A portable spectrometer-based system for detecting adulteration of baijiu

Abstract

The application discloses a kind of liquor adulteration detection systems based on portable spectrometer, including spectral detection module, interference monitoring module, edge computing module, cloud processor and port display module, the present application is in real time by interference monitoring module Multidimensional environmental data, such as temperature, humidity, light intensity etc., interference factor regression model is constructed in combination with edge computing module, the influence of interference on spectral data is quantified, and is dynamically corrected by the interference compensation model of cloud processor, substantially reduce the interference of environmental noise on detection result, improve the detection robustness under complex working condition, by the logic architecture of "dynamic interference monitoring-edge computing-cloud correction-port display", the core pain points of traditional liquor detection, such as large environmental interference, low efficiency and high cost, have high precision, real-time and scalability, provide innovative solutions for food industry intelligent quality control.

Description

Technical Field

[0001] This invention relates to the field of food testing technology, and in particular to a system for detecting adulteration in baijiu (Chinese liquor) based on a portable spectrometer. Background Technology

[0002] Since the foundations of spectroscopy were laid in the 19th century, spectroscopic techniques have gradually evolved from large-scale laboratory equipment to miniaturization and intelligentization. Breakthroughs in near-infrared (NIR), Raman, and ultraviolet-visible (UV-Vis) spectroscopy have made rapid, non-destructive testing possible. With advancements in microelectronics, MEMS (microelectromechanical systems), and optics, traditionally bulky spectrometers have been miniaturized, such as mobile phone-sized micro-spectroscopy instruments (e.g., the Ocean Insight series). Their performance is close to that of laboratory equipment, while also featuring low power consumption and ease of operation.

[0003] Portable spectrometers are increasingly being used in the adulteration of baijiu (Chinese liquor). In production site monitoring, portable spectrometers are deployed near bottling lines to monitor the consistency between raw materials and finished products in real time. In market supervision and enforcement, law enforcement personnel can carry the equipment to conduct surprise inspections of commercially available baijiu and quickly screen for abnormal samples. For consumer applications, by developing external spectral modules for mobile phones (such as SCiO), consumers can scan bottle labels to verify product authenticity.

[0004] Adulteration of baijiu (Chinese liquor) is a hot topic of public concern, with strong-aroma baijiu being the most severely affected. However, existing identification methods based on instrumental analysis and human tasting suffer from high costs and low accuracy. Traditional baijiu detection methods mainly rely on laboratory chemical analysis, such as chromatography and mass spectrometry, which have limitations such as long detection cycles, high equipment costs, and dependence on professional personnel. Especially in complex production environments, environmental interference, such as temperature and humidity fluctuations, unstable light sources, and mechanical vibrations, can significantly affect the accuracy of spectral detection, leading to deviations in the predicted concentration of target compounds, such as adulterants.

[0005] Existing spectral analysis technologies mostly focus on single interference compensation or offline data processing, lacking dynamic interference monitoring and real-time correction capabilities, making it difficult to meet the rapid, high-precision, and interference-resistant quality control requirements of liquor production lines. To address these technical shortcomings, a solution is proposed. Summary of the Invention

[0006] The purpose of this invention is to solve the technical defects of existing spectral analysis technology, which lacks dynamic interference monitoring and real-time correction capabilities and is difficult to meet the requirements of rapid, high-precision, and interference-resistant quality control in liquor production lines.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A system for detecting adulteration in baijiu (Chinese liquor) based on a portable spectrometer includes a spectral detection module, an interference monitoring module, an edge computing module, a cloud processor, and a port display module, wherein the spectral detection module, the interference monitoring module, the edge computing module, the cloud processor, and the port display module are interconnected.

[0009] The spectral detection module is used to collect spectral data of baijiu (Chinese liquor). The spectral detection module collects spectral data of baijiu by setting up a spectrometer. The spectral data of baijiu includes electromagnetic wave wavelength and light signal intensity. The spectral data of baijiu is used to construct a spectral map for baijiu detection.

