A meat product safety monitoring system based on big data

Abstract

This invention discloses a meat product safety monitoring system based on big data, belonging to the field of meat product production monitoring technology. This system monitors water activity, microbial content, oxidation level, and stain area in each production stage in real time, enabling timely identification of potential safety hazards and rapid implementation of corresponding measures to significantly reduce the risk of corrosion and spoilage. Through in-depth analysis of the difference dataset, the system can identify key links affecting meat product quality, thereby optimizing production processes and reducing unnecessary steps. Furthermore, when the system detects that the comprehensive diffusion index Ksz exceeds the risk threshold R, it can quickly generate early warning information and initiate control strategies to adjust production conditions in a timely manner to reduce the impact of corrosion diffusion on the product. By creating and sorting priority lists, the system allows managers to quickly identify the highest-risk monitoring sub-areas, thus concentrating resources and attention on the links requiring the most intervention.

Description

Technical Field

[0001] The present invention relates to the technical field of meat product production monitoring, and in particular to a meat product safety monitoring system based on big data. Background Art

[0002] The processing and production of meat products is a highly complex process involving multiple key links, including raw material selection, processing, cutting, marinating, aging, packaging, etc. In this process, ensuring product safety and improving processing efficiency are crucial. Although traditional meat production technology has been developed for many years, it still has obvious shortcomings in ensuring food safety, optimizing production processes and improving product quality.

[0003] Especially in the raw material processing stage, traditional technologies often rely on experience for selection and processing, and lack systematic quality monitoring. This approach increases the risk of microbial contamination and affects the safety and quality of the final product. In the processing area, the quality of the initial processing of meat directly affects the effect of subsequent cutting, and the water activity is an important factor in the growth of microorganisms. If the moisture content of the raw materials is too high, the cut meat pieces are prone to corruption, which in turn affects the quality of the entire batch of products. During the pickling process, the penetration time of salt and seasonings is also crucial to the pickling effect, and changes in water activity will significantly affect the uniformity and effect of this process, causing corruption in certain parts and damaging the overall quality. Summary of the invention

[0004] In view of the deficiencies of the prior art, the present invention provides a meat product safety monitoring system based on big data to solve the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A meat product safety monitoring system based on big data, comprising:

[0006] The monitoring area segmentation unit is used to segment the meat product production monitoring area into a number of monitoring sub-areas according to the sequential production process; the monitoring sub-areas include the raw material processing area, cutting area, pickling area, maturation area, packaging area and refrigeration area;

[0007] The first detection unit is used to collect the water activity value A in each monitoring sub-area w 、Microbial content M c , peroxide value O x and the spot area S p , and establish water activity data set, microbial data set, oxidation degree data set and stain data set;

[0008] A difference calculation unit is used to calculate the difference in water activity, microbial content, oxidation degree and spot area between adjacent process monitoring sub-areas to form a difference data set;

[0009] An analysis unit is used to construct and train a prediction and recognition model using a convolutional neural network (CNN), and to analyze and evaluate the difference data set to obtain: a comprehensive diffusion index Ksz, and to perform risk assessment on the comprehensive diffusion index Ksz. If the comprehensive diffusion index Ksz exceeds a risk threshold R, a first warning instruction is sent;

[0010] The strategy generation unit, after receiving the first warning instruction, converts the water activity difference in the comprehensive diffusion index Ksz into , Difference in microbial content , oxidation degree difference and the difference in patch area Evaluations are conducted separately to trigger corresponding warning information, and priority is ranked from high to low according to the corresponding values ​​of the corresponding warning information; and corresponding control strategies are generated according to the water activity warning information, microbial content warning information, oxidation degree warning information and stain warning information in the corresponding warning information.

[0011] Preferably, the first detection unit includes a water activity detection module, a microbial content detection module, an oxidation degree detection module and a stain area detection module;

[0012] The water activity detection module is used to install a water activity sensor in each monitoring sub-area to collect the water activity value A of the monitoring sub-area in real time. w , and establish a water activity data set;

[0013] The microbial content detection module is used to install microbial detection equipment in each monitoring sub-area to monitor the microbial content Mc and microbial content M in the monitoring sub-area in real time. c It refers to the concentration of E. coli on the surface of meat products and establishes a microbiological data set;

[0014] The oxidation degree detection module is used to install an electrochemical sensor in each monitoring sub-area to measure the peroxide value O of the monitoring sub-area in real time. x , and establish a data set of oxidation degree;

[0015] The spot area detection module is used to capture the surface image of the meat product in the monitoring sub-area in real time through the image acquisition device, and obtain the spot area S by using the image processing algorithm analysis p , and establish a speckle dataset.

[0016] Preferably, the spot area detection module includes an image acquisition submodule, an image segmentation submodule and an area calculation submodule;

[0017] The image acquisition submodule is used to install image acquisition equipment in each monitoring sub-area, and to take real-time images of the surface of meat products in the monitoring sub-area through the image acquisition equipment;

[0018] The image segmentation submodule is used to convert the meat product surface image into a grayscale image and obtain a binary image through binarization processing. , the specific steps are:

[0019] S11, counting the number of pixels of each gray value in the gray image, generating a histogram, and recording the number of pixels h(i) with gray value i in the histogram;

[0020] S12, calculate the total number of pixels N, ; h(i) represents the number of pixels with gray value i in the histogram;

[0021] S13, applying Otsu's threshold segmentation method, pre-setting the traversal threshold t, maximizing the inter-class variance of the foreground and background after dividing the image by the traversal threshold t, and calculating the foreground probability P1(t) and background probability P2(t) by the following formula:

[0022] ;

[0023] ;

[0024] Where h(i) represents the number of pixels with grayscale value i in the histogram; t represents the traversal threshold, and pixels with grayscale value less than or equal to t are regarded as foreground; foreground probability P1(t) represents the proportion of pixels with grayscale value less than or equal to t to the total pixels, and background probability P2(t) represents the proportion of pixels with grayscale value greater than t to the total pixels;

