Soybean oil processing process online quality detection system and method based on near infrared spectrum technology
Through the online detection system of near-infrared spectroscopy, the artificial error and acid price detection lag in the decolorization process during soybean oil refining are solved, real-time monitoring of the color and acid price of soybean oil is achieved, and production efficiency and product quality are improved.
Patent Information
- Application Number
- CN202411241868.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-05
- Publication Date
- 2024-11-15
AI Technical Summary
During the soybean oil refining process, there are large artificial detection errors in the decolorization process and delayed detection of acid price stability, resulting in untimely quality monitoring during the production process, affecting the quality of the oil.
Using an online quality detection system based on near-infrared spectroscopy technology, by building an online monitoring terminal and a human-computer interactive terminal, a real-time detection model of the color and acid value of soybean oil is established, and data is collected using a spectrometer to calculate the Lab color value and acid value, so as to achieve Real-time inspection and monitoring on site.
It realizes rapid, stable and accurate online quality monitoring of the soybean oil refining process, reduces artificial errors, improves production efficiency and product quality, and meets the real-time control needs of food quality.
Smart Images

Figure CN118961588A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of food detection, and in particular relates to an online quality detection system and method for soybean oil processing based on near-infrared spectroscopy technology. Background Art
[0002] Soybean oil is an important edible oil product with rich nutrients and multiple uses. It is widely consumed worldwide. The soybean oil refining process includes four stages: degumming, deacidification, decolorization, and deodorization. Among them, the decolorization process is one of the important links, and its purpose is to remove the soluble pigments in the crude soybean oil to ensure the good color of the soybean oil and the stability of the acid value of the oil. In the daily soybean oil production process, the decolorization link is mostly for the staff to test the color of the oil, and there is a problem of large human detection errors. In addition, during the decolorization process, in order to stabilize the acid value of soybean oil, the staff needs to take samples for physical and chemical experiments, and there is a lag problem in processing adjustments. Therefore, internal and external quality monitoring of soybean oil during the production process is of great significance. Summary of the invention
[0003] In view of the problems existing in the existing production, the present invention provides an online quality detection system and method for soybean oil processing based on near-infrared spectroscopy technology, which can monitor the quality of the soybean oil refining process, effectively maintain quality and reduce losses, and save manpower, financial resources and material resources.
[0004] The technical solution of the present invention is as follows: the present invention provides an online quality detection system for soybean oil processing based on near-infrared spectroscopy technology, characterized in that an online monitoring terminal is constructed. The online monitoring terminal includes an optical path part and an oil path part. The optical path part includes a visible / near-infrared light source, a connected optical fiber, a monitoring chamber and a spectrometer; the optical path interfaces on both sides of the monitoring chamber are connected to the optical fiber and the optical fibers at both ends are respectively connected to the visible / near-infrared light source and the spectrometer; the data interface of the spectrometer is connected to the computer via USB; the oil path includes small pumps A, B, L-shaped oil pipes, T-type one-way valves, monitoring chambers, solenoid valves and overflow valves; the two ends of the L-shaped oil pipes are respectively connected to the two inlets of the T-type one-way valve and the outlets of small pumps A and B; the top inlet and the side interfaces of the monitoring chamber are respectively connected to the outlet of the T-type one-way valve and the inlet and outlet of the solenoid valve; the overflow valve is connected to the monitoring part pipeline of the monitoring chamber; at the intersection of the oil path and the optical path inside the monitoring chamber, the pipeline of the oil path uses a high-transmittance quartz glass pipeline. The inlet and outlet of the optical path of the monitoring chamber are on one axis and the inlet and outlet of the oil path are also on one axis, and the two axes are perpendicular in one plane.
[0005] The present invention provides an online quality detection method for soybean oil processing based on near infrared spectroscopy technology, comprising the following steps:
[0006] Step 1, establishing a mathematical calculation and correction model for soybean oil Lab color;
[0007] Step 2, establishing a discrimination and correction model for the Y and R color values of soybean oil;
[0008] Step 3, establishing a soybean oil acid value discrimination and correction model;
[0009] Step 4, establishing an on-site real-time detection system for soybean oil color and acid value.