[0010] The interference monitoring module is used to collect spectral interference data: spectral interference data is collected through sensors, including temperature, humidity, light source intensity, vibration intensity, and noise intensity;

[0011] The edge computing module is used to build an interference factor regression model: by preprocessing the spectral data of liquor and the spectral interference data, the corresponding standardized reference values ​​are output; then, a multiple linear regression model is constructed, and the predicted concentration of the target compound is obtained through the detection spectrum of liquor; and an interference factor experimental control group is set up to obtain the concentration deviation of the target compound in the interference factor experiment, thereby constructing the influence function between the spectral interference data and the spectral data of liquor.

[0012] The cloud processor is used to build a spectral interference compensation model: by performing interference correction and standardization on the original spectral data of baijiu, and then substituting it into a multiple linear regression model, a baijiu adulteration warning signal is generated and output.

[0013] The port display module is used to receive the adulteration warning signal for baijiu and display the spectral analysis results of baijiu.

[0014] Furthermore, the standardization process for spectral data of baijiu is as follows:

[0015] Construct a spectral database of target compounds: label the differential compounds to be detected in baijiu, analyze the chemical characteristics of the target compounds, and collect spectral data of baijiu to verify the changes in absorption peaks and color change range;

[0016] The number of target compounds by category is labeled N0, any target compound is labeled A, and the predicted concentration coefficient of target compound A is labeled Cya;

[0017] Using electromagnetic wave wavelength as the abscissa and light signal intensity as the ordinate, a spectral map for baijiu detection is constructed using baijiu spectral data.

[0018] Baseline correction is used to standardize the spectral data of baijiu (Chinese liquor) and obtain standardized reference values ​​for the spectral data.

[0019] Furthermore, the specific process of constructing a multiple linear regression model is as follows:

[0020] The spectral image of the liquor was compared with the spectral database of the target compound to extract the characteristic peak a of the target compound A, as well as its peak intensity Ia, peak area Sa and half-peak width Wa.

[0021] The predicted concentration coefficient Cya of target compound A is obtained by using the peak intensity Ia, peak area Sa, and full width at half maximum (Wa) of characteristic peak a.

[0022] Furthermore, the specific process of the interference factor regression model is as follows:

[0023] By adjusting the parameter values ​​of the spectral interference data, an experimental control group for the interference factor is set up. For the same adulterated liquor sample solution, the corresponding spectral data of the liquor in the experimental control group for the interference factor are collected.

[0024] The number of the control group for the interference factor experiment is labeled as n0, any interference factor experiment is labeled as X, and the spectral data of the liquor in the interference factor experiment X is labeled as GPx.

[0025] The predicted concentration coefficients of the target compounds were obtained from the spectral data GPx of baijiu and integrated and labeled as experimental compound concentration data SYx; the original target compound concentration data of the adulterated baijiu sample solution were labeled as MBx.

[0026] Using the experimental compound concentration data SYx and the target compound concentration data MBx, the concentration deviation coefficients of m0 compounds were obtained and integrated and labeled as the interference deviation set Qe;

[0027] The experimental concentration of compound b is labeled C1, and the actual concentration of compound b is labeled C2. The experimental concentration C1 and the actual concentration C2 of compound b are compared to obtain the concentration deviation coefficient ΔXb of compound b in the interference factor experiment X.

[0028] Furthermore, the specific process for establishing the influence relationship between spectral interference data and baijiu spectral data is as follows:

[0029] The spectral interference data are integrated and labeled as an interference factor set Qd. The interference factor set Qd is then standardized and preprocessed to obtain standardized reference values ​​for the spectral interference data. The interference factors are temperature, humidity, light source intensity, vibration intensity, and noise intensity, respectively.

[0030] The influence function F(y) between the spectral interference data and the spectral data of baijiu was constructed using the interference deviation set Qe:

[0031] Where, ω 0,jLet ω be the baseline intensity of the j-th wavelength in the absence of interference. i,j Let ωj be the linear influence coefficient of the i-th interference factor on the j-th wavelength point, and ωj be the residual term;

[0032] According to the data acquisition cycle T, the spectral data and spectral interference data of baijiu are collected at regular intervals, E is the standardized interference factor matrix, and Zj is the spectral intensity vector of the j-th wavelength.

