CN117670378B

CN117670378B - Food safety monitoring method and system based on big data

Info

Publication number: CN117670378B
Application number: CN202410145824.1A
Authority: CN
Inventors: 李燕; 王忠一; 滕伟; 杨冉冉
Original assignee: Yantai Food And Drug Inspection And Testing Center Yantai Adverse Drug Reaction Monitoring Center Yantai Grain And Oil Quality Testing Center
Current assignee: Yantai Food And Drug Inspection And Testing Center Yantai Adverse Drug Reaction Monitoring Center Yantai Grain And Oil Quality Testing Center
Priority date: 2024-02-02
Filing date: 2024-02-02
Publication date: 2024-04-30
Anticipated expiration: 2044-02-02
Also published as: CN117670378A

Abstract

The invention relates to the technical field of food safety data analysis, in particular to a food safety monitoring method and system based on big data, wherein the method comprises the following steps: based on time series data collected in the food processing process, an autoregressive moving average model and spectrum analysis are adopted to conduct primary trend analysis, a long-term and short-term memory network model is utilized to analyze key parameters in the food processing process, historical data including temperature, humidity and pH value are included, changes in the production process are identified, and key parameter change trend analysis results are generated. According to the invention, the high-dimensional time sequence analysis is performed by the autoregressive moving average model and the spectrum analysis, so that the monitoring of key parameters in the food processing process is more accurate, and the efficiency of resource allocation and the structural optimization of the sample detection flow are obviously improved by the monitoring flow optimized by the branch-and-bound algorithm and the cut plane method. The application of probabilistic risk assessment techniques and logical tree analysis allows for a more comprehensive and systematic risk assessment.

Description

Food safety monitoring method and system based on big data

Technical Field

The invention relates to the technical field of food safety data analysis, in particular to a food safety monitoring method and system based on big data.

Background

The technical field of food safety data analysis is focused on using data analysis techniques to ensure the safety and quality of food. It relates to collecting and analyzing data related to food production, processing, distribution and consumption in order to discover food safety problems in time and to take precautions. Developments in this field benefit from advanced data processing techniques such as big data analysis, artificial intelligence, and machine learning, which are capable of processing and analyzing large amounts of food safety data, thereby improving monitoring efficiency and accuracy.

The purpose of the food safety monitoring method is to ensure the safety and compliance of the food by continuous and systematic monitoring. This includes identifying and assessing substances that pose a threat to consumer health, such as microorganisms, chemical residues, and other potentially harmful substances. Such monitoring methods aim to reduce the incidence of food-borne diseases, ensure public health, and improve consumer confidence in food safety. In addition, the method is also helpful for food manufacturers and distributors to observe relevant laws and regulations, and ensures the market competitiveness of the products.

Although the existing food safety monitoring technology can effectively identify and evaluate substances threatening the health of consumers and reduce the occurrence rate of food-borne diseases to a certain extent, the traditional method still has defects in the aspects of processing complex data and real-time monitoring, the traditional method is difficult to accurately predict and rapidly respond to food safety risks when analyzing high-dimensional and large-scale time sequence data, in addition, the traditional method has defects in the aspects of optimizing and resource allocation efficiency of monitoring processes, such that the traditional method has flexibility and adaptability, the defects in the aspects of resource utilization and monitoring efficiency are caused, the traditional method has difficulty in comprehensively considering multi-factor interaction and risk identification and evaluation for comprehensive evaluation of food safety risks, the traditional method has difficulty in tracking key links of food circulation in the aspect of food safety tracing, the efficient data analysis capability has influence on the speed and accuracy of tracing, and the traditional method has advantages of improving the efficiency and accuracy in identifying risk product batches in the aspect of controlling food quality, and in the aspect of optimizing the whole food monitoring process, and the traditional method has space improvement in the aspects of depth and breadth of large data analysis, especially in the aspect of monitoring process analysis and optimization.

Disclosure of Invention

The application provides a food safety monitoring method and a system based on big data, which solve the problems that although the existing food safety monitoring technology can effectively identify and evaluate substances threatening the health of consumers and reduce the occurrence rate of food-borne diseases to a certain extent, the traditional method still has defects in the aspects of complex data processing and real-time monitoring, the traditional method is difficult to accurately predict and quickly respond to food safety risks when analyzing high-dimensional and large-scale time series data, in addition, the traditional method has the defects of flexibility and adaptability in terms of optimizing and resource allocation efficiency of monitoring processes, the defects of resource utilization and monitoring efficiency are caused, the comprehensive evaluation of the food safety risks is difficult to comprehensively consider multi-factor interaction and risk identification and evaluation, the traditional method has the defects of influencing the speed and accuracy of tracing in terms of food safety sources although the key links of food circulation can be tracked, the traditional method has the problems of improving the efficiency and accuracy in terms of identifying product batches, and finally, the traditional method has the problems of optimizing and optimizing the whole food safety monitoring processes in terms of optimizing and optimizing the whole food safety processes, and improving the data in particular in terms of depth and space monitoring and optimizing the aspects.

In view of the above problems, the application provides a food safety monitoring method and system based on big data.

The application provides a food safety monitoring method based on big data, wherein the method comprises the following steps:

S1: based on time series data collected in the food processing process, adopting an autoregressive moving average model and spectrum analysis to perform primary trend analysis, utilizing a long-term and short-term memory network model to analyze key parameters in the food processing process, including historical data of temperature, humidity and pH value, identifying changes in the production process, and generating a key parameter change trend analysis result;

S2: based on the key parameter change trend analysis result, analyzing the food safety monitoring flow by adopting a branch-and-bound algorithm and a cut plane method, and carrying out structural optimization and efficiency adjustment on a sample detection flow and monitoring resource allocation to generate an optimized food safety monitoring flow scheme;

S3: based on the optimized food safety monitoring flow scheme, performing risk assessment on the monitoring flow by adopting a probability risk assessment technology and logic tree analysis, wherein the risk assessment comprises the steps of identifying a risk source and assessing potential threat caused by the risk source to food safety, and generating a risk assessment result;

s4: based on the risk assessment result, a food safety tracing mechanism is constructed by adopting a data association analysis and link tracking technology, wherein the food safety tracing mechanism comprises the steps of tracking the production, processing and distribution processes of food, and integrating data to generate a food safety tracing data set;

S5: based on the food safety traceability dataset, adopting a spatial clustering and spectral clustering technology based on density, and analyzing food safety test results of multiple batches of products, wherein the analysis comprises the steps of grouping the products according to risk degrees to generate a risk product batch identification result;

S6: based on the identification result of the risk product batch, a Tableau tool is adopted, and a process mining technology and sample processing rate analysis are combined to analyze the detection site workflow in the food monitoring process, wherein the analysis comprises the steps of identifying a process bottleneck and an improvement scheme, and generating a detection site workflow optimization scheme;

s7: based on the key parameter change trend analysis result, the optimized food safety monitoring flow scheme, the risk assessment result, the food safety traceability data set, the risk product batch identification result and the detection station work flow optimization scheme, a data fusion technology is adopted to integrate data, and data analysis and application strategy formulation are performed to generate a food safety monitoring network.

Preferably, the key parameter change trend analysis result comprises a temperature change chart, a humidity fluctuation analysis and a pH value history trend, the optimized food safety monitoring flow scheme comprises a redirected resource allocation scheme, a new sample detection sequence and an adjusted monitoring point layout, the risk assessment result comprises a potential risk point list, a risk grade assessment and a risk result, the food safety traceability data set comprises a raw material batch record, a time mark of a processing link and distribution link tracking information, the risk product batch identification result comprises an emergency attention batch list, a batch classification to be observed and a conventional monitoring batch overview, the detection station workflow optimization scheme comprises an identified workflow bottleneck, an improved measure list and an efficiency improvement prediction, and the food safety monitoring network comprises a data integration structure diagram, a risk management flow chart and a quality control strategy overview.

Preferably, based on time series data collected during food processing, an autoregressive moving average model and spectrum analysis are adopted to perform primary trend analysis, a long-term and short-term memory network model is utilized to analyze key parameters in the food processing, including historical data of temperature, humidity and pH value, and identify changes in the production process, and a step of generating a key parameter change trend analysis result further comprises:

s101: based on time series data collected in the food processing process, adopting an autoregressive moving average model, analyzing autocorrelation and partial autocorrelation in the data by a time series prediction technology in statistical analysis, selecting a hysteresis period number and a moving average term number, and generating a time series preliminary analysis result;

S102: based on the preliminary analysis result of the time sequence, a spectrum analysis method is adopted, and the periodic fluctuation and key frequency components of the data are identified and extracted by calculating the power spectrum density of the time sequence data, and the analysis of the data periodicity and frequency distribution is performed to generate a time sequence spectrum analysis result;

S103: based on the time sequence spectrum analysis result, a long-term and short-term memory network model is adopted, dependency problems in the time sequence data are processed through a circulation unit in a neural network, key parameter historical data in food processing are analyzed, and a key parameter historical analysis result is generated;

S104: based on the key parameter historical analysis result, adopting a statistical analysis method to identify abnormal fluctuation or trend change of the parameter by carrying out statistical analysis on the historical data of the key parameters of temperature, humidity and pH value, and generating a key parameter change trend analysis result.