[0025] S14. Calculate the foreground mean μ1(t) and background mean μ2(t) by the following formula:

[0026] ;

[0027] ;

[0028] In the formula, represents the weighted sum of all gray values ​​i, representing the total brightness of the foreground pixels, h(i) represents the number of pixels with gray value i in the histogram, Represents the total number of foreground pixels; represents the weighted sum of all gray values ​​i, which represents the total brightness of the background pixels; Represents the total number of background pixels;

[0029] S15. Based on the foreground probability P1(t), the background probability P2(t), the foreground mean μ1(t) and the background mean μ2(t), the inter-class variance is calculated using the following formula: :

[0030] ;

[0031] Among them, the larger the inter-class variance, the more obvious the difference between the foreground and the background. When setting each traversal threshold t, if the current inter-class variance If it is greater than the previously recorded maximum inter-class variance, the maximum inter-class variance value and the corresponding optimal threshold X are updated; B represents the pixel value of the background area in the binary image when it is not segmented;

[0032] S16, applying the optimal threshold value X to segment and obtain a binary image :

[0033] ;

[0034] In the formula, Represents the binary image after segmentation, a value of 1 represents the spot foreground area, and a value of 0 represents the background area; Represents the statistics of the pixel value with gray value i in region j;

[0035] The area calculation submodule is used to calculate the area of ​​the binary image. , according to the physical size of each pixel, it is converted into the spot area S by the following formula p :

[0036] ;

[0037] In the formula, is the total number of pixels with value 1 in the binary image, Represents the actual area of ​​each pixel and is calculated based on the number of spot areas S p Create a speckle dataset.

[0038] Preferably, the difference data set includes:

[0039] Water activity difference : ;

[0040] In the formula, represents the water activity value of the k+1th monitoring sub-area, represents the water activity value of the kth monitoring sub-area;

[0041] Difference in microbial content : ;

[0042] In the formula, represents the microbial content value of the k+1th monitoring sub-area, represents the microbial content value of the kth monitoring sub-area;

[0043] Oxidation degree difference : ;

[0044] In the formula, represents the peroxide value of the k+1th monitoring sub-area, represents the peroxide value of the kth monitoring sub-area;

[0045] Patch area difference : ;

[0046] In the formula, represents the spot area value of the k+1th monitoring sub-region, Represents the spot area value of the kth monitoring sub-area.

[0047] Preferably, the analysis unit includes a model building module, a comprehensive analysis module and a first evaluation module;

[0048] After building a model module to preprocess and normalize the difference data set, a convolutional neural network (CNN) is used to build a prediction and recognition model.

[0049] The normalized difference data set is used to train the prediction recognition model; the difference data set is divided into 80% training set, 10% validation set and 10% test set, and the prediction recognition model is trained, validated and tested;

[0050] The comprehensive analysis module is used to extract the water activity difference corresponding to the same monitoring sub-area in the difference data set , Difference in microbial content , oxidation degree difference and the difference in patch area , after dimensionless processing, the comprehensive diffusion index Ksz is generated by the following formula:

[0051] ;

[0052] In the formula, , , and Represents the water activity difference , Difference in microbial content , oxidation degree difference and the difference in patch area is the weight coefficient, and C represents the correction constant.

[0053] Preferably, the first evaluation module is used to preset a risk threshold R, and compare the comprehensive diffusion index Ksz with the risk threshold R to obtain a first evaluation result, including:

[0054] When the comprehensive diffusion index Ksz>risk threshold R, it means that there is a risk of diffusion corrosion of meat products in the current monitoring sub-area, and the first warning instruction is generated;

[0055] When the comprehensive diffusion index Ksz ≤ risk threshold R, it means that there is no diffusion corrosion risk for meat products in the current monitoring sub-area.

[0056] Preferably, the strategy generation unit includes a second evaluation module and a priority module;

[0057] The second evaluation module is used to calculate the water activity difference in the comprehensive diffusion index Ksz after receiving the first warning instruction. , Difference in microbial content , oxidation degree difference and the difference in patch area Conduct assessments respectively to obtain second assessment results, including:

[0058] Preset a first difference threshold, a second difference threshold, a third difference threshold and a fourth difference threshold;

[0059] When the water activity difference > The first difference threshold indicates that the water activity exceeds the normal range, triggering a water activity warning message; when the water activity difference ≤ the first difference threshold, indicating that the water activity does not exceed the normal range and continues to be monitored;

[0060] When the difference in microbial content > The second difference threshold indicates that the microbial content exceeds the normal range, triggering a microbial content warning message; when the microbial content difference ≤ the second difference threshold, indicating that the microbial content does not exceed the normal range and continues to be monitored;

[0061] When the oxidation degree difference > The third difference threshold indicates that the oxidation degree exceeds the safety level, triggering the oxidation degree warning information; when the oxidation degree difference ≤ the third difference threshold, indicating that the degree of oxidation does not exceed the safe level and is continuously monitored;

[0062] When the difference in patch area > The fourth difference threshold indicates that the spot area increases beyond the normal range, triggering the spot warning information; when the spot area difference ≤ the fourth difference threshold, indicating that the increased area of ​​the spot does not exceed the normal range and continues to be monitored.

[0063] Preferably, the priority module is used to create a corresponding list and divide the water activity difference into , Difference in microbial content , oxidation degree difference and the difference in patch area Stored in the corresponding list, and including the corresponding monitoring sub-area identifier, including the k-th monitoring sub-area and the k+1-th monitoring sub-area;

[0064] Sort the list from large to small to obtain a priority list to identify which monitoring sub-area has a greater risk of spread; when the water activity warning information, microbial content warning information, oxidation degree warning information and spot warning information are triggered, the kth monitoring sub-area and the k+1th monitoring sub-area are marked synchronously.

