A food quality detection method and system based on multi-source data

Abstract

The application belongs to the field of food inspection, and specifically relates to a food quality detection method and system based on multi-source data, aiming to solve the problem of insufficient fluid food quality detection precision. The application comprises: acquiring multi-source data on a food production line in real time and dividing data segments; obtaining data fingerprints including thermal accumulation fingerprints, mechanical shear fingerprints and biochemical reaction fingerprints on each data segment according to the multi-source data; mapping the data fingerprints into multiple risk indicators according to categories to form a risk vector, and correcting the risk vector by using a preset coupling coefficient matrix to obtain a coupling risk index; comparing the coupling risk index with a preset safety threshold to determine the quality risk grade of a food batch corresponding to the data segment. The application can instantaneously feed back the production quality of the food production line, fully considers the multi-physical field coupling influence, and improves the real-time performance and accuracy of the food detection process.

Description

Technical Field

[0001] This application belongs to the field of food inspection, and specifically relates to a food quality testing method and system based on multi-source data. Background Technology

[0002] In the continuous production of liquid foods such as dairy products and beer, key quality indicators of the final product, such as the protein and fat content of dairy products and the original wort concentration and flavor compounds of beer, rely on offline sampling and testing in laboratories. The results are severely delayed, making real-time intervention impossible. By the time a non-compliance is detected, the entire batch has already been produced, potentially resulting in significant material and production time losses. These issues make quality monitoring in the food production process neither timely nor effective.

[0003] Most existing online detection technologies are based on empirical threshold judgments of a single physical parameter, without considering the coupling effect of multiple physical fields and the physicochemical changes of food components during processing. This results in a high false judgment rate and poor adaptability, especially in key processes such as high-temperature sterilization, high-pressure homogenization and fermentation, which further limits their application effect and accuracy in actual production. Summary of the Invention

[0004] To address the aforementioned problems in existing technologies, namely insufficient accuracy and poor real-time performance in fluid food quality detection, this application proposes a food quality detection method based on multi-source data, applied to the fluid food production process, comprising:

[0005] Real-time acquisition of multi-source data from the food production line, including at least temperature, pressure, flow rate, and pH. Based on the changing trends of multi-source data, the start and end boundaries of the process section are identified and data segments are divided. Data fingerprints, including thermal accumulation fingerprints, mechanical shear fingerprints, and biochemical reaction fingerprints, are obtained from multi-source data for each data segment. Among them, thermal accumulation fingerprints are obtained by time integration of temperature rate function and piecewise temperature zone weight function; mechanical shear fingerprints are obtained by time integration of instantaneous power, which is obtained based on pressure and flow rate; and biochemical reaction fingerprints are obtained based on biochemical reaction rates determined by pH and temperature. Data fingerprints are mapped to multiple risk indicators according to their categories to form a risk vector. The risk vector, after being corrected by a preset coupling coefficient matrix, is then mapped to a scalar to obtain the coupling risk index. The coupling risk index is compared with a preset safety threshold to determine the quality risk level of the food batch corresponding to the data segment.

[0006] As a preferred implementation, after obtaining the coupling risk index, the method further includes: Based on the changing trend of the coupling risk index within a historical time window, a trend risk factor is calculated, and the coupling risk index is corrected using the trend risk factor; and / or, Based on historical differences from multi-source data and risk indicators, an uncertainty correction factor is calculated, and the coupled risk index is corrected using the uncertainty correction factor.

[0007] As a preferred embodiment, the process of acquiring thermally accumulated fingerprints includes: A temperature-rate function is constructed based on the exponential effect of temperature on the reaction rate. A piecewise temperature zone weighting function is constructed by assigning weighting functions to multiple continuous temperature ranges respectively. The thermal accumulation fingerprint is obtained by integrating the product of the temperature rate function and the piecewise temperature zone weighting function over the time domain of the data segment.

[0008] As a preferred embodiment, the process of obtaining fingerprints by mechanical cutting also includes: The energy conversion coefficient is determined based on fluid characteristic parameters and equipment structural parameters, wherein the fluid characteristic parameters include at least fluid viscosity and the equipment structural parameters include at least equipment clearance. After calculating the product by using the energy conversion coefficient as the coefficient of instantaneous power, the product is integrated over the time domain of the data segment to obtain the mechanical shear fingerprint.

[0009] As a preferred embodiment, the process of obtaining biochemical fingerprints includes: A temperature effect function is constructed based on the deviation between the ambient temperature and the optimum temperature. A pH influence function is constructed based on the deviation between the environmental pH and the optimal pH. A biochemical reaction rate function is constructed by combining the temperature effect function and the acid-base effect function. The biochemical reaction rate function is integrated over the time domain of the data segment to obtain the biochemical reaction fingerprint.

[0010] As a preferred implementation, identifying the start and end boundaries of the process segment and dividing the data into segments includes: When the flow rate exceeds the first flow rate threshold and the duration is greater than the first minimum process time, the process segment is determined to start. When the flow rate drops below the second flow rate threshold and the duration exceeds the second minimum process time, the process segment is considered complete; or... When the pressure change rate exceeds the pressure change rate threshold and the duration is greater than the second minimum process time, the process segment is determined to end.

[0011] As a preferred implementation, the coupling coefficient matrix is ​​an n×n matrix, where n is the number of risk indicators, the diagonal elements of the matrix are preset values, and the off-diagonal elements are preset coupling coefficients between various risk indicators. The coupling coefficients are obtained through offline experiments or historical data calibration.

[0012] As a preferred implementation method, multi-source data from the food production line is acquired in real time, including: Temperature, pressure, flow rate, and pH data are collected by sensors installed in the preset process section. The collected data is sent to the computing node via a communication protocol, and a timestamp based on the same time base is attached to each data point before sending. The timestamp is used to establish a correspondence between the data segment and the food batch by combining the batch traceability identifier.