Preferably, based on the analysis result of the key parameter variation trend, a branch-and-bound algorithm and a cut plane method are adopted to analyze the food safety monitoring flow, and structure optimization and efficiency adjustment are performed on the sample detection flow and the monitoring resource allocation, so as to generate an optimized food safety monitoring flow scheme, and the method further comprises the following steps:

S201: based on the key parameter change trend analysis result, a branch-and-bound algorithm is adopted, a decision tree is constructed to evaluate the path of the monitoring flow, and path selection and pruning are carried out according to cost and efficiency indexes to generate a preliminary optimization scheme of the monitoring flow;

S202: based on the preliminary optimization scheme of the monitoring flow, adopting a cut plane method, and optimizing and restraining a decision path by introducing a solution of a linear programming problem into a decision tree model to generate a secondary optimization scheme of the monitoring flow;

s203: based on the secondary optimization scheme of the monitoring flow, dynamic flow simulation analysis is adopted, and key factors of processing rate and detection accuracy are identified and optimized by simulating the performance response of the sample detection flow under multiple configurations, so that a sample detection flow optimization scheme is generated;

S204: based on the sample detection flow optimization scheme, a resource allocation optimization model is adopted, a resource allocation strategy is identified by analyzing the distribution and efficiency of resources in a monitoring flow, the allocation and the use of the resources are carried out, and an optimized food safety monitoring flow scheme is generated.

Preferably, based on the optimized food safety monitoring flow scheme, a probability risk assessment technology and a logic tree analysis are adopted to perform risk assessment on the monitoring flow, including the steps of identifying a risk source, assessing potential threat caused by the risk source to food safety, and generating a risk assessment result, and further including:

S301: based on the optimized food safety monitoring flow scheme, monte Carlo simulation is adopted, a random variable and probability distribution model is constructed, a scene of risk occurrence is simulated, the occurrence probability and influence degree of potential risk points in the monitoring flow are analyzed, and a risk source identification result is generated;

S302: based on the risk source identification result, fault tree analysis is adopted, a causal relationship chain is established by gradually decomposing risk events, potential threat and influence of each risk point on food safety are evaluated, and a risk point evaluation result is generated;

S303: based on the risk point evaluation result, adopting Bayesian network analysis, quantifying interdependence and conditional probability among risk sources by constructing a probability graph model, evaluating the comprehensive influence of the risk sources, evaluating the risk level, and generating a risk level quantification result;

S304: based on the risk level quantification result, a comprehensive risk assessment model is adopted, all risks in a monitoring process are assessed through analysis of the accumulated effect and interaction of risks, information of all risk points is integrated, risk comprehensive assessment of the whole process is carried out, and a risk assessment result is generated.

Preferably, based on the risk assessment result, a food safety tracing mechanism is constructed by adopting a data association analysis and link tracking technology, including tracking the production, processing and distribution processes of food, integrating data, and generating a food safety tracing dataset, and further including:

S401: based on the risk assessment result, adopting a K-means clustering algorithm, and gathering data in the processes of food production, processing and distribution into various categories by calculating Euclidean distance among data points, identifying key data modes and characteristics in the process of food circulation, and generating a food circulation characteristic map;

S402: based on the food circulation feature mapping, adopting an Apriori algorithm, and revealing the internal connection and mode from food production to distribution multiple links by analyzing and identifying frequent item sets and rules to generate a food link association map;

S403: based on the food link association map, adopting graph theory analysis, identifying key nodes and paths by constructing and analyzing a network map of food circulation, tracking production, processing and distribution paths of food, and generating a food safety path network analysis result;

S404: based on the food safety path network analysis result, a data fusion algorithm is adopted, and the integrity and the accuracy of data are verified by integrating the key data points and the path information of multiple links, so that a food safety traceability data set is generated.

Preferably, based on the food safety traceability dataset, a spatial clustering and spectral clustering technology based on density is adopted to analyze food safety test results of multiple batches of products, including grouping the products according to risk degrees, and generating a risk product batch identification result, and the method further includes:

S501: based on the food safety traceability dataset, a DBSCAN algorithm is adopted, a core point, a boundary point and a noise point are identified by calculating density reachability among data points, data points of a multi-density area are distinguished in a multi-dimensional data space, product batches with multiple risk degrees are preliminarily selected, and a preliminary risk product partitioning result is generated;

S502: based on the primary risk product partitioning result, a spectral clustering algorithm is adopted, a similarity matrix of data is constructed, a spectral decomposition technology is applied, product batches are grouped, product batches with multiple risk grades are refined, and a refined risk product grouping result is generated;

S503: based on the refined risk product grouping result, a support vector machine algorithm is adopted, classification analysis is carried out on risk characteristics of product batches through construction and training of a classification model, risk grade assessment is carried out on multiple batches of products, and a risk degree assessment result is generated;

S504: based on the risk degree evaluation result, a multivariate statistical analysis method is adopted, risk evaluation is carried out on the product batches by integrating multiple types of risk indexes, risk classification of each product batch is matched, and a risk product batch identification result is generated.

Preferably, based on the identification result of the risk product batch, a Tableau tool is adopted, and in combination with a process mining technology and sample processing rate analysis, the analysis is performed on the detection site workflow in the food monitoring process, including the steps of identifying a process bottleneck and an improvement scheme, and generating a detection site workflow optimization scheme, and further including:

S601: based on the risk product batch identification result, adopting a flow analysis method, analyzing workflow data of a detection site through Tableau tools, identifying bottlenecks and potential optimization points in a workflow, and generating a flow efficiency analysis and improvement scheme;

S602: based on the flow efficiency analysis and improvement scheme, reconstructing a working flow by analyzing a flow log by adopting an Alpha algorithm, identifying key activities and path dependencies, finding links needing improvement, and generating a working flow reconstruction and analysis result;

S603: based on the workflow reconstruction and analysis results, adopting time series analysis, analyzing the time consumption and efficiency of a processing link by evaluating the time series data of sample processing, and identifying a flow improvement area to generate a sample processing time efficiency analysis result;

S604: based on the analysis result of the sample processing time efficiency, a multidimensional optimization strategy is adopted, and flow improvement measures and a resource optimization strategy are provided by integrating flow efficiency, resource allocation and time management factors, so that a detection site workflow optimization scheme is generated.

Preferably, based on the key parameter variation trend analysis result, the optimized food safety monitoring flow scheme, the risk assessment result, the food safety traceability dataset, the risk product batch identification result and the detection site workflow optimization scheme, the data integration technology is adopted to perform data integration operation, and the steps of data analysis and application policy formulation are performed to generate the food safety monitoring network, and the method further comprises the steps of:

s701: based on the key parameter change trend analysis result, adopting a multidimensional scale analysis algorithm, revealing the interaction and mode among key parameters through measurement and space configuration of the distance among the parameters, and generating a parameter influence relation analysis result;

S702: based on the optimized food safety monitoring flow scheme and the risk assessment result, adopting a linear programming algorithm, optimizing resource allocation and flow design by establishing a mathematical model, analyzing the efficiency of the flow and the resource allocation, and generating a flow design and resource allocation optimization result;

s703: based on the food safety traceability dataset and the risk product batch identification result, adopting an association rule mining algorithm, and analyzing key factors of a food safety control point by analyzing frequent patterns and associations among data items to generate a food safety control point analysis result;

S704: based on the detection site workflow optimization scheme, parameter influence relation analysis results, flow design and resource allocation optimization results and food safety control point analysis results, a data fusion technology is adopted, and a food safety monitoring framework covering multiple links from production to distribution is constructed by integrating multi-stage key data and analysis results, so that a food safety monitoring network is generated.

The system comprises a time sequence analysis module, a monitoring flow optimization module, a risk assessment module, a traceable data integration module, a risk identification module, a flow efficiency analysis module, a key parameter analysis module and a comprehensive monitoring network construction module;

The time sequence analysis module adopts an autoregressive moving average model to carry out statistical analysis of the time sequence based on time sequence data collected in the food processing process, and comprises calculation of autocorrelation and partial autocorrelation, and selection of hysteresis period number and moving average term number to generate a time sequence statistical analysis result;

the monitoring flow optimizing module adopts a branch-and-bound algorithm and a planar cutting method to evaluate a decision path of the monitoring flow based on a time sequence statistical analysis result, selects and prunes the flow path through an optimizing algorithm, and simultaneously optimizes a sample detection flow and monitoring resource configuration by referring to cost and efficiency indexes and dynamic flow simulation analysis to generate a flow and resource optimizing decision scheme;

The risk assessment module is based on a flow and resource optimization decision scheme, adopts Monte Carlo simulation and fault tree analysis, combines probability risk assessment technology and logic tree analysis, carries out risk assessment, and comprises risk occurrence probability, influence degree simulation and causal relation analysis of key risk points to generate a risk assessment result;

The traceability data integration module is used for carrying out category identification and internal connection analysis on data of food production, processing and distribution links by adopting a K-means clustering algorithm and an Apriori algorithm and a data association analysis and link tracking technology based on a risk assessment result, integrating key data modes and characteristics of food circulation and generating a food safety traceability comprehensive data set;

the risk identification module classifies the product batches and evaluates the risk level based on a food safety traceability comprehensive data set by adopting a spatial clustering and spectral clustering technology based on density and combining a support vector machine algorithm to generate a refined risk product batch classification result;

the flow efficiency analysis module is used for analyzing data of the work flow of the detection site by adopting Tableau tools and combining a flow mining technology based on the refined risk product batch classification result, and comprises flow bottleneck identification and potential optimization point analysis, so as to generate a flow efficiency optimization scheme;

The key parameter analysis module is based on a process efficiency optimization scheme, adopts a multidimensional scale analysis algorithm and a linear programming algorithm to carry out space mapping and resource configuration model construction on key parameters in a monitoring process, analyzes the mutual influence among the parameters and the rationality of resource configuration, and generates a parameter influence and resource configuration result;

The comprehensive monitoring network construction module is used for constructing a monitoring framework covering production to distribution on the basis of a time sequence statistical analysis result, a flow and resource optimization decision scheme, a risk assessment result, a food safety traceability comprehensive data set, a refined risk product batch classification result, a flow efficiency optimization scheme, parameter influence and resource configuration result, comprehensively integrating key data and analysis results of multiple stages by adopting a data fusion technology, and generating a food safety monitoring comprehensive network.

