenoid valve switch 17, start gas washing, remove the sample container 107 from the headspace chamber, replace the sample to be tested, and perform parallel measurements 5 times for each edible vegetable oil sample.

[0103] Step S7. Perform feature-level fusion on the color and odor data collected by this device, and divide this data into a test set and a training set. Use machine learning methods to obtain a predictive model of oil oxidation for subsequent use.

[0104] The sampling workflow diagram in step S5 is as follows: Figure 9 As shown.

[0105] In a preferred embodiment of the present invention, in step S2, the images obtained by the MVS software each day are segmented using Matlab software to uniformly divide the images into small images of 25×25 pixels. The color information of each small image is read and retained as normalized results of red (R), green (G), blue (B), hue (H), saturation (S), a (green to red), and b (blue to yellow).

[0106] In step S4, the electrical signal change data recorded daily by 14 different types of metal oxide semiconductor gas sensors 203 are subtracted from the daily pure air signal baseline, and data for each signal is extracted for 100-104 seconds.

[0107] In a preferred embodiment of the present invention, in step S7, the color and gas data are fused into 21-dimensional feature data: R, G, B, H, S, a, b, TGS2619, WSP2110, MP905, MP503, MP801, MP702, MQ137, MP4, TGS2609, TGS2600, TGS2603, TGS2620, TGS2602, and MQ138.

[0108] In step S7, machine learning uses PLSR, MLP, and SVR as base learners and LR as a meta learner to stack the data.

[0109] The performance parameters of this invention were verified using three accelerated oxidation edible vegetable oil samples (F brand grade 1 soybean oil, grade 1 corn oil, and D brand grade 1 rapeseed oil) at 60°C.

[0110] The accelerated oxidation method employed was the Schaal oven method, which involved placing the oil sample open in a constant-temperature incubator, protecting it from light and keeping it dry, and increasing the temperature to accelerate the reaction rate. The final experimental conditions for accelerated oil oxidation were determined to be: light protection, constant temperature of 60℃, single unblended oil, no added enzymes or catalysts, oxygen concentration equal to air concentration, and neglecting mass transfer effects. Under these conditions, the degradation of hydrogen peroxide (LOOH) and other branching reactions were minimized.

[0111] For the above three types of edible vegetable oils, the device of this invention was used to continuously measure them for twenty days from the date of opening until they became inedible (peroxide value > 0.25g / 100g). The detection accuracy and other performance characteristics finally achieved by this invention are shown in Table 2.

[0112] Table 2 Performance of the regression model of this invention

[0113]

[0114] The detection error for oxidative deterioration of edible vegetable oils using this invention is no more than 0.01 g / 100 g. This value is equivalent to 4% of the maximum control limit for POV (0.25 g / 100 g), which is far superior to the 10% precision requirement in GB 5009.227-2016 "National Food Safety Standard for Vegetable Oils," demonstrating the superior performance of this invention. The method of this invention can be applied to the detection of room temperature oxidative deterioration and frying oxidative deterioration of edible vegetable oils.

[0115] The detection method using a camera and gas sensor in this invention is based on intelligent sensory technology. By simulating human senses, it replaces the human visual and olfactory systems with a sensor system, mapping signals from the physical world to the digital world. When a sample of edible vegetable oil is placed in the detection device, its color and odor information are captured by the camera and gas sensor, resulting in a signal response containing color and odor characteristics—a "fingerprint" information. This "fingerprint" information contains information related to the degree of oxidation and deterioration of the oil.

[0116] Based on the aforementioned intelligent sensory technology, this invention's device acquires image information of an edible vegetable oil sample placed above a surface light source at the bottom of the headspace chamber using a CMOS camera at the top of the headspace chamber. Odor components from the edible vegetable oil sample diffuse into the headspace portion of the headspace chamber and, under the action of a gas sampling pump, enter the olfactory detection device. The odor information is acquired by a gas sensor and converted into a digital signal by a DAM-3055N. The edible vegetable oil image and odor information data acquired by the device are divided into test and training sets. Using PLSR, MLP, and SVR models as base learners and LR as a meta-learner, the data is stacked to obtain a predictive model for oil oxidation. The detection error for edible vegetable oil oxidation and deterioration is no more than 0.01 g / 100 g, equivalent to 4% of the maximum control limit for peroxide value (POV) (0.25 g / 100 g). This device has the advantage of being able to identify the oxidation and deterioration of edible vegetable oils by non-destructively detecting changes in color and odor. It solves the problems of complex sample pretreatment, long processing time, and high technical requirements for testing personnel in existing oil deterioration detection methods.

