Traceability methods, systems, and storage media for food safety

Abstract

The application relates to the technical field of food traceability data processing, and discloses a traceability method and system for food safety and a storage medium. The method comprises the following steps: double random selection is performed on food raw material information to generate a differentiated trust identifier; an asymmetric trust transmission protocol verification node inter-information is constructed to update the trust identifier; three-source cross verification control point information is controlled to form a quality chain and an enhanced trust identifier; and a multi-level information superposition food traceability visualization result is generated by combining consumer preferences and reconstructing a scene. The application solves the problems of lacking an effective trust mechanism in information transmission between traceability chain nodes and over-reliance on a single source in traceability information verification and lack of individualization in a display mode.

Description

Technical Field

[0001] This application relates to the field of food traceability data processing technology, and in particular to a traceability method, system and storage medium for food safety. Background Technology

[0002] Existing food safety traceability technologies primarily focus on establishing information records and connections at each node of the food supply chain. This includes using physical identification technologies such as barcodes and RFID tags to mark food products, and employing distributed ledger technologies like blockchain to ensure data immutability. These technologies can, to a certain extent, enable end-to-end tracking of food from farm to table, helping consumers understand the origin, production process, and quality and safety status of food. However, traditional traceability systems typically rely on centralized databases to store food information, with regulatory agencies or third-party certification bodies verifying and endorsing the information to ensure its reliability.

[0003] However, existing food safety traceability technologies face several limitations. First, traditional traceability labels are easily copied or tampered with, failing to effectively prevent data fraud. Second, the lack of an effective trust mechanism in information transmission between nodes within the traceability system makes it difficult to ensure the integrity and authenticity of data during transmission. Third, the verification of traceability information relies too heavily on a single source, making it difficult to fully guarantee the authenticity of the data. Fourth, the display of traceability information is monotonous and lacks personalization, failing to provide differentiated information services based on the needs of different consumers, thus reducing the practicality and user experience of traceability information. Summary of the Invention

[0004] This application provides a traceability method, system, and storage medium for food safety, which addresses the problem of a lack of effective trust mechanisms for information transmission between traceability chain nodes, as well as the problem of traceability information verification relying too heavily on a single source and lacking personalized display methods.

[0005] Firstly, this application provides a traceability method for food safety, comprising: performing a dual random selection process on basic information tuples, physical feature tuples, and biological feature vectors of food raw materials to obtain differentiated trust identifiers; constructing an asymmetric trust transfer protocol based on the differentiated trust identifiers, performing multi-dimensional verification processing on information transfer between traceability chain nodes to obtain updated differentiated trust identifiers; performing three-source cross-validation processing on food key control point information based on the updated differentiated trust identifiers to generate a quality chain and combining it with the updated differentiated trust identifiers to obtain enhanced trust identifiers; and performing scene reconstruction processing on food quality data based on the enhanced trust identifiers and consumer preference weight vectors to generate a multi-level information overlay food process visualization result.

[0006] Secondly, this application provides a traceability system for food safety, the traceability system for food safety comprising:

[0007] The processing module is used to perform dual random selection processing on the basic information tuples, physical feature tuples and biological feature vectors of food raw materials to obtain differentiated trust labels.

[0008] The construction module is used to construct an asymmetric trust transfer protocol based on the differentiated trust identifier, perform multi-dimensional verification processing on the information transfer between traceability chain nodes, and obtain an updated differentiated trust identifier.

[0009] The verification module is used to perform three-source cross-validation processing on the food critical control point information based on the updated differentiated trust identifier, generate a quality chain and combine it with the updated differentiated trust identifier to obtain an enhanced trust identifier.

[0010] The reconstruction module is used to perform scene reconstruction processing on food quality data based on the enhanced trust identifier and consumer preference weight vector, and generate a food process visualization result with multi-level information superposition.

[0011] Thirdly, a traceability device for food safety is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the traceability device for food safety to execute the aforementioned traceability method for food safety.

[0012] Fourthly, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform the aforementioned traceability method for food safety.

[0013] The technical solution provided in this application obtains differentiated trust identifiers by performing a dual random selection process on the basic information tuples, physical feature tuples, and biometric vectors of food raw materials. This effectively solves the problem of traditional traceability identifiers being easily copied or tampered with, significantly improving the anti-counterfeiting and uniqueness of food identification. Simultaneously, by collecting multi-dimensional food feature data to form a comprehensive digital fingerprint, each batch of food possesses unique identification characteristics, providing a reliable identity foundation for subsequent traceability. Based on the differentiated trust identifiers, an asymmetric trust transfer protocol is constructed to perform multi-dimensional verification processing on information transmission between traceability chain nodes to obtain updated differentiated trust identifiers. This establishes a dynamic verification mechanism based on node trust levels, solving the trust deficiency problem in the information transmission links of traditional traceability systems and ensuring the integrity and authenticity of data during transmission. Furthermore, by distinguishing the trust base values ​​of different types of nodes, differentiated trust transfer rules are constructed, which are more consistent with the different links in the actual food supply chain. The system leverages the trust characteristics of key control points (KCNPs). Based on updated differentiated trust identifiers, it performs three-source cross-validation on food KCNP information, generating a quality chain that is combined with the updated differentiated trust identifiers to obtain an enhanced trust identifier. This innovative multi-source cross-validation mechanism addresses the data reliability issues caused by traditional traceability systems' over-reliance on a single information source. Through mutual verification by autonomous, automatic, and third-party verification sources, the credibility of traceability information is significantly improved. Simultaneously, the verification results of each control point are linked into a quality chain, forming a complete record of the food quality control process. Based on the enhanced trust identifiers and consumer preference weight vectors, food quality data undergoes scenario reconstruction, generating a multi-level information overlay visualization of the food process. This achieves personalized presentation of traceability information, solving the problems of traditional traceability systems' singular and untargeted information display methods. By incorporating consumer preferences for information filtering and reconstruction, the practicality and user experience of traceability information are significantly enhanced.