[0033] The linear influence coefficient ω is estimated using the least squares method. i,j To minimize the sum of squared residuals between the predicted and actual spectra, ωj = (E T E)- 1 E T Zj;

[0034] Therefore, we can deduce that the linear influence weight λ i,j : λ i,j The relative weight of the i-th interference factor with respect to the j-th wavelength, and λ i,j It belongs to the value range [0,1], and linearly influences the weight λ. i,j Evaluate the impact of each interference factor on the spectral signal.

[0035] Furthermore, the specific process of constructing the spectral interference compensation model is as follows:

[0036] Interference correction is performed on the original spectral signal intensity:

[0037] Among them, Zj ref Zj represents the corrected spectral signal intensity. mea The intensity of the spectral signal before correction, thus based on spectral interference data. To achieve interference compensation for spectral signal intensity;

[0038] The intensity of the corrected spectral signal is then standardized:

[0039] Among them, Zj std The standardized value of the corrected spectral signal intensity, max(Zj) ref ) and min(Zj ref () represent the maximum and minimum values ​​of the corrected spectrum, respectively;

[0040] The influence function F(y) is evaluated using the mean squared error of cross-validation (MSE).

[0041]

[0042] Among them, Zj( r ) represents the standardized value of the spectral interference compensation model. These are the standardized values ​​for the spectral data of baijiu (Chinese liquor).

[0043] The spectral signal intensity Zj after standardization std The analysis is performed and substituted into a multiple linear regression model. By using the peak intensity Ia, peak area Sa, and half-peak width Wa of characteristic peak a, the predicted concentration coefficient Cya of target compound A is obtained, thereby outputting a signal indicating adulteration of baijiu and the corresponding spectral analysis results of baijiu.

[0044] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0045] This invention collects spectral interference data of multidimensional environment in real time through interference monitoring module, combines interference factor regression model built by edge computing module to quantify the impact of interference on spectral data of liquor, and dynamically corrects it through interference compensation model of cloud processor, which greatly reduces the interference of environmental noise on detection results and improves detection robustness under complex working conditions.

[0046] This invention achieves localized data preprocessing and model calculation through an edge computing module, reducing cloud transmission latency. Combined with the rapid data acquisition capability of the spectral detection module, it meets the real-time monitoring needs of the production line and improves the detection efficiency of adulterated liquor.

[0047] This invention enables cloud-edge collaborative computing between a multiple linear regression model and an interference correction algorithm through a cloud processor, and provides data-driven decision support for quality control by visually indicating the risk of adulteration through a port display module.

[0048] In summary, this technical solution, through its logical architecture of "dynamic interference monitoring - edge computing - cloud correction - port display," addresses the core pain points of traditional liquor testing, such as high environmental interference, low efficiency, and high cost. It combines high precision, real-time performance, and scalability, providing an innovative solution for intelligent quality control in the food industry. Attached Figure Description

[0049] Figure 1 A connection diagram of the system modules of the present invention is shown;

[0050] Figure 2 A schematic diagram illustrating the steps of the present invention is shown. Detailed Implementation

[0051] 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 some embodiments of the present invention, and not all embodiments. 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.

[0052] Example 1:

[0053] like Figures 1-2 As shown, a system for detecting adulteration in baijiu (Chinese liquor) based on a portable spectrometer includes a spectral detection module, an interference monitoring module, an edge computing module, a cloud processor, and a port display module, wherein the spectral detection module, the interference monitoring module, the edge computing module, the cloud processor, and the port display module are interconnected.

