Visual traceability processing method and system for food deterioration data

Abstract

The invention discloses a visual traceability processing method and system for food deterioration data, and relates to the technical field of data processing and big data resource services. According to food characteristics, a metabolite spectrum under a time sequence is constructed, a radio frequency tag of a food package is taken as a node, four-dimensional Internet of Things networking is carried out, and by coupling a space microenvironment and the metabolite spectrum, the food deterioration data is obtained. The method comprises the following steps: deploying an intelligent contract rule by taking a metamorphic nonlinear acceleration feature as a constraint, carrying out Internet-of-things network access operation through detection of a space microenvironment and a metabolite spectrum, triggering a coding environment cascade decision based on the intelligent contract rule, determining metamorphic data and carrying out uplink storage, and is used for solving the problem of lack of an integrated unified traceability mechanism in the prior art. The technical problem that full-process automatic identification and information uploading from full-period tracking, deterioration generation to responsibility identification are difficult to realize is solved, the food supply chain management and control level is improved, and the traceability and quality transparency of big data resource service are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing and big data resource services, and in particular to a method and system for visually tracing food spoilage data. Background Art

[0002] During storage, transportation, and distribution, food is susceptible to environmental influences such as temperature, humidity, light, and oxygen content, leading to spoilage. Traditional food quality traceability systems rely heavily on batch records, static labels, or quality inspection reports, lacking the ability to provide real-time perception and visual analysis of the dynamic deterioration process throughout the food supply cycle. Furthermore, food spoilage often manifests as changes in microscopic metabolites, but current traceability systems generally fail to structure this metabolic data for traceability analysis. This creates a disconnect in the information chain, making it difficult to support refined management.

[0003] In the existing technology, some food monitoring methods attempt to introduce the judgment of the degree of spoilage, but they mostly focus on static detection or the metabolic state at a certain time point. They lack dynamic capabilities based on time series and cannot accurately express the evolution path of spoilage. At the same time, data uploading and processing mostly rely on centralized systems, which poses a risk of data tampering and affects the credibility of traceability results.

[0004] Therefore, existing technologies lack an integrated unified traceability mechanism, making it difficult to achieve automatic identification and information uploading throughout the entire process, from full-cycle tracking, deterioration occurrence to responsibility identification. Summary of the Invention

[0005] The present application provides a method and system for visual traceability processing of food spoilage data, which is used to solve the technical problems in the existing technology of lacking an integrated unified traceability mechanism and difficulty in achieving automatic identification and information chain-up of the entire process from full-cycle tracking, spoilage occurrence to responsibility identification.

[0006] In view of the above problems, the present application provides a method and system for visual traceability processing of food spoilage data.

[0007] In a first aspect, the present application provides a method for visual traceability processing of food spoilage data, the method comprising: constructing a metabolite spectrum under a time series based on food characteristics, wherein the metabolite spectrum is a low-dimensional directed graph based on the spoilage type; using the radio frequency tags of food packaging as nodes, conducting a four-dimensional Internet of Things network, and deploying smart contract rules by coupling the spatial microenvironment and the metabolite spectrum, with the nonlinear acceleration characteristics of spoilage as constraints, wherein spoilage classification, precursor warning, and dynamic attribution judgment are used as rule-guided; performing Internet of Things access operations based on the detection of the spatial microenvironment and the metabolite spectrum, triggering a coding environment cascade decision based on the smart contract rules, determining the spoilage data and storing it on the chain; wherein the traceability chain includes point traceability based on radio frequency scanning and process traceability based on the coupling analysis of nonlinear acceleration of spoilage between points, and hierarchical display through radio frequency tag recognition.

[0008] In the second aspect, the present application provides a visual traceability processing system for food spoilage data, the system comprising: a construction module for constructing a metabolite spectrum in a time series according to food characteristics, wherein the metabolite spectrum is a low-dimensional directed graph based on the spoilage type; a deployment module for using the radio frequency tags of food packaging as nodes to conduct four-dimensional Internet of Things networking, and by coupling the spatial microenvironment and the metabolite spectrum, and using the nonlinear acceleration characteristics of spoilage as constraints, deploy smart contract rules, wherein spoilage classification, precursor warning, and dynamic attribution judgment are rule-guided; a decision module for performing Internet of Things access operations based on the detection of spatial microenvironment and metabolite spectrum, triggering a coding environment cascade decision based on smart contract rules, determining spoilage data and storing it on the chain; wherein the traceability chain includes point traceability based on radio frequency scanning, and process traceability based on the coupling analysis of nonlinear acceleration of spoilage between points, and hierarchical display through radio frequency tag recognition.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: The embodiment of the present application provides a visual traceability processing method for food spoilage data. According to the characteristics of food, a metabolite spectrum in a time series is constructed, and a four-dimensional Internet of Things network is performed with the radio frequency tag of the food packaging as a node. By coupling the spatial microenvironment and the metabolite spectrum, and taking the nonlinear acceleration characteristics of spoilage as constraints, smart contract rules are deployed, and the Internet of Things access operation is performed based on the detection of the spatial microenvironment and the metabolite spectrum. The coding environment cascade decision based on the smart contract rules is triggered, the spoilage data is determined and stored on the chain. It is used to solve the technical problems in the existing technology that there is a lack of an integrated unified traceability mechanism, and it is difficult to achieve automatic identification and information on the chain for the entire process from full-cycle tracking, spoilage occurrence to responsibility identification. By constructing a low-dimensional metabolite spectrum, deploying a four-dimensional Internet of Things smart contract, creating a two-dimensional traceability display system and implementing precursor warning intelligent routing, efficient processing of food spoilage information, accurate warning traceability and intelligent regulation are achieved, the level of food supply chain management and control is improved, and the traceability and quality transparency of big data resource services are enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This application provides a flow chart of a visual traceability processing method for food spoilage data; Figure 2 This application provides a schematic diagram of the process of constructing a metabolite profile in a method for visual traceability of food spoilage data; Figure 3 A schematic diagram of the structure of a visual traceability processing system for food spoilage data is provided for this application.