[0014] This application employs an improved trust hashing algorithm and a dynamic trust adjustment mechanism in the generation of differentiated trust identifiers, applies similarity calculation and weighted fusion algorithms in information cross-validation, uses association rule mining, cluster analysis, and Bayesian fusion algorithms in consumer preference analysis, and introduces a food science-based decay model in quality prediction. The precise application of these algorithms fully considers the specific functional requirements and application scenario characteristics of food safety traceability. Through intelligent processing of food characteristic data, node trust data, multi-source verification data, and consumer behavior data, the application achieves automation, precision, and personalization of the traceability process, injecting new technological vitality into the field of food safety traceability, significantly improving the reliability, security, and user-friendliness of the traceability system, and providing consumers with more transparent and credible food safety assurance. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of one embodiment of the traceability method for food safety in this application.

[0017] Figure 2 This is a schematic diagram of one embodiment of a traceability system for food safety in this application.

[0018] Figure 3 This is a schematic block diagram of the structure of a traceability device for food safety in an embodiment of the present invention. Detailed Implementation

[0019] This application provides a traceability method, system, and storage medium for food safety. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0020] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the traceability method for food safety in this application includes:

[0021] Step S101: Perform double random selection processing on the basic information tuple, physical feature tuple and biological feature vector of food raw materials to obtain differentiated trust labels.

[0022] Step S102: Construct an asymmetric trust transfer protocol based on the differentiated trust identifier, perform multi-dimensional verification processing on the information transfer between traceability chain nodes, and obtain an updated differentiated trust identifier.

[0023] Step S103: Based on the updated differentiated trust identifier, perform three-source cross-validation on the food critical control point information, generate a quality chain and combine it with the updated differentiated trust identifier to obtain an enhanced trust identifier;

[0024] Step S104: Based on the enhanced trust identifier and consumer preference weight vector, the food quality data is reconstructed to generate a food process visualization result with multi-level information overlay.

[0025] It is understood that the executing entity of this application can be a traceability system for food safety, or it can be a terminal or a server; no specific limitation is made here. This application's embodiments use a server as an example for illustration.

[0026] Specifically, food raw materials undergo a double random selection process. This involves collecting basic information tuples (variety, origin coordinates, harvest time, batch number, and responsible person information) and physical characteristic tuples (weight distribution, shape index, color distribution, hardness, and specific category characteristic parameters) from the raw materials, and extracting biometric fingerprint features to form a biometric vector. The double random selection process uses a first-level random algorithm to select n samples from each batch, and a second-level random algorithm to select m feature points on each sample, forming a sample feature matrix. This double randomness ensures the representativeness and unpredictability of the sampling, effectively preventing data falsification caused by human selection of samples. This information, along with timestamps and random salt values, forms an initial trust data packet. A differentiated trust identifier is calculated using an improved trust hash algorithm. Based on this differentiated trust identifier, an asymmetric trust transfer protocol is constructed to perform multi-dimensional verification of information transmission between traceability chain nodes. First, the traceability chain nodes are classified (farming, primary processing, deep processing, logistics and warehousing, sales, and regulatory nodes), and each node is assigned a triple identity identifier (explicit identifier, implicit identifier, and trust identifier). Node trust coefficients are calculated based on node type and historical behavior, with regulatory nodes having the highest basic trust value, followed by planting / breeding, deep processing, primary processing, logistics / warehousing, and sales nodes. When food is transferred between nodes, the source node combines its differentiated trust identifier with its explicit identifier, timestamp, and target node type to form a transfer request, and signs it with an implicit identifier. After identity verification and trust threshold judgment, the target node verifies based on its trust acceptance coefficient and submits a receipt confirmation, ultimately generating an updated differentiated trust identifier. Based on the updated differentiated trust identifier, three-source cross-validation is performed on food critical control point information. Personalized critical control point lists are established for different food categories, such as raw material acceptance points, marinating points, heating points, cooling points, metal detection points, and packaging points for meat products. An innovative multi-source cross-validation mechanism is introduced, obtaining information from operator-recorded self-validation sources, automatic verification sources from automated monitoring equipment, and independent third-party verification sources, forming information triplets. Three similarity values ​​are obtained through pairwise similarity calculation, and then weighted summation is performed based on weight coefficients determined according to food type and processing stage to obtain an information consistency score. The verification results, consistency scores, quality scores, and other information of each control point are serialized to form a quality chain, which is then combined with the updated differentiated trust identifier to generate an enhanced trust identifier.

[0027] Based on enhanced trust identifiers and consumer preference weight vectors, food quality data undergoes scenario reconstruction. A multi-dimensional quality evaluation system is constructed, encompassing four dimensions: basic safety, nutritional value, sensory experience, and functional characteristics, forming a food quality vector. Consumer-set quality preferences and historical query behaviors are analyzed, and group intelligence optimization is performed in conjunction with attributes such as region and age group to obtain the consumer preference weight vector. Personalized quality scores are obtained through dot product operations, and the current quality status is predicted based on a quality decay model and environmental parameters. Six key scenarios from food cultivation to consumption are parametrically reconstructed, and information from the basic layer, parameter layer, evaluation layer, and knowledge layer is dynamically overlaid, ultimately generating a multi-level information-overlaid visual result of the food journey.

[0028] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0029] The basic information tuples are obtained by collecting and processing the variety, origin coordinates, harvest time, batch number and person in charge of food raw materials;

[0030] The weight distribution, shape index, color distribution, hardness, and specific category characteristic parameters of food raw materials are collected and processed to obtain physical characteristic tuples;

[0031] Biometric fingerprint features are extracted from food raw materials and subjected to feature dimensionality reduction and digitization processing to obtain biometric feature vectors.

[0032] From each batch of food, n samples are selected using a first-level random algorithm, and m feature points are selected for each selected sample using a second-level random algorithm to form a double randomness guarantee, thus obtaining the sample feature matrix.

