Traceability management method and system based on food quality detection

Abstract

The invention discloses a traceability management method and system based on food quality detection, and relates to the technical field of food quality detection, and the method comprises the steps: distributing a unique production serial number for each product, fusing the dimension information of production time, batches, process nodes and the like, and synchronously associating equipment operation parameters and video feature parameters; based on batch and time clustering data, dynamically counting standardized operation intervals of production parameters, generating adaptive thresholds of equipment parameters and dynamic fitting intervals of video features, and eliminating manual setting deviations; after disqualified products are found through spot check, space-time adjacent batches are retrieved according to the serial numbers of the disqualified products, the deviation degree between monitoring parameters of each batch and a standard interval is compared, a weighted production state evaluation value is calculated, and when the evaluation value is lower than a set threshold value, a high-risk serial number is automatically marked, and directional recall is triggered; according to the technical scheme, multi-source data integrated analysis is achieved, and then other food with abnormal detection quality can be located more quickly and accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of food quality detection, and in particular to a traceability management method and system based on food quality detection. Background Art

[0002] At present, food production quality control mainly relies on manual sampling and single-dimensional data traceability, which makes it difficult to accurately locate and quickly respond to quality risks under complex processes. In traditional methods, equipment operating parameters and video surveillance information exist in isolation, resulting in low efficiency in tracing abnormal events and inability to effectively associate the relationship between production data anomalies and specific quality defects. For example, when a product is contaminated with foreign matter, it takes a lot of time to check the equipment logs and video records one by one, and the manually set static parameter thresholds cannot dynamically adapt to equipment aging or environmental changes, which can easily lead to misjudgment or missed detection. In addition, the existing appearance defect detection technology is limited by factors such as uneven lighting and product deformation, and the false alarm rate is high. When unqualified products are found after sampling, the batch-level extensive recall mechanism is likely to lead to increased economic losses. In response to the above problems, it is urgent to develop a traceability management system that integrates multi-source data dynamic modeling, intelligent parameter interval calibration and active risk prediction to achieve refined quality control and precise risk interception of the entire production chain. Summary of the invention

[0003] The purpose of the present invention is to provide a traceability management method and system that can trace and determine the source of food with abnormal quality in combination with multi-source data analysis.

[0004] The present invention discloses a traceability management method based on food quality detection, comprising: A product production serial number is constructed for the production of each product, and the production serial number includes the production time, production batch, production sequence in the production batch and the production step to which it belongs, and a production monitoring parameter group is constructed for each production serial number, and the production monitoring parameter group includes several production equipment operation parameters and monitoring video feature parameters; Based on the production batch and production time in the production serial number, the production monitoring parameter group is classified, and the classified production monitoring parameter group is analyzed in intervals to determine the first standard production operation parameter interval of the production equipment operation parameter and the second standard production operation parameter interval of the monitoring video feature parameter; Finished foods are sampled and foods that do not meet quality standards are identified. Based on the production serial numbers of the foods that do not meet quality standards, other production serial numbers with similar production times and production batches are determined. The production monitoring parameter groups corresponding to other production serial numbers are compared with the first standard production operation parameter range and the second standard production operation parameter range, respectively, and a production status assessment value is calculated. If the production status assessment value is less than or equal to a preset value, the corresponding production serial number is marked as abnormal, and the corresponding food is retrieved for quality inspection.

[0005] In some embodiments disclosed in the present invention, the production equipment operating parameters include: Cleaning machine operating parameters, including water pressure, water temperature, and spray time; Sorting machine operating parameters, including vibration frequency, color sorting accuracy, and rejection rate; Dryer operating parameters, including hot air temperature, humidity, and material residence time; Mixer operating parameters, including speed, torque, current load, and material temperature; Homogenizer operating parameters, including pressure value, number of cycles, and particle size; Fermentation tank operating parameters, including temperature, pH, dissolved oxygen concentration, and agitation rate; Pasteurizer operating parameters, including temperature profile, holding time, flow rate; UHT sterilizer operating parameters, instantaneous temperature, cooling rate, steam pressure; Microwave sterilizer operating parameters, power density, frequency, material penetration depth; Canning machine operating parameters, canning volume error, liquid level height, sealing pressure; Vacuum sealer operating parameters, vacuum degree, sealing temperature, residual oxygen; Labeling machine operating parameters, positioning accuracy, glue coating thickness, label adhesion rate.