[0013] As a preferred implementation, the preset safety threshold is dynamically and adaptively adjusted based on the process stability index of the food production line; the process stability index is calculated based on the degree of fluctuation of the coupled risk index within a historical time window.

[0014] On the other hand, this application proposes a food quality detection system based on multi-source data, applied to the fluid food production process, for executing the aforementioned food quality detection method based on multi-source data, including: The data acquisition module is used to acquire multi-source data from the food production line in real time. The multi-source data includes at least temperature, pressure, flow rate, and pH. The boundary recognition module is used to identify the start and end boundaries of the process section and divide the data into segments based on the changing trends of multi-source data. The fingerprint calculation module is used to obtain data fingerprints, including thermal accumulation fingerprints, mechanical shear fingerprints, and biochemical reaction fingerprints, from multiple sources on each data segment. The thermal accumulation fingerprint is obtained by time integration using a temperature rate function and a piecewise temperature zone weighting function; the mechanical shear fingerprint is obtained by time integration of instantaneous power, which is obtained based on pressure and flow rate; and the biochemical reaction fingerprint is obtained based on the biochemical reaction rate determined by pH and temperature. The coupling risk index acquisition module is used to map data fingerprints into multiple risk indicators according to categories, form a risk vector, and map the risk vector after correction using a preset coupling coefficient matrix into a scalar to obtain the coupling risk index. The risk assessment module is used to compare the coupled risk index with a preset safety threshold to determine the quality risk level of the food batch corresponding to the data segment.

[0015] Compared with the prior art, the technical solution provided in this application has at least one of the following beneficial effects: (1) Real-time data collection of temperature, pressure, flow rate, and pH levels via online sensors, and fingerprint calculation, risk projection, and level determination of the collected data enable the output of risk levels during production and real-time response to detection results. By identifying process segment boundaries and dividing data segments, the production quality status of each batch or process segment can be fed back instantly, allowing operators or control systems to take intervention measures at the early stage of quality deviation, avoiding the scrapping of the entire batch of products, and realizing quality control upgrade from post-detection to in-process interception.

[0016] (2) By constructing three independent yet complementary physicochemical fingerprints—thermal accumulation fingerprint, mechanical shear fingerprint, and biochemical reaction fingerprint—the thermal accumulation effect, fluid shear damage, and biochemical reaction process are quantified respectively. The risk indicators of each physical field are nonlinearly aggregated using the coupling coefficient matrix, which fully considers the synergistic or antagonistic effects between temperature, pressure, and pH. Compared with single-parameter methods, this greatly improves the detection accuracy.

[0017] (3) This application solves the technical problems of traditional food quality testing being lagging, singular, inaccurate, and poorly adaptable. It has made breakthrough progress in real-time performance, accuracy, and reliability, and provides a practical and feasible quality testing technology path for the intelligent manufacturing of fluid foods. Attached Figure Description

[0018] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart of a food quality testing method based on multi-source data provided in one embodiment of this application; Figure 2 This is a system block diagram of a food quality testing system based on multi-source data, provided in one embodiment of this application; Figure 3 This is a schematic diagram of the structure of a computer system used to implement the methods, systems, and electronic devices of this application; The attached figures are labeled as follows: 100. Data Acquisition Module; 200. Boundary Recognition Module; 300. Fingerprint Calculation Module; 400. Coupling Risk Index Acquisition Module; 500. Risk Assessment Module; 301. Central Processing Unit; 302. Read-Only Memory; 303. Random Access Memory; 304. Bus; 305. Input / Output Interface; 306. Input Section; 307. Output Section; 308. Storage Section; 309. Communication Section; 310. Driver; 311. Removable Media. Detailed Implementation

[0019] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0020] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] This application provides a food quality detection method based on multi-source data. It acquires multi-source data from the food production line in real time, including at least temperature, pressure, flow rate, and pH. Based on the changing trends of the multi-source data, it identifies the start and end boundaries of process sections and divides the data into segments. For each data segment, it acquires data fingerprints including thermal accumulation fingerprints, mechanical shear fingerprints, and biochemical reaction fingerprints based on the multi-source data. The thermal accumulation fingerprint is obtained by time integration using a temperature rate function and a piecewise temperature zone weighting function; the mechanical shear fingerprint is obtained by time integration of instantaneous power, which is based on pressure and flow rate; and the biochemical reaction fingerprint is obtained based on the biochemical reaction rate determined by pH and temperature. The data fingerprints are mapped to multiple risk indicators according to their categories, forming a risk vector. The risk vector, corrected using a preset coupling coefficient matrix, is then mapped to a scalar to obtain a coupling risk index. The coupling risk index is compared with a preset safety threshold to determine the quality risk level of the food batch corresponding to the data segment. This application can provide real-time feedback on the production quality of the food production line and fully considers the effects of multi-physics coupling, improving the real-time performance and accuracy of the food detection process.

[0022] To more clearly explain the food quality testing method based on multi-source data proposed in this application, the following will be combined with... Figure 1 The steps in the embodiments of this application are described in detail.

[0023] The first embodiment of this application discloses a food quality detection method based on multi-source data, applicable to the production process of fluid foods, such as dairy products and beer, including steps S10-S50, each step described in detail below: Step S10: Acquire multi-source data from the food production line in real time. The multi-source data includes at least temperature, pressure, flow rate, and pH.

[0024] Optionally, sensors can be deployed in key process sections of the food production line to acquire various physical and / or chemical quantities to form multi-source data.

[0025] In this embodiment of the application, temperature, pressure, flow rate and pH data are collected by sensors installed in a preset process section.

[0026] For example, the preset process sections can be the inlet and outlet of the sterilizer, the front and back of the homogenizer, and different height layers of the fermenter.

[0027] Online sensors such as temperature sensors, pressure sensors, flow meters, and pH electrodes are installed in the aforementioned preset process sections to collect temperature, pressure, flow, and pH data at a set sampling frequency.