[0054] The goal of this project is to design and screen metal ion-mediated color change sensing materials that can specifically recognize the main differential compounds in baijiu (Chinese liquor) through multivariate spectral analysis combined with quantum chemical calculations. Then, by combining various characterization instruments to explore the recognition mechanism and perform targeted optimization, a visual array sensor and model database will be obtained. This will enable rapid and visual identification of strong-aroma baijiu, providing a theoretical basis for quality monitoring and rapid authenticity detection of baijiu and other fermented foods. It will also provide regulatory authorities with a convenient and rapid detection method, which is of great significance for effectively preventing baijiu counterfeiting, avoiding the influx of inferior or counterfeit baijiu into the market, promoting the healthy development of food safety, especially the healthy and sustainable development of high-quality strong-aroma baijiu, and thus generating corresponding social and economic benefits.

[0055] The specific steps of this plan are as follows:

[0056] S1, the spectral detection module collects spectral data of baijiu: the spectral detection module collects spectral data of baijiu by using near-infrared, Raman, and ultraviolet-visible light sources and setting up a spectrometer. The spectral data of baijiu includes electromagnetic wave wavelength and light signal intensity. A spectral map of baijiu detection is constructed by using the spectral data of baijiu.

[0057] Construct a spectral database of target compounds: label the differential compounds to be detected in baijiu, such as methanol, ethyl acetate, acetaldehyde, pyrazines, etc.

[0058] The chemical properties of the target compound are analyzed based on the following chemical principles:

[0059] S1-1, First, determine the functional groups of the target compound, such as hydroxyl, aldehyde, ester, etc., and possible reaction sites, for example:

[0060] Methanol: Contains a hydroxyl group (-OH), which can form coordinate bonds with certain metal ions;

[0061] Ethyl acetate: ester group (-COOR), which may be detected indirectly through hydrolysis;

[0062] Pyrazines: nitrogen-containing heterocycles that can form stable complexes with transition metals;

[0063] S1-2, further screen for metal ions that can produce color characteristics, for example:

[0064] Cu2+: has a strong coordination ability with sulfur-, amino, or hydroxyl-containing compounds;

[0065] Fe3+: Reacts with phenolic and carboxylic acid compounds, which may trigger redox color changes, that is, when Fe3+ is reduced to Fe2+, the color changes from yellowish-brown to light green;

[0066] S1-3, then color change is induced by coordination with the target compound: that is, after the metal ion coordinates with the target compound, the electronic transition changes the absorption spectrum, such as Cu2+ coordinating with ethanolamine and changing from blue to purple.

[0067] S1-4, thereby selecting ligands for metal ions: such as porphyrins, crown ethers, EDTA derivatives and other organic ligands, to enhance the selectivity and stability of metal ions;

[0068] S1-5 uses near-infrared, Raman, and ultraviolet-visible spectral light sources and a spectrometer to collect spectral data of baijiu, thereby verifying the changes in absorption peaks and the range of color changes.

[0069] S2, Interference monitoring module collects spectral interference data: Spectral interference data is collected through sensors, including temperature, humidity, light source intensity, vibration intensity, and noise intensity.

[0070] S2-1, Temperature Data Acquisition:

[0071] Sensor selection: Use high-precision thermocouples or digital temperature sensors (such as DS18B20) with a measurement range of -20℃ to 80℃ and a resolution of ±0.1℃;

[0072] Install sensors near the light source module to monitor the effect of light source heating on spectral stability, and on the surface of optical components (such as gratings and lenses) to detect wavelength drift caused by thermal expansion.

[0073] S2-2, Humidity Data Acquisition:

[0074] Sensor selection: Capacitive humidity sensor (such as SHT30), measuring range 0-100% RH, accuracy ±2%;

[0075] Install sensors inside the spectrometer cavity and in the external space: detect the potential impact of humidity changes in a closed environment on optical devices (such as condensation), and monitor overall humidity fluctuations in a laboratory or production site;

[0076] S2-3, Light Source Intensity Monitoring

[0077] Sensor selection: Silicon photodiode (such as Thorlabs PDA36A), with a response band covering 200-1100nm and a dynamic range of 70dB;

[0078] A beam splitter is installed at the output of the light source to guide 5% of the light intensity to a photodiode, and the light intensity signal is recorded in real time. This signal is then converted into a digital quantity (unit: μW / cm²) by an ADC module. 2 );

[0079] S2-4, Vibration intensity test:

[0080] Sensor selection: Triaxial MEMS accelerometer (such as ADXL345), range ±16g, bandwidth 0-1.6kHz.