[0117] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art can still adjust the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Therefore, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A non-destructive testing device for oxidative deterioration of edible vegetable oils, characterized in that: Includes a visual detection device (1), an olfactory detection device (2), a No. 1 12V DC power supply (3), a No. 1 solenoid valve switch (4), a two-way solenoid valve (5), a No. 1 activated carbon filter (6), a light source controller (7), a computer (8), an RS-485 (9), a DAM-3055N (10), a No. 2 12V DC power supply (11), a three-way solenoid valve (12), a No. 2 activated carbon filter (13), a gas sampling pump (14), a 5V DC power supply (15), a No. 3 activated carbon filter (16), and a No. 2 solenoid valve switch (17); A two-way solenoid valve (5) is provided on one side of the visual inspection device (1). The outlet of the two-way solenoid valve (5) is connected to the visual inspection inlet (102) of the visual inspection device (1) through a gas pipe. The No. 1 activated carbon filter (6) is connected to the inlet of the two-way solenoid valve (5) through a gas pipe. A three-way solenoid valve (12) is provided on the other side of the visual inspection device (1). The normally closed inlet of the three-way solenoid valve (12) is connected to the visual inspection outlet (104) of the visual inspection device (1) through a gas pipe. The normally open inlet of the solenoid valve (12) is connected to the No. 3 activated carbon filter (16) through a gas pipeline. The outlet of the three-way solenoid valve (12) is connected to the olfactory detection inlet (201) of the olfactory detection device (2) through a gas pipeline. The olfactory detection outlet (202) of the olfactory detection device (2) is connected to the No. 2 activated carbon filter (13) through a gas pipeline. The No. 2 activated carbon filter (13) is connected to the inlet of the gas sampling pump (14) through a gas pipeline. The outlet of the gas sampling pump (14) is equipped with a tail gas filter. The first 12V DC power supply (3) is connected to the first solenoid valve switch (4) and the visual inspection device (1) through a circuit. The first solenoid valve switch (4) is connected to the two-way solenoid valve (5) through a circuit. The light source controller (7) is connected to the visual inspection device (1) through a circuit. The second 12V DC power supply (11) is connected to the second solenoid valve switch (17) and the DAM-3055N (10) through a circuit. The second solenoid valve switch (17) is connected to the three-way solenoid valve (12) through a circuit. The 5V DC power supply (15) is connected to the olfactory detection device (2) through a circuit. The visual detection device (1) is connected to the computer (8) via a signal line. The computer (8) is connected to the RS-485 (9) via a signal line. The RS-485 (9) is connected to the DAM-3055N (10) via a signal line. The DAM-3055N (10) is connected to the olfactory detection device (2) via a signal line.

2. The non-destructive testing device for oxidative deterioration of edible vegetable oils according to claim 1, characterized in that: The visual inspection device (1) includes a CMOS camera (101), a visual inspection air inlet (102), a cover (103), a visual inspection air outlet (104), a buckle (105), a lens (106), a sample container (107), a chamber (108), and a light source (109); The cover (103) is located on the top of the visual inspection device (1), and the chamber (108) is located at the bottom of the cover (103). The chamber (108) has a headspace for placing a sample container (107). A light source (109) is embedded at the bottom of the chamber (108). The chamber (108) has a square structure, and adjustable buckles (105) are embedded on the four sides of the outer perimeter of the chamber (108). The top of the cover (103) has a hole for embedding a CMOS camera (101). The CMOS camera (101) is embedded in the hole, and the lens (106) of the CMOS camera (101) extends into the headroom. The top two ends of the cover (103) are respectively provided with a visual inspection air inlet (102) and a visual inspection air outlet (104). The visual inspection air inlet (102) and the visual inspection air outlet (104) are symmetrically distributed on both sides of the CMOS camera (101). The sides of the cover (103) are respectively provided with four slots. The four slots are tightly connected with the buckle (105). A sealing gasket is provided at the junction of the inner wall of the cover (103) and the chamber.

3. The non-destructive testing device for oxidative deterioration of edible vegetable oils according to claim 2, characterized in that: The olfactory detection device (2) is provided with an olfactory detection air inlet (201) and an olfactory detection air outlet (202) at both ends. The olfactory detection device (2) is provided with an air chamber inside. The air chamber contains 14 different types of metal oxide semiconductor gas sensors (203), which are distributed on both sides of the air chamber.