[0033] The basic information tuple, physical feature tuple, biometric feature vector, timestamp, and random salt value are combined according to a preset format to obtain the initial trust data packet;

[0034] An improved trust hash algorithm is applied to the initial trust data packet to calculate the differential trust identifier, which is then combined with a dynamic trust adjustment mechanism.

[0035] Specifically, basic information on food ingredients is collected and processed, including variety identification, origin coordinate location, harvest time recording, batch number allocation, and registration of responsible person information. Variety identification refers to the identification and classification of food ingredients by type and strain; origin coordinates are obtained through GPS positioning to obtain precise latitude and longitude data; harvest time is accurate to the day or even the hour; batch numbers are generated according to preset rules to form unique identification codes; responsible person information includes identity codes and job descriptions. After standardization, these data form basic information tuples, providing basic attribute data for food traceability. The physical characteristics of food ingredients are collected and processed. Weight distribution is obtained through multi-point weighing to obtain statistical data on weight within the sample; shape index is quantified by calculating the aspect ratio, roundness, or other morphological parameters of the object; color distribution uses computer vision technology to obtain spatial distribution data of color; hardness is measured using a professional hardness tester; specific category characteristic parameters are collected for different food types, such as sugar content and acidity of fruits, and muscle fiber orientation of meat. After standardization and normalization, these physical parameters form physical feature tuples, describing the physical properties of food ingredients.

[0036] Extracting bio-fingerprints from food ingredients refers to acquiring molecular or cellular features that uniquely characterize the biological properties of the food ingredient. Common techniques include DNA barcoding, protein mapping, or spectral analysis. Due to the high dimensionality and redundancy of raw bio-fingerprint data, feature reduction and digitization are necessary. Feature reduction employs algorithms such as Principal Component Analysis (PCA) or t-SNE to map high-dimensional data to a low-dimensional space; digitization converts the reduced features into fixed-length binary or numerical vectors. This ultimately generates a bio-feature vector, providing a molecular-level unique identifier for the food. A dual random selection process is implemented: the first-level random algorithm, based on the Monte Carlo method, randomly selects n samples from each batch of food, ensuring sample coverage of the overall batch distribution; the second-level random algorithm randomly determines the sampling positions of m feature points on each selected sample, avoiding data bias that might result from human selection. This dual randomness ensures the representativeness of the sampling points and increases the difficulty of anti-counterfeiting. These two levels of randomization form an n×m dimensional sample feature matrix, recording the typical feature distribution of the batch of food.

[0037] Then, the previously generated basic information tuples, physical feature tuples, and biological feature vectors are combined and concatenated with the current timestamp and randomly generated salt value according to a predetermined format to form the initial trust data packet. The timestamp accurately records the data generation time, and the random salt value increases the uniqueness and unpredictability of the data, preventing collision attacks. The initial trust data packet contains complete feature information of the food ingredients, providing raw data for subsequent trust calculations.

[0038] An improved trust hash algorithm is applied to the initial trust data packet. This algorithm, based on SHA-256, incorporates a weighting mechanism specific to the food industry, assigning different importance weights to different types of information. Simultaneously, a dynamic trust adjustment mechanism is used: the trust value is automatically adjusted according to a time decay function to reflect the food's shelf-life characteristics; the trust enhancement coefficient is increased based on the number of verifications; and a trust penalty mechanism is triggered based on abnormal situations. These calculations generate the final differentiated trust identifier, serving as a unique identifier for this batch of food in the traceability system.

[0039] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0040] The nodes on the traceability chain are classified and assigned triple identity identifiers, resulting in an explicit identifier, an implicit identifier, and a trust identifier for each node.

[0041] The node trust coefficient is obtained by calculating the basic trust value based on the node type and the dynamic trust adjustment coefficient based on the node's historical transaction records.

[0042] The differential trust identifier, the explicit identifier of the source node, the timestamp, and the target node type are combined to form a transmission request, and the implicit identifier of the source node is used to sign it to obtain a signed transmission request.

[0043] The signed transmission request is verified for the source node's identity validity and a trust threshold is determined to obtain a trust transmission credential.

[0044] The trust transfer credential is verified based on the trust acceptance coefficient of the target node, and a receipt confirmation message is submitted to the trust center to obtain the receipt confirmation data.

[0045] The differentiated trust identifier, trust transfer credential, and received confirmation data are hashed together to obtain the updated differentiated trust identifier.

[0046] Specifically, nodes are categorized into six types based on function and location: planting / breeding nodes, primary processing nodes, deep processing nodes, logistics and warehousing nodes, sales nodes, and regulatory nodes. Each node is assigned a triple identity identifier: an explicit identifier, which is publicly visible information such as the company name and unified social credit code, encoded using standard coding rules; an implicit identifier, a private identifier used internally by the system, created by combining node type codes with a high-strength random number generation algorithm to ensure uniqueness and unpredictability; and a trust identifier, reflecting the node's credibility, generated by quantifying the node's reputation within the food safety traceability system. When calculating the basic trust value based on node type, a tiered assignment strategy is used, assigning different basic scores according to the importance of different node types in the food safety responsibility chain. Regulatory nodes receive the highest basic score (e.g., 100 points), followed by planting / breeding nodes (90 points), deep processing nodes (85 points), primary processing nodes (80 points), logistics and warehousing nodes (75 points), and sales nodes (70 points). Meanwhile, the trust dynamic adjustment coefficient is calculated based on the node's historical transaction records. Indicators such as transaction frequency, transaction size, anomaly rate, and complaint rate are analyzed. A behavior score is obtained through a weighted calculation method. This score is then combined with the trust base value to calculate the final node trust coefficient.

[0047] When food products are transferred between nodes in the traceability chain, the source node combines a differentiated trust identifier (i.e., the unique identifier of the food product generated in the previous step), the source node's explicit identifier, the current timestamp, and the target node type into a transmission request data packet according to a predetermined format. The source node then uses its implicit identifier (equivalent to a private key) to digitally sign the data packet, forming a cryptographically secure signed transmission request that cannot be forged or tampered with.