[0010] Furthermore, the soybean oil color discrimination and correction model selects 81 data points in the visible light band from 380nm to 780nm, among which the data points in the 380nm band λ 380 The light power of I 380 is composed of its right adjacent band λ 380j Light intensity I 380i The left adjacent band λ 380i Light intensity I 380j Calculated by the following formula:
[0011]
[0012] Similarly, the light intensity calculation of other corresponding bands can also be done in this way, denoted as I(λ); the white correction value W(λ) and dark correction value D(λ) in the required band of transmittance are obtained by interpolation calculation using the above formula, so that the transmittance τ(λ) in a certain band can be calculated by the following formula, and so on to calculate all corresponding transmittances in these 81 bands.
[0013]
[0014] The soybean oil color Lab calculation model is a mathematical formula as follows:
[0015]
[0016] Where:
[0017] X, Y, Z—the three stimulus values of the CIE1931 standard colorimetric system;
[0018] X n , Y n , Z n (Can be checked) - tristimulus values XYZ of a fully reflecting diffuser;
[0019] L*—brightness;
[0020] a*—positive numbers represent red, negative numbers represent green;
[0021] b*—positive number represents yellow and negative number represents blue;
[0022] HereX n , Y n , Z n The value query shows that: X2 = 109.85, Y2 = 100, Z2 = 35.585. Among them, the stimulus value: X, Y, Z values are calculated according to the following formula:
[0023]
[0024] Where:
[0025] K—normalization coefficient,
[0026] S(λ) (available for inspection)—relative spectral power distribution of standard light source;
[0027] (available for inspection) - Chromaticity function in the XYZ colorimetric system;
[0028] τ(λ)—transmittance of the sample;
[0029] Δλ—the interval between 81 bands, here it is 5nm;
[0030] In this way, we can calculate the L*a*b* value in the CIE1931 standard colorimetric system (2° observer), and we can judge the color index of soybean oil based on the L*a*b* value.
[0031] The correction model is fitted and calibrated based on the parameters obtained from on-site detection and the detection results of the precision colorimeter. The general calibration formula is similar to the formula:
[0032] y=kx+b
[0033] Where:
[0034] x: value before correction
[0035] y: corrected value
[0036] k, b: correction coefficients
[0037] Furthermore, the discrimination and correction model of the soybean oil Y and R color values is based on the R and Y color systems of the Lovibond system, and six color prediction models for drying oil, bleached oil and finished oil are established.
[0038] Taking the Y value of drying oil as an example, it uses a large amount of soybean oil spectrum information collected and the Y value information detected by the Lovibond color system to perform machine learning quasi-regression, derive relevant fitting coefficients to build a detection model. The model expression is:
[0039] Y=a0+a1*H1+a2*H2+a3*H3+a4*H4+a5*H5+…+a n*H n
[0040] In the formula, Y is the test result, a0, a1…a n are model coefficients, H1, H2…H n The spectral variables are screened by the competitive adaptive reweighting algorithm (CARS); the model coefficients are obtained by partial least squares training (PLSR), and a variety of chemometric algorithms are used within the model to update and correct the model.
[0041] The modified model adopts two univariate fitting corrections and performs weighting according to the actual collected data, and the expression is:
[0042] Y=0.6*(0.2678*X+3.6394)+0.4*(4.2435*e (0.0393*X) )
[0043] Where:
[0044] X: value before correction
[0045] Y: Corrected value
[0046] Other discrimination and correction models are not described here.
[0047] Furthermore, the soybean oil acid value discrimination and correction model establishes three acid value prediction models and correction models for drying oil, bleached oil and finished oil. The three prediction and correction model construction methods are the same as the method of claim 4 and will not be repeated here.