[0006] In some embodiments disclosed in the present invention, the method for determining the characteristic parameters of a surveillance video includes: The food assembly line area on the video image is delineated, and the visual analysis technology is used to lock the individual food blocks in the food assembly line area, and the movement rate of each individual food block and the degree of abnormality in the appearance of the individual food block are determined. The movement rate of the individual food block and the degree of abnormality in the appearance of the individual food block are identified as monitoring video feature parameters.

[0007] In some embodiments disclosed in the present invention, the method of determining the degree of abnormality of the appearance of a single food block using visual analysis technology includes: Performing contour edge detection on the food flow area, determining several contour edge lines, and combining different contour edge lines based on the distance between the ends of the contour edge lines, and performing head-to-tail association to form the edge contour of a single food block; A food block direction calibration template is set, the food block direction calibration template includes a standard outline of the food block and a plurality of grayscale correction points in the standard outline, the food block direction calibration template and the edge outline are dynamically transformed and compared for several times, the dynamic change comparison method includes rotating the angle of the food block direction calibration template and translating the position of the food block direction calibration template, so that the standard outline of the food block direction calibration template and the edge outline are overlapped, and comparing whether the grayscale values ​​of the grayscale correction points are the same, if they are the same, the direction indicated by the food block direction calibration template at this time is the direction feature of the edge outline; A food standard grayscale distribution template is set for the edge contour. Based on the determined directional characteristics of the edge contour, the food standard grayscale distribution template is adapted to the edge contour for comparison, the grayscale distribution characteristics within the edge contour and the grayscale distribution matching parameters of the food standard grayscale distribution template are determined, and based on the distribution matching parameter interval to which the grayscale distribution matching parameters belong, the degree of appearance abnormality of the monomer food block corresponding to the edge contour is determined.

[0008] In some embodiments disclosed in the present invention, the method for determining the grayscale distribution characteristics within the edge contour and the grayscale distribution matching parameters of the food standard grayscale distribution template includes: The food standard grayscale distribution template includes a number of contour blocks, and each contour block is set with a standard grayscale value. The food standard grayscale distribution template is compared with the corresponding contour block in the edge contour, and the grayscale value difference corresponding to each contour block is determined; Based on the gray value differences corresponding to all the blocks within the contour, the gray distribution characteristics within the edge contour and the gray distribution matching parameters of the food standard gray distribution template are determined; Among them, the expression for calculating the grayscale distribution matching parameter is: ; Among them, F is the grayscale distribution matching parameter, is the number of blocks within the contour whose grayscale value difference is less than or equal to the preset value. is the total number of blocks within the contour, f is the contour block number ratio impact adjustment coefficient, c is the contour block number ratio impact adjustment constant, The preset maximum grayscale value difference, is the gray value difference of the x-th contour block.

[0009] In some embodiments disclosed in the present invention, the method for performing interval analysis on the classified production monitoring parameter groups includes: A parameter corresponding horizontal axis is constructed, the parameter mapping quantity corresponding to each parameter is determined, a mapping quantity vertical axis is constructed for the parameter mapping quantity and is perpendicular to the parameter corresponding horizontal axis, and a parameter mapping quantity change curve is constructed based on the parameters in a number of production monitoring parameter groups, the midpoint of the parameter mapping quantity change curve is determined, the midpoint of the curve is used as the starting point, the interval critical points are gradually pushed to both ends, and the cumulative sum of the first curve values ​​between the interval critical points and the cumulative sum of the second curve values ​​outside the critical points are calculated; Calculate the cumulative sum ratio of the cumulative sum of the first curve values ​​and the cumulative sum of the second curve values. If the cumulative sum ratio is greater than or equal to the preset value, lock the position of the interval critical point, and identify the interval between the interval critical points at this time as the first standard production operation parameter interval or the second standard production operation parameter interval.

[0010] In some embodiments disclosed in the present invention, the method of comparing the production monitoring parameter groups corresponding to other production serial numbers with the first standard production operation parameter interval and the second standard production operation parameter interval respectively includes: Compare the production equipment operation parameters of the production monitoring parameter group with the corresponding first standard production operation parameter interval, and construct a first adaptation operator based on the comparison result; compare the monitoring video feature parameters with the corresponding second standard production operation parameter interval, and construct a second adaptation operator based on the comparison result; Determining a production status evaluation value based on the first adaptation operator and the second adaptation operator; Among them, the expression for calculating the production status evaluation value is: ; Where P is the production status assessment value, is the adaptation parameter of the i-th parameter, is the preset influence weight coefficient of the i-th parameter, and N is the total number of production equipment operating parameters and monitoring video feature parameters in the production monitoring parameter group; Among them, determine the adaptation parameters The method includes determining whether the parameter belongs to the standard production operation parameter range, and if so, Output 1, if not, then Output 0.