[0028] In this embodiment, temperature sensors are installed at the inlet and outlet of the sterilizer, before and after the homogenizer, and at different heights in the fermenter; pressure sensors are installed at the inlet and outlet of the homogenizer and the outlet of the sterilizer; flow meters are installed at the inlet of each process section; and pH electrodes are installed at the outlet of the fermenter and the sterilizer.

[0029] Each sensor must meet hygiene standards and be made of AISI 316L stainless steel. The sampling frequency can be configured according to the detection mode. For example, in high-precision mode, the temperature sampling frequency is 10Hz, the pressure sampling frequency is 100Hz, the flow sampling frequency is 10Hz, and the pH sampling frequency is 1Hz; in standard mode, the temperature sampling frequency is 1Hz, the pressure sampling frequency is 1Hz, the flow sampling frequency is 1Hz, and the pH sampling frequency is 0.2Hz.

[0030] Furthermore, after collecting the above data, the collected data is sent to the computing node through a communication protocol, and a timestamp based on the same time base is attached to each data before sending. The timestamp is used to establish a correspondence between the data fragment and the food batch by combining the batch traceability identifier.

[0031] As an example, the communication protocol could be an industrial communication protocol, such as Modbus TCP, Profinet, OPCUA, etc.

[0032] In this embodiment of the application, before sending the collected data through the communication protocol, a unified timestamp based on PTP (IEEE 1588) synchronization is attached to each data, with a timestamp accuracy of milliseconds.

[0033] In some embodiments, after receiving multi-source data, the computing node first performs data preprocessing, including time synchronization, outlier removal, and filtering and noise reduction.

[0034] Specifically, time synchronization is achieved through hardware triggering or cubic spline interpolation, with an interpolation window of 5 sampling points before and after; anomaly removal is based on 3 The criterion is that when the value of a data point deviates from the mean of its data series by more than three standard deviations... If a point is considered abnormal, and three consecutive sampling points are abnormal, it is discarded and replaced with linear interpolation. A second-order Butterworth low-pass filter is used for filtering and noise reduction. The cutoff frequency is set according to the parameters and mode. For example, in high-precision mode, the temperature cutoff frequency is 1Hz, the pressure cutoff frequency is 10Hz, the flow rate cutoff frequency is 5Hz, and the pH value cutoff frequency is 0.1Hz. In standard mode, the temperature cutoff frequency is 0.5Hz, the pressure cutoff frequency is 0.5Hz, the flow rate cutoff frequency is 0.5Hz, and the pH value cutoff frequency is 0.1Hz. Zero-phase filtering is used.

[0035] Step S20: Based on the changing trends of multi-source data, identify the start and end boundaries of the process segment and divide the data into segments.

[0036] Optionally, for each process segment, the process segment begins when the flow rate increases from zero to a non-zero value, and ends when the flow rate decreases from a non-zero value to zero. Usually, time needs to be considered in conjunction with the process segment to avoid misjudgment.

[0037] In this embodiment of the application, when the flow rate exceeds a first flow rate threshold and the duration is greater than a first minimum process time, the process segment is determined to start; when the flow rate drops below a second flow rate threshold and the duration is greater than a second minimum process time, the process segment is determined to end; or, when the pressure change rate exceeds a pressure change rate threshold and the duration is greater than a second minimum process time, the process segment is determined to end.

[0038] As an example, the first traffic threshold can be The first minimum process time can be, for example, 5-10 seconds, and the second flow rate threshold can be... The second minimum process time can be, for example, 3-10 seconds, and the pressure change rate threshold can be 0.1-1.0 MPa / s.

[0039] After obtaining the start and end times of each process segment, the multi-source data is divided into multiple data segments according to the timestamps, with each start and end time serving as the segmentation time.

[0040] It should be noted that before each batch of food begins production, a globally unique batch traceability identifier is generated, and this identifier, along with its start time on the production line and expected end time, is recorded in the database. The production time for each batch of the same food is consistent; therefore, the expected end time can be obtained by adding the start time to the production time.

[0041] In some embodiments, continuously produced food products can also be divided into batches by time period, such as the duration of each can of liquid as a batch, or a batch every 30 minutes.

[0042] For each data segment, its time range is compared with the time traversal of each batch to determine whether there is complete inclusion or partial overlap. Data segments that successfully match are added with a batch identifier field and an associated timestamp, and then stored in the relational database.

[0043] Since data segments are usually much smaller than the batch time window, the starting time of the data segment can also be used to determine which batch it belongs to. That is, if the starting time of the data segment falls within the time interval of a certain batch of food, the segment is assigned to that batch.

[0044] Furthermore, when switching batches, there may be a short transition period. A transition interval can be set to avoid erroneous association. For example, if a data segment is located within 5 seconds before or after a batch switch, both batches can be marked simultaneously or marked separately as a transition area.

[0045] Step S30: For each data segment, data fingerprints including thermal accumulation fingerprint, mechanical shear fingerprint, and biochemical reaction fingerprint are obtained based on multi-source data. Among them, the thermal accumulation fingerprint is obtained by time integration of the temperature rate function and the piecewise temperature zone weight function; the mechanical shear fingerprint is obtained by time integration of instantaneous power, which is obtained based on pressure and flow rate; and the biochemical reaction fingerprint is obtained based on the biochemical reaction rate determined by pH and temperature.

[0046] Optionally, for each data segment, a data fingerprint is calculated based on the collected multi-source data. A data fingerprint is a series of quantitative indicators that reflect the cumulative effects of physical and chemical reactions experienced by the corresponding food material within a specific process segment, including at least one or more of thermal cumulative fingerprints, mechanical shear fingerprints, and biochemical reaction fingerprints. By calculating the data fingerprint in real time, the highly coupled and time-varying process environment state within the data segment can be mapped into standardized numerical characteristics directly related to product quality, thereby achieving quantitative measurement and real-time evaluation of the production quality state within the data segment.