[0011] Description of the accompanying drawings: construction module 11, deployment module 12, decision module 13. DETAILED DESCRIPTION

[0012] This application provides a visual traceability processing method and system for food spoilage data to solve the technical problems in the existing technology, such as the lack of an integrated unified traceability mechanism, which makes it difficult to achieve automatic identification and information chain-linking of the entire process from full-cycle tracking, spoilage occurrence to responsibility identification.

[0013] Example 1: Figure 1 As shown, the present application provides a method for visual traceability processing of food spoilage data, the method comprising: S1: Construct a metabolite profile in a time series based on food characteristics, wherein the metabolite profile is a low-dimensional directed graph based on the deterioration type.

[0014] In this embodiment of the present invention, a metabolite profile is first constructed over a time series based on food characteristics to achieve dynamic perception and structured expression of the food spoilage process. These food characteristics refer to the physicochemical properties, biological composition, and metabolic stability exhibited by foods during storage and transportation. Examples include the sensitivity of protein breakdown in dairy products and the generation rate of volatile aldehydes and ketones in fruits. In this step, based on the spoilage sensitivity factors of different foods, a list of metabolic pathways related to spoilage behavior is established, thereby obtaining a continuous spectrum of metabolites changing over time during the spoilage process, that is, the metabolite spectrum under time series.

[0015] Furthermore, to effectively represent the causal dependencies and path evolution characteristics between metabolites, the metabolite profiles were constructed as a low-dimensional directed graph structure based on spoilage types. These types of spoilage refer to the classification of the primary mechanisms of food degradation under different microenvironmental conditions, such as protein denaturation, fat oxidation, and carbohydrate degradation. In this paper, principal component analysis and category clustering techniques were used to train a large amount of sample data to obtain type labels for the spoilage mechanisms.

[0016] In the construction of a directed graph, each metabolite is abstracted as a node in the graph, and the interaction relationships between metabolites, such as generation, transformation, and inhibition, constitute directed edges in the graph. The direction of the edge reflects the temporal sequence of the metabolic process and the interaction link. For example, in the oxidative deterioration of fish food, linoleic acid is oxidized to form peroxides, which are further converted into volatile aldehydes. This process can be modeled as a directed path from linoleic acid to peroxides to aldehydes.

[0017] Furthermore, considering the computing resource constraints of subsequent data processing and traceability system operation, the present invention reduces the dimensionality of the original high-dimensional metabolic graph through feature compression and structure pruning, retaining key metabolic pathways and nodes that are representative and discriminative, thereby forming a lightweight low-dimensional directed graph structure.

[0018] During the actual deployment process, the preset metabolite spectrum is automatically called according to the food type to achieve accurate modeling and visual tracking of the food spoilage process.

[0019] In summary, by constructing a metabolic map that combines food characteristics, time series information and structured expression of metabolic mechanisms, dynamic mapping of the food spoilage process is achieved, providing a high-precision data foundation for subsequent traceability and identification.

[0020] Further, such as Figure 2 As shown, to construct a metabolite profile under a time series, step S1 of this application includes: According to the food characteristics of the target food, clustering based on the deterioration type is performed to determine multiple types of metabolic configurations; the multiple types of metabolic configurations are subjected to dimensionality reduction processing by minimizing the intra-class distance and maximizing the inter-class distance to determine multiple types of lightweight configurations, wherein the dimensionality reduction method includes configuration type dimensionality reduction and configuration local structure dimensionality reduction; based on the multiple types of lightweight configurations, the metabolite spectrum is determined.

[0021] In an embodiment of the present invention, the target food is the food to be traced. In order to construct a metabolite spectrum with discriminative and visual properties, a clustering process of the deterioration type is first performed according to the food characteristics of the target food.

[0022] Among them, the food characteristics include but are not limited to physical and chemical indicators such as protein content (such as >3% in dairy products), moisture content (such as >70% in fresh meat), pH value (commonly 4 to 6 in fruits and vegetables), and may also include perishable factors (such as easily oxidized components) and environmental sensitivity (such as cold chain transportation).

[0023] Specifically, spoilage records of the target food are collected. During the clustering phase, a K-means algorithm is preferably used to perform unsupervised clustering of the metabolic data by spoilage type, such as corresponding to common configurations such as fat oxidation, protein degradation, abnormal sugar metabolism, and microbial growth. Each configuration, i.e., a different spoilage type, contains a temporal representation of several metabolic pathways, including typical pathways such as linoleic acid → lipid peroxides → aldehyde volatiles or lactic acid → acetic acid → ammonia.

[0024] Furthermore, in order to reduce the computational complexity of metabolic configurations and improve the efficiency of subsequent mapping and intelligent discrimination, it is necessary to perform dimensionality reduction on multiple types of metabolic configurations.

[0025] In this paper, the dimensionality reduction method is to minimize the intra-class distance and maximize the inter-class distance. The intra-class distance is calculated using the Euclidean distance metric, which measures the average distance between samples within each configuration, and the inter-class distance is measured using the center point distance between groups. The purpose is to improve the distinction between classes and the representativeness of features within a class.

[0026] In one embodiment, the dimensionality reduction method is preferably a combination of PCA (Principal Component Analysis) and UMAP (Uniform Manifold Approximation and Projection). PCA is used to extract global linear structure, while UMAP is used to preserve nonlinear neighborhood relationships in high-dimensional space. After dimensionality reduction, each type of configuration is compressed into a lightweight configuration of 5-10 dimensions.

[0027] In practice, the present invention employs two dimensionality reduction strategies: configuration-based dimensionality reduction, which preserves the main trends in metabolic expression, and configuration-based local structure dimensionality reduction, which utilizes graph theory methods, such as adjacency matrix sparsification and path importance ranking, for local pruning, prioritizing the preservation of key nodes (e.g., metabolic turning points or high-frequency pathways). For example, in fat oxidation-related deterioration, the chain-like conversion pathways between C18:2 fatty acids and their oxidation products are retained, while parallel, low-frequency pathways are ignored.