[0011] In some embodiments disclosed in the present invention, a traceability management system based on food quality detection is also disclosed, including: The first module is used to construct a product production serial number for each product, and the production serial number includes production time, production batch, production sequence in the production batch and the production step to which it belongs, and a production monitoring parameter group is constructed for each production serial number, and the production monitoring parameter group includes several production equipment operation parameters and monitoring video feature parameters; The second module is used to classify the production monitoring parameter group based on the production batch and production time in the production serial number, and perform interval analysis on the classified production monitoring parameter group to determine the first standard production operation parameter interval of the production equipment operation parameter and the second standard production operation parameter interval of the monitoring video feature parameter; The third module is used to conduct random inspections on finished foods, identify foods that do not meet the quality standards, and based on the production serial numbers of the foods that do not meet the quality standards, determine other production serial numbers with similar production times and production batches, and compare the production monitoring parameter groups corresponding to the other production serial numbers with the first standard production operation parameter interval and the second standard production operation parameter interval, respectively, to calculate the production status evaluation value; if the production status evaluation value is less than or equal to the preset value, the corresponding production serial number is marked as abnormal, and the corresponding food is retrieved for quality inspection.

[0012] The present invention discloses a traceability management method and system based on food quality detection, which relates to the technical field of food quality detection. A unique production serial number is assigned to each product, and dimensional information such as production time, batch, and process node are integrated to synchronously associate equipment operation parameters and video feature parameters. Based on batch and time clustering data, standardized operation intervals of production parameters are dynamically counted to generate adaptive thresholds of equipment parameters and dynamic fitting intervals of video features, thereby eliminating manual setting deviations. After unqualified products are found in random inspections, spatiotemporal neighboring batches are retrieved according to their serial numbers, and the deviations of monitoring parameters of each batch from the standard interval are compared to calculate the weighted production status evaluation value. When the value is lower than the set threshold, high-risk serial numbers are automatically marked and a targeted recall is triggered. The above technical solution of the present invention realizes the integration of multi-source data for comprehensive analysis, thereby realizing faster and more accurate positioning of other foods with abnormal detection quality.

[0013] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a method step diagram of the traceability management method based on food quality detection disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0016] The following will be combined with the accompanying drawings and specific embodiments to clearly and completely describe the technical solution of the present invention. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention and cannot be understood as limiting the scope of protection of the present invention. Those skilled in the art in this field can make some non-essential improvements and adjustments based on the content of the present invention described below. In the present invention, unless otherwise clearly specified and limited, the technical terms used in the present invention should be the common meanings understood by the technical personnel described in the present invention. Example:

[0017] The purpose of the present invention is to provide a traceability management method and system that can trace and determine the source of food with abnormal quality in combination with multi-source data analysis.

[0018] The present invention discloses a traceability management method based on food quality detection, see Figure 1 ,include: Step S100, construct a product production serial number for each product, and the production serial number includes the production time, production batch, production sequence in the production batch and the corresponding production steps, and construct a production monitoring parameter group for each production serial number, the production monitoring parameter group includes several production equipment operating parameters and monitoring video feature parameters.

[0019] Step S100 realizes refined management and quality traceability of the entire production process by constructing a multi-source data association system with a composite coding structure. First, a unique production serial number is generated for each product. The coding content integrates four-dimensional elements: production time (specific time to microsecond accuracy), batch number (covering production line number, date serial number, etc.), sequence code (sequential number within the batch) and the production step identifier (such as cleaning, sorting, sterilization and other process nodes). This serial number uses hash encryption technology to form a uniquely indexed database primary key to ensure that a single product and its corresponding process node can be accurately located during traceability. At the same time, each serial number is mapped to a production monitoring parameter group, which consists of two parts: equipment operation parameters and video feature parameters.

[0020] The equipment operating parameters are directly collected from the sensors and control units of each production equipment, covering dozens of dynamic indicators such as the water flow pressure and spray time of the cleaning machine, the vibration frequency and color sorting accuracy of the sorting machine, the hot air temperature and humidity of the dryer, the pressure and particle size of the homogenizer, etc., forming multi-dimensional equipment status time series data.