[0047] In this embodiment, a temperature rate function is constructed based on the exponential effect of temperature on the reaction rate; a piecewise temperature zone weight function is constructed by assigning weight functions to multiple continuous temperature intervals; and the product of the temperature rate function and the piecewise temperature zone weight function is integrated over the time domain of the data segment to obtain the thermal accumulation fingerprint.

[0048] Specifically, thermal accumulation fingerprints are used to characterize the cumulative effect of the temperature field within a segment on the thermal effects of the material. The calculation process first defines a temperature rate function: ; in, Represents the temperature rate function. The value represents the activation energy (J / mol), and R represents the gas constant, 8.314 J / (mol·K). Indicates the reference temperature (°C).

[0049] It should be noted that activation energy The activation energy range is generally 80,000-100,000 J / mol, and in the embodiments of this application... Take 86500 J / mol; reference temperature The temperature range is generally 80-110℃, and in this embodiment, the value is 85℃.

[0050] It should be noted that the calculation formula for thermal cumulative fingerprints requires the use of Kelvin temperature for meaningful exponent calculation. Therefore, through... Convert to Kelvin temperature.

[0051] Furthermore, a piecewise temperature zone weighting function is constructed. The function distinguishes at least the normal temperature zone (T<50℃), the heating zone (50℃≤T<70℃), the sterilization and heat preservation zone (70℃≤T<110℃), and the high temperature over-limit zone (T≥110℃), and assigns weights to each zone.

[0052] As an example, the piecewise temperature zone weighting function can be obtained using the following formula. : ; Furthermore, calculate the thermal accumulation fingerprint. : ; in, t represents the start time of the data segment, and t represents the end time of the data segment. Represents the time variable during the integration process. Indicates time The temperature below.

[0053] In some embodiments, when the food is a dairy liquid food, a furosine content prediction function can be constructed to verify the accuracy of the thermal accumulation fingerprint calculation model.

[0054] It should be noted that furosine is an early product of the Maillard reaction during the heat treatment of dairy products. Its content is positively correlated with the intensity of heat treatment and can be used as a chemical fingerprint of the heat accumulation effect. By predicting the furosine content, an indirect and real-time auxiliary assessment of food quality can be achieved.

[0055] As an example, a furosine content prediction model can be constructed using the following formula: ; in, This indicates the predicted furosine content (mg / 100g protein). This indicates the initial furosine content (usually 0). Indicates at time The temperature below Dependent on reaction rate, Indicates at time The temperature below The weight of furosine production.

[0056] in, A represents the pre-exponential factor (1 / s), with a range of values ​​of 1 / s. , It is a function representing the change of activation energy with temperature.

[0057] In one embodiment, the activation energy may not be constant but depends on temperature changes, as the reaction mechanism may change with increasing temperature, and the activation energy will change accordingly.

[0058] As an example, the activation energy as a function of temperature can be calculated using the following formula: ; in, This represents the activation energy at temperature T. Indicates the fundamental activation energy. Indicates the range of activation energy change. Indicates the transition temperature. Indicates the width of the change.

[0059] It should be noted that the basic activation energy The value ranges from 70,000 to 80,000 J / mol, and in this embodiment, the value is 76,500; the activation energy variation range The value ranges from 15000 to 25000 J / mol, and in this embodiment, it is taken as 20000 J / mol; transition temperature The value range is 80-95℃, and in this embodiment, the value is 87.5℃; the transformation width This represents half the width of the temperature range in which the activation energy changes significantly; in the embodiments of this application, the transition width... The value ranges from 5 to 15 ℃, and in this embodiment, the value is 10 ℃.

[0060] Furthermore, furosine production weight Also based on temperature changes, as an example, the furosine production weight can be calculated using the following formula: ; After obtaining the predicted furosine content, samples of the corresponding batches of dairy products were taken for actual testing. The measured furosine content was obtained by high performance liquid chromatography (HPLC). The predicted furosine content and the measured furosine content were compared, and the coefficient of determination was calculated to reflect the goodness of fit of the prediction model.

[0061] As an example, the coefficient of determination The calculation formula is: ; in, This represents the measured furosine content of the i-th sample. This represents the predicted furosine content of the i-th sample. This represents the average of all measured furosine contents. Indicates the number of samples. Represents the sum of squared residuals. This represents the total sum of squares.

[0062] According to the coefficient of determination The value of is used to evaluate the goodness of fit of the furosine prediction model; for example, the coefficient of determination. A value between 0.90 and 1.00 indicates strong model interpretability and high prediction accuracy; the coefficient of determination A value between 0.75 and 0.90 indicates that the model has practical value; the coefficient of determination A value between 0.50 and 0.75 indicates that the model has some explanatory power; the coefficient of determination A value <0.50 indicates that the model's explanatory power is insufficient.

[0063] In this embodiment of the application, the mechanical shear fingerprint is obtained by time integration of the instantaneous power, i.e., the product of pressure difference and flow rate.

[0064] As an example, mechanically cut fingerprints can be calculated using the following formula: ; in, This indicates a mechanically cut fingerprint (J). Indicates time Pressure difference (Pa) Indicates time The flow rate below ( ).

[0065] Furthermore, the processing volume is normalized to obtain unit mechanical shear fingerprints. The unit is .

[0066] Mechanical energy input is not completely converted into effective shear energy; frictional losses, turbulent dissipation, and thermal conversion result in energy loss. Differences in equipment model, operating mode, performance, or wear level also lead to variations in the energy conversion coefficient. Therefore, in some embodiments, an energy conversion coefficient can be introduced into the mechanical shear fingerprint acquisition process.

[0067] Specifically, the energy conversion coefficient is determined based on fluid characteristic parameters and equipment structural parameters; after multiplying the energy conversion coefficient as a coefficient of instantaneous power, the product is integrated in the time domain of the data segment to obtain the mechanical shear fingerprint.