[0048] Furthermore, the on-site real-time detection system for soybean oil color and acid value is developed in the Qt6.3 framework using C++ language. The software specifically realizes connection and disconnection with the spectrometer, collection and storage of spectral data, calculation of soybean oil color index based on spectral data, and prediction of soybean oil acid value by the model.
[0049] (1) Connecting and disconnecting the spectrometer: The software provides the function of connecting and disconnecting the XS11639 model Ruhai spectrometer, which is connected to the desktop software through the universal serial port protocol. It is convenient for daily equipment maintenance and other operations.
[0050] (2) Collection and storage of spectral data: For different environments, reference spectra (bright and dark spectra) can be collected for correction to ensure the robustness of model prediction. The collected spectral curve is displayed in real time on the graphic control, which can facilitate operators to find abnormalities. At the same time, the collected spectrum can also be saved to further expand the data and improve the data model.
[0051] (3) Calculate the soybean oil color index Lab based on the spectral data: Write a calculation program based on GB / T3979-2008 to calculate the soybean oil Lab chromaticity value and display it in real time on the interface, so that the operator can adjust the process parameters conveniently.
[0052] (4) Model prediction of soybean oil acid value and Y, R values: The collected spectrum is input into the model for calculation, and the results are displayed on the interface in real time, which greatly facilitates the quality control in soybean oil production
[0053] Finally, the calculation model, prediction model, data collection and oil circuit are all issued and received through the industrial all-in-one computer. The industrial all-in-one computer issues a command to start extracting soybean oil samples, the spectrometer sends the collected data, and the industrial all-in-one calculates the color value and acid value based on the spectral data, and finally displays it on the human-computer interaction terminal, realizing the quality monitoring of the soybean oil processing process.
[0054] Beneficial effects: The present invention can provide a simple, stable and fast online monitoring system for soybean oil refining process to monitor, detect and ensure the quality of soybean oil during the processing, which has important practical significance for satisfying consumers' requirements for food quality and maximizing the benefits of full utilization of raw materials of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] Figure 1 : External structure diagram of the online monitoring terminal of the present invention;
[0056] Figure 2 : Oil circuit structure diagram of the present invention;
[0057] Figure 3 : Schematic diagram of the intersection of the oil path and the light path of the present invention;
[0058] Figure 4 : Schematic diagram of online monitoring terminal;
[0059] In the figure: 1. spectrometer, 2. small pump A, 3. small pump B, 4. L-type oil pipe, 5. T-type one-way valve, 6. solenoid valve, 7. visible / near-infrared light source, 8. optical fiber, 9. monitoring room, 10. overflow valve. DETAILED DESCRIPTION
[0060] The present invention is described in detail below in conjunction with the accompanying drawings.
[0061] Implementation method 1:
[0062] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0063] This application proposes an online quality detection system for soybean oil processing based on near infrared spectroscopy technology, including: an online monitoring terminal and a human-computer interaction terminal. The details are as follows:
[0064] 1. Online monitoring terminal
[0065] This embodiment provides an online quality control method for soybean oil processing based on near infrared spectroscopy technology. Figures 1 to 4 , mainly including a spectrometer 1, a small pump A2, a small pump B3, an L-type oil pipe 4, a T-type one-way valve, a solenoid valve 6, a visible / near-infrared light source 7, an optical fiber 8, a monitoring chamber 9 and a relief valve 10.