[0021] Video feature parameters are extracted through machine vision technology, including: Movement rate of individual food blocks: Use a visual analysis algorithm based on optical flow to track the movement trajectory and speed of food on the production line and identify transmission anomalies (such as blockages or sudden speed changes).

[0022] Abnormality of appearance: Use contour edge detection technology (such as Canny operator) to delineate the contour of a single food, calibrate its posture through dynamic alignment (rotation and translation transformation) with the direction calibration template, and quantify surface defects (such as packaging wrinkles, label defects) based on the grayscale distribution matching parameter (F value, the higher the F value, the lower the abnormality). Among them, grayscale matching is achieved by normalizing the difference in the feature values ​​of the standard template and the detection area, and it is judged as abnormal when the cumulative difference exceeds the threshold.

[0023] The core of this step is the structured integration of spatiotemporal-process multidimensional data: using serial numbers as a link, discrete equipment operation data, continuous visual feature data and spatiotemporal information of production nodes are bound to construct a full life cycle data unit for a single product. This provides basic data structure support for subsequent abnormal detection and refined recall based on dynamic parameter intervals (steps S200-S300), thereby upgrading traditional batch-based extensive quality control to precise traceability at the process level or even the equipment level.

[0024] In some embodiments disclosed in the present invention, the operating parameters of the production equipment include: operating parameters of the cleaning machine, including water flow pressure, water temperature, and spraying time; operating parameters of the sorting machine, including vibration frequency, color sorting accuracy, and rejection rate; operating parameters of the dryer, including hot air temperature, humidity, and material residence time; operating parameters of the mixer, including speed, torque, current load, and material temperature; operating parameters of the homogenizer, including pressure value, number of cycles, and particle size; operating parameters of the fermentation tank, including temperature, pH value, dissolved oxygen concentration, and stirring rate; operating parameters of the pasteurizer, including temperature curve, holding time, and flow rate; operating parameters of the UHT sterilizer, including instantaneous temperature, cooling rate, and steam pressure; operating parameters of the microwave sterilizer, including power density, frequency, and material penetration depth; operating parameters of the canning machine, including canning quantity error, liquid level height, and sealing pressure; operating parameters of the vacuum sealing machine, including vacuum degree, sealing temperature, and residual oxygen; operating parameters of the labeling machine, including positioning accuracy, glue coating thickness, and label adhesion rate.

[0025] In some embodiments disclosed in the present invention, the method for determining the characteristic parameters of a surveillance video includes: Step S101, delineate the food assembly line area on the video image, and use visual analysis technology to lock the individual food blocks in the food assembly line area, and determine the movement rate of each individual food block and the degree of abnormality of the appearance of the individual food block, and identify the movement rate of the individual food block and the degree of abnormality of the appearance of the individual food block as monitoring video feature parameters.

[0026] In some embodiments disclosed in the present invention, the method of determining the degree of abnormality of the appearance of a single food block using visual analysis technology includes: Step S1011, perform contour edge detection on the food flow area, determine a number of contour edge lines, and based on the distance between the ends of the contour edge lines, combine different contour edge lines and perform head-to-tail association to form the edge contour of a single food block.

[0027] Contour extraction and monomer segmentation (step S1011): Based on edge detection algorithms (such as the Canny operator), the initial contour lines of the food in the production line video are identified, and the broken edges are completed through endpoint clustering and region growing algorithms: the Euclidean distance between the edge line endpoints is calculated (the threshold is set to 5 pixels), and the non-closed contour lines are recursively spliced ​​to generate a monomer closed polygon that wraps the food (such as generating the outline of a milk carton). The contour topology is optimized through the minimum bounding rectangle (MBR) to eliminate external noise interference (such as the background of the conveyor belt is eliminated).

[0028] Step S1012, a food block direction correction template is set, wherein the food block direction correction template includes a standard outline of the food block and a plurality of grayscale correction points within the standard outline, and the food block direction correction template and the edge outline are dynamically transformed and compared for several times. The dynamic change comparison method includes rotating the angle of the food block direction correction template and translating the position of the food block direction correction template so that the standard outline of the food block direction correction template and the edge outline coincide with each other, and comparing whether the grayscale values ​​of the grayscale correction points are the same. If they are the same, the direction indicated by the food block direction correction template at this time is taken as the direction feature of the edge outline.