[0048] Upon receiving a signed transfer request, the trust center performs dual verification: First, it verifies the source node's identity by decrypting the signature and comparing it with the source node's publicly available information to confirm that the request indeed originates from the claimed source node. Second, it determines the trust threshold by comparing the source node's trust coefficient with the minimum trust threshold required for this type of transaction; the transaction is only allowed to proceed if the trust coefficient exceeds the threshold. After successful verification, the trust center generates a trust transfer certificate containing transaction details and the verification result.

[0049] The target node verifies the received trust transfer credential according to its own trust acceptance policy, including checking the credential's integrity, validity period, and compatibility with its own business rules. Upon successful verification, the target node submits a receipt confirmation to the trust center, confirming that it has received the batch of food and acknowledged the relevant trust information. The trust center then generates receipt confirmation data containing the transaction completion status. The original differentiated trust identifier, the trust transfer credential, and the receipt confirmation data are hashed using a secure hash algorithm (such as SHA-256) to generate a new hash value. This hash value is then combined with the original differentiated trust identifier to form an updated differentiated trust identifier, recording that the batch of food has completed a valid inter-node transfer.

[0050] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0051] The nodes in the traceability chain are classified into six key node types: planting and breeding nodes, primary processing nodes, deep processing nodes, logistics and warehousing nodes, sales nodes, and regulatory nodes.

[0052] The public information of each node is encoded to obtain the node's explicit identifier;

[0053] Each node is uniquely encoded based on its node type and a random number generation algorithm to obtain a hidden node identifier.

[0054] Based on the node type, a basic trust score is set, and the highest basic score is assigned to the regulated nodes, in the following order: planting / breeding nodes, deep processing nodes, primary processing nodes, logistics and warehousing nodes, and sales nodes, thus obtaining the node type trust foundation table;

[0055] Statistical analysis and processing of historical behavior data of nodes are performed, including the number of transactions, transaction size, anomaly rate and complaint rate, to obtain a behavior score;

[0056] The node trust identifier is obtained by weighting the corresponding score in the node type trust base table with the behavior score.

[0057] Specifically, the nodes in the traceability chain are categorized into six key node types. These include: Planting and Breeding Nodes (the entities responsible for food raw material production, such as crop cultivation or animal breeding); Primary Processing Nodes (the entities that perform preliminary processing of raw materials, such as cleaning, sorting, and cutting); Deep Processing Nodes (the entities that perform deep processing of food, involving core aspects like formulation and technology); Logistics and Warehousing Nodes (the entities responsible for food transportation and storage); Sales Nodes (the entities that sell food directly to consumers); and Regulatory Nodes (government regulatory departments or third-party certification bodies). By analyzing the functional positioning and business scope of each participating entity, its type is determined and assigned a corresponding type code. For example, planting and breeding nodes are coded as "01," primary processing nodes as "02," and so on. The publicly available information of each node is coded to generate a visible node identifier. This publicly available information includes the company name, unified social credit code, business license number, and geographical location. Using standardized coding rules, the company name is first converted into its initials in Pinyin, combined with the last six digits of the Unified Social Credit Code, and then the first four digits of the administrative division code (based on geographical location information) are added to form a fixed-length string. This coding method ensures the uniqueness and readability of the explicit identifier, facilitating identification and citation by all parties.

[0058] Each node is uniquely encoded based on its node type and a random number generation algorithm to obtain a hidden identifier. First, the node type code is used as a prefix, and then a sufficiently long random string is created using a cryptographically secure random number generator. This random string uses a combination of letters, numbers, and special characters to ensure a sufficiently high entropy value. Subsequently, this combined string is subjected to a one-way hash, and the first 16 bits of the hash value are extracted and concatenated with the node type code to form the node's hidden identifier. This hidden identifier is used only internally within the system, serving as the basis for digital signatures and identity verification. A trust base score is set according to the node type, establishing a node type trust base table. Regulatory nodes, due to their authority and impartiality, are assigned the highest base score of 100 points; planting / breeding nodes, as the source of food, are assigned 90 points; deep processing nodes, due to the complexity of their processes and safety risks, are assigned 85 points; primary processing nodes are assigned 80 points; logistics and warehousing nodes are assigned 75 points; and sales nodes are assigned 70 points. This differentiated score setting reflects the importance and trust weight of different types of nodes in food safety assurance, forming a preliminary node trust base table.

[0059] Statistical analysis and processing of historical node behavior data are performed to calculate behavior scores. First, historical transaction data is collected from nodes, extracting four key indicators: transaction frequency (reflecting activity level), transaction size (reflecting business volume), anomaly rate (reflecting the frequency of problems), and complaint rate (reflecting customer satisfaction). Transaction frequency is logarithmically standardized, converting the raw value to a score of 0-10; transaction size is normalized to the industry average, also converting to a score of 0-10; and anomaly rate and complaint rate are scored inversely, with lower values ​​resulting in higher scores, also converting to a score of 0-10. The four scores are then weighted: transaction frequency (0.2 weight), transaction size (0.2 weight), anomaly rate (0.3 weight), and complaint rate (0.3 weight) to calculate a comprehensive behavior score. The corresponding scores from the node type trust baseline table are weighted and calculated with the behavior scores to obtain a node trust identifier. A weighted average method is then used, with the baseline score weighted at 0.7 and the behavior score weighted at 0.3, to calculate the comprehensive trust score. This score is converted into a standardized trust identifier, starting with the letter "T" followed by a numerical value, such as "T92.5" indicating a trust score of 92.5. This trust identifier dynamically reflects the credibility of a node and guides the transfer of trust between nodes during the food traceability process.

[0060] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0061] A personalized list of critical control points is established based on food categories, and critical limits and monitoring methods are set for each critical control point to obtain a critical control point system.

[0062] The information triplet is obtained by forming autonomous verification source information from data recorded by operators, automatic verification source information from data generated by automated monitoring equipment, and third-party verification source information from verification data provided by independent third parties.

