Food safety knowledge graph construction method and system based on neural network

Abstract

The invention discloses a food safety knowledge graph construction method and system based on a neural network, and relates to the technical field of food safety. Food characteristic information data, equipment state information data and external environment state information data are respectively obtained, an equipment surface pollution degree coefficient Xnd and a food surface pollution degree coefficient Xsp of each monitoring time period are calculated and obtained, correlation analysis is carried out, a pollution path correlation index Zwr is calculated, whether a food pollution path comes from processing equipment or not is judged, and if yes, the food pollution path is judged to come from the processing equipment; an atlas construction instruction is sent out; a food safety knowledge graph is constructed by combining a time sequence neural network algorithm, production line equipment which does not meet safety processing requirements is identified through target word retrieval and threshold comparison, and cleaning and disinfection instructions are generated to guarantee food safety.

Description

Technical Field

[0001] The present invention relates to the field of food safety technology, and specifically to a method and system for constructing a food safety knowledge graph based on a neural network. Background Art

[0002] As an important industry related to human health and economic development, the food industry has always received widespread attention. In the modern food processing process, the popularization of industrialization and scale has improved production efficiency, but it has also brought severe challenges to food quality and safety. Microbial contamination and unclean processing environment problems frequently occur in the food production process, which not only poses a threat to consumers' health, but also affects the trust and development of the food industry. In this context, the construction of food safety knowledge graphs has become a research hotspot in the field of food safety. The knowledge graph links multi-source data of food, equipment, microorganisms and environment to form a multi-dimensional dynamic information network, which provides data support and intelligent reasoning capabilities for traceability analysis, pollution prediction and risk assessment.

[0003] Although the food safety knowledge graph constructed at this stage has been proven to be an effective analysis tool, it still has many shortcomings in practical applications, especially in the tracking and tracing of microbial contamination paths. Due to the complex and changeable transmission speed and path of microbial contamination, and the multiple influences of equipment cleaning status, food characteristics and environmental conditions, and the data information contained in the food safety knowledge graph constructed at this stage mostly relies on manual detection and offline analysis, it is difficult to capture the dynamic changes of real-time data, especially the lack of effective methods for quantitative analysis of contamination paths, and it is difficult to achieve intuitive judgment of microbial transmission paths. The abnormal consequences of this deficiency include: the failure to lock the source of contamination in time leads to further spread of contamination, the misjudgment of the risk of food surface contamination leads to unnecessary production shutdowns and waste of resources, and increased health risks for consumers. Therefore, there is an urgent need for a food safety knowledge graph construction method combined with a neural network, which can associate multi-source data of food, equipment, microorganisms and the environment through dynamic monitoring and deep learning to form a multi-dimensional dynamic information network, enhance the credible pollution traceability and judgment level of the food safety knowledge graph, and promote the intelligent and precise development of the food safety field. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a method and system for constructing a food safety knowledge graph based on a neural network, which solves the problems in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for constructing a food safety knowledge graph based on a neural network, comprising the following steps;

[0006] S1. Real-time monitoring is performed based on the dynamic changes of microbial contaminants diffusing from the surface of processing equipment to the surface of food during food processing, and food characteristic information data, equipment status information data and external environment status information data are obtained respectively;

[0007] S2. Analyze the food characteristic information data and the equipment status information data stored in the cloud platform to construct the equipment surface contamination degree coefficient Xnd for several groups of monitoring periods, and analyze the food characteristic information data and the external environment status information data to construct the food surface contamination degree coefficient Xsp for several groups of monitoring periods;

[0008] S3. Generate a pollution path correlation index Zwr based on the equipment surface pollution degree coefficient Xnd and the food surface pollution degree coefficient Xsp of the several monitoring periods, and determine the transmission path of the food surface pollution from the processing equipment according to the value of the pollution path correlation index Zwr, and issue a map construction instruction;

[0009] S4. After receiving the graph construction instruction, a food safety knowledge graph is constructed in combination with a time series neural network algorithm. By searching for target words and performing threshold comparison, the production line equipment that does not meet food safety processing requirements during the current monitoring period is cleaned and disinfected.

[0010] Preferably, the specific steps of S1 include:

[0011] S11. Real-time monitoring is performed based on the dynamic changes of microbial contaminants diffusing from the equipment surface to the food surface during food processing, and food characteristic information data, equipment status information data and external environment status information data are obtained respectively, including:

[0012] S111. During the food processing process, the multiple sets of sensors deployed are used to monitor the state of food surface pollutants in real time during the food processing process, obtain food characteristic information data, and transmit the food characteristic information data to the cloud platform in real time through the network. The food characteristic information data includes the microbial reproduction rate Vfz and the initial food residue concentration Xks in each monitoring period. The multiple sets of sensors include a portable microbial detector and a fluorescence sensor;

[0013] S112, using the cleanliness sensor deployed on the production line equipment to record the pollution change of the equipment after cleaning, obtain the cleaning state decay rate Vsj of each monitoring period, and collect the initial pollution concentration Cwr on the surface of the equipment according to the surface pollution state of the equipment after cleaning. The cleaning state decay rate Vsj of each monitoring period and the initial pollution concentration Cwr construct equipment state information data, and transmit it to the cloud platform in real time through the network;

[0014] S113. Utilize multiple sets of environmental monitoring equipment deployed in the food processing process to monitor the external environment status in real time during the process of microbial contaminants diffusing from the equipment surface to the food surface, obtain external environment status information data, and transmit it to the cloud platform in real time through the network. The external environment status information data includes the average temperature Twd and the average air flow rate Vkq in each monitoring period.

[0015] Preferably, the specific step S1 also includes:

[0016] S12. Preprocess the food characteristic information data, equipment status information data and external environment status information data received by the cloud platform. The preprocessing includes noise removal, missing value filling and data smoothing operations. The missing value filling methods include mean filling, median filling, interpolation filling and regression filling. The preprocessed food characteristic information data, equipment status information data and external environment status information data are stored.

[0017] Preferably, the specific steps of S2 include:

[0018] S21. Extract the characteristics of the food characteristic information data and the equipment status information data stored in the cloud platform, and after dimensionless processing, calculate and obtain the equipment surface contamination degree coefficient Xnd for several groups of monitoring periods. The equipment surface contamination degree coefficient Xnd for the i-th group of monitoring periods is i For example, it is obtained in the following way:

[0019]

[0020] Where Cwr is the initial contamination concentration on the equipment surface, Vfz i Vsj is expressed as the microbial growth rate in the ith monitoring period, i It is expressed as the cleaning state decay rate of the ith monitoring period, e * It is expressed as an exponential function with e as the base, which is used to describe that the change of pollution degree is an exponential growth process. e is the Euler number, and α1 and α2 are both expressed as weight values.