[0081] Install the sensor on the spectrometer base to detect external mechanical vibrations (such as equipment operation or personnel movement);

[0082] S2-5, Noise Intensity Monitoring:

[0083] Sensor selection: digital microphone (such as INMP441), frequency response 20Hz-20kHz, signal-to-noise ratio ≥65dB; place the microphone close to the spectrometer housing and detect the equipment's own noise (such as cooling fan, stepper motor).

[0084] S3, the edge computing module builds a regression model for interference factors:

[0085] S3-1 outputs corresponding standardized reference values ​​by preprocessing the spectral data and spectral interference data of baijiu.

[0086] The spectral interference data are integrated and labeled into an interference factor set Qd, and the interference factor set Qd is preprocessed by standardization to obtain the standardized reference value of the spectral interference data: in, σi represents the standardized interference factor, which follows a distribution with a mean of 0 and a standard deviation of 1; Yi represents the original interference factor set Qd; μi represents the mean of the i-th interference factor; and σi represents the standard deviation of the i-th interference factor.

[0087] The number of target compounds by category is labeled N0, any target compound is labeled A, and the predicted concentration coefficient of target compound A is labeled Cya;

[0088] Using electromagnetic wave wavelength as the abscissa and light signal intensity as the ordinate, a spectral map for baijiu detection is constructed using baijiu spectral data.

[0089] Baseline correction was used to standardize the spectral data of baijiu (Chinese liquor) to obtain standardized reference values. in, The signal is the standardized spectral signal, Zj is the intensity value of the original spectral signal at electromagnetic wavelength j, Baseline(Zj) is the baseline drift, and εj is Gaussian white noise used to prevent overfitting.

[0090] Spectral data often suffers baseline drift due to instrument or environmental interference. Baseline correction removes low-frequency noise, preserves useful high-frequency signals, and adds a small amount of white noise to enhance the robustness of the data model.

[0091] S3-2, then construct a multiple linear regression model, and obtain the predicted concentration of the target compound by detecting the spectrum of liquor;

[0092] The spectral image of the liquor was compared with the spectral database of the target compound to extract the characteristic peak a of the target compound A, as well as its peak intensity Ia, peak area Sa and half-peak width Wa.

[0093] The predicted concentration coefficient Cya of target compound A is obtained by using the peak intensity Ia, peak area Sa, and full width at half maximum (Wa) of characteristic peak a.

[0094] Cya=β0+β1*Ia+β2*Sa+β3*Wa+ε0;

[0095] Wherein, β0, β1, β2, and β3 are the regression coefficients of peak intensity Ia, peak area Sa, and half-peak width Wa, respectively, and β0, β1, β2, and β3 are all greater than 0. ε0 is the error coefficient, and ε0 is obtained by pre-setting after being measured by a large amount of experimental data.

[0096] It should be noted that the characteristic peak position refers to the position of the characteristic absorption or scattering peak in the spectrum of the target compound, such as esters, alcohols, and aldehydes; the signal intensity of the characteristic peak is the signal intensity of the characteristic peak, reflecting the concentration or content of the compound; the peak area refers to the area covered below the characteristic peak, used for quantitative analysis; and the half-maximum width at half-maximum (HWHM) is the width at half the intensity of the characteristic peak, reflecting the purity of the target compound.

[0097] S3-3, set up an experimental control group for the interference factor, obtain the concentration deviation of the target compound in the interference factor experiment, and thus construct the influence function between the spectral interference data and the spectral data of liquor.

[0098] By adjusting the parameter values ​​of the spectral interference data, an experimental control group for the interference factor is set up. For the same adulterated liquor sample solution, the corresponding spectral data of the liquor in the experimental control group for the interference factor are collected.