[0065] Preferably, the strategy generation unit further includes a strategy generation module and a tracing module;

[0066] The strategy generation module is used to generate corresponding strategies based on water activity warning information, microbial content warning information, oxidation degree warning information and spot warning information, including:

[0067] Generate a first strategy based on the water activity warning information, including: regulating the temperature of the kth monitoring sub-area and the k+1th monitoring sub-area to ensure that the storage temperature of the meat products is below 4°C, and the storage temperature difference between the kth monitoring sub-area and the k+1th monitoring sub-area does not exceed ±3°C; regulating the humidity of the kth monitoring sub-area and the k+1th monitoring sub-area to ensure that the relative humidity difference between the kth monitoring sub-area and the k+1th monitoring sub-area does not exceed ±5%;

[0068] The second strategy is generated based on the microbial content warning information, including: testing the microbial content of the kth monitoring sub-area and the k+1th monitoring sub-area, if the E. coli concentration is ≤100 CFU / g, no regulation is performed, once the E. coli concentration is greater than 100 CFU / g, it is indicated that the microorganism exceeds the standard, and the disinfection procedure is initiated;

[0069] A third strategy is generated according to the oxidation degree warning information, including: replacing the kth monitoring sub-area and the k+1th monitoring sub-area with inert gas, including nitrogen and carbon dioxide, to reduce the oxygen concentration in the current monitoring sub-area to below 5%, and ensuring the sealing of the monitoring sub-area to prevent the inflow of external air;

[0070] The fourth strategy is generated according to the spot warning information, including: the spot area difference between the kth monitoring sub-area and the k+1th monitoring sub-area > The meat products with the fourth difference threshold are isolated to avoid cross infection. The same batch of meat products are processed by independent cutting, pickling and packaging equipment, and the corresponding monitoring sub-areas are disinfected. p When the spot is less than 2% of the total area of ​​the meat product, use a knife to remove it;

[0071] For the spot area S pIf it is greater than or equal to 2% of the total area of ​​meat products, it will be scrapped.

[0072] Preferably, the traceability module is used to record the production date, production batch, and raw material source of each batch of meat products; and to record in detail the monitoring process, processing time, and equipment usage data of the kth monitoring sub-area;

[0073] The comprehensive diffusion index Ksz, water activity warning information, microbial content warning information, oxidation degree warning information and stain warning information as well as the corresponding strategies of each batch of meat products are saved, and a unique barcode is generated to facilitate quick query of relevant information of the same batch of meat products.

[0074] The present invention provides a meat product safety monitoring system based on big data, which has the following beneficial effects:

[0075] (1) This meat product safety monitoring system based on big data can detect potential safety hazards in a timely manner and take prompt measures to reduce the risk of corrosion and deterioration by real-time monitoring of the water activity, microbial content, oxidation degree and stain area of ​​each production link, thereby ensuring the safety of meat products. Based on the analysis of the difference data set, the key links that affect the quality of meat products can be identified, the production process can be optimized, unnecessary steps can be reduced, the overall production efficiency can be improved, and the production cost can be reduced.

[0076] (2) This meat product safety monitoring system based on big data can ensure the consistency of the quality of each batch of meat products by analyzing the differences between the monitoring sub-areas, thereby reducing the quality differences caused by fluctuations in the production process.

[0077] (3) This meat product safety monitoring system based on big data can quickly generate early warning information and initiate corresponding control strategies after the monitoring system detects that the comprehensive diffusion index Ksz exceeds the risk threshold R, ensuring timely adjustment of production conditions, thereby reducing the impact of corrosion diffusion on products and reducing losses.

[0078] (4) This meat product safety monitoring system based on big data can quickly identify the highest risk monitoring sub-areas by creating and sorting priority lists. This approach ensures that managers can focus resources and attention on areas that need intervention most, so that timely measures can be taken to reduce risks. The implementation of priority sorting enables managers to quickly locate problem areas when early warning information is triggered, improve the efficiency of responding to emergencies, and reduce potential losses. By marking monitoring sub-areas with greater risk of spread, companies can implement targeted control measures, such as strengthening cleaning and disinfection, adjusting processing technology, or increasing monitoring frequency, thereby improving product safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1This is a schematic diagram of the flow of a meat product safety monitoring system based on big data in the present invention. DETAILED DESCRIPTION

[0080] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0081] Example 1

[0082] See also Figure 1 The present invention provides a meat product safety monitoring system based on big data, comprising:

[0083] The monitoring area segmentation unit is used to segment the meat product production monitoring area into a number of monitoring sub-areas according to the sequential production process; the monitoring sub-areas include the raw material processing area, cutting area, pickling area, maturation area, packaging area and refrigeration area;

[0084] The first detection unit is used to collect the water activity value A in each monitoring sub-area w 、Microbial content M c , peroxide value O x And the spot area S p , and establish water activity data set, microbial data set, oxidation degree data set and stain data set;

[0085] A difference calculation unit is used to calculate the difference in water activity, microbial content, oxidation degree and spot area between adjacent process monitoring sub-areas to form a difference data set;

[0086] An analysis unit is used to construct and train a prediction and recognition model using a convolutional neural network (CNN), and to analyze and evaluate the difference data set to obtain: a comprehensive diffusion index Ksz, and to perform risk assessment on the comprehensive diffusion index Ksz. If the comprehensive diffusion index Ksz exceeds a risk threshold R, a first warning instruction is sent;

[0087] The strategy generation unit, after receiving the first warning instruction, converts the water activity difference in the comprehensive diffusion index Ksz into , Difference in microbial content , oxidation degree difference and the difference in patch area Evaluations are conducted separately to trigger corresponding warning information, and priority is ranked from high to low according to the corresponding values ​​of the corresponding warning information; and corresponding control strategies are generated according to the water activity warning information, microbial content warning information, oxidation degree warning information and stain warning information in the corresponding warning information.