[0063] Based on the three types of verification source information in the information triplet, pairwise similarity functions are calculated to obtain three similarity values;

[0064] The three similarity values ​​are weighted and summed according to the weight coefficients determined by food type and processing stage to obtain the information consistency score;

[0065] The verification results, consistency scores, quality scores, timestamps, and operator identifiers of each key control point are serialized to obtain a quality chain;

[0066] The updated differentiated trust identifier is hashed with the quality chain to obtain the enhanced trust identifier.

[0067] Specifically, a personalized list of critical control points (CCPs) is established based on food categories. CCPs refer to operational steps that can control, eliminate, or reduce food safety hazards to acceptable levels. For different food categories, their unique hazard factors are identified; for example, for meat products, the focus is on microbial contamination; for dairy products, on pathogenic bacteria and antibiotic residues; and for aquatic products, on parasites and heavy metal contamination. For each CCP, critical limits and monitoring methods are clearly defined. CCP limits are the numerical boundaries that distinguish between acceptable and unacceptable values; for example, the core temperature of the meat product heating control point must not be lower than 75°C and must be maintained for at least 15 minutes. Monitoring methods include measuring tools, measurement frequency, measurement location, and recording methods; for example, using a calibrated thermometer to measure the core temperature of each batch of product. This information is systematically organized to form a food-specific CCP system. CCP information is collected from three different sources to form information triplets. Self-verified source information comes from data recorded by operators, such as processing parameters and sensory evaluations filled in by operators through record sheets; automatically verified source information comes from data generated by automated monitoring equipment, such as parameters automatically recorded by equipment such as temperature sensors, humidity monitors, and metal detectors; third-party verified source information comes from verification data provided by independent third parties, such as test reports from official testing agencies or third-party laboratories. These three types of information at each critical control point are organized into an information tripartite structure to ensure the diversity and independence of data sources.

[0068] Based on the three types of verification source information in the information triplet, pairwise similarity functions are calculated to obtain three similarity values. The similarity function quantifies the degree of consistency between information from different sources, employing methods such as cosine similarity or Euclidean distance. First, the similarity Sim(Is,Ia) between the self-verified source and the automatic verification source is calculated, reflecting the consistency between manual records and equipment measurements. Second, the similarity Sim(Is,It) between the self-verified source and the third-party verification source is calculated, reflecting the consistency between internal records and external verification. Finally, the similarity Sim(Ia,It) between the automatic verification source and the third-party verification source is calculated, reflecting the consistency between equipment measurements and professional testing. A similarity value closer to 1 indicates greater consistency, while a value closer to 0 indicates greater discrepancy. The three similarity values ​​are weighted and summed according to weight coefficients determined by food type and processing stage to obtain the information consistency score. The weighting coefficients are dynamically determined based on the characteristics of the food type and processing stage. For example, for cooked food processing, higher weights are assigned to the similarity between automated verification sources and third-party verification sources at the heating point (CCP3); and higher weights are assigned to the similarity between autonomous verification sources and third-party verification sources at the packaging point (CCP6). Weights w1, w2, and w3 are set such that w1 + w2 + w3 = 1. The three similarity values ​​are multiplied by their respective weights and then summed to obtain the final information consistency score, which reflects the reliability of the critical control point information.

[0069] The verification results, consistency scores, quality scores, timestamps, and operator identifiers for each critical control point are serialized to obtain a quality chain. The verification result is a Boolean value indicating whether the control point meets the standard requirements; the consistency score is the consistency rating of the information source calculated in the previous step; the quality score is a quantitative evaluation of the actual product quality, calculated based on the deviation between measured parameters and ideal values; the timestamp records the precise time the data was generated; and the operator identifier is associated with the identity information of the personnel performing the operation. This information is organized according to a predetermined format to form a data structure containing complete traceability information, and linked according to the chronological order of the critical control points to constitute a quality chain reflecting the quality control status of the entire food process.

[0070] The updated differentiated trust identifier and the quality chain are hashed together to obtain the enhanced trust identifier. The differentiated trust identifier, derived from the previous steps, represents the identity information of the food batch; the quality chain contains the quality control information of that batch at each critical control point. The two are concatenated according to a pre-defined format, and then hashed using the secure hash algorithm SHA-256 to generate a fixed-length hash value. This hash value is combined with the original differentiated trust identifier to form the enhanced trust identifier, which retains the unique identity information of the food while also including verification of the integrity of the quality control process.

[0071] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0072] A multi-dimensional evaluation system for food quality is constructed, including basic safety, nutritional value, sensory experience, and functional characteristics, resulting in a food quality vector.

[0073] The consumer preference weight vector is obtained by analyzing the quality dimension preferences and historical query behavior set by consumers, and by combining the region and age group for group intelligence optimization.

[0074] The food quality vector and the consumer preference weight vector are processed by performing a dot product operation to obtain a personalized quality score.

[0075] Based on the quality decay model in the food type library and the real-time collected storage environment parameters, the change law of quality over time and environment is predicted and calculated to obtain the current quality status.

[0076] The six key scenarios of food production, from planting and breeding, processing, testing, storage and transportation, sales to consumption, are standardized, adjusted, and reconstructed using parameters to obtain virtual scenario models.

[0077] By dynamically overlaying basic layer information, parameter layer information, evaluation layer information, and knowledge layer information onto a virtual scene model, a multi-level information overlay visualization result of the food process is obtained.