4. The non-destructive testing device for oxidative deterioration of edible vegetable oils according to claim 3, characterized in that: The CMOS camera (101) is a color camera with a resolution of not less than 1920×1080. The lens (106) of the CMOS camera (101) has low distortion and achromatic design. The CMOS camera (101) is connected to the computer (8) via a GigE cable.

5. The non-destructive testing device for oxidative deterioration of edible vegetable oils according to claim 4, characterized in that: The lid (103) and the compartment (108) are made of black light-blocking organic plastic around their perimeter, and the bottom of the compartment (108) is made of transparent organic plastic. The headspace has a diameter of 100mm and a height of 100mm. The chamber (108) is a square structure of 140mm×140mm. The diameter of the hole that can embed the CMOS camera (101) is 27.8mm.

6. The non-destructive testing device for oxidative deterioration of edible vegetable oils according to claim 5, characterized in that: The light source (109) is a surface light source, the light source is white, the color temperature of the light source is 6000-7500K, the maximum output current of the power supply of the light source is 1.2A, and the illuminance is fixed at 3000lx. The light source (109) is connected to the light source controller (7) through a circuit. The sample container (107) is a transparent cup with an outer dimension of Φ100×90mm. The sample container (107) is used to hold the edible vegetable oil sample to be tested.

7. The non-destructive testing device for oxidative deterioration of edible vegetable oils according to claim 6, characterized in that: The 14 different types of metal oxide semiconductor gas sensors (203) are model numbers TGS2619, WSP2110, MP905, MP503, MP801, MP702, MQ137, MP4, TGS2609, TGS2600, TGS2603, TGS2620, TGS2602 and MQ138, respectively. The outer dimensions of the air chamber are 240mm×80mm×40mm, the inner dimensions of the air chamber are 170mm×30mm×40mm, and the wall thickness of the air chamber is 5mm.

8. A non-destructive method for detecting the oxidative deterioration of edible vegetable oils, characterized in that, Includes the following steps: Step S1. Turn on the power, connect the MVS software to the CMOS camera (101) of the vision inspection device (1), and connect the DAM-3000M software to the DAM-3055N (10). The gas sampling pump (14) has a flow rate of 600 mL / min. It is calibrated with an LZB-3WB glass rotor flow meter. The gas is washed and preheated for at least 1 hour. Step S2. Take 200 mL of the oxidized edible vegetable oil sample to be tested and place it in the headspace sample container (107). Secure the lid (103) and capture and save the image information of the sample in the MVS software. Step S3. MVS performs sampling, and the white balance is fixed at 1682-1024-1526. After the image stabilizes, the sample is saved. The white balance value is the camera white balance result when no edible vegetable oil sample is added to this device. Step S4. Wait 90 minutes, and at the same time open the first solenoid valve switch (4) and the second solenoid valve switch (17) to change the gas path. At the same time, collect 240 seconds of odor data in the DAM-3000M software. Step S5. DAM-3000M performs sampling. Select the COM port connected to the acquisition module, search for the acquisition module at the preset baud rate, click the search result, enable saving when starting sampling, and click Start Sampling; Step S6. Simultaneously close the first solenoid valve switch (4) and the second solenoid valve switch (17), start gas washing, remove the sample container (107) from the headspace chamber, replace the sample to be tested, and perform parallel measurements 5 times for each edible vegetable oil sample. Step S7. Perform feature-level fusion on the color and odor data collected by this device, and divide this data into a test set and a training set. Use machine learning methods to obtain a predictive model of oil oxidation for subsequent use.

9. The method as described in claim 8, characterized in that: In step S2, the images obtained by the MVS software each day are segmented using Matlab software. The images are evenly divided into small images of 25×25 pixels. The color information of each small image is read and preserved as normalized results of red (R), green (G), blue (B), hue (H), saturation (S), a (green to red), and b (blue to yellow). In step S4, the electrical signal change data recorded daily by 14 different types of metal oxide semiconductor gas sensors (203) are subtracted from the daily pure air signal baseline, and data for each signal is extracted for 100-104 seconds.

10. The method as described in claim 9, characterized in that: In step S7, the color and gas data are fused into 21-dimensional feature data: R, G, B, H, S, a, b, TGS2619, WSP2110, MP905, MP503, MP801, MP702, MQ137, MP4, TGS2609, TGS2600, TGS2603, TGS2620, TGS2602, and MQ138. In step S7, machine learning uses PLSR, MLP, and SVR as base learners and LR as a meta learner to stack the data.

Images