[0021] Preferably, the specific step S2 also includes:

[0022] S22, extracting features from the food characteristic information data and the external environment status information data, and calculating and obtaining the food surface contamination degree coefficient Xsp of several groups of monitoring periods after dimensionless processing, and taking the food surface contamination degree coefficient Xsp of the i-th group of monitoring periods as i For example, it is obtained in the following way:

[0023]

[0024] Where Xks is the initial residue concentration in food, Twd i Expressed as the average temperature of the ith monitoring period, Vkq i is the average air velocity in the i-th monitoring period, γ is the diffusion sensitivity factor, which is used to describe the effect of temperature and air flow on the diffusion rate, e * It is expressed as an exponential function with e as the base, which is used to describe that the change of pollution degree is an exponential growth process. e is the Euler number, and β1 and β2 are both expressed as weight values.

[0025] Preferably, the specific steps of S3 include:

[0026] S31, based on the equipment surface contamination degree coefficient Xnd of several groups of monitoring periods and the food surface contamination degree coefficient Xsp of several groups of monitoring periods, and combined with the statistical averaging algorithm, generate the contamination path association index Zwr, the contamination path association index Zwr is obtained by the following formula:

[0027]

[0028] Where Xnd i It is expressed as the equipment surface contamination coefficient of the i-th monitoring period, Xsp i It is expressed as the food surface contamination degree coefficient of the i-th group monitoring period, It is expressed as the mean value of the equipment surface contamination coefficient during the monitoring period. It is expressed as the mean value of the food surface contamination coefficient within the monitoring period, i = 1, 2, 3, ..., n, where n is the monitoring period.

[0029] Preferably, the specific step S3 also includes:

[0030] S32. Preset the correlation range [G1, G2], compare the pollution path correlation index Zwr with the correlation range [G1, G2], analyze the correlation between the equipment surface pollution and the food surface pollution, and determine whether the transmission path of the food surface pollution comes from the processing equipment. The specific contents are as follows:

[0031] If the pollution path correlation index Zwr is greater than G2, it means that there is a direct correlation between the equipment surface pollution and the food surface pollution, that is, the change direction of the two is consistent, indicating that the equipment surface pollution has an impact on the food surface pollution, and at the same time, it means that the transmission path of the food surface pollution comes from the processing equipment. At this time, the map construction instruction will be issued;

[0032] If the contamination path correlation index Zwr is in the correlation range [G1, G2], it means that there is no direct correlation between the equipment surface contamination and the food surface contamination, and it cannot be said that the transmission path of the food surface contamination comes from the processing equipment;

[0033] If the contamination path correlation index Zwr is less than G1, it means that there is no correlation between the equipment surface contamination and the food surface contamination, that is, the two change independently, and it cannot be explained that the transmission path of food surface contamination comes from the processing equipment.

[0034] Preferably, the specific steps of S4 include:

[0035] S41. After receiving the graph construction instruction, the food characteristic information data, equipment status information data and external environment status information data received by the cloud platform are input into the time series neural network, and the local features are extracted by performing convolution operation in the convolution layer to generate feature quantities, and the feature quantities are input into the pooling layer to downsample the feature quantities, and then the downsampled feature quantities are input into the fully connected layer to perform a full connection operation, and the probability distribution of each timing parameter information is output, and finally the probability distribution of each timing parameter information is input into the output layer, and the probability distribution of each timing parameter information is decoded in the output layer, and the decoded timing parameter information is aligned according to the timing of the monitoring period, and several groups of aligned data belonging to the same monitoring period are obtained, and the equipment surface contamination degree coefficient Xnd and the food surface contamination degree coefficient Xsp of several groups of monitoring periods are mapped to the aligned data of the corresponding monitoring period to construct a food safety knowledge graph.

[0036] Preferably, the specific step S4 also includes:

[0037] S42. Based on the food safety knowledge graph constructed in step S41, the food surface contamination coefficient Xsp is used as the search target word. After data retrieval and establishment of entity relationships, the threshold value comparison is performed on the food surface contamination coefficient Xsp in each monitoring period during the food processing process in chronological order. When the food surface contamination coefficient Xsp exceeds the threshold value, it means that the surface cleanliness of the corresponding processing equipment in the current monitoring period does not meet the food safety processing requirements, and the production line equipment in the current period is cleaned and disinfected. When the food surface contamination coefficient Xsp does not exceed the threshold value, it means that the surface cleanliness of the processing equipment in the food processing process meets the food safety processing requirements, and the next stage of food processing operations is continued.

[0038] A food safety knowledge graph construction system based on neural network, including a monitoring module, a pollution analysis module, a path association module and a graph construction module;

[0039] The monitoring module is used to perform real-time monitoring according to the dynamic changes of microbial contaminants diffusing from the surface of the processing equipment to the surface of the food during the food processing process, and obtain food characteristic information data, equipment status information data and external environment status information data respectively;

[0040] The pollution analysis module is used to analyze the food characteristic information data and the equipment status information data stored in the cloud platform to construct the equipment surface pollution degree coefficient Xnd for several groups of monitoring periods, and to analyze the food characteristic information data and the external environment status information data to construct the food surface pollution degree coefficient Xsp for several groups of monitoring periods;

[0041] The path association module is used to generate a pollution path association index Zwr based on the equipment surface pollution degree coefficient Xnd of several groups of monitoring periods and the food surface pollution degree coefficient Xsp of several groups of monitoring periods, and to determine the transmission path of food surface pollution from the processing equipment according to the value of the pollution path association index Zwr, and to issue a map construction instruction;

[0042] The graph construction module is used to construct a food safety knowledge graph in combination with a temporal neural network algorithm after receiving a graph construction instruction, and to clean and disinfect production line equipment that does not meet food safety processing requirements during the current monitoring period by performing target word retrieval and threshold comparison.