[0088] In this embodiment, by dividing each link of meat product production into monitoring sub-areas (such as raw material processing area, cutting area, pickling area, etc.), the system can monitor each link in real time. This regional monitoring method enables each step of the production process to be accurately tracked, ensuring that potential problems are discovered in a timely manner. The first detection unit can collect key parameters such as water activity, microbial content, peroxide value and stain area in real time. The systematic management of these data not only improves the accuracy of monitoring, but also provides a basis for subsequent data analysis. Using convolutional neural network CNN to analyze the difference data set, abnormal situations can be effectively identified, and a comprehensive diffusion index Ksz can be generated to ensure a scientific assessment of the safety of meat products. By establishing a risk threshold, the system can issue an early warning in time when the comprehensive diffusion index Ksz exceeds the risk threshold R. This early warning mechanism helps to take quick measures to prevent the problem from further expanding, thereby reducing losses and potential health risks.

[0089] The strategy generation unit automatically generates corresponding control strategies based on monitoring data to ensure that environmental conditions such as temperature and humidity during the production process are always kept within the optimal range. For example, by adjusting storage conditions to control water activity and microbial content, the shelf life of products can be effectively extended, and the safety and market competitiveness of meat products can be improved. The data-driven monitoring system reduces reliance on manual experience, avoids human errors in traditional methods, and improves overall production efficiency. At the same time, real-time monitoring and rapid response mechanisms reduce stagnation time in the production process, ensuring a smoother production process.

[0090] Example 2

[0091] This example is an explanation of Example 1. Figure 1 ,Specifically, the first detection unit includes a water activity detection module, a microbial content detection module, an oxidation degree detection module and a stain area detection module;

[0092] The water activity detection module is used to install a water activity sensor in each monitoring sub-area to collect the water activity value A of the monitoring sub-area in real time. w , and establish a water activity data set;

[0093] The microbial content detection module is used to install microbial detection equipment in each monitoring sub-area to monitor the microbial content M of the monitoring sub-area in real time. c , microbial content Mc It refers to the concentration of E. coli on the surface of meat products and establishes a microbiological data set;

[0094] The oxidation degree detection module is used to install an electrochemical sensor in each monitoring sub-area, measure the peroxide value Ox of the monitoring sub-area in real time, and establish an oxidation degree data set; the electrochemical sensor detects peroxides in oils and fats through current changes or potential changes. When peroxides react with electrodes, a measurable electrical signal is generated. The electrodes in the sensor react with peroxides to generate current, and the magnitude of this current is related to the concentration of peroxides in the sample. Through precise measurement and analysis, the specific concentration of peroxides in the sample can be obtained, thereby evaluating the oxidation degree of oils and fats or food.

[0095] The spot area detection module is used to capture the surface image of the meat product in the monitoring sub-area in real time through the image acquisition device, and obtain the spot area S by using the image processing algorithm analysis p , and establish a speckle dataset.

[0096] In this embodiment, the installation of each monitoring module ensures the real-time collection of key parameters of the monitoring sub-area. The application of water activity sensors, microbial detection equipment, oxidation degree sensors and image acquisition equipment makes data collection more accurate and can timely reflect changes in the production process. Through real-time monitoring and rapid feedback, each link in the production process can be adjusted in time according to the test results, thereby reducing the production of unqualified products, improving production efficiency, and ensuring the consistency and stability of the final product.

[0097] Example 3

[0098] This example is explained in Example 2. Figure 1 ,Specifically, the spot area detection module includes an image acquisition submodule, an image segmentation submodule and an area calculation submodule;

[0099] The image acquisition submodule is used to install image acquisition equipment in each monitoring sub-area, and to take real-time images of the surface of meat products in the monitoring sub-area through the image acquisition equipment;

[0100] The image segmentation submodule is used to convert the meat product surface image into a grayscale image and obtain a binary image through binarization processing. , the specific steps are:

[0101] S11, counting the number of pixels of each gray value in the gray image, generating a histogram, and recording the number of pixels h(i) with gray value i in the histogram;

[0102] S12, calculate the total number of pixels N, ; h(i) represents the number of pixels with gray value i in the histogram;

[0103] S13, applying Otsu's threshold segmentation method, pre-setting the traversal threshold t, maximizing the inter-class variance of the foreground and background after dividing the image by the traversal threshold t, and calculating the foreground probability P1(t) and background probability P2(t) by the following formula:

[0104] ;

[0105] ;

[0106] Where h(i) represents the number of pixels with grayscale value i in the histogram; t represents the traversal threshold, and pixels with grayscale value less than or equal to t are regarded as foreground; foreground probability P1(t) represents the proportion of pixels with grayscale value less than or equal to t to the total pixels, and background probability P2(t) represents the proportion of pixels with grayscale value greater than t to the total pixels;

[0107] S14. Calculate the foreground mean μ1(t) and background mean μ2(t) by the following formula:

[0108] ;

[0109] ;

[0110] In the formula, represents the weighted sum of all gray values ​​i, representing the total brightness of the foreground pixels, h(i) represents the number of pixels with gray value i in the histogram, Represents the total number of foreground pixels; represents the weighted sum of all gray values ​​i, which represents the total brightness of the background pixels; Represents the total number of background pixels;

[0111] S15. Based on the foreground probability P1(t), the background probability P2(t), the foreground mean μ1(t) and the background mean μ2(t), the inter-class variance is calculated using the following formula: :

[0112] ;

[0113] Among them, the larger the inter-class variance, the more obvious the difference between the foreground and the background. When setting each traversal threshold t, if the current inter-class variance If it is greater than the previously recorded maximum inter-class variance, the maximum inter-class variance value and the corresponding optimal threshold X are updated; B represents the pixel value of the background area in the binary image when it is not segmented;

[0114] S16, applying the optimal threshold value X to segment and obtain a binary image :

[0115] ;

[0116] In the formula, Represents the binary image after segmentation, a value of 1 represents the spot foreground area, and a value of 0 represents the background area; Represents the statistics of the pixel value with gray value i in region j;

[0117] The area calculation submodule is used to calculate the area of ​​the binary image. , according to the physical size of each pixel, it is converted into the spot area S by the following formula p :

[0118] ;

[0119] In the formula, is the total number of pixels with value 1 in the binary image, Represents the actual area of ​​each pixel and is calculated based on the number of spot areas S p Create a speckle dataset.