[0078] Specifically, after completing three-source cross-validation and obtaining enhanced trust labels, food quality is evaluated in multiple dimensions, constructing an evaluation system encompassing four core dimensions. The basic safety dimension focuses on food safety indicators, including parameters mandated by regulations such as microbiological indicators, pesticide residues, and additive use. The nutritional value dimension focuses on the nutritional composition of food, including the content of nutrients such as protein, vitamins, and minerals. The sensory experience dimension assesses characteristics directly perceived by consumers, such as appearance, odor, and taste. The functional characteristics dimension considers added value such as shelf life, target population, and special effects. For each dimension, corresponding evaluation indicators and weights are set according to food type, and the indicator values ​​are standardized to obtain dimension scores. The scores of the four dimensions combine to form a food quality vector, describing the quality status of the food in each dimension. Consumer preference information collection and processing is a key step in personalized traceability. Through a mobile application interface, consumer-initiated quality dimension preferences, such as the degree of importance placed on safety, nutrition, and taste, are collected to form an initial set of preference weights. Simultaneously, historical consumer query and purchase behavior data are analyzed to extract implicit preference features, such as frequently queried indicator types and product characteristics of interest, to construct a set of behavioral preference weights. Furthermore, based on consumers' geographical location and age group information, consumer groups with similar attributes are selected from the database, and their common preference trends are analyzed for group intelligence optimization. A Bayesian fusion algorithm is used to integrate explicit setting preferences, implicit behavioral preferences, and group statistical preferences to generate comprehensive preference weights. After normalization processing to ensure that the sum of the weights of the four dimensions is 1, the final consumer preference weight vector is obtained.

[0079] The food quality vector and consumer preference weight vector are multiplied by a dot product, which involves multiplying the quality score of each dimension by its corresponding preference weight and then summing the results. Specifically, the calculation process is as follows: first, the score for the basic safety dimension is multiplied by the consumer's preference weight for safety; then, the score for the nutritional value dimension is multiplied by the consumer's preference weight for nutrition; next, the score for the sensory experience dimension is multiplied by the consumer's preference weight for sensory experiences; and finally, the score for the functional characteristics dimension is multiplied by the consumer's preference weight for functionality. These four products are then summed to obtain the final personalized quality score. This dot product operation achieves a precise match between quality evaluation and personal preferences. The resulting personalized quality score reflects both the objective quality status of the food and takes into account the consumer's subjective preferences, providing consumers with evaluation results that meet their individual needs. Food quality is not static but dynamically changes with time and environment. Based on a pre-established quality decay model in a food type database, combined with real-time collected storage environment parameters, the variation law of quality over time and environment is predicted and calculated. The quality decay model is based on food science theory and describes the change function of quality parameters for different types of food under specific conditions. Real-time environmental parameters, including temperature, humidity, light intensity, and gas composition, are collected through IoT sensors. The current time point, initial quality data, and historical environmental parameter sequences are input into the decay model to calculate the predicted quality status at the current time point, updating the food quality vector and achieving dynamic quality assessment.

[0080] To intuitively display information throughout the entire food lifecycle, six key scenarios from planting and breeding to consumption are visualized. First, a standardized model library is established for these six scenarios: planting / breeding, processing, testing, storage and transportation, sales, and consumption. For each specific food product, parameterized reconstruction is performed based on its traceability data, adjusting model parameters to match actual conditions. For example, parameters for the planting / breeding scenario are adjusted according to the actual planting environment and growth conditions; parameters for the processing scenario are adjusted according to the actual technological process and equipment parameters. This parameterized reconstruction ensures consistency between the scenario model and reality, while establishing logical connections between scenarios to support smooth and continuous scenario transitions, ultimately forming a complete virtual scenario model.

[0081] This system overlays multi-level information on a virtual scene model, displaying traceability information in layers. The base layer presents basic scene information, such as location, time, and responsible person; the parameter layer displays key quality parameters and control indicators, such as temperature, humidity, and additive content; the evaluation layer displays professional evaluation results and consumer feedback; and the knowledge layer provides relevant food science knowledge and health information. This overlay information display strategy allows users to dynamically switch between different information layers according to their needs, achieving personalized and multi-dimensional information acquisition. Ultimately, it generates a multi-level overlay visualization of the food's journey, providing consumers with an intuitive and detailed food safety traceability experience.

[0082] In one specific embodiment, the process of executing step S106 may specifically include the following steps:

[0083] The quality dimension preferences set by consumers through mobile applications are collected and processed to obtain an initial set of preference weights;

[0084] Association rule mining is performed on consumer historical query and purchase behavior data to extract implicit preference features and obtain a set of behavioral preference weights;

[0085] Based on the consumers' location and age group, a reference group set is obtained by filtering consumer groups with similar attributes from the consumer database.

[0086] Cluster analysis is performed on the preference data of the reference group set to identify the main preference patterns and obtain the group preference model;

[0087] Based on the initial set of preference weights, the set of behavioral preference weights, and the group preference model, a Bayesian fusion algorithm is used to integrate and calculate the comprehensive preference weights.

[0088] The overall preference weights are normalized to ensure that the sum of the weights of the basic safety dimension, nutritional value dimension, sensory experience dimension, and functional characteristic dimension is 1, thus obtaining the consumer preference weight vector.

[0089] Specifically, through the mobile application's interactive interface, a rating slider or questionnaire format is designed to allow consumers to explicitly express their level of concern for different quality dimensions of food. Consumers assign a score of 1-10 to the basic safety dimension, nutritional value dimension, sensory experience dimension, and functional characteristic dimension using a slider or numerical input method. These explicitly expressed preference scores are standardized and converted into weight values ​​between 0 and 1, forming an initial preference weight set. For example, if a consumer rates the safety dimension as 9 points, the nutrition dimension as 7 points, the sensory dimension as 5 points, and the functional dimension as 3 points, after standardization, an initial preference weight set is formed [0.375, 0.292, 0.208, 0.125]. Association rule mining is then performed on consumers' historical query and purchase behavior data to extract implicit preference features. Association rule mining is a data mining technique used to discover the relationships between different items in a dataset. In this method, consumers' food traceability query records and purchase records from the past 3 months are collected first, and behavioral features such as the most frequently queried indicator types, the information pages with the longest dwell time, and the most frequently purchased food types are extracted. Then, the Apriori algorithm is used to analyze this behavioral data, calculating the support and confidence levels between different indicators and behaviors to identify strong association rules. For example, it was found that the consumer frequently checks pesticide residue information and spends a long time on it, inferring a high implicit preference for the basic safety dimension; frequently checking nutrition facts but rarely paying attention to taste evaluations indicates a higher focus on nutritional value than sensory experience. Based on the strength of the strong association rules, weight values ​​are assigned to the four dimensions to form a set of behavioral preference weights.