[0043] The present invention provides a method and system for constructing a food safety knowledge graph based on a neural network, which has the following beneficial effects:

[0044] (1) The method for constructing a food safety knowledge graph based on a neural network can associate multi-source data of food, equipment, microorganisms and the environment through dynamic monitoring and deep learning to form a multi-dimensional dynamic information network, enhance the credible pollution traceability and judgment level of the food safety knowledge graph, and promote the intelligent and precise development of the food safety field. Compared with the traditional method that relies on manual detection and lacks quantitative analysis of pollution paths, it can capture dynamically changing data in real time, quickly generate knowledge graphs and perform intelligent reasoning, providing more comprehensive support for pollution traceability, risk assessment and equipment optimization. Especially in scenarios where the microbial transmission path is complex and affected by multiple factors, it can accurately calculate the pollution path correlation index Zwr, thereby realizing the automatic judgment and processing strategy of the pollution transmission path, and combining the time series neural network algorithm to construct a food safety knowledge graph, and through target word retrieval and threshold comparison, identify production line equipment that does not meet the safety processing requirements, and generate cleaning and disinfection instructions to ensure food safety. Finally, it effectively reduces the spread of food contamination incidents, optimizes production processes, and improves the credibility of food safety management, providing reliable support for the intelligent and precise development of the food industry, while reducing the downtime losses and consumer health risks caused by misjudgment.

[0045] (2) Through real-time monitoring of multiple groups of dynamic data in the food processing process, the correlation between the equipment surface contamination coefficient Xnd and the food surface contamination coefficient Xsp in each monitoring period was comprehensively analyzed, and the contamination path correlation index Zwr was constructed. The relationship between equipment surface contamination and food surface contamination was quantified through the contamination path correlation index Zwr. When the contamination path correlation index Zwr is greater than G2, it can accurately identify that the contamination path comes from the processing equipment. When the contamination path correlation index Zwr is in the correlation range [G1, G2] or less than G1, it can be determined that the contamination path has no direct correlation with the processing equipment. The introduction of this quantitative analysis not only improves the scientificity and credibility of the contamination path determination, but also reduces the errors caused by manual determination. In complex production environments, including scenarios where multiple foods share equipment, the source of contamination can be quickly identified, thereby avoiding further spread of contamination. At the same time, through real-time monitoring and automated analysis, the time for problem investigation can be shortened, the efficiency and accuracy of pollution traceability can be improved, and strong support can be provided for early warning and rapid disposal of food safety accidents.

[0046] (3) The food safety knowledge graph constructed through dynamic updating and deep learning of time-series neural networks can analyze the changing trends of equipment surface cleaning status and food surface contamination risks in real time, and automatically generate optimization suggestions for equipment cleaning and disinfection. It provides innovative solutions in production process optimization and intelligent decision support. When the food surface contamination coefficient Xsp exceeds the set threshold, it means that the corresponding equipment surface cleanliness level during the current monitoring period does not meet the safety processing of food, and the production line equipment in the current period is cleaned and disinfected to avoid contamination affecting subsequent food processing links. This data-driven dynamic optimization method not only reduces the unnecessary cleaning frequency of equipment, but also improves production efficiency under resource-constrained conditions. In addition, the consideration of external environmental factors makes the risk assessment of the production environment more comprehensive, which helps enterprises to achieve refined risk control in food safety management. This full-process intelligent decision support can effectively reduce the risk of production shutdown and resource waste, and enhance the food safety assurance capabilities of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic diagram of a method for constructing a food safety knowledge graph based on a neural network according to the present invention;

[0048] Figure 2 This is a block diagram of a food safety knowledge graph construction system based on a neural network in the present invention. DETAILED DESCRIPTION

[0049] 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.

[0050] Example 1

[0051] See also Figure 1 , the present invention provides a method for constructing a food safety knowledge graph based on a neural network, comprising the following steps;

[0052] S1. Real-time monitoring is performed based on the dynamic changes of microbial contaminants diffusing from the surface of processing equipment to the surface of food during food processing, and food characteristic information data, equipment status information data and external environment status information data are obtained respectively;

[0053] S2. Analyze the food characteristic information data and the equipment status information data stored in the cloud platform to construct the equipment surface contamination degree coefficient Xnd for several groups of monitoring periods, and analyze the food characteristic information data and the external environment status information data to construct the food surface contamination degree coefficient Xsp for several groups of monitoring periods;

[0054] S3. Generate a pollution path correlation index Zwr based on the equipment surface pollution degree coefficient Xnd and the food surface pollution degree coefficient Xsp of the several monitoring periods, and determine the transmission path of the food surface pollution from the processing equipment according to the value of the pollution path correlation index Zwr, and issue a map construction instruction;

[0055] S4. After receiving the graph construction instruction, a food safety knowledge graph is constructed in combination with a time series neural network algorithm. By searching for target words and performing threshold comparison, the production line equipment that does not meet food safety processing requirements during the current monitoring period is cleaned and disinfected.

[0056] In this embodiment, through real-time monitoring, dynamic modeling and intelligent analysis, the problems of poor real-time performance, difficult traceability and insufficient dynamic adaptability in traditional food safety monitoring are effectively solved; compared with the current food safety knowledge graph that relies on manual detection and offline analysis, the special advantage of this method is that it combines multi-source dynamic data and time series neural networks to achieve accurate determination of contamination paths and real-time updating of dynamic food safety knowledge graphs; first, by monitoring the dynamic changes of microbial contaminants diffusing from the equipment surface to the food surface during food processing, it can capture food characteristics, equipment status and external environment status data, effectively overcoming the limitation of traditional methods that are difficult to dynamically capture real-time changes; secondly, this method constructs the equipment surface contamination degree coefficient Xnd and the food surface contamination degree coefficient Xnd. The number Xsp is calculated, and through the quantitative analysis of the pollution path association index Zwr, it is accurately determined whether the source of pollution comes from the processing equipment, realizing the scientific tracing of the complex pollution transmission path; finally, the food safety knowledge graph constructed in combination with the time series neural network not only provides an intuitive presentation of the pollution risk, but also automatically identifies the equipment that does not meet the food safety processing conditions through target word retrieval and threshold comparison, and generates cleaning and disinfection treatment; this dynamic, accurate and intelligent food safety knowledge graph effectively makes up for the shortcomings of traditional graph construction methods in the quantitative analysis of microbial transmission paths, significantly reduces food safety accidents caused by untimely pollution tracing, and at the same time avoids the waste of production resources caused by misjudgment, further improving the level of intelligent and precise management in the field of food safety.