[0068] Among them, the fluid characteristic parameters include at least the fluid viscosity, and the equipment structural parameters include at least the equipment clearance.

[0069] As an example, the energy conversion factor can be calculated using the following formula: ; in, Indicates the energy conversion coefficient. This represents the standard energy conversion factor under the reference operating condition. Indicates fluid viscosity. This indicates the fluid viscosity under reference operating conditions. Indicates pressure difference, This indicates the pressure difference under the reference operating condition. Indicates the clearance dimensions of the equipment. This indicates the equipment clearance under the reference operating conditions. An exponential parameter representing the viscosity of a fluid. An exponential parameter representing the pressure difference. An exponential parameter representing equipment clearance.

[0070] In the embodiments of this application, fluid viscosity The value range is 1-1000 mPa·s, pressure difference The value range is 5-30MPa, equipment clearance. The value range is 0.05-0.5 mm. The reference values ​​and index parameters under the baseline operating conditions are determined through experiments or engineering calibration. For example, the calibration of low-viscosity fluids... Medium viscosity fluid High viscosity fluid .

[0071] After introducing the energy conversion coefficient, the mechanical shear fingerprint can be calculated using the following formula: .

[0072] In this embodiment, the biochemical reaction fingerprint is obtained by the biochemical reaction rate affected by temperature and pH.

[0073] Specifically, a temperature influence function is constructed using the Arrhenius form or the Ratkowsky temperature growth model; an acid-base influence function is constructed using a bell-shaped function centered on the optimal pH; a biochemical reaction rate function is constructed by combining the temperature influence function and the acid-base influence function; and the biochemical reaction rate function is integrated over the time domain of the data segment to obtain the biochemical reaction fingerprint.

[0074] As one possible implementation, a temperature influence function is constructed based on the deviation between the ambient temperature and the optimum temperature: ; in, Represents the effect of temperature. Indicates the optimal temperature.

[0075] As one possible implementation method, an acid-base influence function is constructed based on the deviation between the environmental acidity / alkalinity and the optimal acidity / alkalinity: ; in, This represents the effect function of acidity and alkalinity. Indicates the optimal pH value. Indicates pH sensitivity.

[0076] Furthermore, construct the biochemical reaction rate function: ; in, Represents the rate function of a biochemical reaction. The maximum growth rate is represented by S, and the substrate concentration is represented by S. Let X represent the saturation constant and X represent the biomass. This indicates the decay rate.

[0077] In the embodiments of this application, the optimal temperature The value range is 38-42℃, and the maximum growth rate is... The value range is 0.1-1.0 h - ¹, saturation constant The value range is 0.5-5.0 g / L.

[0078] It should be noted that the substrate concentration S and biomass X among the above parameters can be measured online. For example, the substrate concentration can be monitored online using infrared spectroscopy or electrochemical probes, and the biomass can be monitored in real time using online turbidity sensors or capacitance methods. Other parameters need to be calibrated experimentally, such as the maximum growth rate. The maximum increase in biomass per unit time was obtained by batch culture measurement; saturation constant. The optimal pH value was obtained by linear fitting of substrate concentration-growth rate data using the double reciprocal method. and pH sensitivity Through multiple pH gradient cultivation experiments, the peak growth rate corresponded to the optimal pH value. The full width at half maximum (FWHM) of the curve reflects pH sensitivity. Optimal temperature and activation energy The saturation constant was obtained by linearizing and fitting the growth rate at different temperatures using the Arrhenius equation; Obtained by measuring the natural decay rate of biomass during static culture (without substrate limitations).

[0079] Furthermore, calculate the biochemical fingerprint: ; in, Fingerprints represent biochemical reactions.

[0080] Because under high-temperature processes, such as pasteurization or UHT ultra-high temperature instantaneous sterilization, any target biochemical reaction has stopped or its rate is negligible due to the high temperature, the calculation model of biochemical reaction fingerprint can be disabled or shielded, that is, the biochemical reaction fingerprint is set to 0.

[0081] Specifically, the process stage to which the current data segment belongs is determined based on temperature or timestamp, and the fingerprint data is calculated by calling the fingerprint that matches that stage.

[0082] As an example, based on temperature sensor data, it is determined that the current temperature T exceeds a preset biological inactivation threshold. When the current data segment is determined to be in the high-temperature processing stage, the biochemical reaction fingerprint is set to 0. At this point, the comprehensive risk assessment mainly relies on the thermal accumulation fingerprint. and mechanically cut fingerprints The calculation results are used to monitor the risk of thermal damage such as protein denaturation and excessive Maillard reaction.

[0083] As an example, the biological inactivation threshold can be set to a value that is... .

[0084] Step S40: Map the data fingerprints to multiple risk indicators according to their categories to form a risk vector, and map the risk vector after correction using a preset coupling coefficient matrix to a scalar to obtain the coupling risk index.

[0085] Optionally, the three types of data fingerprints—thermal accumulation fingerprints, mechanical shear fingerprints, and biochemical reaction fingerprints—are mapped to three risk indicators, which are then transformed into risk indicators with unified dimensions to form a risk vector. The coupling effects between multiple physics fields are considered, and the coupling coefficient matrix is ​​corrected. Finally, a single coupled risk index is aggregated to quantify the comprehensive quality risk of the food batch containing the data segment.

[0086] Specifically, firstly, each data fingerprint is mapped to a risk index value for a single physics field using a normalization formula through security limits: ; in, This represents the standardized risk value of the j-th data fingerprint. This represents the j-th data fingerprint. Let represent the minimum value of the j-th data fingerprint. This represents the maximum threshold value of the j-th data fingerprint.

[0087] In this embodiment of the application, there are three data fingerprints, i.e., j=1, 2, 3. As an example, when j=1, Indicates thermal accumulation fingerprint Corresponding heat accumulation risk When j=2, Indicates mechanically cut fingerprint Corresponding mechanical shear risk When j=3, Fingerprints representing biochemical reactions Corresponding biochemical reaction risks .