[0029] Direction correction and template matching (step S1012): Load a preset food standard direction template (such as the outline of a complete yogurt cup), and calibrate N key grayscale correction points (usually distributed in the label printing area) inside it. Dynamically adjust the template's rotation angle (1° step within ±30°) and translation offset (±20 pixels) through the iterative closest point (ICP) algorithm to maximize the overlap between the template contour and the detection contour (using the principle of minimizing the difference area). At the same time, verify the grayscale matching of the grayscale correction points (such as CMYK channel difference ≤10). If all key points pass the verification (such as 8 / 10 points meet), record the rotation transformation matrix at this time and output the standardized direction of the single food (horizontal rotation correction offset angle θ).

[0030] Step S1013, a food standard grayscale distribution template is set for the edge contour, and based on the determined directional characteristics of the edge contour, the food standard grayscale distribution template is adapted to the edge contour for comparison, the grayscale distribution characteristics within the edge contour and the grayscale distribution matching parameters of the food standard grayscale distribution template are determined, and based on the distribution matching parameter interval to which the grayscale distribution matching parameters belong, the degree of appearance abnormality of the monomer food block corresponding to the edge contour is determined.

[0031] Grayscale distribution defect detection (step S1013): For the single food after direction correction, a mask is superimposed to intercept its effective area. A block-based grayscale comparison strategy is applied.

[0032] In some embodiments disclosed in the present invention, the method for determining the grayscale distribution characteristics within the edge contour and the grayscale distribution matching parameters of the food standard grayscale distribution template includes: Step S10131, the food standard grayscale distribution template includes a plurality of contour blocks, and each contour block is set with a standard grayscale value, the food standard grayscale distribution template is compared with the corresponding contour block in the edge contour, and the grayscale value difference corresponding to each contour block is determined; Step S10132, based on the gray value differences corresponding to the blocks within all the contours, determining the gray distribution characteristics within the edge contour and the gray distribution matching parameters of the food standard gray distribution template; Among them, the expression for calculating the grayscale distribution matching parameter is: ; Among them, F is the grayscale distribution matching parameter, is the number of blocks within the contour whose grayscale value difference is less than or equal to the preset value. is the total number of blocks within the contour, f is the contour block number ratio impact adjustment coefficient, c is the contour block number ratio impact adjustment constant, The preset maximum grayscale value difference, is the gray value difference of the x-th contour block.

[0033] Step S200, classify the production monitoring parameter group based on the production batch and production time in the production serial number, and perform interval analysis on the classified production monitoring parameter group to determine the first standard production operation parameter interval of the production equipment operation parameters and the second standard production operation parameter interval of the monitoring video feature parameters.

[0034] Step S200 constructs a standardized parameter range that adapts to changes in the production environment through dynamic data modeling and parameter benchmark self-calibration strategies. First, based on the batch and time tags in the production serial number, the equipment operating parameters and video feature parameter groups under the same production conditions are systematically aggregated to form a data set cluster divided by time and space dimensions. The specific process includes the following two core operations: Equipment parameter interval calibration: Using the hierarchical sliding window statistical method, the equipment parameters in the same batch (such as the vibration frequency of the sorting machine and the pasteurization temperature) are grouped according to the time window (such as 30-minute slices), and the parameter distribution probability density function is calculated by kernel density estimation (KDE). Take the confidence interval with a confidence level of α=0.05 (such as the sterilization temperature interval [72.4℃, 75.8℃]) to cover the normal production data fluctuation range and replace the threshold set by manual experience. In order to address the attenuation of equipment performance, a dynamic calibration mechanism is introduced, and the Gaussian mixture model (GMM) is used to analyze the short-term (such as the past week) and long-term (nearly three months) distribution differences of the newly added batch data, and the boundary value of the confidence interval is automatically corrected.

[0035] Video feature interval modeling: Based on the dynamic baseline of visual features (such as the F value of the appearance abnormality of a single food) and the correlation between batches, a time series recursive model is established. The LSTM network is used to learn the feature fluctuation pattern in the continuous production process within the batch (such as the change pattern of the packaging integrity score over time), predict the feature threshold of the next period and evaluate the deviation tolerance (such as the grayscale matching tolerance ±10%); at the same time, the feature data of adjacent batches (such as the batch 30 minutes before the production line) are integrated to ensure the adaptability and robustness of the video features.