[0057] Example 2

[0058] Please refer to Figure 1 , specifically: S1 specific steps include:

[0059] S11. Real-time monitoring is performed based on the dynamic changes of microbial contaminants diffusing from the equipment surface to the food surface during food processing, and food characteristic information data, equipment status information data and external environment status information data are obtained respectively, including:

[0060] S111. During the food processing process, the multiple sets of sensors deployed are used to monitor the state of food surface pollutants in real time during the food processing process, obtain food characteristic information data, and transmit the food characteristic information data to the cloud platform in real time through the network. The food characteristic information data includes the microbial reproduction rate Vfz and the initial food residue concentration Xks in each monitoring period. The multiple sets of sensors include a portable microbial detector and a fluorescence sensor;

[0061] It should be noted that the microbial reproduction rate Vfz is obtained by a portable microbial detector combined with a deep learning algorithm; during the food processing process, the portable microbial detector collects the changes in the number of microorganisms on the food surface in real time, uses fluorescent labeling technology and quantitative PCR technology to accurately detect the number of microorganisms, and records its change trend at different time points; combined with the deep learning algorithm, the time series data is analyzed to fit the reproduction rate of microorganisms on the food surface, which is expressed as the change in the number of bacteria per unit time (CFU / h), reflecting the growth ability of microorganisms on the surface of specific foods. It is the core parameter of the dynamics of pollution transmission. By obtaining the microbial reproduction rate Vfz in real time, the degree of threat of microorganisms to food contamination can be quantified, and a scientific basis can be provided for the optimization of equipment cleaning frequency and processing environment conditions. At the same time, the microbial reproduction rate Vfz is also a key variable for calculating the degree of contamination on the equipment surface and the degree of contamination on the food surface, which directly affects the accuracy of pollution traceability and risk assessment.

[0062] The initial food residue concentration Xks is obtained by calculation after measurement by a fluorescent sensor. First, a surface diffusion test is performed and the diffusion behavior of pollutants on the food surface is monitored by using fluorescent markers. The coverage area and diffusion time of pollutant diffusion are recorded in real time by using fluorescent sensors. The collected data are input into a deep learning algorithm to analyze the diffusion characteristics of different foods under specific environmental conditions. Finally, the initial food residue concentration Xks is obtained, and the rate of change of the diffusion area over time (m 2 / s) indicates that the initial residue concentration Xks of food reflects the propagation speed of pollutants on the food surface and is an important parameter for dynamic modeling of pollution diffusion. By real-time monitoring of the initial residue concentration Xks of food, the scope and trend of microbial contamination propagation on the food surface can be analyzed, thereby predicting the propagation path. Accurate acquisition of the initial residue concentration Xks of food plays a vital role in optimizing processing procedures, adjusting environmental conditions and preventing pollution diffusion. It also provides scientific support for the analysis of food contamination diffusion paths in the construction of food safety knowledge graphs.

[0063] S112, using the cleanliness sensor deployed on the production line equipment to record the pollution change of the equipment after cleaning, obtain the cleaning state decay rate Vsj of each monitoring period, and collect the initial pollution concentration Cwr on the surface of the equipment according to the surface pollution state of the equipment after cleaning. The cleaning state decay rate Vsj of each monitoring period and the initial pollution concentration Cwr construct equipment state information data, and transmit it to the cloud platform in real time through the network;

[0064] It should be noted that the cleaning state decay rate Vsj is obtained by recording in real time the rate of re-accumulation of pollutants on the equipment surface after cleaning through the cleanliness sensor. By detecting the change in the concentration of pollutants on the equipment surface and combining the ATP bioluminescence method, the decay rate of the cleaning effect per unit time is calculated according to the time interval after the cleaning is completed. The cleaning state decay rate Vsj reflects the duration of the cleaning effect on the equipment surface and the re-accumulation trend of pollutants. It is an important parameter for evaluating the cleaning cycle and efficiency. By monitoring the cleaning state decay rate Vsj, the cleaning plan is optimized, the impact of equipment pollution on the food processing process is reduced, and at the same time, key input data is provided for the calculation of the degree of pollution on the equipment surface.

[0065] The initial contamination concentration Cwr is obtained by detecting the residual concentration of contaminants on the equipment surface through a cleanliness sensor immediately after the equipment is cleaned, and is uploaded in real time through an ATP detector to accurately record the initial state of the contaminants after cleaning. The cleaning state decay rate Vsj is the basic parameter for calculating the degree of contamination on the equipment surface, and directly determines the initial state of the equipment contamination trend.

[0066] S113. Utilize multiple sets of environmental monitoring equipment deployed in the food processing process to monitor the external environment status in real time during the process of microbial contaminants diffusing from the equipment surface to the food surface, obtain external environment status information data, and transmit it to the cloud platform in real time through the network. The external environment status information data includes the average temperature Twd and the average air flow rate Vkq in each monitoring period.

[0067] It should be noted that the average temperature Twd is collected in real time by the environmental temperature sensors deployed in the processing area. Multiple sensors are evenly distributed in the processing environment, and the temperature data is recorded regularly. The temperature values ​​in a specific monitoring period are averaged and uploaded to the cloud platform to ensure the accuracy and timeliness of the data. The average temperature Twd is an important environmental parameter that affects the microbial reproduction rate and pollution spread. By real-time monitoring of the average temperature Twd, the role of the processing environment in promoting the growth and spread of microorganisms is evaluated, providing a scientific basis for environmental regulation and control, and serving as one of the core variables for pollution spread path analysis;

[0068] The average air flow rate Vkq is measured by a wind speed sensor deployed in the processing environment. The air flow rate is monitored in real time, and the data within the monitoring period is averaged and the results are uploaded to the cloud platform through the network to form dynamic air flow rate status information. The average air flow rate Vkq is an important determinant of the pollutant diffusion path and rate. By monitoring the average air flow rate Vkq, the impact of air flow on the spread of pollutants from the equipment surface to the food surface is evaluated, providing key data support for the prediction of pollution diffusion risk.

[0069] Specifically, the specific steps of S1 also include:

[0070] S12. Preprocess the food characteristic information data, equipment status information data and external environment status information data received by the cloud platform. The preprocessing includes noise removal, missing value filling and data smoothing operations. The missing value filling methods include mean filling, median filling, interpolation filling and regression filling. The preprocessed food characteristic information data, equipment status information data and external environment status information data are stored.