[0028] Finally, the multiple lightweight configurations obtained from the dimensionality reduction process are structured to construct a metabolite profile of the target food's spoilage behavior. This profile is represented as a lightweight directed graph, with each node representing a key metabolite and each edge indicating a specific reaction direction, along with attribute information such as time and concentration change rate. This metabolite profile is then used for subsequent microenvironmental coupling assessment and visual traceability.

[0029] S2: Using the radio frequency tags on food packaging as nodes, a four-dimensional IoT network is established. By coupling the spatial microenvironment with the metabolite spectrum and taking the nonlinear acceleration characteristics of spoilage as constraints, smart contract rules are deployed. Among them, spoilage classification, precursor warning, and dynamic attribution judgment are the rule-oriented.

[0030] In the implementation of this invention, based on the need for unique identification of food units, radio frequency identification (RFID) tags on food packaging are preferably used as network node entry points to build a multi-dimensional IoT architecture that supports real-time environmental perception and data linkage. These RFID tags not only perform basic identity recognition and batch numbering functions but also serve as data anchors in the interaction between food and the environment.

[0031] In specific applications, there is no restriction on the data collection method. An ultra-high frequency RFID (UHF RFID) with reading and writing functions or an RFID system with an environmental sensing expansion module can be selected. External sensing can also be used, such as external sensors in cold chain containers, to collect the micro-environmental parameters of the food location (such as temperature, humidity, air pressure, light, etc.).

[0032] The four dimensions include three-dimensional spatial positioning and one-dimensional temporal information, which are used to calibrate the food's location, such as the coordinates of the transport compartment, and the environmental trends over time. Specifically, each RFID tag node is configured with a spatial-temporal coordinate system. Its location can be obtained through a warehouse positioning system (such as UWB or three-dimensional Bluetooth positioning), and the temporal dimension is synchronized using timestamps. By connecting multiple RFID tag nodes, a spatial topology can be formed, creating a network structure that maps physical locations to data paths.

[0033] After completing the network, we further introduced a joint analysis mechanism for the spatial microenvironment and metabolite profile. The spatial microenvironment refers to a collection of real-time physical parameter data related to the food storage or transportation environment. These parameters are sampled and recorded at a frequency of minutes or hours using temperature, humidity, and gas sensors. The metabolite profile, based on the aforementioned construction steps, represents the metabolic state of food under specific spoilage mechanisms. In practice, since the spoilage process of food proceeds at the expected rate under the expected environment, such as the time evolution of the metabolite spectrum, but due to the incomplete controllability under the specific environment, there are certain differences between the actual and the expected. The present invention introduces the nonlinear acceleration feature of spoilage, that is, the acceleration of the metabolite spectrum by environmental variables, taking into account process tracing and improving the actual fit of the data.

[0034] In the smart contract rules of this application, the present invention uses the nonlinear acceleration characteristics of deterioration as a constraint condition to establish an environmental response relationship based on the dynamics of food degradation. For example, when it is detected that the rate of increase of spatial temperature exceeds a threshold, the growth curve of the metabolite concentration will cause an obvious nonlinear slope change (such as exceeding its 3σ range), which will cause accelerated deterioration.

[0035] Based on the above constraints, three types of core rule guidance are set in the smart contract: first, the spoilage classification rule, which is used to classify the current node food into a specific spoilage category label, such as protein decomposition type, oxidation type, moisture imbalance type, etc.; second, the precursor warning rule, which is used to automatically output a warning mark and send it to the IoT management platform when the accumulation of specific precursor substances (such as volatile sulfides, amines, etc.) exceeds the standard in the metabolic pathway; third, the dynamic attribution judgment rule, after the actual spoilage event of food, the on-chain comparison and node responsibility attribution judgment are carried out according to parameters such as the similarity of environmental data and metabolic evolution between nodes, so as to realize the intelligent positioning of spoilage responsibility.

[0036] In summary, the system effectively realizes dynamic perception, early warning and responsibility determination of food spoilage process, and has significant advantages in food safety traceability, such as clear structure, timely response and strong scalability.

[0037] Furthermore, to deploy smart contract rules, step S2 of this application includes: The coding mode is used as the contract environment, and the discriminant mode is deployed based on the adversarial training principle. The discriminant mode includes a first discriminant node based on deterioration classification, a second discriminant node based on precursor warning, and a third discriminant node based on dynamic attribution judgment. The mode is trained with a generative-discriminative architecture and deployed with a decomposed adversarial architecture. The dynamic attribution is used to determine the responsibility for deterioration. The smart contract is deployed according to the discriminant mode.

[0038] In the implementation of this invention, to automatically identify food spoilage and determine liability, we propose constructing an intelligent discrimination mechanism based on adversarial training principles, using a coding model as the contract environment. The coding model converts environmental data and metabolite profiles into a hash-coded state, which in this application is a traceability code, i.e., a structured data input suitable for smart contract processing.

[0039] In this coding environment, adversarial training principles are employed to design a discriminant model to enhance the accuracy of identifying spoilage behavior and the ability to generalize rules. Adversarial training employs a generative-discriminative network, using both real and simulated spoilage samples to train the discriminant model. This allows the model to discriminate between actual spoilage states and precursory characteristics in complex situations.

[0040] During the specific implementation, the construction of the first discriminant node based on spoilage classification begins with determining sample data: actual spoilage type samples, generated simulated samples, and classified samples. Supervised training is then performed under a generator-discriminator architecture until convergence. The trained generator-discriminator architecture is then further decomposed, with the discriminator retained as the first discriminant node. Preferably, the generator constructs mimic samples based on existing metabolic profile data, such as an approximate spoilage path constructed by perturbing temperature and humidity values with normal samples. The discriminator then identifies valid spoilage behavior based on real-world sensory data.

[0041] Similarly, the construction of the second discriminant node and the third discriminant node is executed, wherein the training steps of the first discriminant node, the second discriminant node and the third discriminant node are consistent, but the sample data are different.