[0099] The number of the control group for the interference factor experiment is labeled as n0, any interference factor experiment is labeled as X, and the spectral data of the liquor in the interference factor experiment X is labeled as GPx.

[0100] The predicted concentration coefficients of the target compounds were obtained from the spectral data GPx of baijiu and integrated and labeled as experimental compound concentration data SYx; the original target compound concentration data of the adulterated baijiu sample solution were labeled as MBx.

[0101] Using the experimental compound concentration data SYx and the target compound concentration data MBx, the concentration deviation coefficients of m0 compounds were obtained and integrated and labeled as the interference deviation set Qe;

[0102] In this process, any compound is labeled as b. The actual compound concentration of compound b is extracted from the original target compound concentration data of the adulterated liquor sample solution. The experimental compound concentration of compound b is extracted from the experimental compound concentration data of the interference factor experimental control group.

[0103] The experimental concentration of compound b is labeled C1, and the actual concentration of compound b is labeled C2. The experimental concentration C1 and the actual concentration C2 of compound b are compared to obtain the concentration deviation coefficient ΔXb of compound b in the interference factor experiment X:

[0104] S3-4, construct the influence function F(y) between the spectral interference data and the spectral data of baijiu using the interference deviation set Qe:

[0105] Where, ω 0,j Let ω be the baseline intensity of the j-th wavelength in the absence of interference. i,j Let ωj be the linear influence coefficient of the i-th interference factor on the j-th wavelength point, and ωj be the residual term;

[0106] According to the data acquisition cycle T, the spectral data and spectral interference data of baijiu are collected at regular intervals, E is the standardized interference factor matrix, and Zj is the spectral intensity vector of the j-th wavelength.

[0107] The linear influence coefficient ω is estimated using the least squares method. i,j To minimize the sum of squared residuals between the predicted and actual spectra, ωj = (E T E)- 1 E T Zj;

[0108] Therefore, the linear influence weight λ is derived. i,j : λ i,j The relative weight of the i-th interference factor with respect to the j-th wavelength, and λ i,j It belongs to the range [0,1];

[0109] By linearly influencing the weight λ i,j The influence of each interference factor on the spectral signal is assessed, and the local importance of the interference factor is quantified.

[0110] S4, cloud processor builds spectral interference compensation model: by performing interference correction and standardization on the original baijiu spectral data, it is then substituted into the multiple linear regression model to output the baijiu spectral analysis results;

[0111] Interference correction is performed on the original spectral signal intensity:

[0112] Among them, Zj ref Zj represents the corrected spectral signal intensity. mea The intensity of the spectral signal before correction, thus based on spectral interference data. To achieve interference compensation for spectral signal intensity;

[0113] The intensity of the corrected spectral signal is then standardized:

[0114] Among them, Zj std The standardized value of the corrected spectral signal intensity, max(Zj) ref ) and min(Zj ref () represent the maximum and minimum values ​​of the corrected spectrum, respectively;

[0115] The influence function F(y) is evaluated using the mean squared error of cross-validation (MSE).

[0116]

[0117] Among them, Zj( r ) represents the standardized value of the spectral interference compensation model. To standardize the spectral data of baijiu, the two sets of datasets Zj collected above are used as the basis for the calculation. r )and Divide the model into k subsets at the same time point, where Nr is the number of samples in the r-th subset. Train the model using k-1 subsets in turn, and validate it using the remaining 1 subset. Calculate the mean squared error (MSE). The higher the MSE, the lower the reliability of the model. The closer the MSE is to 0, the higher the reliability of the model.

[0118] The spectral signal intensity Zj after standardization std The analysis is performed and substituted into a multiple linear regression model. The predicted concentration coefficient Cya of the target compound A is obtained by using the peak intensity Ia, peak area Sa and half-peak width Wa of the characteristic peak a. When the target compound is detected, a signal indicating adulteration of the liquor is output and the concentration of the target compound is marked. The concentration results of all detected compounds are integrated and marked as the spectral analysis results of the liquor.