[0036] In some embodiments disclosed in the present invention, the method for performing interval analysis on the classified production monitoring parameter groups includes: Step S201, construct a horizontal axis corresponding to the parameters, determine the parameter mapping quantity corresponding to each parameter, construct a mapping quantity vertical axis perpendicular to the parameter corresponding horizontal axis, and construct a parameter mapping quantity change curve based on the parameters in a number of production monitoring parameter groups, determine the midpoint of the parameter mapping quantity change curve, take the midpoint of the curve as the starting point, gradually advance the interval critical points to both ends, calculate the cumulative sum of the first curve values ​​between the interval critical points and the cumulative sum of the second curve values ​​outside the critical points.

[0037] Step S202, calculate the cumulative sum ratio of the cumulative sum of the first curve values ​​and the cumulative sum of the second curve values. If the cumulative sum ratio is greater than or equal to a preset value, lock the interval critical point position, and identify the interval between the interval critical points at this time as the first standard production operation parameter interval or the second standard production operation parameter interval.

[0038] Step S300, conduct random inspection on finished food products, identify food with unqualified quality, and based on the production serial numbers of the unqualified food, determine other production serial numbers with similar production time and production batches, and compare the production monitoring parameter groups corresponding to the other production serial numbers with the first standard production operation parameter interval and the second standard production operation parameter interval, respectively, and calculate the production status evaluation value. If the production status evaluation value is less than or equal to the preset value, the corresponding production serial number is marked as abnormal, and the corresponding food is retrieved for quality inspection.

[0039] Step S300 achieves efficient tracing and accurate recall of quality issues based on spatiotemporal diffusion risk modeling and multimodal evidence fusion mechanism.

[0040] Proximal batch selection: parse the production serial number of the defective product, extract its production timestamp, batch number and process node identifier, and construct spatiotemporal proximity screening rules: Temporal proximity: batches produced on the same production line with a time interval ≤ H hours (e.g. 2 hours); Logical proximity: following the process sequence across processes (e.g. when a defective product is in the sterilization process, it is automatically associated with the batch of its predecessor cleaning process).

[0041] In some embodiments disclosed in the present invention, the method of comparing the production monitoring parameter groups corresponding to other production serial numbers with the first standard production operation parameter interval and the second standard production operation parameter interval respectively includes: Step S301, compare the production equipment operating parameters of the production monitoring parameter group with the corresponding first standard production operating parameter interval, and construct a first adaptation operator based on the comparison result; compare the monitoring video feature parameters with the corresponding second standard production operating parameter interval, and construct a second adaptation operator based on the comparison result.

[0042] Step S302: Determine a production status evaluation value based on the first adaptation operator and the second adaptation operator.

[0043] Among them, the expression for calculating the production status evaluation value is: ; Where P is the production status assessment value, is the adaptation parameter of the i-th parameter, is the preset influence weight coefficient of the i-th parameter, and N is the total number of production equipment operating parameters and monitoring video feature parameters in the production monitoring parameter group.

[0044] Among them, determine the adaptation parameters The method includes determining whether the parameter belongs to the standard production operation parameter range, and if so, Output 1, if not, then Output 0.

[0045] In some embodiments disclosed in the present invention, a traceability management system based on food quality detection is also disclosed, including: The first module is used to construct a product production serial number for each product, and the production serial number includes production time, production batch, production sequence in the production batch and the production step to which it belongs, and a production monitoring parameter group is constructed for each production serial number, and the production monitoring parameter group includes several production equipment operation parameters and monitoring video feature parameters; The second module is used to classify the production monitoring parameter group based on the production batch and production time in the production serial number, and perform interval analysis on the classified production monitoring parameter group to determine the first standard production operation parameter interval of the production equipment operation parameter and the second standard production operation parameter interval of the monitoring video feature parameter; The third module is used to conduct random inspections on finished foods, identify foods that do not meet the quality standards, and based on the production serial numbers of the foods that do not meet the quality standards, determine other production serial numbers with similar production times and production batches, and compare the production monitoring parameter groups corresponding to the other production serial numbers with the first standard production operation parameter interval and the second standard production operation parameter interval, respectively, to calculate the production status evaluation value; if the production status evaluation value is less than or equal to the preset value, the corresponding production serial number is marked as abnormal, and the corresponding food is retrieved for quality inspection.