[0042] The first discriminant node is a discriminator based on spoilage classification, responsible for classifying the spoilage mechanism of the current sample, such as distinguishing between protein degradation and oxidation. The second discriminant node is a discriminator based on precursor warning, used to detect whether key precursor metabolites such as indole, trimethylamine, and ketone compounds show abnormal accumulation trends in the metabolic profile. The third discriminant node is a discriminator based on dynamic attribution. It targets the current spoilage event and combines historical node data with similar environmental data to output a specific node number and spoilage responsibility weight score. Dynamic attribution is the determination of spoilage responsibility, achieving a three-dimensional joint localization of space, time, and responsibility in the food spoilage process.

[0043] After completing the aforementioned adversarial training, a decomposition-based deployment strategy is used to modularize the adversarial architecture into executable smart contract code by function. Each discriminant node is deployed as a lightweight discriminant service module on-chain or on an edge node, supporting rapid response and distributed decision-making in edge computing scenarios. For example, the early warning module can be deployed on cold chain transport vehicles, while the dynamic attribution module can be deployed on a cloud-based collaborative chain node.

[0044] Ultimately, the core smart contract logic of the food traceability system is constructed based on the contract structure generated by this discrimination model and embedded in blockchain nodes, ensuring that the entire process of spoilage monitoring, early warning, and accountability is verifiable and tamper-resistant. Through this approach, the present invention effectively integrates environmental perception, metabolic analysis, and smart contract rules to construct a food spoilage traceability system with adaptive discrimination capabilities.

[0045] Furthermore, taking the coding mode as the contract environment, step S2 of this application includes: A first Hash chain and a second Hash chain are introduced, wherein the first Hash chain is associated with the spatial microenvironment, and the second Hash chain is associated with the metabolite spectrum; microenvironment sensing is performed to convert the first code based on the first Hash chain; according to the upstream metabolite spectrum, the second code is converted to the second Hash chain; and a traceability code is generated based on the first code and the second code.

[0046] In an embodiment of the present invention, in order to ensure the source verifiability and chain information integrity of food spoilage data, it is proposed to introduce a double-chain structure of the first hash chain and the second hash chain, which are used to irreversibly encode spatial microenvironment data and metabolite spectrum information, respectively, thereby realizing multidimensional data integration and traceability code generation throughout the entire life cycle of food.

[0047] Specifically, the first hash chain is used to record microenvironmental data related to the food's environment. This microenvironmental data includes, but is not limited to, parameters such as temperature, humidity, air pressure, oxygen concentration, and light intensity. During each sampling period, the currently collected microenvironmental data set is used as input. After data regularization and standardization, it is input into a hash function to generate a unique summary. This summary is then concatenated with the hash value from the previous period to form a time-series hash chain. This chain structure, known as the first hash chain, ensures the non-repudiation and temporal consistency of microenvironmental perception data within the traceability system.

[0048] Meanwhile, a second hash chain is used to manage metabolite profile evolution data during food spoilage. A metabolite profile is a low-dimensional directed graph constructed based on the food's spoilage mechanism, time series, and metabolic pathways. Its data structure can include the type, concentration, temporal rate of change of each key metabolite, and its connection to previous and subsequent metabolites. In actual processing, the structural characteristics of the metabolic graph (such as path encoding, node weights, and edge direction information) are extracted as an ordered data vector and input into a hash function to generate a fixed-length hash digest value. This is also chained together with the previous spectrogram hash value through a chain structure to construct a second hash chain, ensuring that metabolic changes have a verifiable historical trace on the chain.

[0049] Furthermore, a multi-source data fusion code generation method is proposed. First, the latest digest value of the first hash chain in the current cycle is extracted as the first code representing the current microenvironmental state. Then, the latest digest value of the second hash chain in the current cycle is extracted as the second code representing the internal metabolic state of the food. These two codes, derived from the external environmental perception chain and the internal metabolic state chain, respectively, are highly independent and unforgeable.

[0050] Finally, the first and second codes are combined to generate the traceability code. This traceability code is used to bind the spatial-metabolic complex state of the current food unit at that time point, and serves as the core index value for smart contract identification, on-chain evidence storage, and visual tracking.

[0051] In summary, the complete spatiotemporal binding and unique identity identification of food spoilage data are achieved, ensuring that every change in the environment and internal status can be recorded and verified in a chain.

[0052] Furthermore, after deploying the smart contract rules, step S2 of this application includes: Using radio frequency tags as nodes, a traceability chain is generated by networking the Internet of Things based on three-dimensional spatial dimensions and time dimensions; wherein the traceability chain is updated according to the decision output of the smart contract rules.

[0053] In the implementation of this invention, to achieve continuous tracking and verifiable management of food spoilage behavior under different spatiotemporal conditions, we propose constructing an Internet of Things (IoT) network structure that integrates three-dimensional spatial and temporal dimensions, using radio frequency identification (RFID) tags as node units. This structure then generates a traceability chain for food spoilage. The RFID tags are information carriers attached to food packaging, providing unique identification capabilities, supporting wireless inductive reading, and collaborating with microenvironmental sensors for dynamic data collection. Each RFID tag is considered an independent node in the network, corresponding one-to-one with the food sample to which it is attached.

[0054] This invention further constructs an IoT networking system encompassing three-dimensional space and time, ensuring that food status data possesses spatial positioning capabilities and historical evolution traceability. Specifically, the three-dimensional spatial dimension is achieved through a positioning reference system established within the warehousing or transportation system, such as an XYZ coordinate system or ultra-wideband (UWB) / RF ranging positioning to obtain the spatial position of tag nodes. The temporal dimension uses a unified timestamp management mechanism (such as the NTP network time protocol) to record the instants of each node's data collection and uploading. Thus, each RFID node can be defined as a state unit within the three-dimensional structure of time, space, and data. All nodes in the network form an updateable IoT topology map based on spatial adjacency and state evolution relationships.

[0055] Based on this four-dimensional network, the present invention constructs a traceability chain to store the state evolution records of each food node over different time periods. The traceability chain is a chain structure composed of multiple node state data. Each chain records the microenvironment code (derived from the first hash chain), metabolic state code (derived from the second hash chain), and smart contract judgment results (such as deterioration type, precursor level, and attribution responsibility) of a food unit within a certain period. It is chain-linked to the previous state node through a timestamp and hash value, forming a complete chain record of the continuous changes in the food life cycle.