[0090] Based on the demographic characteristics of consumers, a reference group with similar attributes is selected from the consumer database. First, attribute information such as consumer location (e.g., northern urban area), age group (e.g., 25-35 years old), gender, and occupation is extracted. Then, similarity calculation rules are set: 10 points for a perfect regional match, 1 point for every 10% overlap in age groups, 5 points for a matching gender, and 3 points for similar occupations. Similarity scores are calculated for all consumers in the database, and consumers with scores exceeding a preset threshold (e.g., 20 points) are selected to form a reference group set. For example, 1000 users with attributes similar to the target consumer are selected as the reference group for subsequent analysis. Cluster analysis is performed on the preference data of the reference group set to identify the main preference patterns. The K-means clustering algorithm is used to cluster the preference vectors of the reference group members in four-dimensional space. First, the optimal number of clusters K is determined by calculating the silhouette coefficient under different K values, and the K value with the largest coefficient is selected. Then, K-means clustering is performed on the preference data of the reference group, grouping similar preference patterns together, and calculating the cluster center as a representative preference pattern. Finally, based on the distance between the target consumer and each cluster center, the most likely preference group to which the target consumer belongs is determined, and the cluster center of this group is used as the group preference model.

[0091] Based on the initial preference weight set, the behavioral preference weight set, and the group preference model, a Bayesian fusion algorithm is used for integrated calculation. The Bayesian fusion algorithm, based on Bayesian theory, integrates multi-source information according to its credibility. First, prior credibility is assigned to the three preference sources: initial preference credibility is 0.5 (consumer self-perception may differ from actual behavior), behavioral preference credibility is 0.7 (based on actual behavior, but may be influenced by short-term factors), and group preference credibility is 0.4 (high reference value, but significant individual differences). Then, the weighted average of the three sources on each dimension is calculated, and corrected according to the correlation between dimensions to obtain the comprehensive preference weight. The comprehensive preference weight is normalized to ensure that the sum of the weights of the four dimensions is 1. Specifically, the weight value of each dimension is divided by the sum of the weight values ​​of all dimensions to obtain a standardized weight ratio. For example, the initial calculated comprehensive weights are [0.62, 0.48, 0.25, 0.15], which become [0.413, 0.320, 0.167, 0.100] after normalization. This is the final consumer preference weight vector, which intuitively reflects the distribution of the consumer's attention to the four quality dimensions of food.

[0092] The above describes the traceability method for food safety in the embodiments of this application. The following describes the traceability system for food safety in the embodiments of this application. Please refer to [link / reference]. Figure 2One embodiment of the traceability system for food safety in this application includes:

[0093] Processing module 201 is used to perform dual random selection processing on the basic information tuple, physical feature tuple and biological feature vector of food raw materials to obtain differentiated trust identifiers.

[0094] Construction module 202 is used to construct an asymmetric trust transfer protocol based on the differentiated trust identifier, perform multi-dimensional verification processing on the information transfer between traceability chain nodes, and obtain an updated differentiated trust identifier;

[0095] The verification module 203 is used to perform three-source cross-validation processing on the food critical control point information based on the updated differentiated trust identifier, generate a quality chain and combine it with the updated differentiated trust identifier to obtain an enhanced trust identifier.

[0096] The reconstruction module 204 is used to perform scene reconstruction processing on food quality data based on the enhanced trust identifier and consumer preference weight vector, and generate a food process visualization result with multi-level information superposition.

[0097] above Figure 2 The traceability system for food safety in the embodiments of the present invention will be described in detail from the perspective of modular functional entities. The traceability device for food safety in the embodiments of the present invention will be described in detail from the perspective of hardware processing.

[0098] Figure 3 This is a schematic diagram of a traceability device for food safety provided in an embodiment of the present invention. The traceability device 300 for food safety can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) for storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the traceability device 300 for food safety. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the traceability device 300 for food safety to implement the steps of the aforementioned traceability method for food safety.

[0099] The traceability device 300 for food safety may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The illustrated traceability device structure for food safety does not constitute a limitation on the traceability device for food safety provided by the present invention. It may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0100] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the traceability method for food safety.

[0101] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0102] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a traceability device for food safety (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0103] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A traceability method for food safety, characterized in that, The method includes: A differentiated trust identifier is obtained by performing a dual random selection process on the basic information tuple, physical feature tuple, and biometric vector of food raw materials. This process includes: collecting and processing the variety, origin coordinates, harvest time, batch number, and responsible person information of the food raw materials to obtain a basic information tuple; collecting and processing the weight distribution, shape index, color distribution, hardness, and specific category characteristic parameters of the food raw materials to obtain a physical feature tuple; extracting biometric fingerprint features from the food raw materials and performing feature dimensionality reduction and digitization processing to obtain a biometric vector; selecting n samples from each batch of food using a first-level random algorithm, and selecting m feature points from the basic information tuple, physical feature tuple, and biometric vector for each selected sample using a second-level random algorithm to form a dual randomness guarantee, resulting in a sample feature matrix; combining the sample feature matrix, timestamp, and random salt value according to a preset format to obtain an initial trust data packet; and applying an improved trust hash algorithm to the initial trust data packet, combined with a dynamic trust adjustment mechanism, to obtain the differentiated trust identifier. An asymmetric trust transfer protocol is constructed based on the differentiated trust identifier to perform multi-dimensional verification processing on information transmission between traceability chain nodes, resulting in an updated differentiated trust identifier. This includes: classifying nodes on the traceability chain and assigning them triple identity identifiers to obtain an explicit identifier, an implicit identifier, and a trust identifier for each node; calculating a basic trust value based on node type and a dynamic trust adjustment coefficient based on the node's historical transaction records to obtain a node trust coefficient; combining the differentiated trust identifier, the source node's explicit identifier, a timestamp, and the target node type to form a transfer request, and signing it using the source node's implicit identifier to obtain a signed transfer request; verifying the source node's identity validity and determining a trust threshold on the signed transfer request to obtain a trust transfer credential; verifying the received trust transfer credential according to the target node's own trust acceptance strategy, including checking the credential's integrity, validity period, and compatibility with its own business rules; submitting a receipt confirmation to the trust center after successful verification to obtain receipt confirmation data; and performing a hash calculation on the differentiated trust identifier, the trust transfer credential, and the receipt confirmation data to obtain the updated differentiated trust identifier. Based on the updated differentiated trust identifier, the food critical control point information is subjected to three-source cross-validation to generate a quality chain, which is then combined with the updated differentiated trust identifier to obtain an enhanced trust identifier. Based on the enhanced trust identifier and consumer preference weight vector, the food quality data is reconstructed to generate a multi-level information overlay visualization of the food journey.