[0071] In this embodiment, the implementation of step S1 significantly improves the comprehensiveness, real-timeness and reliability of data collection, and effectively solves the problems of insufficient data dynamics and large errors caused by reliance on manual detection and offline analysis in traditional food safety monitoring; specifically, multiple sets of sensors and deep learning algorithms are used to obtain food characteristic information data, equipment status information data and external environment status information data in real time, which not only effectively overcomes the limitation that traditional detection methods are difficult to capture the dynamic diffusion changes of microbial contaminants, but also ensures the multi-dimensionality and accuracy of data collection; the combination of portable microbial detector and fluorescent sensor improves the high-precision monitoring of food surface contamination status, while the cleanliness sensor provides dynamic change information on the degree of contamination on the equipment surface, making the analysis of equipment status more scientific. At the same time, environmental monitoring equipment is used to capture in real time the external environmental parameters of the average temperature Twd and the average air flow rate Vkq in each monitoring period, which effectively solves the deficiency of traditional methods in ignoring the impact of environmental factors on pollution propagation; more importantly, through data preprocessing steps, including denoising, missing value filling and data smoothing operations, it ensures that all data input into the cloud platform are of high quality and consistency, providing a solid foundation for the subsequent construction of a food safety knowledge graph; overall, the special advantage of this method is that it realizes real-time monitoring and high-quality data management of food, equipment and environment, effectively making up for the deficiencies of traditional systems in dynamic data capture capabilities, and providing higher accuracy and intelligent support for pollution traceability and risk assessment.

[0072] Example 3

[0073] Please refer to Figure 1 , specifically: S2 specific steps include:

[0074] S21. Extract the characteristics of the food characteristic information data and the equipment status information data stored in the cloud platform, and after dimensionless processing, calculate and obtain the equipment surface contamination degree coefficient Xnd for several groups of monitoring periods. The equipment surface contamination degree coefficient Xnd for the i-th group of monitoring periods is i For example, it is obtained in the following way:

[0075]

[0076] Where Cwr is the initial contamination concentration on the equipment surface, Vfz i Vsj is expressed as the microbial growth rate in the ith monitoring period, i It is expressed as the cleaning state decay rate of the ith monitoring period, e * It is expressed as an exponential function with e as the base, which is used to describe that the change of pollution degree is an exponential growth process. e is the Euler number, which is about 2.71828. α1 and α2 are both expressed as weight values.

[0077] Specifically, the specific steps of S2 also include:

[0078] S22, extracting features from the food characteristic information data and the external environment status information data, and calculating and obtaining the food surface contamination degree coefficient Xsp of several groups of monitoring periods after dimensionless processing, and taking the food surface contamination degree coefficient Xsp of the i-th group of monitoring periods as i For example, it is obtained in the following way:

[0079]

[0080] Where Xks is the initial residue concentration in food, Twd i Expressed as the average temperature of the ith monitoring period, Vkq i is the average air velocity in the i-th monitoring period, γ is the diffusion sensitivity factor, which is used to describe the effect of temperature and air flow on the diffusion rate, e * It is expressed as an exponential function with e as the base, which is used to describe that the change of pollution degree is an exponential growth process. e is the Euler number, which is about 2.71828. β1 and β2 are both expressed as weight values.

[0081] In this embodiment, the implementation of step S2 plays a key role in the construction of the food safety knowledge graph, significantly improves the quantitative analysis capability of the dynamic changes of pollution propagation, and effectively solves the deficiency of the traditional method in the quantitative analysis of the microbial propagation path and the pollution degree; based on the food characteristic information data, the equipment status information data and the external environment status information data, the equipment surface pollution degree coefficient Xnd and the food surface pollution degree coefficient Xsp are generated in each monitoring period respectively; this method based on feature extraction and mathematical calculation realizes the time period division and dynamic quantification of the pollution concentration, especially in the calculation of the equipment pollution concentration Cnd, combined with the microbial reproduction rate Vfz and the clean state decay rate Vsj, accurately describes the exponential growth law of the pollution concentration, and provides a basis for the complex It provides a scientific basis for the prediction of pollution trends of various processing equipment; similarly, in the calculation of the food surface contamination degree coefficient Xsp, the initial food residue concentration Xks and external environmental condition parameters are introduced, and the effects of temperature, humidity and air flow on pollution diffusion are considered, making the transmission dynamics of food contamination clearer; the special advantages of this dynamic concentration modeling and calculation are: it can not only accurately quantify the impact of pollution sources, but also predict pollution trends and risk points through segmented analysis in the time dimension, laying the foundation for the accurate determination of pollution paths in the food processing process; it effectively overcomes the problems of data staticity and one-sided analysis, provides more accurate and scientific support for pollution tracing and risk assessment, and effectively reduces the uncertainty of pollution transmission and the potential health risks it brings.

[0082] Example 4

[0083] Please refer to Figure 1 , specifically: S3 specific steps include:

[0084] S31, based on the equipment surface contamination degree coefficient Xnd of several groups of monitoring periods and the food surface contamination degree coefficient Xsp of several groups of monitoring periods, and combined with the statistical averaging algorithm, generate the contamination path association index Zwr, the contamination path association index Zwr is obtained by the following formula:

[0085]

[0086] Where Xnd i It is expressed as the equipment surface contamination coefficient of the i-th monitoring period, Xsp i It is expressed as the food surface contamination degree coefficient of the i-th group monitoring period, It is expressed as the mean value of the equipment surface contamination coefficient during the monitoring period. It is expressed as the mean value of the food surface contamination coefficient within the monitoring period, i = 1, 2, 3, ..., n, where n is the monitoring period.

[0087] Specifically, the S3 specific steps also include:

[0088] S32. Preset the correlation range [G1, G2], compare the pollution path correlation index Zwr with the correlation range [G1, G2], analyze the correlation between the equipment surface pollution and the food surface pollution, and determine whether the transmission path of the food surface pollution comes from the processing equipment. The specific contents are as follows:

[0089] If the pollution path correlation index Zwr is greater than G2, it means that there is a direct correlation between the equipment surface pollution and the food surface pollution, that is, the change direction of the two is consistent, indicating that the equipment surface pollution has an impact on the food surface pollution, and at the same time, it means that the transmission path of the food surface pollution comes from the processing equipment. At this time, the map construction instruction will be issued;

[0090] If the contamination path correlation index Zwr is in the correlation range [G1, G2], it means that there is no direct correlation between the equipment surface contamination and the food surface contamination, and it cannot be said that the transmission path of the food surface contamination comes from the processing equipment;

[0091] If the contamination path correlation index Zwr is less than G1, it means that there is no correlation between the equipment surface contamination and the food surface contamination, that is, the two change independently, and it cannot be explained that the transmission path of food surface contamination comes from the processing equipment.