[0088] Furthermore, the risk vector is composed of the above three risk indicators. The risk vector is then modified using the coupling coefficient matrix, and the modified risk vector is denoted as the coupling risk vector.

[0089] Specifically, the coupling coefficient matrix is ​​an n×n matrix, where n is the number of risk indicators. The diagonal elements in the matrix are preset values, and the off-diagonal elements are preset coupling coefficients between various risk indicators. The coupling coefficients are obtained through offline experiments or historical data calibration.

[0090] As an example, the default value can be 1.

[0091] In this embodiment, there are three risk indicators, and the generated coupling coefficient matrix is ​​3×3, which can be represented as: ; Wherein, the coupling coefficient matrix Each element in the table represents the coupling coefficient between the corresponding risk indicators, which is determined through offline calibration. express and The coupling coefficient between them express and The coupling coefficient between them express and The coupling coefficient between them express and The coupling coefficient between them express and The coupling coefficient between them express and The coupling coefficient between them.

[0092] In this embodiment of the application, the coupling risk vector , is a 3×1 vector.

[0093] It should be noted that when biochemical fingerprints... When set to 0, its corresponding risk indicator It is also 0. At this time, the input risk vector The form is It remains a 3x1 vector. Since there is no longer a coupling effect on other risks when the biochemical reaction fails, therefore, in the biochemical reaction fingerprint... When set to 0, the coupling coefficient matrix will be linked to the biochemical reaction fingerprint. All relevant coupling coefficients were set to 0 to eliminate all effects of the biochemical reactions.

[0094] In the embodiments of this application, when biochemical reaction fingerprints When set to 0, the coupling coefficient matrix In , , and The values ​​are all 0.

[0095] It should be noted that the coupling coefficient quantifies the coupling strength between different risk indicators. A positive coupling coefficient indicates that the two risk indicators promote each other; the presence of one risk indicator amplifies the damage caused by the other. For example, in dairy processing, the high mechanical shear force brought about by high-pressure homogenization can damage the fat globule membrane structure, exposing the milk fat. If the fluid is also at a high temperature, the oxidative rancidity rate of the exposed milk fat under high-temperature catalysis will increase exponentially, far exceeding the oxidation rate under no-shear or low-temperature conditions. A negative coupling coefficient indicates that the two risk indicators inhibit each other; the enhancement of one risk indicator may inhibit the occurrence of the other. For example, in a specific fermentation process, as the biochemical reaction progresses, the acidic metabolites produced gradually lower the pH value, while the reaction rate of adverse high-temperature side reactions such as the Maillard reaction decreases significantly with decreasing pH. Therefore, this acidic environment can inhibit certain adverse side reactions caused by high temperature, thus reducing the risk of biochemical reactions. Risk of heat accumulation They exhibit a negative coupling relationship. However, this continuous coupling relationship weakens when the temperature exceeds the biological inactivation threshold. It becomes ineffective at this time, and at this time it reacts with the biochemical fingerprint. All relevant coupling coefficients are set to 0.

[0096] By cross-quantifying the risks in fluid food production using the coupling coefficient matrix, and considering the mutual promotion or inhibition between different parameters, the complex food chemical and biological changes under multi-physics fields are more realistically reflected. This approach can identify the complex cross-effects of food chemical reactions that cannot be captured by simple linear weighting.

[0097] Furthermore, by mapping the coupling risk vector to a scalar, we obtain the coupling risk index after superimposing the influence of multiple physical fields.

[0098] As an example, the coupling risk index is obtained by taking the offset geometric mean of each element in the coupling risk vector.

[0099] In the embodiments of this application, the offset geometric mean is a geometric mean with an offset, which is an improved algorithm of the traditional geometric mean. By applying a fixed positive offset to each risk element, the input value is mapped to the positive real number domain, avoiding the problem of calculation failure or result distortion of the traditional geometric mean when the risk value is 0. After the calculation is completed, the offset is deducted to restore the original dimensions.

[0100] Specifically, in this embodiment, an offset of 1 is selected, and the coupling risk index is calculated using the following formula: ; Wherein, CRI represents the coupling risk index. Represents the coupling risk vector The k-th element in.

[0101] In other embodiments, a scalar mapping can be performed by taking the geometric mean of each element in the coupling risk vector or by taking the norm of the coupling risk vector to obtain the coupling risk index.

[0102] In some embodiments, to further improve accuracy, trend risk correction and / or uncertainty risk correction may be added after obtaining the coupling risk index.

[0103] Specifically, based on the changing trend of the coupling risk index within a historical time window, a trend risk factor is calculated, and the coupling risk index is corrected using the trend risk factor; and / or, Based on historical differences from multi-source data and risk indicators, an uncertainty correction factor is calculated, and the coupled risk index is corrected using the uncertainty correction factor.

[0104] In this embodiment of the application, the historical time window can be 30 seconds, and the trend risk factor... It can be calculated using the following formula: ; in, This represents the average risk within the window. Indicates the rate of change of risk. Indicates the trend sensitivity coefficient. This indicates the prediction time step.

[0105] As an example, the trend sensitivity coefficient The value range is 2.0-5.0, and the prediction time step is... It can be 10 seconds.

[0106] As one possible implementation method, the coupling risk index can be modified using a trend risk factor: ,in, This represents the coupled risk index after adjustment by the trend risk factor.

[0107] In the embodiments of this application, the measurement uncertainty of each sensing parameter is determined by Bayesian estimation or sensor calibration, and then used to calculate the uncertainty correction factor.

[0108] As an example, the uncertainty correction factor can be calculated using the following formula. : ; in, This represents the measurement uncertainty corresponding to the j-th data fingerprint. This represents the uncertainty sensitivity coefficient.

[0109] As an example, the uncertainty sensitivity coefficient The value range is 0.5-1.5.