[0056] Furthermore, the present invention incorporates the decision output of smart contract rules as a triggering mechanism for dynamic updates to the traceability chain. When the system detects a change in the state of a radio frequency node (such as a dramatic fluctuation in the microenvironment or an abnormal shift in the metabolic pathway), a smart contract deployed on an edge computing node or blockchain platform is executed in real time. This contract evaluates and determines the current node state based on deterioration classification rules, precursor warning rules, or attribution judgment rules, outputting results such as deterioration grade, warning information, or responsible node labels. This result is then written as a new field into the status record of the corresponding node in the traceability chain, completing the chain update.

[0057] For example, in a cold chain logistics scenario, if a radio frequency node detects a temperature exceeding a set threshold for more than two hours, this will cause a certain degree of metabolic acceleration and a nonlinear accumulation trend of volatile amines in the metabolic profile. The smart contract can determine that the node has entered a state of mild deterioration and automatically update the node's record in the traceability chain, adding a new status field such as: {Warning Level: Level II, Deterioration Type: Oxidative Decomposition, Recommended Disposition: Nearest Isolation}. The traceability chain is then updated, forming an evolutionary chain segment with status annotations, providing the data foundation for subsequent visualization, food recalls, and responsibility allocation.

[0058] In summary, the present invention constructs a food status traceability chain through the spatial and temporal joint networking of radio frequency nodes, and dynamically updates the chain based on smart contract decisions, realizing a closed-loop mechanism from data perception, rule execution to information chain generation.

[0059] S3: IoT access is performed through the detection of spatial microenvironment and metabolite profiles, triggering cascade decision-making based on the coding environment of smart contract rules, determining the deterioration data and storing it on the chain; Among them, the traceability chain includes point traceability based on radio frequency scanning and process traceability based on inter-point metamorphic nonlinear acceleration coupling analysis, and is displayed in layers through radio frequency tag identification.

[0060] In further implementations of this invention, the combined detection results of the spatial microenvironment and metabolite profile are proposed as network access conditions, enabling data access and dynamic identification operations for IoT nodes. The spatial microenvironment refers to a real-time environmental dataset collected by environmental sensors distributed on food packaging or storage containers, including indicators such as temperature, humidity, oxygen concentration, and light intensity. The metabolite profile refers to a metabolic pathway map reflecting the biochemical reaction state within the food, generated based on the food's deterioration mechanism and its temporal evolution.

[0061] In a preferred embodiment, by setting a threshold judgment rule, when the microenvironment data change reaches the set sensitivity threshold and the concentration or structural change of key metabolites in the metabolic spectrum shows a nonlinear trend, it is considered to be a valid condition for network access, indicating that the node has the status trigger qualification to participate in the construction of the traceability chain.

[0062] Furthermore, the preset smart contract rules are called to carry out a cascade decision-making mechanism based on the coding environment. That is, the coding environment is composed of the traceability code generated by the metabolic spectrum hash of the first hash chain and the second hash chain constructed in the previous step, which is tamper-proof and synchronized in time and space.

[0063] Among them, based on the traceability code as the index key value, the smart contract rule set applicable to the current node is retrieved and matched, including modules such as deterioration classification, warning level determination, and attribution responsibility allocation. The execution is triggered in sequence through chain logic, thus forming a cascade judgment link of input traceability code → matching contract rules → multi-level judgment output.

[0064] Furthermore, the node status of data identified as valid spoilage data is structured and stored on-chain simultaneously. This on-chain storage, based on blockchain technology, combines the food unit's node identification, spatial location information, timestamp, microenvironmental data summary, metabolic profile summary, contract judgment results, and other information into a structured data package. This data is then packaged into blocks and written to the blockchain, ensuring that spoilage behavior has a verifiable historical record and on-chain status.

[0065] The traceability chain structure generated by the present invention has a two-layer traceability mechanism: the first is point traceability based on radio frequency scanning, that is, the unique code and binding data of the current node are obtained through the RFID reader, so as to quickly realize the static status query of a single food unit at a specific time point; the second is process traceability based on the nonlinear acceleration coupling analysis of inter-point deterioration. This part tracks the state change trends of multiple nodes in the time dimension and spatial adjacent relationship, and uses statistical modeling or graph neural network methods to evaluate whether the deterioration behavior has conductivity or local aggregation. For example, if a cold storage area experiences environmental temperature rise due to equipment failure, causing synchronous metabolic spectrum acceleration changes in adjacent nodes, it can be determined as a regional nonlinear deterioration event.

[0066] Furthermore, to enhance user interaction and improve data management efficiency, a hierarchical display mechanism is provided. When users access data through code scanning, card swiping, or radio frequency sensing, the base layer data associated with the tag (such as food identity information and current environmental status) is first displayed. Then, as needed, higher-level information layers are called up, such as spoilage type determination, precursor level analysis, and process link backtracking. This is presented using an expandable hierarchical structure or graphical display interface, achieving a multi-level traceability view.

[0067] In summary, complete tracking and visual management of food units during the dynamic deterioration process is achieved.

[0068] Furthermore, the coding environment cascade decision-making based on the smart contract rules is triggered to determine the deteriorated data and store it on the chain. Step S3 of this application includes: When a radio frequency tag scan is performed or there is a change in the spatial microenvironment, a dynamic update node is generated; an updated traceability code of the dynamic update node is obtained, wherein the updated traceability code is the traceability code updated in real time; for the updated traceability code, a cascade discrimination based on the smart contract rules is triggered to determine the deteriorated data and store it on the chain.

[0069] In the further implementation process of the present invention, in view of the state changes that may occur at any time during the storage, transportation or sales cycle of food, a mechanism is proposed to trigger dynamic node generation and traceability chain update through radio frequency tag scanning or spatial microenvironment change monitoring, thereby realizing real-time perception, dynamic judgment and on-chain evidence storage of food spoilage information.