[0119] By supporting model iteration and multi-terminal collaboration through cloud processors, it can be adapted to liquor production lines of different sizes and extended to adulteration detection scenarios for other liquid foods (such as rice wine and beverages).

[0120] S5, the port display module receives the adulteration warning signal for baijiu and displays the spectral analysis results of baijiu;

[0121] The spectral analysis results of baijiu include whether the target compound is detected and its concentration, and generate corresponding adulteration warning signals for baijiu, thus displaying the spectral analysis results of baijiu.

[0122] In summary, this invention collects multi-dimensional environmental data (temperature, humidity, light intensity, etc.) in real time through an interference monitoring module, and combines it with an interference factor regression model constructed by an edge computing module to quantify the impact of interference on spectral data. Furthermore, it dynamically corrects the interference through an interference compensation model in a cloud processor, thereby significantly reducing the interference of environmental noise on the detection results and improving the robustness of detection under complex working conditions.

[0123] Furthermore, the edge computing module enables localized data preprocessing and model calculation, reducing cloud transmission latency. Combined with the rapid data acquisition capability of the spectral detection module, it meets the real-time monitoring needs of the production line and improves the detection efficiency of adulterated liquor.

[0124] This invention enables cloud-edge collaborative computing between a multiple linear regression model and an interference correction algorithm through a cloud processor, and provides data-driven decision support for quality control by visually indicating the risk of adulteration through a port display module.

[0125] This invention solves the core pain points of traditional liquor testing, such as large environmental interference, low efficiency, and high cost, through a logical architecture of "dynamic interference monitoring - edge computing - cloud correction - port display". It combines high precision, real-time performance, and scalability, providing an innovative solution for intelligent quality control in the food industry.

[0126] The size of the interval and threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value.

[0127] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0128] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A system for detecting adulteration in baijiu (Chinese liquor) based on a portable spectrometer, characterized in that: It includes a spectral detection module, an interference monitoring module, an edge computing module, a cloud processor, and a port display module, wherein the spectral detection module, the interference monitoring module, the edge computing module, the cloud processor, and the port display module are interconnected. The spectral detection module is used to collect spectral data of baijiu (Chinese liquor). The spectral detection module collects spectral data of baijiu by setting up a spectrometer. The spectral data of baijiu includes electromagnetic wave wavelength and light signal intensity. The spectral data of baijiu is used to construct a spectral map for baijiu detection. The interference monitoring module is used to collect spectral interference data: spectral interference data is collected through sensors, including temperature, humidity, light source intensity, vibration intensity, and noise intensity; The edge computing module is used to build an interference factor regression model: by preprocessing the spectral data of liquor and the spectral interference data, the corresponding standardized reference values ​​are output; then, a multiple linear regression model is constructed, and the predicted concentration of the target compound is obtained through the detection spectrum of liquor; and an interference factor experimental control group is set up to obtain the concentration deviation of the target compound in the interference factor experiment, thereby constructing the influence function between the spectral interference data and the spectral data of liquor. The cloud processor is used to build a spectral interference compensation model: by performing interference correction and standardization on the original spectral data of baijiu, and then substituting it into a multiple linear regression model, a baijiu adulteration warning signal is generated and output. The port display module is used to receive the adulteration warning signal for baijiu and display the spectral analysis results of baijiu.

2. The system for detecting adulteration of liquor based on a portable spectrometer according to claim 1, characterized in that: The standardization process for spectral data of Baijiu is as follows: Construct a spectral database of target compounds: label the differential compounds to be detected in baijiu, analyze the chemical characteristics of the target compounds, and collect spectral data of baijiu to verify the changes in absorption peaks and color change range; The number of target compounds by category is labeled N0, any target compound is labeled A, and the predicted concentration coefficient of target compound A is labeled Cya; Using electromagnetic wave wavelength as the abscissa and light signal intensity as the ordinate, a spectral map for baijiu detection is constructed using baijiu spectral data. Baseline correction is used to standardize the spectral data of baijiu (Chinese liquor) and obtain standardized reference values ​​for the spectral data.