[0092] It should be noted that G1 and G2 in the correlation range [G1, G2] represent the lower limit and upper limit, respectively, which are calculated and obtained through the distribution characteristics of the pollution path correlation index Zwr in the experimental sample, specifically:

[0093] G1=μ-k*σ, G2=μ+k*σ;

[0094] In the formula, k is a constant, usually with a value of 1-3, corresponding to different confidence levels. The specific value is set by the user (according to the actual situation), μ is the mean, and σ is the standard deviation;

[0095] In this embodiment, the pollution path association index Zwr is calculated and analyzed to accurately determine the propagation path of food surface contamination, providing a scientific basis for the construction of food safety knowledge graphs and pollution tracing. Compared with the traditional method of constructing food safety knowledge graphs, the special advantage of this step is that it introduces quantitative analysis based on the statistical mean algorithm, which can extract key features of pollution path association from dynamic data, and effectively solves the problem that pollution propagation paths are difficult to quantify and intuitively determine in traditional methods. Specifically, the pollution path association index Zwr is calculated through multi-period analysis of the equipment surface contamination degree coefficient Xnd and the food surface contamination degree coefficient Xsp, and based on this, the degree of influence of equipment surface contamination on food surface contamination is clarified. When the pollution path association index Zwr is greater than G2, it can be accurately determined that equipment surface contamination is a direct result of food contamination. The method connects to the source and issues a knowledge graph construction instruction to quickly complete pollution tracing; when the pollution path correlation index Zwr is in the correlation range [G1, G2] or less than G1, it can identify that the pollution propagation path does not come from the processing equipment, thereby avoiding resource waste and misjudgment risks; compared with the traditional method that relies on manual experience and offline analysis, this method uses dynamic data-driven quantitative analysis, which not only improves the efficiency and accuracy of pollution tracing, but also reduces production shutdowns and resource waste caused by unknown pollution sources; more importantly, this step provides a strong scientific basis for rapid decision-making through the intuitive and quantified pollution path correlation index Zwr, significantly reduces the risk of spread of food contamination incidents, sets new standards for intelligent management and precise traceability of food processing, and effectively makes up for the shortcomings of traditional food safety monitoring in dynamic analysis capabilities and scientific decision-making support.

[0096] Example 5

[0097] Please refer to Figure 1 , specifically: S4 specific steps include:

[0098] S41. After receiving the graph construction instruction, the food characteristic information data, equipment status information data and external environment status information data received by the cloud platform are input into the time series neural network, and the local features are extracted by performing convolution operation in the convolution layer to generate feature quantities, and the feature quantities are input into the pooling layer to downsample the feature quantities, and then the downsampled feature quantities are input into the fully connected layer to perform a full connection operation, and the probability distribution of each timing parameter information is output, and finally the probability distribution of each timing parameter information is input into the output layer, and the probability distribution of each timing parameter information is decoded in the output layer, and the decoded timing parameter information is aligned according to the timing of the monitoring period, and several groups of aligned data belonging to the same monitoring period are obtained, and the equipment surface contamination degree coefficient Xnd and the food surface contamination degree coefficient Xsp of several groups of monitoring periods are mapped to the aligned data of the corresponding monitoring period to construct a food safety knowledge graph.

[0099] It should be noted that the time series neural network algorithm is a deep learning algorithm specifically used to process time series data. By capturing the dependencies and dynamic change characteristics of data in the time dimension, it provides a powerful tool for the prediction and analysis of complex systems. In the construction of the food safety knowledge graph, the time series neural network algorithm extracts input data through the convolution layer, including local features of food characteristic information data, equipment status information data and external environment status information data, and combines the pooling layer to downsample the features to retain key information. Then, the fully connected layer is used to fuse the features to generate the probability distribution of each time series parameter. Finally, these distributions are decoded and time-series aligned by the output layer to form a dynamic data structure associated with a specific monitoring period. The algorithm is also used to uniformly map multi-source data in the time dimension, including the equipment surface contamination degree coefficient Xnd and the food surface contamination degree coefficient Xsp of several groups of monitoring periods, to provide accurate data association and prediction capabilities for the construction of a dynamic food safety knowledge graph. Through the time series neural network algorithm, the dynamic laws of pollution propagation are revealed, and the pollution diffusion path and risk evolution trend in food processing are intuitively presented, providing scientific support for pollution source tracing, risk assessment and real-time decision-making.

[0100] Specifically, the specific steps of S4 also include:

[0101] S42. Based on the food safety knowledge graph constructed in step S41, the food surface contamination coefficient Xsp is used as the search target word. After data retrieval and establishment of entity relationships, the threshold value comparison is performed on the food surface contamination coefficient Xsp in each monitoring period during the food processing process in chronological order. When the food surface contamination coefficient Xsp exceeds the threshold value, it means that the surface cleanliness of the corresponding processing equipment in the current monitoring period does not meet the food safety processing requirements, and the production line equipment in the current period is cleaned and disinfected. When the food surface contamination coefficient Xsp does not exceed the threshold value, it means that the surface cleanliness of the processing equipment in the food processing process meets the food safety processing requirements, and the next stage of food processing operations is continued.

[0102] In this embodiment, through the constructed food safety knowledge graph and intelligent threshold comparison, in-depth analysis and real-time decision-making of the contamination status in the food processing process are realized, which provides an important guarantee for improving the efficiency and accuracy of food safety management; compared with the traditional static construction method of the food safety knowledge graph, the advantage lies in the dynamic integration of the multi-dimensional data of food, equipment and external environment by combining the time series neural network algorithm, extracting local features through the convolution layer, downsampling by the pooling layer to retain key information, and feature fusion by the fully connected layer, finally constructing an accurate and real-time updated food safety knowledge graph; this method can map the equipment surface contamination degree coefficient Xnd and the food surface contamination degree coefficient Xsp to the corresponding monitoring period data during the time series registration process, which provides intuitive support for the multi-dimensional dynamic presentation of the contamination path; at the same time, the food surface contamination degree coefficient is used as the target word for intelligent It can search and compare thresholds. When the food surface contamination coefficient Xsp exceeds the threshold, it can automatically issue cleaning and disinfection instructions to optimize the production line equipment in the current period. This automated analysis and real-time intervention capability effectively solves the defects of traditional methods in dynamic contamination monitoring and real-time processing, especially in food processing environments with complex multivariate interactions, and significantly improves the efficiency and accuracy of contamination judgment. By constructing a dynamic food safety knowledge graph and monitoring the cleanliness of equipment in real time, this method effectively reduces food safety accidents caused by the spread of contamination, and effectively avoids the waste of resources caused by excessive or unnecessary cleaning. Overall, the S4 step deeply integrates intelligent technology into the monitoring and management of food processing, which not only realizes the accurate judgment of the source of contamination transmission, but also promotes the food processing industry to develop in a more intelligent, real-time and efficient direction.