[0066] Among them, the optical path part is implemented as a visible / near-infrared light source 7 connected to the monitoring room 9 through an optical fiber 8, and the visible / near-infrared light is injected into the intersection of the oil path and the optical path inside the monitoring room 9. The visible / near-infrared light passes through the high-transparency quartz oil pipe at the intersection and is received by the optical fiber on the other side and then transmitted to the connected spectrometer 1. After receiving the spectral data, the spectrometer transmits it to the industrial all-in-one computer via USB to calculate the color of soybean oil and predict the acid value, and finally displays it on the human-computer interaction terminal. In this embodiment, the model of the spectrometer 1 can be referenced as: Ruhai Spectrometer XS11639, the optical fiber model can be referenced as: Ruhai Optical Fiber UV-VIS, and the visible / near-infrared light source can be referenced as: HL2000-ADJ. In order to ensure that the spectral signal during monitoring is not interfered by the outside world, the surface of the intersection of the optical path and the oil path inside the monitoring room 9 is covered with tin and is opaque. When installing the optical fibers at both ends, it is necessary to ensure that they are on the same axis and perpendicular to the monitored oil pipe.
[0067] The oil circuit is implemented as a suction port of a small pump A2 connected to a soybean oil processing pipeline to extract soybean oil samples in the refining process. The discharge port of the small pump A2 is connected to a monitoring chamber 9 through one end of a T-type one-way valve 5, and the soybean oil sample is sent to the monitoring chamber 9 for monitoring while preventing the monitored soybean oil from flowing back. The suction port of the small pump B3 is connected to a cleaning liquid pool to extract cleaning liquid to wash away the soybean oil residue detected last time in the pipeline. The discharge port of the small pump B3 is connected to the monitoring chamber 9 through the other end of a T-type one-way valve 5, and the cleaning liquid is sent to a high-transparency quartz glass tube to clean the residual soybean oil. In addition, an overflow valve 10 is provided at the outlet of the soybean oil monitoring pipeline of the monitoring chamber 9 to prevent the high-transparency quartz glass pipeline at the monitoring pipeline from being damaged due to excessive pressure. A solenoid valve 6 is also provided at the outlet of the soybean oil monitoring pipeline of the monitoring chamber 9 to control the inlet and outlet of soybean oil to ensure the stability of the soybean oil state during detection. In this embodiment, the small pump A2 sucks the soybean oil sample into the detection chamber. The sucking process needs to last for a period of time to ensure that the new soybean oil washes away the soybean oil left over from the previous detection. The electronic valve 6 is switched to a closed state. If the pressure in the pipeline is too high, the pressure in the pipeline can be maintained at about 0.1Mpa by adjusting the overflow valve 10. After the detection is completed, the electronic valve 6 is switched to an open state to allow the soybean oil that has been detected to flow back to the pipeline. The entire pipeline is a one-way street, and there is no backflow to pollute the refined soybean oil.
[0068] In this embodiment, the USB interface for data transmission is a USB 3.0 interface, which ensures fast and stable data transmission.
[0069] 2. Human-computer interaction terminal
[0070] In this embodiment, the human-computer interaction terminal is a desktop software developed based on the Qt6.0 framework, which can support running on win10 and win11 systems.
[0071] In this embodiment, the wavelength of the calculated and predicted visible / near infrared spectrum is between 348.352 and 1068.44 nm, including the visible light band of 380 nm to 780 nm. Each single spectrum contains 2048 data points at different wavelengths, and the wavelength is not an integer. The Lab chromaticity value is calculated according to GB / T3979-2008. First, 81 data points need to be interpolated from the visible light band, and the data points and their band sizes are the same. In order to obtain higher accuracy, the wavelength interval is selected as 5 nm. For example, 81 data points, including those located in the 780 nm band (λ 380 )'s light intensity 380 is the right adjacent band (λ 380j ) of the light intensity I 780i The left adjacent band (λ 380i ) of the light intensity I 780j Calculated by the following formula (1.1):
[0072]
[0073] Similarly, the light intensity calculation of other corresponding bands can also be done in this way, denoted as I(λ). The white correction value (W(λ)) and dark correction value (D(λ)) in the required band of transmittance can be obtained by interpolation calculation using formula (1.1). In this way, the transmittance τ(λ) in a certain band can be calculated using the following formula (1.2), and all corresponding transmittances in these 81 bands can be calculated in the same way.