One or more technical schemes provided by the application have at least the following technical effects or advantages:

The high-dimensional time series analysis performed by the autoregressive moving average model and the spectrum analysis enables the monitoring of key parameters in the food processing process to be more accurate, and potential food safety risks can be found and early warned in time. The efficiency of resource allocation and the structural optimization of the sample detection flow are obviously improved through the branch-and-bound algorithm and the monitoring flow optimized by the cut plane method. The probability risk assessment technology and the logic tree analysis are applied, so that the risk assessment is more comprehensive and systematic, the risk sources in food production can be accurately identified, and the potential threat to food safety can be effectively assessed. In the aspect of food safety tracing, the combination of data association analysis and link tracing technology provides powerful technical support for constructing an efficient food safety tracing mechanism. The tracing speed of the food from raw materials to each link of consumption is accelerated, and the tracing accuracy is improved. And the product is subjected to food safety test result analysis by using a spatial clustering and spectral clustering technology based on density, so that the identification of risk product batches is more accurate, and the effect of food quality control is effectively improved. In addition, the monitoring process optimization based on Tableau tools and process mining technology not only identifies the bottleneck in the process, but also provides effective improvement measures, and further enhances the efficiency and effect of the monitoring process.

The foregoing description is only an overview of the present application, and is intended to be implemented in accordance with the teachings of the present application in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present application more readily apparent.

Drawings

Fig. 1 is a schematic overall flow chart of a food safety monitoring method based on big data;

fig. 2 is a schematic diagram of a specific flow of S1 of a food safety monitoring method based on big data according to the present invention;

FIG. 3 is a schematic diagram of a specific flow of S2 of a food safety monitoring method based on big data according to the present invention;

fig. 4 is a schematic diagram of a specific flow of S3 of a food safety monitoring method based on big data according to the present invention;

FIG. 5 is a schematic diagram showing a specific flow of S4 of a food safety monitoring method based on big data according to the present invention;

FIG. 6 is a schematic diagram showing a specific flow of S5 of a food safety monitoring method based on big data according to the present invention;

FIG. 7 is a schematic diagram showing a specific flow of S6 of a food safety monitoring method based on big data according to the present invention;

FIG. 8 is a schematic diagram of a specific flow of S7 of a food safety monitoring method based on big data according to the present invention;

fig. 9 is a block diagram of a food safety monitoring system based on big data according to the present invention.

Detailed Description

The application provides a food safety monitoring method and system based on big data.

Summary of the application:

Aiming at the technical problems, the technical scheme provided by the application has the following overall thought:

As shown in fig. 1, the present application provides a food safety monitoring method based on big data, wherein the method comprises:

S2: based on the key parameter change trend analysis result, adopting a branch-and-bound algorithm and a cut plane method to analyze the food safety monitoring flow, and carrying out structural optimization and efficiency adjustment on the sample detection flow and the monitoring resource configuration to generate an optimized food safety monitoring flow scheme;

s5: based on a food safety traceability dataset, adopting a spatial clustering and spectral clustering technology based on density, and analyzing food safety test results of multiple batches of products, wherein the analysis comprises the steps of grouping the products according to the risk degree to generate a risk product batch identification result;

S6: based on the identification result of the risk product batch, adopting Tableau tools, combining a process mining technology and sample processing rate analysis, analyzing the detection site work flow in the food monitoring process, including identifying a process bottleneck and an improvement scheme, and generating a detection site work flow optimization scheme;

s7: based on key parameter change trend analysis results, optimized food safety monitoring flow schemes, risk assessment results, food safety traceability data sets, risk product batch identification results and detection site workflow optimization schemes, a data fusion technology is adopted to conduct data integration operation, data analysis and application strategy formulation are conducted, and a food safety monitoring network is generated.

The key parameter change trend analysis results comprise a temperature change chart, humidity fluctuation analysis and pH value history trend, the optimized food safety monitoring flow scheme comprises a redirected resource allocation scheme, a new sample detection sequence and an adjusted monitoring point layout, the risk assessment results comprise a potential risk point list, a risk grade assessment and a risk result, the food safety traceability data set comprises a raw material batch record, a time mark of a processing link and distribution link tracking information, the risk product batch identification result comprises an emergency attention batch list, a batch classification to be observed and a conventional monitoring batch overview, the detection station workflow optimization scheme comprises an identified workflow bottleneck, an improved measure list and an efficiency improvement prediction, and the food safety monitoring network comprises a data integration structure diagram, a risk management flow chart and a quality control strategy overview.

In step S1, time series data such as temperature, humidity, pH, etc. during food processing are collected. The data format is continuous time sequence value, record the historical change of each key parameter. And performing preliminary trend analysis on the time series data by adopting an autoregressive moving average model and spectrum analysis. The autoregressive moving average model predicts future parameter changes by identifying the autocorrelation and hysteresis characteristics of the data. Spectral analysis is then used to identify periodic fluctuations and critical frequency components in the data. In addition, long and short term memory network models are used to analyze these time series data in depth to identify minor variations in the production process. Finally, key parameter change trend analysis results, such as a temperature change chart, humidity fluctuation analysis and pH value historical trend, are generated in the step, and basic data is provided for subsequent risk assessment.

S2, based on the analysis result, the step adopts a branch-and-bound algorithm and a cutting plane method to carry out structural optimization on the food safety monitoring flow. The branch-and-bound algorithm evaluates and optimizes the path selection of the monitoring flow by building a decision tree model, while the cut plane algorithm further optimizes and constrains the decision path. The purpose of this step is to reduce the complexity of the sample detection procedure and to increase the efficiency of monitoring the resource allocation. The optimized food safety monitoring flow scheme comprises a redirected resource allocation scheme, a new sample detection sequence and an adjusted monitoring point layout.

And S3, carrying out risk assessment by adopting a probability risk assessment technology and logic tree analysis on the basis of the optimized monitoring flow. Through probabilistic risk assessment techniques, potential threats posed by risk sources to food safety can be quantified, while logical tree analysis is used to further analyze and identify risk sources. The risk evaluation result generated by the step comprises a potential risk point list, risk grade evaluation and risk results, and an important basis is provided for formulating a coping strategy.

In the step S4, a food safety tracing mechanism is constructed by utilizing a data association analysis and link tracking technology. Data correlation analysis is used to identify key data patterns and features during food production, processing and distribution, while link tracking techniques track the entire circulation path of food products. The output of the step is a food safety traceability data set which comprises raw material batch records, time marks of processing links and distribution link trace information.

In the step S5, based on the food safety traceability dataset, the food safety test results of the multi-batch products are analyzed by adopting a spatial clustering and spectral clustering technology based on the density. Spatial clustering based on density (such as DBSCAN algorithm) identifies data points of different density regions, and spectral clustering technology is used to group products according to risk degree. This process generates a risk product lot identification result, including an emergency attention lot list and lot classifications to be observed.

In step S6, analyzing the work flow of the detection site in the food monitoring process by utilizing Tableau tools and a flow mining technology and combining sample processing rate analysis. The purpose of this step is to identify bottlenecks and potential improvements in the flow. Through data visualization and flow analysis, the generated detection site workflow optimization scheme includes identified workflow bottlenecks, improvement measure lists, and efficiency improvement predictions.

In the step S7, the analysis result and the optimization scheme of the previous step are synthesized, and the data integration technology is adopted for data integration. This step creates a comprehensive food safety monitoring network by analyzing and applying policy formulation. The network includes data integration block diagrams, risk management flow charts, and quality control strategy summaries, which are intended to provide a comprehensive view to monitor the overall food production and distribution process.

Specifically, as shown in fig. 2, based on time series data collected during food processing, an autoregressive moving average model and spectrum analysis are adopted to perform primary trend analysis, a long-term memory network model is utilized to analyze key parameters including historical data of temperature, humidity and pH value during food processing, and identify changes during production, and a step of generating a key parameter change trend analysis result further comprises:

S103: based on the time sequence spectrum analysis result, a long-term and short-term memory network model is adopted, the dependence problem in the time sequence data is processed through a circulation unit in a neural network, key parameter historical data in food processing is analyzed, and a key parameter historical analysis result is generated;

S104: based on the historical analysis results of the key parameters, adopting a statistical analysis method, and identifying abnormal fluctuation or trend change of the parameters by carrying out statistical analysis on the historical data of the key parameters of temperature, humidity and pH value to generate the analysis results of the change trend of the key parameters.

In the sub-step S101, time series data collected during food processing is initially analyzed by an autoregressive moving average model. The model combines both autoregressive and moving average methods to facilitate capturing the dynamics of the time series. Specifically, the model first identifies the autocorrelation and partial autocorrelation of the data, and determines the lag period number and the moving average term number by analyzing the autocorrelation function and partial autocorrelation function graphs. After the model parameters are determined, the parameters of the autoregressive moving average model are estimated by using a least square method or a maximum likelihood estimation method, so that a preliminary analysis result of the time sequence is generated. The step can effectively reveal the basic structure and dynamic change trend of the time sequence, and provides a basis for subsequent deep analysis.

In the sub-step S102, the data is further analyzed by a spectral analysis method based on the preliminary analysis result of the time series. Spectral analysis is a fourier transform-based method for detecting periodic fluctuations and critical frequency components in time series data. By calculating the power spectral density of the time series, the main frequency components and periodicity characteristics of the data can be revealed. This step includes performing a fourier transform on the time series data to obtain a spectrogram, and then analyzing peaks in the spectrogram to identify the dominant period and frequency distribution characteristics of the data. Spectral analysis results are helpful in understanding the periodic variations and key frequency components in food processing data, and are critical for subsequent pattern recognition and trend prediction.