[0046] The present invention discloses a traceability management method and system based on food quality detection, which relates to the technical field of food quality detection. A unique production serial number is assigned to each product, and dimensional information such as production time, batch, and process node are integrated to synchronously associate equipment operation parameters and video feature parameters. Based on batch and time clustering data, standardized operation intervals of production parameters are dynamically counted to generate adaptive thresholds of equipment parameters and dynamic fitting intervals of video features, thereby eliminating manual setting deviations. After unqualified products are found in random inspections, spatiotemporal neighboring batches are retrieved according to their serial numbers, and the deviations of monitoring parameters of each batch from the standard interval are compared to calculate the weighted production status evaluation value. When the value is lower than the set threshold, high-risk serial numbers are automatically marked and a targeted recall is triggered. The above technical solution of the present invention realizes the integration of multi-source data for comprehensive analysis, thereby realizing faster and more accurate positioning of other foods with abnormal detection quality.

[0047] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present invention can be implemented by hardware, or by software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present invention.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A traceability management method based on food quality testing, characterized in that: include: A product production serial number is constructed for the production of each product, and the production serial number includes the production time, production batch, production sequence in the production batch and the production step to which it belongs, and a production monitoring parameter group is constructed for each production serial number, and the production monitoring parameter group includes several production equipment operation parameters and monitoring video feature parameters; Based on the production batch and production time in the production serial number, the production monitoring parameter group is classified, and the classified production monitoring parameter group is analyzed in intervals to determine the first standard production operation parameter interval of the production equipment operation parameter and the second standard production operation parameter interval of the monitoring video feature parameter; Finished foods are sampled and foods that do not meet quality standards are identified. Based on the production serial numbers of the foods that do not meet quality standards, other production serial numbers with similar production times and production batches are determined. The production monitoring parameter groups corresponding to other production serial numbers are compared with the first standard production operation parameter range and the second standard production operation parameter range, respectively, and a production status assessment value is calculated. If the production status assessment value is less than or equal to a preset value, the corresponding production serial number is marked as abnormal, and the corresponding food is retrieved for quality inspection.

2. The traceability management method based on food quality detection according to claim 1 is characterized in that: Production equipment operating parameters include: Cleaning machine operating parameters, including water pressure, water temperature, and spray time; Sorting machine operating parameters, including vibration frequency, color sorting accuracy, and rejection rate; Dryer operating parameters, including hot air temperature, humidity, and material residence time; Mixer operating parameters, including speed, torque, current load, and material temperature; Homogenizer operating parameters, including pressure value, number of cycles, and particle size; Fermentation tank operating parameters, including temperature, pH, dissolved oxygen concentration, and agitation rate; Pasteurizer operating parameters, including temperature profile, holding time, flow rate; UHT sterilizer operating parameters, instantaneous temperature, cooling rate, steam pressure; Microwave sterilizer operating parameters, power density, frequency, material penetration depth; Canning machine operating parameters, canning volume error, liquid level height, sealing pressure; Vacuum sealer operating parameters, vacuum degree, sealing temperature, residual oxygen; Labeling machine operating parameters, positioning accuracy, glue coating thickness, label adhesion rate.

3. The traceability management method based on food quality detection according to claim 1 is characterized in that: Methods for determining characteristic parameters of surveillance video include: The food assembly line area on the video image is delineated, and the visual analysis technology is used to lock the individual food blocks in the food assembly line area, and the movement rate of each individual food block and the degree of abnormality in the appearance of the individual food block are determined. The movement rate of the individual food block and the degree of abnormality in the appearance of the individual food block are identified as monitoring video feature parameters.

4. The traceability management method based on food quality detection according to claim 3 is characterized in that: Methods for determining the degree of abnormality in the appearance of individual food blocks using visual analysis technology include: Performing contour edge detection on the food flow area, determining several contour edge lines, and combining different contour edge lines based on the distance between the ends of the contour edge lines, and performing head-to-tail association to form the edge contour of a single food block; A food block direction calibration template is set, the food block direction calibration template includes a standard outline of the food block and a plurality of grayscale correction points in the standard outline, the food block direction calibration template and the edge outline are dynamically transformed and compared for several times, the dynamic change comparison method includes rotating the angle of the food block direction calibration template and translating the position of the food block direction calibration template, so that the standard outline of the food block direction calibration template and the edge outline are overlapped, and comparing whether the grayscale values ​​of the grayscale correction points are the same, if they are the same, the direction indicated by the food block direction calibration template at this time is the direction feature of the edge outline; A food standard grayscale distribution template is set for the edge contour. Based on the determined directional characteristics of the edge contour, the food standard grayscale distribution template is adapted to the edge contour for comparison, the grayscale distribution characteristics within the edge contour and the grayscale distribution matching parameters of the food standard grayscale distribution template are determined, and based on the distribution matching parameter interval to which the grayscale distribution matching parameters belong, the degree of appearance abnormality of the monomer food block corresponding to the edge contour is determined.