[0110] As one possible implementation method, the coupling risk index is corrected using an uncertainty correction factor: ,in, This represents the coupling risk index after adjustment by the uncertainty correction factor.

[0111] As another possible implementation, the coupled risk index can be modified by combining trend risk factors and uncertainty correction factors: ,in, This indicates a modified coupling risk index.

[0112] By introducing trend risk factors and uncertainty correction factors, the changes in risk and the uncertainty of sensing measurements within the historical window are compensated, effectively avoiding misjudgments caused by sensor drift, noise, or sudden disturbances, and improving the accuracy of risk identification.

[0113] Step S50: Compare the coupling risk index with a preset safety threshold to determine the quality risk level of the food batch corresponding to the data segment.

[0114] Optionally, multiple safety thresholds can be set to divide the coupled risk index into multiple risk intervals, each corresponding to a risk level. By determining the risk interval in which the coupled risk index is located, the quality risk level of the food batch in which the current data segment is located can be determined.

[0115] As an example, the risk is divided into four zones using three safety thresholds: Normal level At the warning level, For the classification level, This is the shutdown level.

[0116] Based on the range of the coupled risk index, the corresponding risk level is determined. If it is normal, the corresponding batch of food is being produced normally; if it is warning, it indicates that the corresponding batch of food may have some risk, and a prompt to check is displayed; if it is diversion, it indicates that the corresponding batch of food may have a greater risk, and the product needs to be diverted and the batch needs to be marked; if it is shutdown, it indicates that the corresponding batch of food has a serious abnormality, and an immediate emergency shutdown and cutting off of feed are required.

[0117] In some embodiments, the preset safety threshold is dynamically and adaptively adjusted based on the process stability index of the food production line; the process stability index is calculated based on the degree of fluctuation of the coupled risk index within a historical time window.

[0118] As an example, the security threshold can also be adjusted using the following formula: ; in, This represents the adaptively adjusted safety threshold. Indicates the basic security threshold. Indicates the stability sensitivity coefficient. This indicates the process stability index.

[0119] ; in, This represents the variance of the coupled risk index within a historical time window. This represents the target process reference variance.

[0120] In this embodiment, the volatility of the coupling risk index is characterized by the variance of the coupling risk index within a historical time window.

[0121] The larger the variance of the coupled risk index within the historical time window, the greater the fluctuation of the process segment and the more unstable the process. The adaptive threshold is reduced by the above formula, tightening the judgment criteria, so that early warning can be triggered earlier and more sensitively when the fluctuation intensifies, thereby enhancing the intelligent management of adaptive risk control.

[0122] Please see Figure 2 The second embodiment of this application provides a food quality testing system based on multi-source data, used to execute the above-described food quality testing method based on multi-source data, comprising: a data acquisition module 100, a boundary recognition module 200, a fingerprint calculation module 300, a coupling risk index acquisition module 400, and a risk assessment module 500.

[0123] The data acquisition module 100 is used to acquire multi-source data from the food production line in real time. The multi-source data includes at least temperature, pressure, flow rate, and pH. Boundary recognition module 200 is used to identify the start and end boundaries of the process section and divide the data segments based on the changing trends of multi-source data; The fingerprint calculation module 300 is used to acquire data fingerprints, including thermal accumulation fingerprints, mechanical shear fingerprints, and biochemical reaction fingerprints, based on multi-source data for each data segment. The thermal accumulation fingerprint is obtained by time integration using a temperature rate function and a piecewise temperature zone weighting function. The mechanical shear fingerprint is obtained by time integration of instantaneous power, which is obtained based on pressure and flow rate. The biochemical reaction fingerprint is obtained based on the biochemical reaction rate determined by pH and temperature. The coupling risk index acquisition module 400 is used to map data fingerprints into multiple risk indicators according to categories, form a risk vector, and map the risk vector after correction using a preset coupling coefficient matrix into a scalar to obtain the coupling risk index. The risk assessment module 500 is used to compare the coupled risk index with a preset safety threshold to determine the quality risk level of the food batch corresponding to the data segment.

[0124] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the system described above can be found in the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0125] It should be noted that the food quality testing method and system based on multi-source data provided in the above embodiments are only illustrative examples of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of this application can be further decomposed or combined. For example, the modules in the above embodiments can be merged into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of this application are only for distinguishing the various modules or steps and are not considered as an improper limitation of this application.

[0126] A device according to a third embodiment of this application includes: At least one processor; and a memory communicatively connected to at least one of the processors; The memory stores instructions that can be executed by the processor to implement the aforementioned food quality detection method based on multi-source data.

[0127] A fourth embodiment of this application provides a computer-readable storage medium storing computer instructions, which are executed by the computer to implement the above-described food quality detection method based on multi-source data.

[0128] A computer program product according to the fifth embodiment of this application, when run on an electronic device, causes the electronic device to execute the above-described food quality detection method based on multi-source data.

[0129] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes and related descriptions of the electronic devices, computer-readable storage media, and computer program products described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0130] The following is for reference. Figure 3 It shows a schematic diagram of the structure of a computer system for implementing embodiments of the systems, methods, and electronic devices of this application. Figure 3The server shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0131] like Figure 3 As shown, the computer system includes a Central Processing Unit (CPU) 301, which can perform various appropriate actions and processes based on programs stored in Read Only Memory (ROM) 302 or programs loaded from storage section 308 into Random Access Memory (RAM) 303. The RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An Input / Output (I / O) interface 305 is also connected to the bus 304.

[0132] The following components are connected to I / O interface 305: an input section 306 including a keyboard, mouse, etc.; an output section 307 including a cathode ray tube (CRT), liquid crystal display (LCD), and speakers, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card and a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to I / O interface 305 as needed. Removable media 311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 310 as needed so that computer programs read from them can be installed into storage section 308 as needed.