In the step S103, based on the result of the time series spectrum analysis, a long-short-term memory network model is applied, and the long-short-term memory network model is a special type of cyclic neural network and is very suitable for processing the long-term dependence problem in the time series data. In this step, the long-short term memory network model processes time series data through its unique gating mechanism (including forgetting gate, input gate and output gate), so that complex relationships and dynamic changes of key parameters in the food processing process can be captured. By training the long-term and short-term memory network model, historical analysis results of key parameters can be generated, and the dynamic change modes of parameters such as temperature, humidity, pH value and the like along with time are revealed. These analytical results are critical to understanding and predicting key parameter changes in food processing.

In the step S104, based on the key parameter historical analysis result generated by the long-short-period memory network model, the variation trend of the parameters is further analyzed by adopting a statistical analysis method. By carrying out detailed statistical analysis on historical data of key parameters such as temperature, humidity, pH value and the like, abnormal fluctuation or trend change of the parameters can be identified. This includes calculating statistical indicators of mean, standard deviation, skewness, kurtosis, etc. for each parameter, and performing correlation analysis and trend line fitting. The result of this step is a detailed key parameter trend analysis report that reveals not only the general pattern of change for each parameter, but also identifies abnormal fluctuations or significant trend changes, which is of great importance for the establishment of food safety control measures and for the improvement of the production flow.

The following time series data were assumed to be collected during food processing: average temperature per day (°c), average humidity (%) and average pH. These data items are respectively: temperature [22, 23, 24, 23, 22, 21, 20, 21, 22, 23], humidity [30, 32, 35, 33, 31, 29, 28, 30, 32, 34], pH [5.5, 5.6, 5.7, 5.6, 5.5, 5.4, 5.3, 5.4, 5.5, 5.6]. In step S101, these data are analyzed by an autoregressive moving average model to determine the hysteresis period number and the moving average term number, for example, to derive autoregressive and moving average models for the temperature data. Then, the model parameters are estimated by using a least square method, a preliminary analysis result of the time series is generated, and the basic variation trend of the temperature, the humidity and the pH value is revealed. In step S102, fourier transform is performed on these data, resulting in a power spectral density map for each parameter. For example, the power spectrum of the temperature data shows the main frequency component of a one-week period. In step S103, a long-short term memory network model is trained to capture the complex relationship of temperature, humidity and pH over time. For example, long and short term memory network models reveal the effect of temperature on pH under specific humidity conditions. In step S104, statistical analysis is performed, statistical indexes of each parameter, such as an average value, a standard deviation, and the like of temperature, are calculated, and abnormal fluctuations or trend changes are identified. For example, analysis shows an abnormal rise in temperature under certain humidity conditions. Through the steps, a comprehensive food processing parameter analysis report can be generated, a dynamic change mode of temperature, humidity and pH value and any abnormal or remarkable trend are provided, and basis is provided for food safety control and production flow improvement.

Specifically, as shown in fig. 3, based on the analysis result of the key parameter variation trend, a branch-and-bound algorithm and a cut plane method are adopted to analyze the food safety monitoring flow, and structure optimization and efficiency adjustment are performed on the sample detection flow and the monitoring resource allocation, so as to generate an optimized food safety monitoring flow scheme, and the method further comprises the steps of:

S202: based on a monitoring flow primary optimization scheme, adopting a cut plane method, and optimizing and restraining a decision path by introducing a solution of a linear programming problem into a decision tree model to generate a monitoring flow secondary optimization scheme;

S203: based on a secondary optimization scheme of the monitoring flow, adopting dynamic flow simulation analysis, and identifying and optimizing key factors of processing rate and detection accuracy by simulating performance response of the sample detection flow under multiple configurations to generate a sample detection flow optimization scheme;

In the sub-step S201, the path of the food safety monitoring flow is optimally analyzed by a branch-and-bound algorithm. First, a decision tree is constructed representing the various paths in the monitoring process. Each node of the decision tree represents a decision point, for example, checking a sample or adjusting a temperature parameter at a specific point. And (5) assigning cost and efficiency indexes to each decision point based on the key parameter change trend analysis result. The tree is then traversed using a branch-and-bound algorithm. In this process, the algorithm evaluates the cumulative cost and efficiency of each path, and by setting thresholds and constraints, eliminates over-cost or over-efficiency paths. The algorithm continues to iterate until an optimal path is found, i.e., the least costly and most efficient combination. The preliminary optimization scheme of the monitoring flow generated in the step comprises a series of operation steps and parameter settings, which together form a cost-effective and maximized monitoring flow.

In the sub-step S202, the monitoring flow is further optimized by using the facet method. Based on the preliminary optimization scheme generated in S201, this step optimizes and constrains the decision path by introducing a solution to the linear programming problem in the decision tree model. The cut plane method gradually improves the solution of the linear programming by adding a linear inequality. The algorithm first solves a relaxed linear programming problem and then analyzes the integer nature of the solution, if the solution is not an integer, then adds a cutting plane, i.e., a linear inequality, to exclude the current non-integer solution. And iterating the process until an integer solution meeting all conditions is found, so as to generate a secondary optimization scheme of the monitoring flow. The scheme provides more refined monitoring flow adjustment, such as adjustment of detection frequency, reallocation of resources and the like, and aims to further improve monitoring efficiency and reduce unnecessary cost.

In the sub-step S203, dynamic flow simulation analysis is applied to optimize the sample detection flow. This step identifies and optimizes the critical factors of process rate and detection accuracy by modeling the performance response of the sample detection process under different configurations. The simulation model considers various operating conditions, such as equipment capabilities, personnel configuration, detection methods, etc., and simulates the flow runs under these conditions. And identifying the bottleneck and the low-efficiency links in the flow by evaluating indexes such as processing time, error rate, resource utilization rate and the like under different conditions. These links are then adjusted and optimized, such as to add personnel configuration at critical detection points, to improve detection techniques, or to adjust the order of detection steps. Finally, a specific sample detection flow optimization scheme is generated that specifies how to adjust the flow to improve overall efficiency and accuracy.

In the S204 substep, the resource allocation optimization model is adopted to carry out final optimization on the monitoring flow. This step focuses on analyzing and optimizing the resource distribution and efficiency in the monitoring process. By evaluating the use of resources in a monitoring process, such as the allocation of personnel, equipment and funds, imbalances and waste in resource allocation are identified. The resources are then reconfigured using an optimization model, such as a linear or nonlinear programming model. The model considers the cost, benefit and constraint condition of the resource, and obtains the optimal resource allocation scheme through optimization calculation. The scheme not only improves the use efficiency of resources, but also ensures the smoothness and high efficiency of the monitoring flow. The finally generated optimized food safety monitoring flow scheme is a comprehensive document, and the allocation and use condition of resources in each step and the expected efficiency improvement effect are described in detail.

Assuming that in a food processing plant, the monitoring process includes routine monitoring of temperature, humidity and pH, the cost and efficiency indicators for each test have been derived from prior analysis. For example, the cost of temperature detection is 100 yuan per time, with an efficiency score of 90; the cost of humidity detection is 80 yuan each time, and the efficiency score is 85; the cost of the pH detection was 120 yuan per time and the efficiency score was 95. In step S201, a decision tree is constructed containing different combinations and frequencies of the three detections. Using the branch-and-bound algorithm, the cost threshold is set to 300 yuan per day and the efficiency threshold is 85. By iterative calculation, the detection combination with the lowest cost and highest efficiency is found, such as twice daily temperature detection, once humidity detection and once pH value detection. In step S202, a planar cutting method is further applied to the solution, and constraints are added to optimize the detection frequency and sequence, such as first performing temperature detection, and then performing pH detection if the temperature is abnormal. In step S203, the processing time and accuracy under different detection sequences and frequencies are evaluated by dynamic flow simulation analysis, and the optimization point is identified, such as increasing the frequency of pH detection under high temperature conditions. In step S204, personnel and equipment resources are reallocated according to the optimized detection flow, so that each detection point is ensured to have enough resource support, and resource waste is avoided. Through the steps, a detailed monitoring flow optimization scheme is finally generated, wherein the detailed monitoring flow optimization scheme comprises the optimal frequency, sequence and corresponding resource allocation plan of each item of detection, and aims to improve the overall monitoring efficiency and reduce unnecessary cost.

Specifically, as shown in fig. 4, based on the optimized food safety monitoring flow scheme, a probability risk assessment technology and a logic tree analysis are adopted to perform risk assessment on the monitoring flow, including the steps of identifying a risk source, assessing a potential threat caused by the risk source to food safety, and generating a risk assessment result, and further including:

S301: based on the optimized food safety monitoring flow scheme, adopting Monte Carlo simulation, simulating a scene of risk occurrence by constructing a random variable and probability distribution model, analyzing occurrence probability and influence degree of potential risk points in the monitoring flow, and generating a risk source identification result;

S303: based on the risk point evaluation result, adopting Bayesian network analysis, quantifying the interdependence and conditional probability among risk sources by constructing a probability graph model, evaluating the comprehensive influence of the risk sources, evaluating the risk level, and generating a risk level quantification result;

S304: based on the risk level quantification result, a comprehensive risk assessment model is adopted, all risks in a monitoring process are assessed through analysis of the accumulated effect and interaction of the risks, information of all risk points is integrated, comprehensive risk assessment of the whole process is carried out, and a risk assessment result is generated.

In S301 substep, based on the optimized food safety monitoring flow scheme, a monte carlo simulation is used for risk assessment. First, risk events occurring in the monitoring process, such as equipment failure, data loss, or operation errors, are defined. Each risk event is assigned a set of random variables that describe the uncertainty of the occurrence of the risk, such as the frequency of occurrence and the degree of influence. A probability distribution model, such as a normal distribution or poisson distribution, is then constructed to model the probability of occurrence and the extent of impact of these risk events. Through a large number of random samples, the Monte Carlo simulation repeatedly generates scenes in which risks occur, and the result of each simulation is recorded. And analyzing the simulation results, calculating the average occurrence probability and influence degree of each risk event, and generating a risk source identification result. The result helps identify key risk points in the monitoring flow and provides basis for formulating a risk management strategy.