3. The system for detecting adulteration of liquor based on a portable spectrometer according to claim 2, characterized in that: The specific process of constructing a multiple linear regression model is as follows: The spectral image of the liquor was compared with the spectral database of the target compound to extract the characteristic peak a of the target compound A, as well as its peak intensity Ia, peak area Sa and half-peak width Wa. The predicted concentration coefficient Cya of target compound A is obtained by using the peak intensity Ia, peak area Sa, and full width at half maximum (Wa) of characteristic peak a.

4. The system for detecting adulteration of liquor based on a portable spectrometer according to claim 3, characterized in that: The specific process of the interference factor regression model is as follows: By adjusting the parameter values ​​of the spectral interference data, an experimental control group for the interference factor is set up. For the same adulterated liquor sample solution, the corresponding spectral data of the liquor in the experimental control group for the interference factor are collected. The number of the control group for the interference factor experiment is labeled as n0, any interference factor experiment is labeled as X, and the spectral data of the liquor in the interference factor experiment X is labeled as GPx. The predicted concentration coefficients of the target compounds were obtained from the spectral data GPx of baijiu (Chinese liquor) and integrated and labeled as the experimental compound concentration data SYx. The original target compound concentration data of the adulterated liquor sample solution are labeled as MBx; Using the experimental compound concentration data SYx and the target compound concentration data MBx, the concentration deviation coefficients of m0 compounds were obtained and integrated and labeled as the interference deviation set Qe; The experimental concentration of compound b is labeled C1, and the actual concentration of compound b is labeled C2. The experimental concentration C1 and the actual concentration C2 of compound b are compared to obtain the concentration deviation coefficient of compound b in the interference factor experiment X. .

5. The system for detecting adulteration of liquor based on a portable spectrometer according to claim 4, characterized in that: The specific process for establishing the relationship between spectral interference data and baijiu spectral data is as follows: The spectral interference data are integrated and labeled as an interference factor set Qd. The interference factor set Qd is then standardized and preprocessed to obtain standardized reference values ​​for the spectral interference data. The interference factors are temperature, humidity, light source intensity, vibration intensity, and noise intensity. An influence function F(y) between the spectral interference data and the baijiu (Chinese liquor) spectral data is constructed using the interference deviation set Qe. ;in, Let J be the baseline intensity of the j-th wavelength in the absence of interference. Let be the linear influence coefficient of the i-th interference factor on the j-th wavelength point. The residual term is defined as follows: Baijiu spectral data and spectral interference data are collected periodically according to the data acquisition cycle T; E is the standardized interference factor matrix; Zj is the spectral intensity vector at the j-th wavelength; the linear influence coefficient is estimated using the least squares method. To minimize the sum of squared residuals between the predicted and actual spectra, then Therefore, it can be deduced that the linear influence weights : ; The relative weight of the i-th interference factor with respect to the j-th wavelength is... Belonging to the value range [0,1], the weights are affected linearly. Evaluate the impact of each interference factor on the spectral signal.

6. The system for detecting adulteration of liquor based on a portable spectrometer according to claim 5, characterized in that: The specific process of constructing the spectral interference compensation model is as follows: Interference correction is performed on the original spectral signal intensity: ,in, The corrected spectral signal intensity, The intensity of the spectral signal before correction, thus based on spectral interference data. To compensate for interference in the spectral signal intensity; and to standardize the corrected spectral signal intensity: ,in, The standardized value of the spectral signal intensity after correction. and These are the maximum and minimum values ​​of the corrected spectrum, respectively; The influence function F(y) is evaluated using the mean squared error of cross-validation (MSE). ;in, These are the standardized values ​​for the spectral interference compensation model. The standardized values ​​of the spectral data of baijiu; the intensity of the spectral signal after standardization. The analysis is performed and substituted into a multiple linear regression model. By using the peak intensity Ia, peak area Sa, and half-peak width Wa of characteristic peak a, the predicted concentration coefficient Cya of target compound A is obtained, thereby outputting a signal indicating adulteration of baijiu and the corresponding spectral analysis results of baijiu.

Images