[0133] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by central processing unit (CPU) 301, it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0134] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0135] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0136] The terms “first”, “second”, etc., are used to distinguish similar objects, not to describe or indicate a specific order or sequence.

[0137] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent in such process, method, article, or apparatus / device.

[0138] The technical solutions of this application have been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of this application is obviously not limited to these specific embodiments. Without departing from the principles of this application, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of this application.

Claims

1. A food quality detection method based on multi-source data, applied to the production process of fluid food, characterized in that, include: Real-time acquisition of multi-source data from the food production line, wherein the multi-source data includes at least temperature, pressure, flow rate, and pH. Based on the changing trends of the multi-source data, the start and end boundaries of the process segment are identified and data segments are divided. For each data segment, data fingerprints including thermal accumulation fingerprints, mechanical shear fingerprints, and biochemical reaction fingerprints are obtained based on the multi-source data. The thermal accumulation fingerprint is obtained by time integration using a temperature rate function and a piecewise temperature zone weighting function. The mechanical shear fingerprint is obtained by time integration of instantaneous power, which is obtained based on the product of pressure difference and flow rate. The biochemical reaction fingerprint is obtained based on the biochemical reaction rate determined by pH and temperature. The data fingerprints are mapped into multiple risk indicators according to their categories to form a risk vector. The risk vector, after being corrected by a preset coupling coefficient matrix, is then mapped into a scalar to obtain the coupling risk index. The coupling risk index is compared with a preset safety threshold to determine the quality risk level of the food batch corresponding to the data segment; The process of acquiring the thermally accumulated fingerprint includes: The temperature-rate function is constructed based on the exponential effect of temperature on the reaction rate; The segmented temperature zone weighting function is constructed by assigning weighting functions to multiple continuous temperature ranges respectively; The thermal accumulation fingerprint is obtained by integrating the product of the temperature rate function and the piecewise temperature zone weighting function over the time domain of the data segment. The process of obtaining the biochemical fingerprint includes: A temperature effect function is constructed based on the deviation between the ambient temperature and the optimum temperature. A pH influence function is constructed based on the deviation between the environmental pH and the optimal pH. A biochemical reaction rate function is constructed by combining the temperature effect function and the pH effect function; The biochemical reaction rate function is integrated over the time domain of the data segment to obtain the biochemical reaction fingerprint.

2. The food quality detection method based on multi-source data according to claim 1, characterized in that, After obtaining the coupling risk index, it also includes: Based on the changing trend of the coupling risk index within a historical time window, a trend risk factor is calculated, and the coupling risk index is corrected using the trend risk factor; and / or, Based on the historical differences of the multi-source data and the risk indicators, an uncertainty correction factor is calculated, and the coupled risk index is corrected using the uncertainty correction factor.

3. The food quality detection method based on multi-source data according to claim 1, characterized in that, The process of obtaining fingerprints by mechanical cutting also includes: The energy conversion coefficient is determined based on fluid characteristic parameters and equipment structural parameters, wherein the fluid characteristic parameters include at least fluid viscosity and the equipment structural parameters include at least equipment clearance. The mechanical shear fingerprint is obtained by multiplying the energy conversion coefficient as the coefficient of the instantaneous power and integrating it over the time domain of the data segment.

4. The food quality detection method based on multi-source data according to claim 1, characterized in that, The process of identifying the start and end boundaries of the process segment and dividing the data into segments includes: When the flow rate exceeds the first flow rate threshold and the duration is greater than the first minimum process time, the process segment is determined to start. The process segment is determined to end when the flow rate drops below the second flow rate threshold and the duration is greater than the second minimum process time; or, the process segment is determined to end when the pressure change rate exceeds the pressure change rate threshold and the duration is greater than the second minimum process time.

5. The food quality detection method based on multi-source data according to claim 1, characterized in that, The coupling coefficient matrix is ​​an n×n matrix, where n is the number of risk indicators. The diagonal elements of the matrix are preset values, and the off-diagonal elements are preset coupling coefficients between various risk indicators. The coupling coefficients are obtained through offline experiments or historical data calibration.

6. The food quality detection method based on multi-source data according to claim 1, characterized in that, The real-time acquisition of multi-source data from the food production line includes: Temperature, pressure, flow rate, and pH data are collected by sensors installed in the preset process section. The collected data is sent to the computing node via a communication protocol, and a timestamp based on the same time base is attached to each data point before sending. The timestamp is used to establish a correspondence between the data segment and the food batch in combination with the batch traceability identifier.

7. The food quality detection method based on multi-source data according to claim 1, characterized in that, The preset safety threshold is dynamically and adaptively adjusted based on the process stability index of the food production line. The process stability index is calculated based on the degree of fluctuation of the coupled risk index within a historical time window.

8. A food quality detection system based on multi-source data, applied in the production process of fluid food, for executing the food quality detection method based on multi-source data as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to acquire multi-source data from the food production line in real time, wherein the multi-source data includes at least temperature, pressure, flow rate and pH. The boundary recognition module is used to identify the start and end boundaries of the process section and divide the data segments based on the changing trends of the multi-source data. The fingerprint calculation module is used to acquire data fingerprints, including thermal accumulation fingerprints, mechanical shear fingerprints, and biochemical reaction fingerprints, based on the multi-source data for each data segment; wherein, the thermal accumulation fingerprint is acquired by time integration using a temperature rate function and a piecewise temperature zone weighting function; the mechanical shear fingerprint is acquired by time integration of instantaneous power, which is obtained based on pressure and flow rate; and the biochemical reaction fingerprint is acquired based on the biochemical reaction rate determined by pH and temperature. The coupling risk index acquisition module is used to map the data fingerprint into multiple risk indicators according to categories, form a risk vector, and map the risk vector after correction using a preset coupling coefficient matrix into a scalar to obtain the coupling risk index. The risk assessment module is used to compare the coupled risk index with a preset safety threshold to determine the quality risk level of the food batch corresponding to the data segment.

Images