In the sub-step S302, based on the risk source recognition result of S301, each risk point is deeply evaluated using fault tree analysis. Fault tree analysis is a graphical analysis method for creating a causal relationship chain for risk events. First, top-level events that cause food safety problems, such as product contamination or non-compliance with safety standards, are identified. These top-level events are then broken down step by step, identifying the immediate causes and potential factors that lead to their occurrence. Each factor is considered a node in the tree, and the connections between the nodes represent causal relationships. And analyzing the fault tree, evaluating potential threat and influence of each risk point on food safety, and generating a risk point evaluation result. This result provides a detailed understanding of the potential risks in the monitoring process, helping to determine which risk points require priority management and control.

In the sub-step S303, based on the risk point evaluation result of S302, a bayesian network analysis is employed to quantify the interdependencies and conditional probabilities between risk sources. A bayesian network is a probabilistic graph model that can represent conditional dependencies between variables. In this step, a network is constructed in which nodes represent risk sources and edges represent probabilistic relationships between nodes. Statistical data or expert knowledge is used to estimate conditional probability tables in the network. And calculating the occurrence probability of other risk sources when the specific risk source occurs through Bayesian reasoning, and evaluating the comprehensive influence of the risk source. And finally, evaluating the risk level to generate a risk level quantification result. This result provides an important basis for risk management, helping to determine which risk sources are critical points in the monitoring process, requiring special attention.

In the step S304, based on the risk level quantification result of S303, the comprehensive risk assessment model is used to evaluate all risks in the monitoring flow. The model analyzes the cumulative effects and interactions of risks and integrates the assessment of individual risk points into one comprehensive risk assessment. The model considers the probability of occurrence, the severity and the influence of each risk point on each other. And generating a comprehensive risk assessment report of the whole process through comprehensive assessment, wherein the comprehensive risk assessment report comprises detailed assessments and overall risk levels of all the identified risk points. The report provides a comprehensive view for formulating effective risk management strategies and countermeasures, and contributes to improving the overall effect of food safety monitoring.

It is assumed that in one food processing monitoring procedure, equipment failure, environmental parameter anomalies, and operational errors are identified as major risk points. The specific data show that the annual probability of occurrence of equipment faults is 5%, wherein the probability of occurrence of power faults is 2%, and the probability of occurrence of mechanical faults is 3%. The occurrence probability of the environmental parameter abnormality, particularly the temperature abnormality, is 10% for the month, and the occurrence probability of the operation error for the day is 1%. With these data, a monte carlo simulation is run in step S301, repeated thousands of times to estimate the overall impact of each risk. Next, in step S302, a fault tree analysis is applied, refining the class of equipment faults, finding the proportion of power supply faults and mechanical faults contributing to the overall risk. Further, in step S303, the constructed bayesian network analysis shows that the power failure increases the occurrence probability of the temperature abnormality to 15%. Finally, the comprehensive evaluation in step S304 reveals that the contribution rates of equipment failure, environmental parameter abnormality, and operation error to the monitoring process risk are 40%, 35%, and 25%, respectively. The analysis results are combined into an exhaustive risk assessment report, and important data support and insight are provided for improving the monitoring flow and risk management. By the method, key risk points in the food processing process can be more effectively identified and relieved, and the overall safety and efficiency are improved.

Specifically, as shown in fig. 5, based on the risk assessment result, a food safety tracing mechanism is constructed by adopting a data association analysis and link tracking technology, including tracking the production, processing and distribution processes of food, integrating data, and generating a food safety tracing dataset, and further including:

S401: based on a risk assessment result, a K-means clustering algorithm is adopted, data in the processes of food production, processing and distribution are aggregated into various categories by calculating Euclidean distances among data points, key data modes and features in the process of food circulation are identified, and a food circulation feature map is generated;

S402: based on food circulation feature mapping, adopting an Apriori algorithm, and revealing the internal connection and modes from food production to distribution and multiple links through analyzing and identifying frequent item sets and rules to generate a food link association map;

S403: based on the food link association graph, adopting graph theory analysis, identifying key nodes and paths by constructing and analyzing a network graph of food circulation, tracking production, processing and distribution paths of food, and generating a food safety path network analysis result;

s404: based on the analysis result of the food safety path network, a data fusion algorithm is adopted, and the integrity and the accuracy of data are verified by integrating the key data points and the path information of multiple links, so that a food safety traceability data set is generated.

In the S401 substep, a K-means clustering algorithm is applied to the construction of a food safety traceability mechanism. The data formats include various parameters of food production, processing and distribution, such as temperature, humidity, pH, date of manufacture, lot number, etc. The K-means clustering algorithm aggregates these parameters into a variety of categories by calculating Euclidean distances between data points. First, the number of clusters K is selected, and K cluster centers are randomly initialized. The algorithm then iteratively performs the steps of: each data point is assigned to the nearest cluster center, and the position of each cluster center is updated according to the data points in the clusters. This process is repeated until the position of the cluster center stabilizes or a preset number of iterations is reached. In this way, key data patterns and features during food production, processing, and distribution are identified and classified, resulting in a food circulation feature map. This mapping reveals similarities and differences between different batches or product lines, providing a basis for subsequent food tracking and quality control.

In a substep S402, the Apriori algorithm is employed to analyze the inherent links of food product production to distribution based on the food circulation profile map. The Apriori algorithm is mainly used for identifying frequent item sets and association rules in large data sets. The algorithm first calculates the frequency of all individual items and then gradually builds a set of items that contains more items, leaving only those that meet the minimum support threshold. Next, association rules are generated for these frequent item sets and the support and confidence of each rule is calculated. In this way, the Apriori algorithm reveals frequent and meaningful patterns of correlations between food production, processing, and distribution links, generating a food link correlation map. This map helps understand key points of relevance in the overall circulation of food from raw materials to consumers, providing insight into food safety management and optimizing supply chain strategies.

In the S403 substep, based on the food link correlation map, graph theory analysis is employed to further trace and analyze the circulation path of the food. In this process, a graph representing a food circulation network is constructed, nodes representing key points in circulation, such as production lots, process factories, distribution centers, and edges representing paths of product flow. Key nodes and paths in the network are identified using graph theory algorithms, such as shortest path algorithms or network flow analysis. This analysis reveals key control points and potential bottlenecks in food circulation, producing a food safety path network analysis result. These results not only provide a thorough understanding of the structure of the food circulation network, but also help optimize the process and reduce risks in the supply chain.

In the step S404, based on the analysis result of the food safety path network, a data fusion algorithm is adopted to integrate the key data points and the path information of multiple links. The data fusion process includes verifying the integrity and accuracy of the data, integrating the data from different links and sources. This involves operations such as data cleansing, alignment, merging, etc. For example, for data from different production lots, standardized data formats are required, time stamps are aligned, and then merged into one unified data set. After these steps are completed, a food safety traceability data set is generated, which contains comprehensive information from raw materials to final products. The generation of the data set enables related production batches or distribution nodes to be tracked rapidly when quality problems are found, and emergency response capacity and risk management efficiency of food safety accidents are improved effectively.

Assume a food production and distribution network comprising 100 production lots, 5 processing facilities, and 10 distribution centers. The data of the production lot are exemplified as follows: batch 1 had a temperature of 20 ℃, a humidity of 30% and a pH of 5.5; batch 2 had a temperature of 22 ℃, a humidity of 35% and a pH of 5.6; and so on. All of this data is input into the K-means clustering algorithm of step S401, where K is set to 5. The clustering results classify products with similar production conditions into the same category, e.g., one category includes products with a temperature of 21 ℃ to 23 ℃ and a humidity of 33% to 36%. Continuing to step S402, the Apriori algorithm analyzes the clustered data, finding some frequent patterns. For example, the data shows that high temperatures (above 25 ℃) are associated with low pH values (below 5.5). These modes reveal combinations of parameters that require special attention during production. At step S403, graph theory analysis is used to reveal the network structure of the process plant and distribution center. Analysis results indicate that process plant B is a critical node because multiple product lines pass through the plant. For example, lot 1, lot 5, and lot 7 are sent to distribution centers X and Y after processing at process plant B. Finally, in step S404, all of these information are combined to generate a food safety traceability dataset. This dataset details the complete path of each product from production, processing to distribution. For example, the data set indicates that lot 1 is distributed to distribution center X after processing at process plant B, while lot 2 is sent to distribution center Y after processing at process plant C. The data set provides comprehensive traceability for food safety, so that affected production batches, processing factories and distribution centers can be quickly tracked when quality problems occur, and the efficiency and accuracy for coping with food safety accidents are effectively improved.

Specifically, as shown in fig. 6, based on the food safety traceability dataset, the method adopts spatial clustering and spectral clustering technology based on density, and performs analysis of food safety test results on multiple batches of products, including grouping the products according to risk degrees, and generating a risk product batch identification result, and further includes:

S501: based on a food safety traceability dataset, a DBSCAN algorithm is adopted, core points, boundary points and noise points are identified by calculating density reachability among data points, data points of multi-density areas are distinguished in a multi-dimensional data space, product batches with multiple risk degrees are preliminarily selected, and a preliminary risk product partitioning result is generated;

S503: based on the refined risk product grouping result, a support vector machine algorithm is adopted, classification analysis is carried out on the risk characteristics of the product batches through construction and training of a classification model, risk grade assessment is carried out on the multiple batches of products, and a risk degree assessment result is generated;

S504: based on the risk degree evaluation result, a multivariate statistical analysis method is adopted, the risk evaluation is carried out on the product batches by integrating multiple types of risk indexes, the risk classification of each product batch is matched, and a risk product batch identification result is generated.