[0103] Example 6

[0104] Please refer to Figure 1 and Figure 2 ,Specifically: A food safety knowledge graph construction system based on neural network, including a monitoring module, a pollution analysis module, a path association module and a graph construction module;

[0105] The monitoring module is used to perform real-time monitoring according to the dynamic changes of microbial contaminants diffusing from the surface of the processing equipment to the surface of the food during the food processing process, and obtain food characteristic information data, equipment status information data and external environment status information data respectively;

[0106] The pollution analysis module is used to analyze the food characteristic information data and the equipment status information data stored in the cloud platform to construct the equipment surface pollution degree coefficient Xnd for several groups of monitoring periods, and to analyze the food characteristic information data and the external environment status information data to construct the food surface pollution degree coefficient Xsp for several groups of monitoring periods;

[0107] The path association module is used to generate a pollution path association index Zwr based on the equipment surface pollution degree coefficient Xnd of several groups of monitoring periods and the food surface pollution degree coefficient Xsp of several groups of monitoring periods, and to determine the transmission path of food surface pollution from the processing equipment according to the value of the pollution path association index Zwr, and to issue a map construction instruction;

[0108] The graph construction module is used to construct a food safety knowledge graph in combination with a temporal neural network algorithm after receiving a graph construction instruction, and to clean and disinfect production line equipment that does not meet food safety processing requirements during the current monitoring period by performing target word retrieval and threshold comparison.

[0109] In this embodiment, the problem of insufficient real-time, dynamic and intelligent performance of traditional food safety monitoring is effectively solved through the deep integration of modular design and intelligent technology. The monitoring module collects food characteristic information data, equipment status information data and external environment status information data in real time, and combines the equipment surface contamination degree coefficient Xnd and food surface contamination degree coefficient Xsp constructed by the pollution analysis module to effectively realize dynamic modeling and accurate analysis of pollution concentration. The path association module uses the pollution path association index Zwr to quantitatively analyze the propagation path of food surface contamination, and quickly determine whether the contamination comes from the processing equipment, thereby greatly improving the efficiency and accuracy of pollution tracing. The graph construction module combines the time series neural network technology to map multi-source dynamic data into the food safety knowledge graph, effectively realizing the intuitive presentation and real-time update of the contamination path, and accurately identifies the production line equipment that does not meet the safety standards through target word retrieval and threshold comparison, and performs cleaning and disinfection. Compared with the traditional method that relies on manual detection, the system has higher real-time and intelligent levels, which not only improves the response efficiency of pollution events, but also effectively reduces the risk of pollution spread and resource waste, providing strong technical support for the safety and management efficiency of food processing, and at the same time promoting the transformation and development of the food safety field towards intelligence and dynamics.

[0110] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for constructing a food safety knowledge graph based on a neural network, characterized in that: The steps include: S1. Real-time monitoring is performed based on the dynamic changes of microbial contaminants diffusing from the surface of processing equipment to the surface of food during food processing, and food characteristic information data, equipment status information data and external environment status information data are obtained respectively; S2. Analyze the food characteristic information data and the equipment status information data stored in the cloud platform to construct the equipment surface contamination degree coefficient Xnd for several groups of monitoring periods, and analyze the food characteristic information data and the external environment status information data to construct the food surface contamination degree coefficient Xsp for several groups of monitoring periods; S3. Generate a pollution path correlation index Zwr based on the equipment surface pollution degree coefficient Xnd and the food surface pollution degree coefficient Xsp of the several monitoring periods, and determine the transmission path of the food surface pollution from the processing equipment according to the value of the pollution path correlation index Zwr, and issue a map construction instruction; S4. After receiving the graph construction instruction, a food safety knowledge graph is constructed in combination with a time series neural network algorithm. By searching for target words and performing threshold comparison, the production line equipment that does not meet food safety processing requirements during the current monitoring period is cleaned and disinfected.

2. A method for constructing a food safety knowledge graph based on a neural network according to claim 1, characterized in that: The specific steps of S1 include: S11. Real-time monitoring is performed based on the dynamic changes of microbial contaminants diffusing from the equipment surface to the food surface during food processing, and food characteristic information data, equipment status information data and external environment status information data are obtained respectively, including: S111. During the food processing process, the multiple sets of sensors deployed are used to monitor the state of food surface pollutants in real time during the food processing process, obtain food characteristic information data, and transmit the food characteristic information data to the cloud platform in real time through the network. The food characteristic information data includes the microbial reproduction rate Vfz and the initial food residue concentration Xks in each monitoring period. The multiple sets of sensors include a portable microbial detector and a fluorescence sensor; S112, using the cleanliness sensor deployed on the production line equipment to record the pollution change of the equipment after cleaning, obtain the cleaning state decay rate Vsj of each monitoring period, and collect the initial pollution concentration Cwr on the surface of the equipment according to the surface pollution state of the equipment after cleaning. The cleaning state decay rate Vsj of each monitoring period and the initial pollution concentration Cwr construct equipment state information data, and transmit it to the cloud platform in real time through the network; S113. Utilize multiple sets of environmental monitoring equipment deployed in the food processing process to monitor the external environment status in real time during the process of microbial contaminants diffusing from the equipment surface to the food surface, obtain external environment status information data, and transmit it to the cloud platform in real time through the network. The external environment status information data includes the average temperature Twd and the average air flow rate Vkq in each monitoring period.

3. A method for constructing a food safety knowledge graph based on a neural network according to claim 2, characterized in that: The specific steps of S1 also include: S12. Preprocess the food characteristic information data, equipment status information data and external environment status information data received by the cloud platform. The preprocessing includes noise removal, missing value filling and data smoothing operations. The missing value filling methods include mean filling, median filling, interpolation filling and regression filling. The preprocessed food characteristic information data, equipment status information data and external environment status information data are stored.

4. A method for constructing a food safety knowledge graph based on a neural network according to claim 3, characterized in that: The specific steps of S2 include: S21. Extract the characteristics of the food characteristic information data and the equipment status information data stored in the cloud platform, and after dimensionless processing, calculate and obtain the equipment surface contamination degree coefficient Xnd for several groups of monitoring periods. The equipment surface contamination degree coefficient Xnd for the i-th group of monitoring periods is i For example, it is obtained in the following way: Where Cwr is the initial contamination concentration on the equipment surface, Vfz i Vsj is expressed as the microbial growth rate in the ith monitoring period, i It is expressed as the cleaning state decay rate of the ith monitoring period, e * It is expressed as an exponential function with e as the base, which is used to describe that the change of pollution degree is an exponential growth process. e is the Euler number, and α1 and α2 are both expressed as weight values.