[0120] In this embodiment, the automated image segmentation and threshold selection process reduces the need for manual intervention and improves detection efficiency. Through Otsu's threshold segmentation method, the system can automatically select the optimal threshold to ensure accurate extraction of the stain area. By counting the pixels of the binary image, the stain is converted into the actual area. , get the spot area S p , providing quantitative indicators for subsequent quality assessment. This indicator can directly reflect the surface quality of meat products and help to detect potential spoilage risks. The accumulated stain data set can be used for historical trend analysis to help companies identify the root causes of problems, optimize production processes and raw material selection, thereby improving overall production efficiency and product quality.

[0121] Example 4

[0122] This example is an explanation of Example 1. Figure 1 ,Specifically, the difference data set includes:

[0123] Water activity difference : ;

[0124] In the formula, represents the water activity value of the k+1th monitoring sub-area, represents the water activity value of the kth monitoring sub-area;

[0125] Difference in microbial content : ;

[0126] In the formula, represents the microbial content value of the k+1th monitoring sub-area, represents the microbial content value of the kth monitoring sub-area; the microbial content value is the concentration of the detected Escherichia coli concentration;

[0127] Oxidation degree difference : ;

[0128] In the formula, represents the peroxide value of the k+1th monitoring sub-area, represents the peroxide value of the kth monitoring sub-area;

[0129] Patch area difference : ;

[0130] In the formula, represents the spot area value of the k+1th monitoring sub-region, Represents the spot area value of the kth monitoring sub-area.

[0131] The difference data set makes it easy to see the changes in quality indicators during the production process. In some cases, the calculated difference may result in negative values, which may lead to misinterpretation of the data. Using absolute values ​​can eliminate this problem and ensure that all differences are positive, which is convenient for subsequent analysis. By analyzing the differences between different monitoring sub-areas, potential quality fluctuation trends can be identified, providing a basis for optimizing the production process. Monitoring the differences in different areas helps to identify the risk of cross-contamination. For example, if the difference in microbial content between the cutting area and the pickling area is abnormal, it means that there may be a risk of microorganisms being transferred from the cutting area to the pickling area, which can guide the implementation of prevention and control measures in a timely manner. Analyzing the difference data can reveal the bottleneck links in the production process. For example, if the difference in water activity is large, it means that the previous area is not properly handled, which may lead to uneven pickling in the subsequent links, thereby affecting product quality. This provides a specific direction for improving the production process. Through difference monitoring, companies can adjust resource allocation in a timely manner. For example, when the difference in oxidation level in a certain area is too high, it may be necessary to increase refrigeration resources or adjust the processing temperature to extend the shelf life of the product.

[0132] The analysis unit includes a model building module, a comprehensive analysis module and a first evaluation module;

[0133] After building a model module to preprocess and normalize the difference data set, a convolutional neural network (CNN) is used to build a prediction and recognition model.

[0134] The normalized difference data set is used to train the prediction recognition model; the difference data set is divided into 80% training set, 10% validation set and 10% test set, and the prediction recognition model is trained, validated and tested;

[0135] The comprehensive analysis module is used to extract the water activity difference corresponding to the same monitoring sub-area in the difference data set , Difference in microbial content , oxidation degree difference and the difference in patch area , after dimensionless processing, the comprehensive diffusion index Ksz is generated by the following formula:

[0136] ;

[0137] In the formula, , , and Represents the water activity difference , Difference in microbial content , oxidation degree difference and the difference in patch area The weight coefficient of , , , , The weight coefficient is also adjusted and set by the user according to the size of the factors affecting moisture, microorganisms, oxidation and stain area of ​​different types of meat products. C represents the correction constant.

[0138] The first assessment module is used to preset a risk threshold R, and compare the comprehensive diffusion index Ksz with the risk threshold R to obtain a first assessment result, including:

[0139] When the comprehensive diffusion index Ksz>risk threshold R, it means that there is a risk of diffusion corrosion of meat products in the current monitoring sub-area, and the first warning instruction is generated;

[0140] When the comprehensive diffusion index Ksz ≤ risk threshold R, it means that there is no diffusion corrosion risk for meat products in the current monitoring sub-area.

[0141] In this embodiment, the comprehensive analysis module extracts the difference data of different monitoring sub-areas and integrates them into the comprehensive diffusion index Ksz, so that the influence of various factors can be comprehensively considered. This multi-dimensional evaluation can fully reflect the quality status of the production process and avoid the one-sided judgment that may be caused by a single indicator. By monitoring and analyzing the comprehensive diffusion index Ksz, a more effective early warning mechanism can be established. When Ksz exceeds the set risk threshold R, the system can automatically issue an early warning to remind the operator to take corresponding measures.

[0142] Example 5

[0143] This example is an explanation of Example 1. Figure 1 ,Specifically, the strategy generation unit includes a second evaluation module and a priority ,module;

[0144] The second evaluation module is used to calculate the water activity difference in the comprehensive diffusion index Ksz after receiving the first warning instruction. , Difference in microbial content , oxidation degree difference and the difference in patch area Conduct assessments respectively to obtain second assessment results, including:

[0145] Preset a first difference threshold, a second difference threshold, a third difference threshold and a fourth difference threshold;

[0146] When the water activity difference > The first difference threshold indicates that the water activity exceeds the normal range, triggering a water activity warning message; when the water activity difference ≤ the first difference threshold, indicating that the water activity does not exceed the normal range and continues to be monitored;

[0147] When the difference in microbial content > The second difference threshold indicates that the microbial content exceeds the normal range, triggering a microbial content warning message; when the microbial content difference ≤ the second difference threshold, indicating that the microbial content does not exceed the normal range and continues to be monitored;

[0148] When the oxidation degree difference > The third difference threshold indicates that the oxidation degree exceeds the safety level, triggering the oxidation degree warning information; when the oxidation degree difference ≤ the third difference threshold, indicating that the degree of oxidation does not exceed the safe level and is continuously monitored;

[0149] When the difference in patch area > The fourth difference threshold indicates that the spot area increases beyond the normal range, triggering the spot warning information; when the spot area difference ≤ the fourth difference threshold, indicating that the increased area of ​​the spot does not exceed the normal range and continues to be monitored.