In S501 substep, the food safety traceability dataset is analyzed by means of a DBSCAN algorithm to identify risk product batches. A DBSCAN (distributed base station) spatial clustering algorithm based on density is suitable for a large-scale spatial database with noise. The data set includes various parameters collected during production, processing, and distribution, such as temperature, humidity, pH, date of manufacture, and the like. In the DBSCAN algorithm, two parameters are first set: neighborhood radius (epsilon) and minimum points (MinPts). The algorithm then traverses each data point, marking it as a core point if there are more than MinPts in the ε neighborhood of one point. Next, each core point or points with a density of core points are clustered, and points that cannot fall into these clusters are regarded as noise. The process can distinguish data points of multiple density areas in a multidimensional data space, and initially identify product batches with different risk degrees. By the method, a preliminary risk product partitioning result is generated, and a basis is provided for further risk analysis and management.

In the S502 substep, a spectral clustering algorithm is adopted for refining based on the primary risk product partitioning result of the DBSCAN. Spectral clustering algorithms group data by constructing a similarity matrix of the data and applying spectral decomposition techniques. Firstly, calculating the similarity between every two pairs of data points according to food safety related parameters to form a similarity matrix. Then, cluster analysis is performed using the eigenvalues and eigenvectors of this matrix. Specifically, feature vectors corresponding to the first K largest feature values of the similarity matrix are selected to form a new feature space, and a conventional clustering method (such as K-means) is applied to the space. In this way, spectral clustering can effectively cluster under the condition that the original data is not spherically distributed, and more finely distinguish product batches with different risk levels. The generated grouping result of the refined risk products provides more accurate division of the risks of the food batch, and facilitates subsequent risk management and countermeasure establishment.

In the S503 substep, the risk characteristics of the product batch are further analyzed by adopting a support vector machine algorithm based on the refined risk product grouping result of the spectral clustering. The support vector machine is a method for supervised learning and is used for classifying data. In this step, the risk level of each product lot is first marked according to the food safety test results and the risk assessment indicators. These labeled data are then used to train a support vector machine classification model that separates the different classes of data by building one or more hyperplanes. The goal of the support vector machine is to find a hyperplane that maximizes the margin between the different classes. And after training, carrying out risk grade assessment on the unlabeled product batch by using the model. In this way, the support vector machine algorithm helps to accurately identify product batches with different risk degrees, generates a risk degree evaluation result, and provides effective risk classification for food safety supervision.

In the step S504, based on the risk level evaluation result of the support vector machine, a multivariate statistical analysis method is applied to comprehensively evaluate the risk of the product batch. Multivariate statistical analysis involves a variety of statistical techniques, such as principal component analysis, factor analysis, and cluster analysis, to process and interpret relationships between a plurality of variables. In this step, first, a plurality of risk indicators, such as microbiological test results, chemical residual levels, product storage conditions, etc., are integrated, and the risk of each product batch is comprehensively evaluated. Then, the product batches are classified into different risk levels according to the evaluation results. The comprehensive assessment method not only considers single risk factors, but also considers interaction between the single risk factors, and provides a more comprehensive risk view. The generated risk product batch identification result is a detailed description of the risk of the whole product line, is of great importance to monitoring and management of food safety, and is helpful for guiding policy formulation and execution of food safety.

Assume that there is a food safety traceability dataset comprising 200 product batches. Specific data include temperature per batch (range 20 ℃ to 30 ℃), humidity (30% to 60%), pH (5.0 to 7.0), date of manufacture (from 1 month 2023 to 12 months 2023), and specific microbiological test results and chemical residual levels. For example, the data for batch 1 are: the temperature is 24 ℃, the humidity is 40%, the pH value is 5.8, the production date is 2023, 1 month and 5 days, the microorganism test result is qualified, and the chemical residual level is low; whereas the data for batch 2 are: the temperature is 29 ℃, the humidity is 55%, the pH value is 6.5, the production date is 2023, the production date is 1 month and 10 days, the microorganism test result is unqualified, and the chemical residual level is high. In step S501, the DBSCAN algorithm identifies batches with abnormal temperatures (e.g., 5 consecutive batches with temperatures exceeding 28 ℃) as high risk using parameter settings of epsilon=3 ℃ and minpts=5. The spectral clustering algorithm in step S502 further distinguishes batches with abnormal temperature and pH values based on the similarity matrix, and finds that the batches are mainly concentrated in a specific production line. In step S503, the SVM algorithm classifies all batches based on the marked risk levels (high risk and low risk), and the result shows that about 30 batches are classified as high risk. Finally, in step S504, a multivariate statistical analysis method is applied to integrate a plurality of risk factors such as temperature, humidity, pH, microbiological test and chemical residual level, and detailed risk assessment is performed for each batch. For example, lot 1 is rated as low risk, while lot 2 is rated as high risk. These analysis results provide specific risk management guidelines for food companies, such as enhancing quality control for high risk batches and auditing the operational flow of a particular production line.

Specifically, as shown in fig. 7, based on the identification result of the batch of risk products, a Tableau tool is used to analyze the workflow of the detection site in the process of monitoring food in combination with the process mining technology and the sample processing rate analysis, including the steps of identifying the bottleneck of the process and the improvement scheme, and generating the optimization scheme of the workflow of the detection site, and further including:

S601: based on the identification result of the risk product batch, adopting a flow analysis method, analyzing workflow data of a detection site through Tableau tools, identifying bottlenecks and potential optimization points in the workflow, and generating a flow efficiency analysis and improvement scheme;

S604: based on a sample processing time efficiency analysis result, a multidimensional optimization strategy is adopted, and flow improvement measures and a resource optimization strategy are provided by integrating flow efficiency, resource allocation and time management factors, so that a detection site workflow optimization scheme is generated.

In the sub-step S601, workflow data of a detection site in the food monitoring process is analyzed by means of Tableau tools and a flow analysis method. The data format includes processing time, number of detections, detection results, operator information, and the like for each detection site. First, tableau are used to visualize these data, revealing the distribution of processing time, the trend of error rates, and the performance differences for each inspection site. These data are then analyzed in depth to identify bottlenecks and potential optimization points in the workflow, such as excessive processing time or abnormally high error rates for certain inspection sites. These analysis results generate process efficiency analyses and improvements that provide specific improvement suggestions, such as adding hands to a particular site or optimizing operational steps, to increase the efficiency and accuracy of the overall process.

In S602 substep, alpha algorithm is used to reconstruct and analyze the workflow of the detection site. Alpha algorithm is a process mining technique that enables workflows to be found from event logs. The algorithm first parses log data, extracts the relationship between event sequences and activities, and then builds a workflow model. By means of the Alpha algorithm, key activities, path dependencies and potential improvement points in the detection process can be identified. For example, it has been found that the order of execution of certain steps may lead to processing delays or quality problems. Based on these analyses, workflow reconstruction and analysis results are generated, providing a clear direction and basis for further optimization.

In the sub-step S603, a time-series analysis method is applied to evaluate time-series data of the detection site processing sample. This step focuses on the length of the processing time, the trend of variation and the periodic fluctuations. The main influencing factors and the change patterns of the processing time are analyzed by using a statistical method such as an autoregressive model. For example, it was found that certain time periods were inefficient to process or that the error rate increased under certain conditions. Based on the findings, a sample processing time efficiency analysis result is generated, a key region for improving processing efficiency and accuracy is revealed, and data support is provided for optimizing a flow.

In the step S604, the previous analysis results are synthesized, and a multidimensional optimization strategy is adopted to propose specific flow improvement measures and resource optimization strategies. This includes reallocating resources, adjusting workflows, optimizing operational steps, and the like. For example, personnel configuration is adjusted according to the efficiency analysis results, and the operation schedule is optimized according to the time series analysis results. Finally, a detection site workflow optimization scheme is generated that details how to adjust the workflow of each detection site to improve overall efficiency and accuracy while reducing resource waste.

Assume that in one particular food monitoring network embodiment, three detection sites A, B and C are included, each site processing 100 different types of samples per day. Site a handles temperature-sensitive and pH-tested samples, site B handles humidity-sensitive samples, and site C is responsible for chemical residue-tested samples. In the sub-step S601, sample processing data for each site is analyzed by Tableau tools. The average treatment time for station a was found to be 30 minutes per sample, while stations B and C were 20 minutes and 25 minutes, respectively. This finding indicates that site a has efficiency problems. Subsequently, in S602 substep, the Alpha algorithm is applied to reconstruct and analyze the workflow of site a. Analysis revealed that there was a repeated data entry step at site a, resulting in unnecessary delay in conducting the pH test. In the sub-step S603, time series analysis is performed, and it is found that the processing time of site a increases to 35 minutes during the period of 2 to 4 pm, and the other period averages 25 minutes. This time series analysis indicates a specific period and cause of inefficiency. Based on these analyses, in a sub-step S604, a specific optimization scheme is proposed. Including the step of adding automation equipment at site a to reduce manual data entry and adjusting staff schedules for the afternoon shift to ensure adequate operator during peak hours. These improvements are expected to reduce the average processing time of site a to the same level as sites B and C, i.e., about 20 minutes, significantly improving the processing efficiency and response speed of the overall network. Furthermore, these improvements are expected to reduce error rates and improve overall sample processing quality. This embodiment demonstrates how to efficiently identify and solve efficiency and quality problems in a food monitoring network by comprehensively analyzing and applying different algorithms and techniques.

Specifically, as shown in fig. 8, based on the key parameter variation trend analysis result, the optimized food safety monitoring flow scheme, the risk assessment result, the food safety traceability dataset, the risk product batch identification result, and the detection site workflow optimization scheme, the data integration technology is adopted to perform data integration operation, and the steps of data analysis and application policy formulation are performed, so as to generate a food safety monitoring network, and the method further comprises:

s702: based on the optimized food safety monitoring flow scheme and risk assessment result, adopting a linear programming algorithm, optimizing resource allocation and flow design by establishing a mathematical model, analyzing the flow and the efficiency of the resource allocation, and generating a flow design and resource allocation optimization result;

S703: based on a food safety traceability dataset and a risk product batch identification result, adopting an association rule mining algorithm, and analyzing key factors of a food safety control point by analyzing frequent patterns and associations among data items to generate a food safety control point analysis result;

s704: based on a detection site workflow optimization scheme, parameter influence relation analysis results, flow design and resource configuration optimization results and food safety control point analysis results, a data fusion technology is adopted, and a food safety monitoring framework covering multiple links from production to distribution is constructed by integrating multi-stage key data and analysis results, so that a food safety monitoring network is generated.