2. The traceability method for food safety according to claim 1, characterized in that, The process of classifying and assigning triple identity identifiers to nodes on the traceability chain yields an explicit identifier, a implicit identifier, and a trust identifier for each node, including: The nodes in the traceability chain are classified into six key node types: planting and breeding nodes, primary processing nodes, deep processing nodes, logistics and warehousing nodes, sales nodes, and regulatory nodes. The public information of each node is encoded to obtain the node's explicit identifier; Each node is uniquely encoded based on its node type and a random number generation algorithm to obtain a hidden node identifier. Based on the node type, a basic trust score is set, and the highest basic score is assigned to the regulated nodes, in the following order: planting / breeding nodes, deep processing nodes, primary processing nodes, logistics and warehousing nodes, and sales nodes, thus obtaining the node type trust foundation table; Statistical analysis and processing of historical behavior data of nodes are performed, including the number of transactions, transaction size, anomaly rate and complaint rate, to obtain a behavior score; The node trust identifier is obtained by weighting the corresponding score in the node type trust base table with the behavior score.

3. The traceability method for food safety according to claim 1, characterized in that, The process involves performing three-source cross-validation on food critical control point information based on the updated differentiated trust identifier, generating a quality chain, and combining it with the updated differentiated trust identifier to obtain an enhanced trust identifier, including: A personalized list of critical control points is established based on food categories, and critical limits and monitoring methods are set for each critical control point to obtain a critical control point system. The information triplet is obtained by forming autonomous verification source information from data recorded by operators, automatic verification source information from data generated by automated monitoring equipment, and third-party verification source information from verification data provided by independent third parties. Based on the three types of verification source information in the information triplet, pairwise similarity functions are calculated to obtain three similarity values; The three similarity values ​​are weighted and summed according to the weight coefficients determined by food type and processing link to obtain the information consistency score; The verification results, consistency scores, quality scores, timestamps, and operator identifiers of each key control point are serialized to obtain the quality chain; The updated differentiated trust identifier is hashed with the quality chain to obtain the enhanced trust identifier.

4. The traceability method for food safety according to claim 1, characterized in that, The process involves reconstructing the food quality data based on the enhanced trust identifier and consumer preference weight vector to generate a multi-level information overlay visualization of the food journey, including: A multi-dimensional evaluation system for food quality is constructed, including basic safety, nutritional value, sensory experience, and functional characteristics, resulting in a food quality vector. The consumer preference weight vector is obtained by analyzing the quality dimension preferences and historical query behavior set by consumers, and by combining the region and age group for group intelligence optimization. The food quality vector and the consumer preference weight vector are multiplied by a dot product to obtain a personalized quality score. Based on the quality decay model in the food type library and the real-time collected storage environment parameters, the change law of quality over time and environment is predicted and calculated to obtain the current quality status. The six key scenarios of food production, from planting and breeding, processing, testing, storage and transportation, sales to consumption, are standardized, adjusted, and reconstructed using parameters to obtain virtual scenario models. Basic layer information, parameter layer information, evaluation layer information, and knowledge layer information are dynamically superimposed on the virtual scene model to obtain a food process visualization result with the superimposed multi-level information.

5. The traceability method for food safety according to claim 4, characterized in that, The process involves analyzing consumer-defined quality dimension preferences and historical query behavior, and then performing group intelligence optimization based on the consumer's region and age group to obtain the consumer preference weight vector, which includes: The quality dimension preferences set by consumers through mobile applications are collected and processed to obtain an initial set of preference weights; Association rule mining is performed on consumer historical query and purchase behavior data to extract implicit preference features and obtain a set of behavioral preference weights; Based on the consumers' location and age group, a reference group set is obtained by filtering consumer groups with similar attributes from the consumer database. Cluster analysis is performed on the preference data of the reference group set to identify the main preference patterns and obtain the group preference model; Based on the initial set of preference weights, the set of behavioral preference weights, and the group preference model, a Bayesian fusion algorithm is used to integrate and calculate the comprehensive preference weights. The comprehensive preference weights are normalized to ensure that the sum of the weights of the basic safety dimension, nutritional value dimension, sensory experience dimension, and functional characteristic dimension is 1, thus obtaining the consumer preference weight vector.

6. A traceability system for food safety, characterized in that, For implementing the traceability method for food safety as described in any one of claims 1 to 5, the traceability system for food safety comprises: The processing module is used to perform dual random selection processing on the basic information tuples, physical feature tuples and biological feature vectors of food raw materials to obtain differentiated trust labels. The construction module is used to construct an asymmetric trust transfer protocol based on the differentiated trust identifier, perform multi-dimensional verification processing on the information transfer between traceability chain nodes, and obtain an updated differentiated trust identifier. The verification module is used to perform three-source cross-validation processing on the food critical control point information based on the updated differentiated trust identifier, generate a quality chain and combine it with the updated differentiated trust identifier to obtain an enhanced trust identifier. The reconstruction module is used to perform scene reconstruction processing on food quality data based on the enhanced trust identifier and consumer preference weight vector, and generate a food process visualization result with multi-level information superposition.

7. A traceability device for food safety, characterized in that, It includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the computer program to implement the traceability method for food safety as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by a processor, it causes the processor to perform the traceability method for food safety as described in any one of claims 1 to 5.

Images