[0070] Specifically, when a user is detected scanning a radio frequency tag—for example, using an RFID reader, NFC mobile terminal, or other device to identify a food tag at a transfer station, or when deployed microenvironmental sensors detect changes in spatial microenvironmental parameters such as sudden temperature changes, excessive humidity, or abnormal lighting that exceed a set threshold—a dynamic node is automatically generated and recorded as a dynamic update node. This type of node is essentially a temporal node added to the existing traceability chain, representing a snapshot of the state of a food unit after a specific event, such as a temperature rise of more than 3°C or a user scanning a code to open it.

[0071] At the same time, based on the latest collected environmental data and metabolic status, an updated traceability code corresponding to that node is recalculated and generated. This updated traceability code is based on the state summary information synchronously derived from the first hash chain (microenvironment) and the second hash chain (metabolic profile), combined with auxiliary data such as the current node's timestamp and spatial location. A unique identifier is generated using a hash function, ensuring that the code fully represents the food's baseline state of spoilage risk at that moment. Compared to the initial traceability code, this updated traceability code offers real-time performance and traceability of state differences, enabling chain-based state evolution analysis and responsibility attribution.

[0072] Subsequently, using the update traceability code as an index, the corresponding smart contract rules are triggered to perform cascade discrimination processing. This cascade discrimination refers to the sequential matching and calculation of the discrimination hierarchy within the smart contract, including but not limited to the following levels: whether it belongs to a known deterioration type; whether the precursor warning trigger conditions are met; whether there is a similarity or coupling relationship between nodes for attribution of responsibility; and whether the on-chain storage threshold conditions are met. The contract operation process adopts a chain logic structure to ensure that the judgment outputs at different levels can be triggered in conjunction with status information, avoiding false positives or missed judgments.

[0073] Once the cascaded discriminant output confirms that a dynamic node exhibits a significant deterioration trend or has entered a suspicious state, the system immediately performs an on-chain storage operation. This on-chain operation writes structured information to the blockchain node, including the node's unique identifier, updated traceability code, spatial / temporal identifier, deterioration determination, precursor level, contract execution path, and whether an early warning is required, forming a complete on-chain record data package. This data package not only ensures the immutability of the information but also provides an on-chain basis for subsequent abnormal recalls, cold chain verification, and food responsibility audits.

[0074] For example, during a cold chain transportation process, the humidity sensor at a node in the middle area of transportation for a batch of dairy products detected that the humidity briefly jumped to above 85%, and was subsequently accompanied by an increase in the accumulation rate of aldehyde metabolites. The system generated an updated traceability code for the node, triggered the smart contract cascade judgment, and finally determined it to be a mild oxidative deterioration trend node, and marked it as warning level II on the chain, forming an on-chain record with time and space labels.

[0075] In summary, the automatic response mechanism and high-frequency status monitoring capability of food spoilage data in practical applications are realized.

[0076] Furthermore, step S3 of the present application includes: if the spoilage data contains precursor warning data, generating a nearest routing instruction; and performing nearest routing processing on the batch of food according to the nearest routing instruction.

[0077] In the further implementation of the present invention, in order to improve the response efficiency and disposal timeliness of potentially spoiled food, a local routing mechanism based on precursor warning data is constructed to trigger a chain response when the spoilage trend first appears, guiding the real-time adjustment and optimization management of the food circulation path.

[0078] Specifically, after executing the smart contract rules and identifying the dynamic nodes, if the spoilage data contains identifiable precursor warning data, that is, the food has not yet completely spoiled but has shown a clear abnormal metabolic pathway trend or abnormal characteristics of microenvironmental parameters, the node will be regarded as a potential risk node and the nearest disposal strategy will be further initiated.

[0079] The precursor warning data includes, but is not limited to, the following characteristics: abnormal accumulation of certain key metabolites in the food metabolic profile (such as volatile amino compounds, thiols, aldehydes, and ketones), with their change curves exhibiting nonlinear mutations exceeding conventional thresholds; spatial microenvironmental data, such as sudden temperature increases exceeding 3°C, humidity exceeding 80%, and a sharp drop in oxygen concentration, reflecting a sudden change in the storage and transportation environment; or historical spoilage events of similar foods with similar characteristic values, which have been confirmed on the chain. Once the above characteristic conditions are identified, the food unit can be deemed to have spoilage precursor attributes.

[0080] In response to this precursor state, the present invention has constructed a mechanism for generating nearby routing instructions. That is, based on the spatial positioning coordinates of the current node, the batch it is in, the transportation route, the target area and other information, combined with the configured schedulable resource nodes, such as nearby cold storage, quality re-inspection stations, near-expiry sorting areas, allocation temporary storage points, etc., a shortest processing path or minimum cost transfer path is selected.

[0081] In terms of implementation, a graph optimization method can be adopted to construct a path map based on the food Internet of Things network topology map, determine the nearest transferable destination node, and finally generate a nearest routing instruction in the form of {starting point node ID, target disposal node ID, recommended path segment ID set, priority parameter}.

[0082] Furthermore, batches of food are scheduled and routed in real time according to the nearest routing instructions, that is, food marked with warning marks is separated from the original distribution flow on a priority basis without affecting the overall logistics path, and is redirected to the designated node to complete secondary testing, isolated storage or pre-disposal.

[0083] For example, if a batch of dairy products with metabolic abnormalities is identified in a cold chain delivery vehicle, an instruction is generated to transfer the batch to the inspection station at the second logistics node along the line for review. The dispatching platform completes the physical separation and direction adjustment of the batch through automatic sorting devices or manual prompt systems to achieve rapid response.

[0084] In summary, this step achieves active transfer and risk control of batch-level food units in the early stages of spoilage through early identification of precursory spoilage characteristics, path analysis, and deployment of routing instruction generation mechanisms.

[0085] Furthermore, after the on-chain storage is completed, the application steps include: A multi-factor identification code is introduced, and the traceability chain is marked based on the multi-factor identification code, wherein the traceability chain includes a coding identification layer and a source data layer; the coding identification layer is displayed through radio frequency tag recognition, and the data of the source data layer is retrieved and displayed through inter-layer mapping association.