[0074]
[0075] In this embodiment, the value of L*a*b* in the CIELAB color space can be calculated according to the following formula (1.3):
[0076]
[0077] Where:
[0078] X, Y, Z—the three stimulus values of the CIE1931 standard colorimetric system;
[0079] X n , Y n , Z n (Can be checked) - tristimulus values XYZ of a fully reflecting diffuser;
[0080] HereX n , Y n , Z n The values of X2 = 109.85, Y2 = 100, Z2 = 35.585 can be found by querying the values of X2 = 109.85, Y2 = 100, Z2 = 35.585. Among them, the stimulus values: X, Y, Z are calculated according to the following formula (1.4):
[0081]
[0082] Where:
[0083] K—normalization coefficient,
[0084] S(λ) (available for inspection)—relative spectral power distribution of standard light source;
[0085] (available for inspection) - Chromaticity function in the XYZ colorimetric system;
[0086] τ(λ)—transmittance of the sample;
[0087] Δλ—the interval between 81 bands, here it is 5nm;
[0088] In this embodiment, we can calculate the L*a*b* value in the CIE1931 standard colorimetric system (2° observer), and can judge the color index of soybean oil according to the L*a*b* value.
[0089] In this embodiment, a variable regression model is constructed using a machine learning regression algorithm, and the expression is as follows:
[0090] Y=a0+a1*H1+a2*H2+a3*H3+a4*H4+a5*H5+…+a n *H n
[0091] Where:
[0092] Y is the test result, a0, a1…a n are model coefficients, H1, H2…H n It is the spectral variable that has been screened by the competitive adaptive reweighting algorithm (CARS);
[0093] In this embodiment, the constructed soybean oil color calculation model is embedded in the program in the method of program class, which is convenient for instantiating objects to call its methods. The model parameters of the constructed acid value prediction model are saved in a text document and placed in the program folder for easy calling and calculation.
[0094] Implementation method 2:
[0095] The present invention measures the change of the color value Lab of soybean oil, and uses the measurement result of a standard colorimeter as a reference to ensure the accuracy of the color detection of the detection system. First, the self-built soybean oil quality detection system and the ColorQuestXE laboratory-grade spectrophotometer are used to collect the L*a*b values of 36 soybean oil samples as shown in Table 1 below, and the relative deviation is less than 1%. Overall, the regression effect is good, indicating that the self-built soybean oil quality detection system and calculation model can quickly and accurately detect the color quality of soybean oil online.
[0096] Table 1 Soybean oil L*a*b value
[0097]
[0098]
[0099] Implementation method three:
[0100] In the R value detection, there is a certain deviation as shown in Table 2. However, the calculated results are somewhat different from those of the Lovibond colorimeter. However, through observation, it is found that except for some abnormal values, they are basically systematic errors. The detection accuracy can be further improved by fitting and correcting the following formula, thereby improving production efficiency and product quality.
[0101] Y=0.6*(0.2678*X+3.6394)+0.4*(4.2435*e (0.0393*X) )
[0102] Table 2 R value measurement
[0103]
[0104]
[0105] In short, the rapid non-destructive color testing system is relatively stable and the detection accuracy is within an appropriate range to basically meet production needs.
[0106] In the acid value detection, the detection results of the present invention are shown in Table 3 below, and it can be seen that the detection accuracy can be almost accurate to two decimal places. The mean square error (MSE) between the predicted value and the actual value is 0.0001, the root mean square error (RMSE) is 0.0121, the determination coefficient (R-square) is 0.9571, and the mean absolute percentage error (MAPE) is 5.6%. In summary, MSE and RMSE are almost close to 0, and R-square is close to 1, indicating that the model has a high degree of fit and performs well in prediction.
[0107] Table 3 Detection of the acid value of soybean oil on the production line
[0108]
[0109]
[0110] The above embodiments are only used to illustrate the design ideas and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications made based on the principles and design ideas disclosed by the present invention are within the protection scope of the present invention.