In the step S701, the key parameter variation trend analysis result is analyzed by a multidimensional scaling analysis algorithm. The data in this step includes time series data of key parameters such as temperature, humidity, pH, etc., reflecting the trend of change in the course of food processing. The multidimensional scaling algorithm reveals interactions and patterns between different parameters by calculating the distances between these parameters and configuring them in a multidimensional space. For example, the algorithm finds a trend in pH decrease as temperature increases. This analysis helps identify which combinations of parameters have the greatest impact on product quality and the resulting analysis of the parameter impact relationship is used to guide the control and adjustment of parameters during the manufacturing process.

In the step S702, a linear programming algorithm is applied to optimize resource allocation and process design based on the optimized food safety monitoring process scheme and risk assessment result. The data of this step includes usage information of production equipment, human resources, material consumption, etc., and costs thereof. Linear programming algorithms aim to minimize cost or maximize efficiency by building mathematical models while taking into account various constraints, such as resource limitations, time limitations, etc. In this way, the algorithm proposes improved flow design and resource allocation schemes, such as adjusting the work order of the production line or reallocating staff, to improve overall efficiency and reduce cost.

In the step S703, key factors are analyzed by adopting an association rule mining algorithm based on the food safety traceability dataset and the risk product lot identification result. The data set encompasses production, processing, and distribution information of the food product, including time, place, processing method, etc. Association rule mining algorithms analyze frequent patterns and associations between these data items, such as where a certain processing method is associated with a high risk product lot. Such analysis reveals key factors in food safety control and the resulting analysis of the food safety control points can be used to further optimize production and processing flows.

In the step S704, a comprehensive food safety monitoring network is constructed by adopting a data fusion technology based on the previous analysis result and the detection site workflow optimization scheme. This step integrates multi-stage critical data from production to distribution with analytical results such as parameter impact relationships, flow design, resource allocation, and food safety control points. The food safety monitoring network after data fusion provides a comprehensive view for a decision maker, supports more effective risk management and decision making, and improves the safety and efficiency of the whole food supply chain.

It is assumed that in one food production and distribution network embodiment, 50 production lots, 3 processing factories, and 5 distribution centers are involved. Examples of data for production lots include: batch 1 had a temperature of 22 ℃ at process plant a, a humidity of 40%, and a pH of 5.5; batch 2 had a temperature of 24 ℃ at process plant B, a humidity of 45% and a pH of 5.7. The multidimensional scaling algorithm in step S701 reveals that at temperatures exceeding 23 ℃, the frequency of pH values below 5.5 increases, indicating that this is a critical risk factor. The linear programming algorithm in step S702 analyzes the human resource usage of process plants a and B, and finds that the human resource usage of plant a is 90% and that of plant B is only 60%. The algorithm suggests transferring part of the workload from factory a to factory B, which is expected to improve overall production efficiency and reduce costs. In step S703, the association rule mining algorithm analyzes the food safety traceability dataset, and finds that the occurrence rate of food safety problems of a specific product lot increases when the humidity exceeds 50%, which becomes a key control point in the production process. Finally, in step S704, the analysis results of all the above steps are integrated, and a comprehensive food safety monitoring network is constructed, which covers all links from production to distribution. For example, the network can indicate increased risk at a particular combination of temperature and humidity and recommend adjustments to the production plan to avoid these risk conditions, thereby significantly improving the food safety level and efficiency of the overall supply chain.

Specifically, as shown in fig. 9, the food safety monitoring system based on big data comprises a time sequence analysis module, a monitoring flow optimization module, a risk assessment module, a traceability data integration module, a risk identification module, a flow efficiency analysis module, a key parameter analysis module and a comprehensive monitoring network construction module;

The time sequence analysis module adopts an autoregressive moving average model to carry out statistical analysis of the time sequence based on time sequence data collected in the food processing process, and comprises calculation of autocorrelation and partial autocorrelation, and selection of a hysteresis period number and a moving average term number, so as to generate a time sequence statistical analysis result;

The monitoring flow optimizing module adopts a branch-and-bound algorithm and a cut plane method to evaluate a decision path of the monitoring flow based on a time sequence statistical analysis result, selects and prunes the flow path through an optimizing algorithm, and simultaneously optimizes a sample detection flow and monitoring resource configuration by referring to cost and efficiency indexes and dynamic flow simulation analysis to generate a flow and resource optimizing decision scheme;

The risk assessment module is based on a flow and resource optimization decision scheme, adopts Monte Carlo simulation and fault tree analysis, combines probability risk assessment technology and logic tree analysis, carries out risk assessment, and comprises simulation of risk occurrence probability and influence degree and causal relation analysis of key risk points, so as to generate a risk assessment result;

The risk identification module classifies the product batches and evaluates the risk level based on the food safety traceability comprehensive data set by adopting a spatial clustering and spectral clustering technology based on density and combining a support vector machine algorithm to generate a refined risk product batch classification result;

The flow efficiency analysis module is used for analyzing the data of the work flow of the detection site by adopting Tableau tools and combining a flow mining technology based on the refined risk product batch classification result, and comprises flow bottleneck identification and potential optimization point analysis, so as to generate a flow efficiency optimization scheme;

The comprehensive monitoring network construction module is used for comprehensively integrating key data and analysis results of multiple stages by adopting a data fusion technology based on a time sequence statistical analysis result, a flow and resource optimization decision scheme, a risk assessment result, a food safety traceability comprehensive data set, a refined risk product batch classification result, a flow efficiency optimization scheme, a parameter influence and a resource configuration result, constructing a monitoring framework covering production to distribution, and generating a food safety monitoring comprehensive network.

The time series analysis module uses an autoregressive moving average model to statistically analyze key parameters such as temperature, humidity, and pH during food processing. The analysis is helpful for identifying and predicting trends and abnormal conditions in the food processing process, so that the production process is more controllable, and the occurrence of food safety accidents is reduced.

The monitoring flow optimization module combines a branch-and-bound algorithm and a cut plane method, so that the sample detection flow and the monitoring resource allocation are effectively optimized. The optimization not only improves the efficiency of food detection, but also reduces the operation cost, and ensures the optimal utilization of resources.

The risk assessment module provides detailed risk assessment through Monte Carlo simulation and fault tree analysis, which is helpful for timely identifying and coping with risk points, reducing occurrence probability of food safety accidents and improving effectiveness of overall food safety management.

The traceability data integration module integrates key data by using a K-means clustering algorithm and an Apriori algorithm, so that the monitoring capability of each link of food circulation is enhanced. By comprehensively tracking the production, treatment and distribution of the food, the module greatly improves the traceability and transparency of food safety accidents.

The risk identification module is applied to accurately evaluate the risk level of the food batch, and timely identify high-risk products, so that proper food safety control measures are adopted.

The flow efficiency analysis module analyzes the data through Tableau tools and a flow mining technology and helps identify bottlenecks and optimization points in the workflow, so that the workflow is improved, and the overall operation efficiency is improved.

The key parameter analysis module applies a multidimensional scale analysis algorithm and a linear programming algorithm to further optimize key parameters and resource configuration in the monitoring flow. This not only improves the accuracy of parameter control during food processing, but also ensures efficient utilization of resources and maximization of cost effectiveness.

The comprehensive monitoring network construction module constructs a comprehensive food safety monitoring network by integrating analysis results of all the modules. This network not only covers each link from production, processing to distribution of the food product, but also ensures the integrity and consistency of food safety management.

Although the application has been described in connection with specific features and embodiments thereof, it will be apparent that various modifications and combinations can be made without departing from the spirit and scope of the application. Accordingly, the specification and drawings are merely exemplary illustrations of the present application as defined in the appended claims and are considered to cover any and all modifications, variations, combinations, or equivalents that fall within the scope of the application. It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the scope of the application. Thus, it is intended that the present application cover the modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

Claims

1. A method for monitoring food safety based on big data, the method comprising:

Based on time series data collected in the food processing process, adopting an autoregressive moving average model and spectrum analysis to perform primary trend analysis, utilizing a long-term and short-term memory network model to analyze key parameters in the food processing process, including historical data of temperature, humidity and pH value, identifying changes in the production process, and generating a key parameter change trend analysis result;

Based on the key parameter change trend analysis result, analyzing the food safety monitoring flow by adopting a branch-and-bound algorithm and a cut plane method, and carrying out structural optimization and efficiency adjustment on a sample detection flow and monitoring resource allocation to generate an optimized food safety monitoring flow scheme;

Based on the optimized food safety monitoring flow scheme, performing risk assessment on the monitoring flow by adopting a probability risk assessment technology and logic tree analysis, wherein the risk assessment comprises the steps of identifying a risk source and assessing potential threat caused by the risk source to food safety, and generating a risk assessment result;

Based on the risk assessment result, a food safety tracing mechanism is constructed by adopting a data association analysis and link tracking technology, the food safety tracing mechanism comprises the steps of tracking the production, processing and distribution processes of food, integrating data and generating a food safety tracing data set, and the food safety tracing mechanism further comprises:

Based on the risk assessment result, adopting a K-means clustering algorithm, and gathering data in the processes of food production, processing and distribution into various categories by calculating Euclidean distance among data points, identifying key data modes and characteristics in the process of food circulation, and generating a food circulation characteristic map;