[0150] In this embodiment, by setting thresholds and conducting real-time monitoring, the module can issue early warning information in a timely manner when indicators such as water activity, microbial content, oxidation degree, and stain area are abnormal. This rapid response mechanism can effectively reduce product safety risks and ensure the quality of meat products. When the indicators do not exceed the threshold, the system can continue to monitor to ensure that they remain alert within the normal range. This dynamic monitoring mechanism helps to track and manage potential risks in the long term and prevent the accumulation and expansion of problems.

[0151] Example 6

[0152] This example is explained in Example 5. Figure 1 Specifically, the priority module is used to create a corresponding list and divide the water activity difference into , Difference in microbial content , oxidation degree difference and the difference in patch area Stored in the corresponding list, and including the corresponding monitoring sub-area identifier, including the k-th monitoring sub-area and the k+1-th monitoring sub-area;

[0153] Sort the list from large to small to obtain a priority list to identify which monitoring sub-area has a greater risk of spread; when the water activity warning information, microbial content warning information, oxidation degree warning information and spot warning information are triggered, the kth monitoring sub-area and the k+1th monitoring sub-area are marked synchronously.

[0154] In this embodiment, by creating and sorting a priority list, the system can quickly identify the monitoring sub-areas with the highest risk. This approach ensures that managers can focus resources and attention on areas that need intervention most, so that timely measures can be taken to reduce risks. The implementation of priority sorting enables managers to quickly locate problem areas when early warning information is triggered, improve the efficiency of responding to emergencies, and reduce potential losses. By marking monitoring sub-areas with greater risk of spread, companies can implement targeted control measures, such as strengthening cleaning and disinfection, adjusting processing technology, or increasing monitoring frequency, thereby improving product safety.

[0155] Example 7

[0156] This example is explained in Example 6. Figure 1 ,Specifically, the strategy generation unit also includes a strategy generation module and a ,tracing module;

[0157] The strategy generation module is used to generate corresponding strategies based on water activity warning information, microbial content warning information, oxidation degree warning information and spot warning information, including:

[0158] Generate a first strategy based on the water activity warning information, including: regulating the temperature of the kth monitoring sub-area and the k+1th monitoring sub-area to ensure that the storage temperature of the meat products is below 4°C, and the storage temperature difference between the kth monitoring sub-area and the k+1th monitoring sub-area does not exceed ±3°C; regulating the humidity of the kth monitoring sub-area and the k+1th monitoring sub-area to ensure that the relative humidity difference between the kth monitoring sub-area and the k+1th monitoring sub-area does not exceed ±5%;

[0159] The second strategy is generated based on the microbial content warning information, including: testing the microbial content of the kth monitoring sub-area and the k+1th monitoring sub-area, if the E. coli concentration is ≤100 CFU / g, no regulation is performed, once the E. coli concentration is greater than 100 CFU / g, it is indicated that the microorganism exceeds the standard, and the disinfection procedure is initiated;

[0160] A third strategy is generated according to the oxidation degree warning information, including: replacing the kth monitoring sub-area and the k+1th monitoring sub-area with inert gas, including nitrogen and carbon dioxide, to reduce the oxygen concentration in the current monitoring sub-area to below 5%, and ensuring the sealing of the monitoring sub-area to prevent the inflow of external air;

[0161] The fourth strategy is generated according to the spot warning information, including: the spot area difference between the kth monitoring sub-area and the k+1th monitoring sub-area > The meat products with the fourth difference threshold are isolated to avoid cross infection. The same batch of meat products are processed by independent cutting, pickling and packaging equipment, and the corresponding monitoring sub-areas are disinfected. p When the spot is less than 2% of the total area of ​​the meat product, use a knife to remove it;

[0162] For the spot area S p If it is greater than or equal to 2% of the total area of ​​meat products, it will be scrapped.

[0163] The traceability module is used to record the production date, production batch, and raw material source of each batch of meat products; and to record in detail the monitoring process, processing time, and equipment usage data of the kth monitoring sub-area;

[0164] The comprehensive diffusion index Ksz, water activity warning information, microbial content warning information, oxidation degree warning information and stain warning information as well as the corresponding strategies of each batch of meat products are saved, and a unique barcode is generated to facilitate quick query of relevant information of the same batch of meat products.

[0165] In this embodiment, corresponding strategies are generated according to different early warning information, so that the response measures are more targeted and can ensure that specific problems can be effectively solved. For example, when the water activity warning is issued, the risk of microbial growth is reduced by adjusting the temperature and humidity. By monitoring the microbial content and timely disinfection measures, the potential content of pathogenic microorganisms in meat products can be reduced, thereby improving the safety of the product. Strict control of storage temperature and humidity ensures that meat products are stored in a suitable environment, effectively extending their shelf life and reducing the risk of corruption. Through inert gas replacement, the oxygen concentration is reduced, the oxidation reaction of meat products is reduced, and the freshness and flavor of the product are maintained. Meat products with excessive stain area are isolated and processed to effectively reduce the risk of cross contamination and ensure the safety of other products. Different treatment measures are taken according to the proportion of the stain area to ensure the rational use of resources and avoid unnecessary losses.

[0166] The data records in the traceability module help enterprises make continuous improvements, analyze and identify problems based on historical data, and continuously optimize production and management processes.

[0167] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technicians in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0168] The above formulas are obtained by collecting a large amount of data for software simulation and selecting a formula that is close to the actual value. The coefficients in the formula are set by technical personnel in this field according to actual conditions. The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited to this. Any technical personnel familiar with the technical field within the technical scope disclosed by the present invention, according to the technical solution and the inventive concept of the present invention, make equivalent replacement or change, which should be covered within the protection scope of the present invention.