Claims
1. An online quality detection system for soybean oil processing based on near infrared spectroscopy technology, characterized in that: An online monitoring terminal is constructed, the online monitoring terminal comprises an optical path part and an oil path part, the optical path part comprises a visible / near infrared light source (7), a connected optical fiber (8), a monitoring chamber (9) and a spectrometer (1); the optical path interfaces on both sides of the monitoring chamber (9) are connected to the optical fiber (8), and the optical fibers (8) at both ends are respectively connected to the visible / near infrared light source (7) and the spectrometer (1); the data interface of the spectrometer (1) is connected to a computer via a USB; the oil path part comprises a small pump A (2), a small pump B (3), an L-shaped oil pipe (4), a T-shaped one-way valve (5), a monitoring chamber (9), a solenoid valve (6) and a relief valve (10); the L-shaped oil pipe (4) is two, and the two One end of the L-shaped oil pipe (4) is connected to the two inlets of the T-shaped one-way valve (5), and the other ends of the two L-shaped oil pipes (4) are respectively connected to the outlets of the small pump A (2) and the small pump B (3); the top inlet of the monitoring chamber (9) is connected to the outlet of the T-shaped one-way valve (5), and the interface on the side of the monitoring chamber (9) is connected to the inlet and outlet of the electromagnetic valve (6); the overflow valve (10) is connected to the monitoring part pipeline of the monitoring chamber (9); at the intersection of the oil circuit and the optical path inside the monitoring chamber (9), the pipeline of the oil circuit is a high-transmittance quartz glass pipeline, the inlet and outlet of the optical path of the monitoring chamber (9) are on the same axis and the inlet and outlet of the oil circuit are also on the same axis, and the two axes are perpendicular in a plane.
2. A method for online quality detection of soybean oil processing based on near infrared spectroscopy technology, characterized in that: The following steps are involved: Step 1, establishing a mathematical calculation and correction model for soybean oil Lab color; Step 2, establishing a discrimination and correction model for the Y and R color values of soybean oil; Step 3, establishing a soybean oil acid value discrimination and correction model; Step 4, establishing an on-site real-time detection system for soybean oil color and acid value.
3. The method for online quality detection of soybean oil processing based on near infrared spectroscopy technology according to claim 2 is characterized in that: The mathematical calculation and correction model of soybean oil Lab color is based on the selection of 81 data points in the visible light band from 380nm to 780nm, among which the λ 380 The light power of I 380 is composed of its right adjacent band λ 380j Light intensity I 380i Left adjacent band λ 380i Light intensity I 380j Calculated by the following formula: Similarly, the light intensity calculation of other corresponding bands can also be done in this way, recorded as I(λ); the white correction value W(λ) and dark correction value D(λ) in the required band of transmittance can be obtained by interpolation calculation using the above formula, so that the transmittance in a certain band can be calculated by the following formula τ (λ), and so on to calculate all the corresponding transmittances in these 81 bands. The soybean oil color Lab calculation model is a mathematical formula as follows: Where: X, Y, Z—the three stimulus values of the CIE1931 standard colorimetric system; X n , Y n , Z n —The tristimulus values XYZ of a fully reflecting diffuser; L*—brightness; a*—positive numbers represent red, negative numbers represent green; b*—positive numbers represent yellow and negative numbers represent blue.
4. The method for online quality detection of soybean oil processing based on near infrared spectroscopy technology according to claim 3 is characterized in that: The X n , Y n , Z n The value query shows: X2=109.85, Y2=100, Z2=35.585, where the stimulus values: X, Y, and Z are calculated using the following formula: Where: K—normalization coefficient, S(λ)—Relative spectral power distribution of standard light source; — Chromaticity function in the XYZ colorimetry system; τ (λ)—transmittance of the sample; Δλ—the interval between 81 bands, here it is 5nm; In this way, the L*a*b* value in the CIE1931 standard colorimetric system can be calculated, and then the color index of soybean oil can be judged according to the L*a*b* value; The correction model is fitted and calibrated based on the parameters obtained from the on-site detection and the detection results of the precision colorimeter. The general calibration formula is similar to the formula: y=kx+b Where: x: value before correction; y: corrected value; k, b: correction coefficients.