Based on the food circulation feature mapping, adopting an Apriori algorithm, and revealing the internal connection and mode from food production to distribution multiple links by analyzing and identifying frequent item sets and rules to generate a food link association map;

based on the food link association map, adopting graph theory analysis, identifying key nodes and paths by constructing and analyzing a network map of food circulation, tracking production, processing and distribution paths of food, and generating a food safety path network analysis result;

Based on the food safety path network analysis result, adopting a data fusion algorithm, and verifying the integrity and accuracy of data by integrating key data points and path information of multiple links to generate a food safety traceability data set;

Based on the food safety traceability dataset, adopting spatial clustering and spectral clustering technology based on density, carrying out food safety test result analysis on multiple batches of products, including grouping the products according to risk degree, and generating a risk product batch identification result, and further including:

based on the food safety traceability dataset, a DBSCAN algorithm is adopted, a core point, a boundary point and a noise point are identified by calculating density reachability among data points, data points of a multi-density area are distinguished in a multi-dimensional data space, product batches with multiple risk degrees are preliminarily selected, and a preliminary risk product partitioning result is generated;

based on the primary risk product partitioning result, a spectral clustering algorithm is adopted, a similarity matrix of data is constructed, a spectral decomposition technology is applied, product batches are grouped, product batches with multiple risk grades are refined, and a refined risk product grouping result is generated;

Based on the refined risk product grouping result, a support vector machine algorithm is adopted, classification analysis is carried out on risk characteristics of product batches through construction and training of a classification model, risk grade assessment is carried out on multiple batches of products, and a risk degree assessment result is generated;

Based on the risk degree evaluation result, adopting a multivariate statistical analysis method, performing risk evaluation on the product batches by integrating multiple types of risk indexes, and matching the risk classification of each product batch to generate a risk product batch identification result;

based on the risk product batch identification result, a Tableau tool is adopted, a process mining technology and sample processing rate analysis are combined, the detection site workflow in the food monitoring process is analyzed, the method comprises the steps of identifying a process bottleneck and an improvement scheme, and generating a detection site workflow optimization scheme, and further comprises:

based on the risk product batch identification result, adopting a flow analysis method, analyzing workflow data of a detection site through Tableau tools, identifying bottlenecks and potential optimization points in a workflow, and generating a flow efficiency analysis and improvement scheme;

Based on the flow efficiency analysis and improvement scheme, reconstructing a working flow by analyzing a flow log by adopting an Alpha algorithm, identifying key activities and path dependencies, finding links needing improvement, and generating a working flow reconstruction and analysis result;

Based on the workflow reconstruction and analysis results, adopting time series analysis, analyzing the time consumption and efficiency of a processing link by evaluating the time series data of sample processing, and identifying a flow improvement area to generate a sample processing time efficiency analysis result;

Based on the analysis result of the sample processing time efficiency, adopting a multidimensional optimization strategy, and providing a flow improvement measure and a resource optimization strategy by integrating flow efficiency, resource allocation and time management multiple factors to generate a detection site workflow optimization scheme;

Based on the key parameter change trend analysis result, the optimized food safety monitoring flow scheme, the risk assessment result, the food safety traceability data set, the risk product batch identification result and the detection station work flow optimization scheme, a data fusion technology is adopted to integrate data, and data analysis and application strategy formulation are performed to generate a food safety monitoring network.

2. The big data based food safety monitoring method of claim 1, wherein: the key parameter change trend analysis results comprise a temperature change chart, humidity fluctuation analysis and pH value historical trend, the optimized food safety monitoring flow scheme comprises a redirected resource allocation scheme, a new sample detection sequence and an adjusted monitoring point layout, the risk assessment results comprise a potential risk point list, a risk grade assessment and a risk result, the food safety traceability data set comprises a raw material batch record, a time mark of a processing link and distribution link tracking information, the risk product batch identification result comprises an emergency attention batch list, a batch classification to be observed and a conventional monitoring batch overview, the detection station workflow optimization scheme comprises an identified workflow bottleneck, an improved measure list and an efficiency improvement prediction, and the food safety monitoring network comprises a data integration structure diagram, a risk management flow chart and a quality control strategy overview.

3. The big data based food safety monitoring method of claim 1, wherein: based on time series data collected in the food processing process, adopting an autoregressive moving average model and spectrum analysis to perform preliminary trend analysis, utilizing a long-term and short-term memory network model to analyze key parameters in the food processing process, including historical data of temperature, humidity and pH value, identifying changes in the production process, and generating a key parameter change trend analysis result, and further comprising the steps of:

Based on time series data collected in the food processing process, adopting an autoregressive moving average model, analyzing autocorrelation and partial autocorrelation in the data by a time series prediction technology in statistical analysis, selecting a hysteresis period number and a moving average term number, and generating a time series preliminary analysis result;

Based on the preliminary analysis result of the time sequence, a spectrum analysis method is adopted, and the periodic fluctuation and key frequency components of the data are identified and extracted by calculating the power spectrum density of the time sequence data, and the analysis of the data periodicity and frequency distribution is performed to generate a time sequence spectrum analysis result;

Based on the time sequence spectrum analysis result, a long-term and short-term memory network model is adopted, dependency problems in the time sequence data are processed through a circulation unit in a neural network, key parameter historical data in food processing are analyzed, and a key parameter historical analysis result is generated;

Based on the key parameter historical analysis result, adopting a statistical analysis method to identify abnormal fluctuation or trend change of the parameter by carrying out statistical analysis on the historical data of the key parameters of temperature, humidity and pH value, and generating a key parameter change trend analysis result.

4. The big data based food safety monitoring method of claim 1, wherein: based on the analysis result of the key parameter variation trend, a branch-and-bound algorithm and a cut plane method are adopted to analyze the food safety monitoring flow, and the structure optimization and the efficiency adjustment are carried out on the sample detection flow and the monitoring resource allocation, so that an optimized food safety monitoring flow scheme is generated, and the method further comprises the following steps:

Based on the key parameter change trend analysis result, a branch-and-bound algorithm is adopted, a decision tree is constructed to evaluate the path of the monitoring flow, and path selection and pruning are carried out according to cost and efficiency indexes to generate a preliminary optimization scheme of the monitoring flow;

Based on the preliminary optimization scheme of the monitoring flow, adopting a cut plane method, and optimizing and restraining a decision path by introducing a solution of a linear programming problem into a decision tree model to generate a secondary optimization scheme of the monitoring flow;

based on the secondary optimization scheme of the monitoring flow, dynamic flow simulation analysis is adopted, and key factors of processing rate and detection accuracy are identified and optimized by simulating the performance response of the sample detection flow under multiple configurations, so that a sample detection flow optimization scheme is generated;

Based on the sample detection flow optimization scheme, a resource allocation optimization model is adopted, a resource allocation strategy is identified by analyzing the distribution and efficiency of resources in a monitoring flow, the allocation and the use of the resources are carried out, and an optimized food safety monitoring flow scheme is generated.

5. The big data based food safety monitoring method of claim 1, wherein: based on the optimized food safety monitoring flow scheme, adopting a probability risk assessment technology and logic tree analysis to carry out risk assessment on the monitoring flow, including the steps of identifying a risk source, assessing potential threat caused by the risk source to food safety, and generating a risk assessment result, and further including:

Based on the optimized food safety monitoring flow scheme, monte Carlo simulation is adopted, a random variable and probability distribution model is constructed, a scene of risk occurrence is simulated, the occurrence probability and influence degree of potential risk points in the monitoring flow are analyzed, and a risk source identification result is generated;

Based on the risk source identification result, fault tree analysis is adopted, a causal relationship chain is established by gradually decomposing risk events, potential threat and influence of each risk point on food safety are evaluated, and a risk point evaluation result is generated;

Based on the risk point evaluation result, adopting Bayesian network analysis, quantifying interdependence and conditional probability among risk sources by constructing a probability graph model, evaluating the comprehensive influence of the risk sources, evaluating the risk level, and generating a risk level quantification result;

Based on the risk level quantification result, a comprehensive risk assessment model is adopted, all risks in a monitoring process are assessed through analysis of the accumulated effect and interaction of risks, information of all risk points is integrated, risk comprehensive assessment of the whole process is carried out, and a risk assessment result is generated.

6. The big data based food safety monitoring method of claim 1, wherein: based on the key parameter change trend analysis result, the optimized food safety monitoring flow scheme, the risk assessment result, the food safety traceability dataset, the risk product batch identification result and the detection site workflow optimization scheme, the data integration technology is adopted to integrate data, data analysis and application strategy formulation are performed, and the step of generating a food safety monitoring network further comprises the following steps:

Based on the key parameter change trend analysis result, adopting a multidimensional scale analysis algorithm, revealing the interaction and mode among key parameters through measurement and space configuration of the distance among the parameters, and generating a parameter influence relation analysis result;

based on the optimized food safety monitoring flow scheme and the risk assessment result, adopting a linear programming algorithm, optimizing resource allocation and flow design by establishing a mathematical model, analyzing the efficiency of the flow and the resource allocation, and generating a flow design and resource allocation optimization result;

Based on the food safety traceability dataset and the risk product batch identification result, adopting an association rule mining algorithm, and analyzing key factors of a food safety control point by analyzing frequent patterns and associations among data items to generate a food safety control point analysis result;

Based on the detection site workflow optimization scheme, parameter influence relation analysis results, flow design and resource allocation optimization results and food safety control point analysis results, a data fusion technology is adopted, and a food safety monitoring framework covering multiple links from production to distribution is constructed by integrating multi-stage key data and analysis results, so that a food safety monitoring network is generated.

7. A food safety monitoring system based on big data, which is characterized in that: the big data based food safety monitoring method according to any of claims 1-6, wherein the system comprises a time sequence analysis module, a monitoring flow optimization module, a risk assessment module, a traceability data integration module, a risk identification module, a flow efficiency analysis module, a key parameter analysis module, and a comprehensive monitoring network construction module;