Claims

1. A meat product safety monitoring system based on big data, characterized in that: include: The monitoring area segmentation unit is used to segment the meat product production monitoring area into a plurality of monitoring sub-areas according to the production process sequence; the monitoring sub-areas include a raw material processing area, a cutting area, a pickling area, a maturation area, a packaging area and a refrigeration area; The first detection unit is used to collect the water activity value A in each monitoring sub-area w 、Microbial content M c , peroxide value O x and the spot area S p , and establish water activity data set, microbial data set, oxidation degree data set and stain data set; A difference calculation unit is used to calculate the difference in water activity, microbial content, oxidation degree and spot area between adjacent process monitoring sub-areas to form a difference data set; An analysis unit is used to construct and train a prediction and recognition model using a convolutional neural network (CNN), analyze and evaluate the difference data set to obtain a comprehensive diffusion index Ksz, and perform risk assessment on the comprehensive diffusion index Ksz. If the comprehensive diffusion index Ksz exceeds a risk threshold R, a first warning instruction is sent; The analysis unit includes a model building module, a comprehensive analysis module and a first evaluation module; The model building module is used to preprocess and normalize the difference data set, and then use the convolutional neural network (CNN) to build a prediction and recognition model; And use the normalized difference data set to train the prediction recognition model; The difference data set is divided into 80% training set, 10% validation set and 10% test set to train, validate and test the prediction recognition model; The comprehensive analysis module is used to extract the water activity difference corresponding to the same monitoring sub-area in the difference data set , Difference in microbial content , oxidation degree difference and the difference in patch area , after dimensionless processing, the comprehensive diffusion index Ksz is generated by the following formula: ； In the formula, , , and Represents the water activity difference , Difference in microbial content , oxidation degree difference and the difference in patch area The weight coefficient, C represents the correction constant; The strategy generation unit, after receiving the first warning instruction, converts the water activity difference in the comprehensive diffusion index Ksz into , Difference in microbial content , oxidation degree difference and the difference in patch area Evaluate them respectively to trigger corresponding warning information, and sort them by priority from high to low according to the corresponding values ​​of the corresponding warning information; And generate corresponding control strategies based on the water activity warning information, microbial content warning information, oxidation degree warning information and spot warning information in the corresponding warning information.

2. The meat product safety monitoring system based on big data according to claim 1 is characterized in that: The first detection unit includes a water activity detection module, a microbial content detection module, an oxidation degree detection module and a stain area detection module; The water activity detection module is used to install a water activity sensor in each monitoring sub-area to collect the water activity value A of the monitoring sub-area in real time. w , and establish a water activity data set; The microbial content detection module is used to install microbial detection equipment in each monitoring sub-area to monitor the microbial content M of the monitoring sub-area in real time. c , microbial content M c It refers to the concentration of E. coli on the surface of meat products and establishes a microbiological data set; The oxidation degree detection module is used to install an electrochemical sensor in each monitoring sub-area to measure the peroxide value of the monitoring sub-area in real time. x , and establish a data set of oxidation degree; The stain area detection module is used to take real-time images of the surface of the meat product in the monitoring sub-area through an image acquisition device, and obtain the stain area S by using an image processing algorithm for analysis. p , and establish a speckle dataset.

3. The meat product safety monitoring system based on big data according to claim 2 is characterized in that: The stain area detection module includes an image acquisition submodule, an image segmentation submodule and an area calculation submodule; The image acquisition submodule is used to install an image acquisition device in each monitoring sub-area, and to capture the surface image of the meat product in the monitoring sub-area in real time through the image acquisition device; The image segmentation submodule is used to convert the meat product surface image into a grayscale image, and obtain a binary image by binarization processing. , the specific steps are: S11, counting the number of pixels of each gray value in the gray image, generating a histogram, and recording the number of pixels h(i) with gray value i in the histogram; S12, calculate the total number of pixels N, ; h(i) represents the number of pixels with gray value i in the histogram; S13, applying Otsu's threshold segmentation method, pre-setting the traversal threshold t, maximizing the inter-class variance of the foreground and background after dividing the image by the traversal threshold t, and calculating the foreground probability P1(t) and background probability P2(t) by the following formula: ； ； Where h(i) represents the number of pixels with grayscale value i in the histogram; t represents the traversal threshold, and pixels with grayscale value less than or equal to t are regarded as foreground; The foreground probability P1(t) represents the proportion of pixels with grayscale values ​​less than or equal to t to the total pixels, and the background probability P2(t) represents the proportion of pixels with grayscale values ​​greater than t to the total pixels; S14. Calculate the foreground mean μ1(t) and background mean μ2(t) by the following formula: ； ； In the formula, represents the weighted sum of all gray values ​​i less than or equal to the traversal threshold t, represents the total brightness of the foreground pixels, h(i) represents the number of pixels with gray value i in the histogram, Represents the total number of foreground pixels; It represents the weighted sum of the gray values ​​i greater than the traversal threshold t, which represents the total brightness of the background pixels; Represents the total number of background pixels; S15. Based on the foreground probability P1(t), the background probability P2(t), the foreground mean μ1(t) and the background mean μ2(t), the inter-class variance is calculated using the following formula: : ； If the current between-class variance If it is greater than the maximum inter-class variance recorded previously, the maximum inter-class variance value and the corresponding optimal threshold X are updated, and B represents the pixel value of the background area in the binary image when it is not segmented; S16, applying the optimal threshold value X to segment and obtain a binary image : ； In the formula, Represents the binary image after segmentation, a value of 1 represents the spot foreground area, and a value of 0 represents the background area; Represents the statistics of the pixel value with gray value i in region j; The area calculation submodule is used to calculate the area of ​​the binary image. , according to the physical size of each pixel, it is converted into the spot area S by the following formula p : ； In the formula, is the total number of pixels with value 1 in the binary image, Represents the actual area of ​​each pixel and is calculated based on the number of spot areas S p Create a speckle dataset.