5. The traceability management method based on food quality detection according to claim 4 is characterized in that: The method for determining the grayscale distribution characteristics within the edge contour and the grayscale distribution matching parameters of the food standard grayscale distribution template includes: The food standard grayscale distribution template includes a number of contour blocks, and each contour block is set with a standard grayscale value. The food standard grayscale distribution template is compared with the corresponding contour block in the edge contour, and the grayscale value difference corresponding to each contour block is determined; Based on the gray value differences corresponding to all the blocks within the contour, the gray distribution characteristics within the edge contour and the gray distribution matching parameters of the food standard gray distribution template are determined; Among them, the expression for calculating the grayscale distribution matching parameter is: ； Among them, F is the grayscale distribution matching parameter, is the number of blocks within the contour whose grayscale value difference is less than or equal to the preset value. is the total number of blocks within the contour, f is the contour block number ratio impact adjustment coefficient, c is the contour block number ratio impact adjustment constant, The preset maximum grayscale value difference, is the gray value difference of the x-th contour block.

6. The traceability management method based on food quality detection according to claim 1 is characterized in that: The method of performing interval analysis on the classified production monitoring parameter groups includes: A parameter corresponding horizontal axis is constructed, the parameter mapping quantity corresponding to each parameter is determined, a mapping quantity vertical axis is constructed for the parameter mapping quantity and is perpendicular to the parameter corresponding horizontal axis, and a parameter mapping quantity change curve is constructed based on the parameters in a number of production monitoring parameter groups, the midpoint of the parameter mapping quantity change curve is determined, the midpoint of the curve is used as the starting point, the interval critical points are gradually pushed to both ends, and the cumulative sum of the first curve values ​​between the interval critical points and the cumulative sum of the second curve values ​​outside the critical points are calculated; Calculate the cumulative sum ratio of the cumulative sum of the first curve values ​​and the cumulative sum of the second curve values. If the cumulative sum ratio is greater than or equal to the preset value, lock the position of the interval critical point, and identify the interval between the interval critical points at this time as the first standard production operation parameter interval or the second standard production operation parameter interval.

7. The video detection-based traceability management method according to claim 1 is characterized in that: The method of comparing the production monitoring parameter groups corresponding to other production serial numbers with the first standard production operation parameter interval and the second standard production operation parameter interval respectively includes: Compare the production equipment operation parameters of the production monitoring parameter group with the corresponding first standard production operation parameter interval, and construct a first adaptation operator based on the comparison result; compare the monitoring video feature parameters with the corresponding second standard production operation parameter interval, and construct a second adaptation operator based on the comparison result; Determining a production status evaluation value based on the first adaptation operator and the second adaptation operator; Among them, the expression for calculating the production status evaluation value is: ； Where P is the production status assessment value, is the adaptation parameter of the i-th parameter, is the preset influence weight coefficient of the i-th parameter, and N is the total number of production equipment operating parameters and monitoring video feature parameters in the production monitoring parameter group; Among them, determine the adaptation parameters The method includes determining whether the parameter belongs to the standard production operation parameter range, and if so, Output 1, if not, then Output 0.

8. The traceability management system based on food quality inspection is characterized by: The method for implementing the traceability management method according to any one of claims 1 to 7 comprises: The first module is used to construct a product production serial number for each product, and the production serial number includes production time, production batch, production sequence in the production batch and the production step to which it belongs, and a production monitoring parameter group is constructed for each production serial number, and the production monitoring parameter group includes several production equipment operation parameters and monitoring video feature parameters; The second module is used to classify the production monitoring parameter group based on the production batch and production time in the production serial number, and perform interval analysis on the classified production monitoring parameter group to determine the first standard production operation parameter interval of the production equipment operation parameter and the second standard production operation parameter interval of the monitoring video feature parameter; The third module is used to conduct random inspections on finished foods, identify foods that do not meet the quality standards, and based on the production serial numbers of the foods that do not meet the quality standards, determine other production serial numbers with similar production times and production batches, and compare the production monitoring parameter groups corresponding to the other production serial numbers with the first standard production operation parameter interval and the second standard production operation parameter interval, respectively, to calculate the production status evaluation value; if the production status evaluation value is less than or equal to the preset value, the corresponding production serial number is marked as abnormal, and the corresponding food is retrieved for quality inspection.

Images