5. The method for constructing a food safety knowledge graph based on a neural network according to claim 3, characterized in that: The specific steps of S2 also include: S22, extracting features from the food characteristic information data and the external environment status information data, and calculating and obtaining the food surface contamination degree coefficient Xsp of several groups of monitoring periods after dimensionless processing, and taking the food surface contamination degree coefficient Xsp of the i-th group of monitoring periods as i For example, it is obtained in the following way: Where Xks is the initial residue concentration in food, Twd i Expressed as the average temperature of the ith monitoring period, Vkq i is the average air velocity in the i-th monitoring period, γ is the diffusion sensitivity factor, which is used to describe the effect of temperature and air flow on the diffusion rate, e * It is expressed as an exponential function with e as the base, which is used to describe that the change of pollution degree is an exponential growth process. e is the Euler number, and β1 and β2 are both expressed as weight values.

6. A method for constructing a food safety knowledge graph based on a neural network according to claim 5, characterized in that: The specific steps of S3 include: S31, based on the equipment surface contamination degree coefficient Xnd of several groups of monitoring periods and the food surface contamination degree coefficient Xsp of several groups of monitoring periods, and combined with the statistical averaging algorithm, generate the contamination path association index Zwr, the contamination path association index Zwr is obtained by the following formula: Where Xnd i It is expressed as the equipment surface contamination coefficient of the i-th monitoring period, Xsp i It is expressed as the food surface contamination degree coefficient of the i-th group monitoring period, It is expressed as the mean value of the equipment surface contamination coefficient during the monitoring period. It is expressed as the mean value of the food surface contamination coefficient within the monitoring period, i = 1, 2, 3, ..., n, where n is the monitoring period.

7. The method for constructing a food safety knowledge graph based on a neural network according to claim 6, characterized in that: The specific steps of S3 also include: S32. Preset the correlation range [G1, G2], compare the pollution path correlation index Zwr with the correlation range [G1, G2], analyze the correlation between the equipment surface pollution and the food surface pollution, and determine whether the transmission path of the food surface pollution comes from the processing equipment. The specific contents are as follows: If the pollution path correlation index Zwr is greater than G2, it means that there is a direct correlation between the equipment surface pollution and the food surface pollution, that is, the change direction of the two is consistent, indicating that the equipment surface pollution has an impact on the food surface pollution, and at the same time, it means that the transmission path of the food surface pollution comes from the processing equipment. At this time, the map construction instruction will be issued; If the contamination path correlation index Zwr is in the correlation range [G1, G2], it means that there is no direct correlation between the equipment surface contamination and the food surface contamination, and it cannot be said that the transmission path of the food surface contamination comes from the processing equipment; If the contamination path correlation index Zwr is less than G1, it means that there is no correlation between the equipment surface contamination and the food surface contamination, that is, the two change independently, and it cannot be explained that the transmission path of food surface contamination comes from the processing equipment.

8. The method for constructing a food safety knowledge graph based on a neural network according to claim 7, characterized in that: The specific steps of S4 include: S41. After receiving the graph construction instruction, the food characteristic information data, equipment status information data and external environment status information data received by the cloud platform are input into the time series neural network, and the local features are extracted by performing convolution operation in the convolution layer to generate feature quantities, and the feature quantities are input into the pooling layer to downsample the feature quantities, and then the downsampled feature quantities are input into the fully connected layer to perform a full connection operation, and the probability distribution of each timing parameter information is output, and finally the probability distribution of each timing parameter information is input into the output layer, and the probability distribution of each timing parameter information is decoded in the output layer, and the decoded timing parameter information is aligned according to the timing of the monitoring period, and several groups of aligned data belonging to the same monitoring period are obtained, and the equipment surface contamination degree coefficient Xnd and the food surface contamination degree coefficient Xsp of several groups of monitoring periods are mapped to the aligned data of the corresponding monitoring period to construct a food safety knowledge graph.

9. A method for constructing a food safety knowledge graph based on a neural network according to claim 8, characterized in that: The specific steps of S4 also include: S42. Based on the food safety knowledge graph constructed in step S41, the food surface contamination coefficient Xsp is used as the search target word. After data retrieval and establishment of entity relationships, the threshold value comparison is performed on the food surface contamination coefficient Xsp in each monitoring period during the food processing process in chronological order. When the food surface contamination coefficient Xsp exceeds the threshold value, it means that the surface cleanliness of the corresponding processing equipment in the current monitoring period does not meet the food safety processing requirements, and the production line equipment in the current period is cleaned and disinfected. When the food surface contamination coefficient Xsp does not exceed the threshold value, it means that the surface cleanliness of the processing equipment in the food processing process meets the food safety processing requirements, and the next stage of food processing operations is continued.

10. A food safety knowledge graph construction system based on a neural network, used to implement a food safety knowledge graph construction method based on a neural network as described in any one of claims 1 to 9, characterized in that: It includes monitoring module, pollution analysis module, path association module and graph construction module; The monitoring module is used to perform real-time monitoring according to the dynamic changes of microbial contaminants diffusing from the surface of the processing equipment to the surface of the food during the food processing process, and obtain food characteristic information data, equipment status information data and external environment status information data respectively; The pollution analysis module is used to analyze the food characteristic information data and the equipment status information data stored in the cloud platform to construct the equipment surface pollution degree coefficient Xnd for several groups of monitoring periods, and to analyze the food characteristic information data and the external environment status information data to construct the food surface pollution degree coefficient Xsp for several groups of monitoring periods; The path association module is used to generate a pollution path association index Zwr based on the equipment surface pollution degree coefficient Xnd of several groups of monitoring periods and the food surface pollution degree coefficient Xsp of several groups of monitoring periods, and to determine the transmission path of food surface pollution from the processing equipment according to the value of the pollution path association index Zwr, and to issue a map construction instruction; The graph construction module is used to construct a food safety knowledge graph in combination with a temporal neural network algorithm after receiving a graph construction instruction, and to clean and disinfect production line equipment that does not meet food safety processing requirements during the current monitoring period by performing target word retrieval and threshold comparison.

Images