[0086] In the further implementation of the present invention, in order to enhance the structural clarity, identification traceability and visualization friendliness of food spoilage data in the chain management process, a multi-level identification code is introduced to systematically mark and display different data levels of the food traceability chain.

[0087] The multi-dimensional identification code refers to the use of different and unique identification methods for traceability data with different characteristics to improve intuitiveness.

[0088] Specifically, the traceability chain structure constructed in this invention comprises two logical layers: a coding identification layer and a source data layer. The coding identification layer records the coded information corresponding to each status node in the traceability chain and includes the aforementioned identification codes, such as the node's primary ID (generated by concatenating the RFID tag ID and a timestamp), a status type label (e.g., normal, warning, deteriorated), an associated traceability code (generated by a hash chain), node location information, and a summary of the node's source path. This layer, with its streamlined structure and clear identification, is primarily used for rapid display and preliminary screening within the visualization system's front-end interface.

[0089] The source data layer stores the original perception data and structured judgment results under the node status, including microenvironment raw data (temperature, humidity, air pressure, etc.), metabolic spectrum coding structure, smart contract execution log, historical status records, data sampling cycle records, etc. The data volume is large and the structure is complex. It is used to support internal responsibility attribution and regulatory review.

[0090] During operation, the data field information of the coding identification layer is displayed first by identifying the radio frequency tag bound to the food packaging. By clicking, expanding or filtering, the provided inter-layer mapping mechanism is called, and the node ID, traceability code, etc. recorded in the coding identification layer are used as index key values to trigger the retrieval and display of data in the source data layer.

[0091] Preferably, the mapping mechanism supports calling the corresponding node original data from the blockchain or traceability database through structured query language, for example, on-chain structured retrieval, and displaying detailed content including the temperature change curve under the node, metabolite concentration sequence, judgment node output path, contract execution details, etc., in the form of charts or forms, to achieve a visual closed loop from label identification to data traceability.

[0092] In summary, the logical decoupling and physical mapping of the coding identification layer and the source data layer are achieved, providing a complete data identification framework and calling logic for multi-level visual tracking of food spoilage data.

[0093] This application provides a method for visually tracing food spoilage data, which has the following technical effects: 1. Based on food characteristics, a low-dimensional directed metabolite profile is constructed in a time series. Complex spoilage information is processed through clustering and dimensionality reduction. Compared to traditional high-dimensional data, this reduces analysis complexity and improves the efficiency of spoilage feature identification, laying the foundation for subsequent accurate traceability. A four-dimensional IoT network is constructed using radio frequency tags as nodes, coupling the spatial microenvironment with the metabolite profile to deploy smart contract rules. This enables full temporal state traceability for both static traceability nodes and dynamic traceability processes, improving the completeness of link information. Utilizing a discriminative model based on adversarial training, spoilage classification, early warning, and dynamic attribution are implemented, enabling early detection of spoilage risks, precise location of the source of responsibility, and enhanced food supply chain management capabilities. A coding environment is employed to enhance data chain security.

[0094] 2. Two-dimensional traceability display, encompassing both point and process traceability chains, uses multi-layered identification codes to display both the coded identification layer and the source data layer. This protects the privacy of core data while enabling on-demand access to source data, providing intuitive and clear traceability information, facilitating regulatory and consumer inquiries, and enhancing transparency and credibility of food quality. Intelligent routing based on early warnings: If early warning data for spoilage is detected, routing instructions are automatically generated to the nearest batch of food. This mechanism can promptly prevent the spread of potential spoilage risks, reduce food loss, optimize supply chain logistics resource allocation, and lower enterprise operating costs.

[0095] Example 2: Based on the same inventive concept as the method for visually tracing food spoilage data in the above embodiment, Figure 3 As shown, the present application provides a visual traceability processing system for food spoilage data, the system comprising: A construction module 11 is used to construct a metabolite profile in a time series according to food characteristics, wherein the metabolite profile is a low-dimensional directed graph based on the deterioration type; Deployment module 12 is used to establish a four-dimensional IoT network using the radio frequency tags on food packaging as nodes. By coupling the spatial microenvironment with the metabolite spectrum and using the nonlinear acceleration characteristics of spoilage as constraints, smart contract rules are deployed. Among them, spoilage classification, precursor warning, and dynamic attribution judgment are the rule-oriented guidelines; Decision module 13 is used to perform IoT network access operations based on the detection of spatial microenvironment and metabolite spectrum, trigger the coding environment cascade decision based on smart contract rules, determine the deterioration data and store it on the chain; Among them, the traceability chain includes point traceability based on radio frequency scanning and process traceability based on inter-point metamorphic nonlinear acceleration coupling analysis, and is displayed in layers through radio frequency tag identification.

[0096] Furthermore, the construction module 11 is used to perform the following steps: according to the food characteristics of the target food, clustering is performed based on the deterioration type to determine multiple types of metabolic configurations; by minimizing the intra-class distance and maximizing the inter-class distance, the multiple types of metabolic configurations are subjected to dimensionality reduction processing to determine multiple types of lightweight configurations, wherein the dimensionality reduction method includes configuration type dimensionality reduction and configuration local structure dimensionality reduction; based on the multiple types of lightweight configurations, the metabolite spectrum is determined.

[0097] Furthermore, the deployment module 12 is configured to perform the following steps: The coding mode is used as the contract environment, and the discriminant mode is deployed based on the adversarial training principle. The discriminant mode includes a first discriminant node based on deterioration classification, a second discriminant node based on precursor warning, and a third discriminant node based on dynamic attribution determination. The mode is trained using a generative adversarial architecture and deployed using a decomposed adversarial architecture part. The dynamic attribution is used to determine the responsibility for deterioration. The smart contract is deployed based on the discriminant mode.

[0098] Furthermore, the deployment module 12 is configured to perform the following steps: A first Hash chain and a second Hash chain are introduced, wherein the first Hash chain is associated with the spatial microenvironment, and the second Hash chain is associated with the metabolite spectrum; microenvironment sensing is performed to convert the first code based on the first Hash chain; according to the upstream metabolite spectrum, the second code is converted to the second Hash chain; and a traceability code is generated based on the first code and the second code.