4. The meat product safety monitoring system based on big data according to claim 1 is characterized in that: The difference data set includes: Water activity difference : ; In the formula, represents the water activity value of the k+1th monitoring sub-area, represents the water activity value of the kth monitoring sub-area; Difference in microbial content : ; In the formula, represents the microbial content value of the k+1th monitoring sub-area, represents the microbial content value of the kth monitoring sub-area, and the microbial content value is the concentration of the detected Escherichia coli concentration; Oxidation degree difference : ; In the formula, represents the peroxide value of the k+1th monitoring sub-area, represents the peroxide value of the kth monitoring sub-area; Patch area difference : ; In the formula, represents the spot area value of the k+1th monitoring sub-region, Represents the spot area value of the kth monitoring sub-area.

5. The meat product safety monitoring system based on big data according to claim 1 is characterized in that: The first evaluation module is used to preset a risk threshold R, and compare the comprehensive diffusion index Ksz with the risk threshold R to obtain a first evaluation result, including: When the comprehensive diffusion index Ksz>risk threshold R, it means that there is a risk of diffusion corrosion of meat products in the current monitoring sub-area, and the first warning instruction is generated; When the comprehensive diffusion index Ksz ≤ risk threshold R, it means that there is no diffusion corrosion risk for meat products in the current monitoring sub-area.

6. The meat product safety monitoring system based on big data according to claim 5 is characterized in that: The strategy generation unit includes a second evaluation module and a priority module; The second evaluation module is used to calculate the water activity difference in the comprehensive diffusion index Ksz after receiving the first warning instruction. , Difference in microbial content , oxidation degree difference and the difference in patch area Conduct assessments respectively to obtain second assessment results, including: Preset a first difference threshold, a second difference threshold, a third difference threshold and a fourth difference threshold; When the water activity difference > The first difference threshold indicates that the water activity exceeds the normal range, triggering a water activity warning message; when the water activity difference ≤ the first difference threshold, indicating that the water activity does not exceed the normal range and continues to be monitored; When the difference in microbial content > The second difference threshold indicates that the microbial content exceeds the normal range, triggering a microbial content warning message; when the microbial content difference ≤ the second difference threshold, indicating that the microbial content does not exceed the normal range and continues to be monitored; When the oxidation degree difference > The third difference threshold indicates that the oxidation degree exceeds the safety level, triggering the oxidation degree warning information; when the oxidation degree difference ≤ the third difference threshold, indicating that the degree of oxidation does not exceed the safe level and is continuously monitored; When the difference in patch area > The fourth difference threshold indicates that the spot area increases beyond the normal range, triggering the spot warning information; when the spot area difference ≤ the fourth difference threshold, indicating that the increased area of ​​the spot does not exceed the normal range and continues to be monitored.

7. The meat product safety monitoring system based on big data according to claim 6 is characterized in that: The priority module is used to create a corresponding list and divide the water activity difference into , Difference in microbial content , oxidation degree difference and the difference in patch area Stored in the corresponding list, and including the corresponding monitoring sub-area identifier, including the k-th monitoring sub-area and the k+1-th monitoring sub-area; According to the corresponding values ​​of the corresponding warning information, the priority is sorted from high to low to obtain a priority sorting list, which is used to identify which monitoring sub-area has a greater risk of spread; When the water activity warning information, the microbial content warning information, the oxidation degree warning information and the stain warning information are triggered, the kth monitoring sub-area and the k+1th monitoring sub-area are marked synchronously.

8. The meat product safety monitoring system based on big data according to claim 7 is characterized in that: The strategy generation unit also includes a strategy generation module and a tracing module; The strategy generation module is used to generate corresponding strategies according to water activity warning information, microbial content warning information, oxidation degree warning information and spot warning information, including: Generate a first strategy based on the water activity warning information, including: regulating the temperature of the kth monitoring sub-area and the k+1th monitoring sub-area to ensure that the storage temperature of the meat products is below 4°C, and the storage temperature difference between the kth monitoring sub-area and the k+1th monitoring sub-area does not exceed ±3°C; regulating the humidity of the kth monitoring sub-area and the k+1th monitoring sub-area to ensure that the relative humidity difference between the kth monitoring sub-area and the k+1th monitoring sub-area does not exceed ±5%; The second strategy is generated based on the microbial content warning information, including: testing the microbial content of the kth monitoring sub-area and the k+1th monitoring sub-area, if the E. coli concentration is ≤100 CFU / g, no regulation is performed, once the E. coli concentration is greater than 100 CFU / g, it is indicated that the microorganism exceeds the standard, and the disinfection procedure is initiated; A third strategy is generated according to the oxidation degree warning information, including: replacing the kth monitoring sub-area and the k+1th monitoring sub-area with inert gas, including nitrogen and carbon dioxide, to reduce the oxygen concentration in the current monitoring sub-area to below 5%, and ensuring the sealing of the monitoring sub-area to prevent the inflow of external air; The fourth strategy is generated according to the spot warning information, including: the spot area difference between the kth monitoring sub-area and the k+1th monitoring sub-area > The meat products with the fourth difference threshold are isolated to avoid cross infection. The same batch of meat products are processed by independent cutting, pickling and packaging equipment, and the corresponding monitoring sub-areas are disinfected. p When the stain is less than 2% of the total area of ​​the meat product, the stain is removed with a knife; p If it is greater than or equal to 2% of the total area of ​​meat products, it will be scrapped.

9. The meat product safety monitoring system based on big data according to claim 8, characterized in that: The traceability module is used to record the production date, production batch, and raw material source of each batch of meat products; and to record in detail the monitoring process, processing time, and equipment usage data of the kth monitoring sub-area; The comprehensive diffusion index Ksz, water activity warning information, microbial content warning information, oxidation degree warning information and stain warning information as well as the corresponding strategies of each batch of meat products are saved, and a unique barcode is generated to facilitate quick query of relevant information of the same batch of meat products.

Images