5. The method for online quality detection of soybean oil processing based on near infrared spectroscopy technology according to claim 2, characterized in that: The soybean oil Y, R color value discrimination and correction model is based on the R and Y color system of the Lovibond system, and six color prediction models for drying oil, bleached oil and finished oil are established.
6. The method for online quality detection of soybean oil processing based on near infrared spectroscopy technology according to claim 5, characterized in that: The specific process is: using the large amount of soybean oil spectrum information collected and the Y value information detected by the Lovibond color system to perform machine learning quasi-regression, deriving the relevant fitting coefficients to build a detection model. The model expression is: <h2 style=";text-align:left;direction:ltr">Y = a0 + a1 * H1 + a2 * H2 + a3 * H3 + a4 * H4 + a5 * H5 +… + a<h2 style=";text-align:left;direction:ltr"> n <h2 style=";text-align:left;direction:ltr"> *H<h2 style=";text-align:left;direction:ltr"> n In the formula, Y is the test result, a0, a1…a n are model coefficients, H1, H2…H n It is a spectral variable that has been screened by the competitive adaptive reweighting algorithm CARS. The model coefficients are obtained by partial least squares PLSR training, and a variety of chemometric algorithms are used within the model to update and correct the model. The modified model adopts two univariate fitting corrections and performs weighting according to the actual collected data, and the expression is: Y=0.6*(0.2678*X+3.6394)+0.4*(4.2435*e (0.0393*X) ) Where: X: value before correction; Y: Corrected value.
7. The method for online quality detection of soybean oil processing based on near infrared spectroscopy technology according to claim 2, characterized in that: The soybean oil acid value discrimination and correction model establishes three acid value prediction models and correction models for drying oil, bleached oil and finished oil. The construction methods of the three prediction and correction models are the same as the method of the soybean oil Y, R color value discrimination and correction model.
8. The method for online quality detection of soybean oil processing based on near infrared spectroscopy technology according to claim 2, characterized in that: The on-site real-time detection system for soybean oil color and acid value is established as follows: the detection system is secondary developed in the Qt6.3 framework using C++ language. The software specifically realizes the connection and disconnection with the spectrometer, the collection and storage of spectral data, the calculation of soybean oil color index based on spectral data and the prediction of soybean oil acid value by the model, specifically: Connecting and disconnecting the spectrometer: The software provides the function of connecting and disconnecting the XS11639 model Ruhai spectrometer, and connects with the desktop software through the universal serial port protocol, which is convenient for daily equipment maintenance and other operations; Collection and preservation of spectral data: Reference spectra are collected for correction in different environments to ensure the robustness of model prediction. The collected spectral curves are displayed in real time on the graphic control, which is convenient for operators to find abnormalities. The collected spectra are also saved to further expand the data to improve the data model. Calculate the soybean oil color index Lab based on spectral data: Write a calculation program based on GB / T3979-2008 to calculate the soybean oil Lab chromaticity value and display it in real time on the interface, making it convenient for operators to adjust process parameters; The model predicts the acid value and Y and R values of soybean oil: the collected spectrum is input into the model for calculation, and the results are displayed on the interface in real time, which greatly facilitates the quality control in soybean oil production.
9. The method for online quality detection of soybean oil processing based on near infrared spectroscopy technology according to claim 2, characterized in that: In the steps 1-4, the industrial all-in-one computer issues instructions and receives data; the industrial all-in-one computer issues instructions to start extracting soybean oil samples, the spectrometer (1) sends the collected data, the industrial all-in-one computer calculates the color value and acid value based on the spectral data, and finally displays them on the human-computer interaction terminal, thereby realizing quality monitoring during the soybean oil processing process.