[0099] Furthermore, the deployment module 12 is configured to perform the following steps: Using radio frequency tags as nodes, a traceability chain is generated by networking the Internet of Things based on three-dimensional spatial dimensions and time dimensions; wherein the traceability chain is updated according to the decision output of the smart contract rules.

[0100] Furthermore, the decision module 13 performs the following steps: When a radio frequency tag scan is performed or there is a change in the spatial microenvironment, a dynamic update node is generated; an updated traceability code of the dynamic update node is obtained, wherein the updated traceability code is the traceability code updated in real time; for the updated traceability code, a cascade discrimination based on the smart contract rules is triggered to determine the deteriorated data and store it on the chain.

[0101] Furthermore, the decision module 13 executes the following steps: if the spoilage data contains precursor warning data, generating a nearest routing instruction; and performing nearest routing processing on the batch of food according to the nearest routing instruction.

[0102] Furthermore, the system is further configured to perform the following steps: A multi-factor identification code is introduced, and the traceability chain is marked based on the multi-factor identification code, wherein the traceability chain includes a coding identification layer and a source data layer; the coding identification layer is displayed through radio frequency tag recognition, and the data of the source data layer is retrieved and displayed through inter-layer mapping association.

[0103] Through the detailed description of the visual traceability processing method for food spoilage data in the foregoing description, those skilled in the art can clearly understand the visual traceability processing method and system for food spoilage data in this embodiment. As for the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0104] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A visual traceability processing method for food spoilage data, characterized in that: The method comprises: Constructing a metabolite profile in a time series according to food characteristics, wherein the metabolite profile is a low-dimensional directed graph based on the deterioration type; Using radio frequency tags on food packaging as nodes, a four-dimensional IoT network is established. By coupling the spatial microenvironment with the metabolite spectrum and using the nonlinear acceleration characteristics of spoilage as constraints, smart contract rules are deployed. Among them, spoilage classification, precursor warning, and dynamic attribution judgment are the rule-guided methods. The IoT is connected to the network through the detection of spatial microenvironment and metabolite spectrum, triggering cascade decision-making based on the coding environment of smart contract rules, determining the deterioration data and storing it on the chain; Among them, the traceability chain includes point traceability based on radio frequency scanning and process traceability based on inter-point metamorphic nonlinear acceleration coupling analysis, and is displayed in layers through radio frequency tag identification.

2. The method according to claim 1, wherein Construct a metabolite profile under time series, including: According to the food characteristics of the target food, clustering based on the spoilage type is performed to determine multiple metabolic profiles; By minimizing the intra-class distance and maximizing the inter-class distance, the multiple metabolic configurations are subjected to dimensionality reduction processing to determine multiple lightweight configurations, wherein the dimensionality reduction method includes configuration type dimensionality reduction and configuration local structure dimensionality reduction; The metabolite profile is determined based on the multiple types of lightweight configurations.

3. The method according to claim 1, wherein Deploy smart contract rules, including: Using the coding model as the contract environment and adversarial training principles, a discriminant model is deployed. The discriminant model includes a first discriminant node based on deterioration classification, a second discriminant node based on precursor warning, and a third discriminant node based on dynamic attribution determination. Training is performed using a generative adversarial architecture and deployment is performed using a decomposed adversarial architecture. Dynamic attribution is used to determine deterioration responsibility. Deploy the smart contract according to the judgment mode.

4. The method according to claim 3, wherein The contract environment is coded in code, including: Introducing a first Hash chain and a second Hash chain, wherein the first Hash chain is associated with the spatial microenvironment, and the second Hash chain is associated with the metabolite profile; By performing microenvironment sensing, converting into a first code based on a first hash chain; According to the upstream metabolite spectrum, converting into a second code based on the second hash chain; A traceability code is generated according to the first code and the second code.

5. The method according to claim 1, wherein After deploying the smart contract rules, including: Using radio frequency tags as nodes, a traceability chain is generated by networking the Internet of Things based on three-dimensional spatial and temporal dimensions; Wherein, the traceability chain is updated according to the decision output of the smart contract rules.

6. The method according to claim 5, wherein Triggering cascading decisions in the coding environment based on smart contract rules to identify deteriorating data and store it on-chain, including: When performing RFID tag scanning or there are changes in the spatial microenvironment, a dynamic update node is generated; Obtaining an updated traceability code of the dynamically updated node, wherein the updated traceability code is the traceability code updated in real time; For the updated traceability code, a cascade discrimination based on the smart contract rules is triggered to determine the deteriorated data and store it on the chain.

7. The method according to claim 6, wherein If there is any precursor warning data in the deteriorated data, a nearest routing instruction is generated; According to the nearest routing instruction, the batch of food is routed nearest.

8. The method according to claim 1, wherein After the on-chain storage, it includes: Introducing a multi-element identification code and marking the traceability chain based on the multi-element identification code, wherein the traceability chain includes a coding identification layer and a source data layer; The coded identification layer is displayed through radio frequency tag identification, and the data of the source data layer is retrieved and displayed through inter-layer mapping association.

9. A visual traceability processing system for food spoilage data, characterized in that: A system for executing a method for visual traceability of food spoilage data according to any one of claims 1 to 8, comprising: A construction module is used to construct a metabolite profile in a time series according to food characteristics, wherein the metabolite profile is a low-dimensional directed graph based on the deterioration type; The deployment module is used to establish a four-dimensional IoT network using radio frequency tags on food packaging as nodes. By coupling the spatial microenvironment with the metabolite spectrum and using the nonlinear acceleration characteristics of spoilage as constraints, smart contract rules are deployed. Among them, spoilage classification, precursor warning, and dynamic attribution judgment are the rule-oriented guidelines. The decision-making module is used to perform IoT network access operations based on the detection of spatial microenvironment and metabolite spectrum, trigger the coded environment cascade decision based on smart contract rules, determine the deterioration data and store it on the chain; Among them, the traceability chain includes point traceability based on radio frequency scanning and process traceability based on inter-point metamorphic nonlinear acceleration coupling analysis, and is displayed in layers through radio frequency tag identification.

Images