A slaughter whole-process traceability management system based on big data

Abstract

The application discloses a kind of slaughter whole-process traceability management systems based on big data, it is related to slaughtering traceability technical field, including the acquisition terminal of each batch livestock's whole life cycle original data set is obtained by Internet of Things;After data cleaning and standardization conversion, the standardized traceability dataset with global unique batch traceability code is generated;Import big data distributed storage cluster and establish cross-link chain index in time sequence to form traceability main chain;Based on traceability main chain, start batch processing computing task, multidimensional correlation analysis excavates key control point data, and matches risk rule library to generate quality safety audit log;Enhanced traceability archives are generated by merging audit log and standardized traceability dataset, and write into block chain storage evidence node.The system realizes slaughter whole-process traceability coherent and traceable, guarantees that traceability information is authentic and credible, and is suitable for slaughtering industry quality safety traceability management.

Description

Technical Field

[0001] This invention belongs to the field of slaughter traceability technology, specifically a big data-based full-process traceability management system for slaughter. Background Technology

[0002] Currently, traceability management in the slaughtering industry largely relies on traditional technologies. This involves deploying simple data collection devices to acquire basic data from various stages of the slaughtering process. This data is stored in local databases or small servers, using a single encoding method. The traceability process depends on manual searching of scattered data from each stage, lacking a unified data integration and correlation mechanism. Existing systems only perform simple, single-dimensional statistical analysis, providing only basic organization of raw data without in-depth data correlation and mining. Quality and safety audits rely on manual verification against basic standards, resulting in a lack of effective integration between traceability data and audit results. Evidence is primarily stored locally, lacking reliable anti-tampering mechanisms.

[0003] Traditional storage methods cannot handle the massive amounts of multi-stage data generated throughout the entire slaughtering process. Dispersed data storage leads to gaps in traceability information across different stages, preventing the formation of a coherent traceability chain, resulting in low efficiency and frequent data loss in traceability queries. Single-dimensional data analysis fails to capture key risk points in the slaughtering production process, and audits lack specificity, making it difficult to comprehensively control product quality and safety. Local evidence storage is susceptible to external factors, and data is easily tampered with or lost, failing to guarantee the authenticity and integrity of traceability information and making it difficult to meet the traceability management needs of the slaughtering industry's high-quality development.

[0004] It is necessary to achieve unified coding and time-series associated storage of data throughout the entire slaughtering process to solve the problems of data dispersion and broken traceability chains at each stage; it is necessary to mine key production data through multi-dimensional correlation analysis to complete accurate audit verification, and to achieve secure storage of traceability data and audit results to solve the problems of inaccurate storage and easy tampering. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in the prior art;

[0006] Therefore, this invention proposes a big data-based traceability management system for the entire slaughtering process, comprising:

[0007] The data acquisition module acquires raw data sets for the entire life cycle of each batch of livestock through IoT acquisition terminals deployed at various operational stages of the slaughtering production line.

[0008] The data processing module performs data cleaning and standardization transformation on the raw data set of the entire life cycle to generate a standardized traceability dataset with unified coding. Each record in the standardized traceability dataset carries a globally unique batch traceability code.

[0009] The storage index module imports the standardized traceability dataset into a big data distributed storage cluster, and establishes a cross-stage chain index relationship according to the temporal relationship of data generation to form the main traceability chain;

[0010] The analysis and calculation module, based on the traceability main chain, initiates a batch processing calculation task to perform multi-dimensional correlation analysis on the standardized traceability dataset and mine key control point data in the slaughtering production process;

[0011] The audit and verification module uses the key control point data to perform matching and verification in a pre-set risk rule base and generates a quality and safety audit log for the corresponding batch of livestock.

[0012] The document storage module merges the quality and safety audit logs with the corresponding standardized traceability dataset to generate an enhanced traceability document, and writes the enhanced traceability document into the blockchain storage node.

[0013] Furthermore, the original dataset throughout its entire lifecycle is cleaned and standardized to generate a standardized traceability dataset with unified coding, including:

[0014] The complete lifecycle raw data set includes breeding record data, entry quarantine data, slaughtering and processing data, and cold chain logistics data;

[0015] The ear tag numbers, immunization records, and feed feeding records in the breeding records are deduplicated and logically verified to remove contradictory or missing records.

[0016] The animal quarantine certificate number, clinical examination results, and laboratory test reports in the entry quarantine data are standardized in format, and unstructured text is converted into structured fields;

[0017] The knife disinfection records, carcass weighing data, and by-product classification information in the slaughtering and processing data are normalized to a unit, and the operator's employee number information is added.

[0018] The temperature and humidity sensor readings in the cold chain logistics data are synchronized with the geographical coordinates in time to generate a transportation environment trajectory with timestamps.

[0019] For each data record after data cleaning and standardization, a globally unique batch traceability code is generated, which includes the place of origin code, slaughterhouse code, production date and random serial number. All data records with attached batch traceability codes are combined into the standardized traceability dataset.

[0020] Furthermore, the standardized traceability dataset is imported into a big data distributed storage cluster, and a cross-stage chain index relationship is established according to the temporal relationship of data generation to form the main traceability chain, including:

[0021] In the big data distributed storage cluster, an independent storage space partition is created for each batch traceability code;

[0022] The breeding record data, entry quarantine data, slaughtering and processing data and cold chain logistics data belonging to the same batch traceability code are written into the storage space partition in the order of data collection time.

[0023] During the data writing process, a hash pointer is automatically generated that points to the data record of the previous stage, thus linking the data records of all stages together.

[0024] When any data record in any link changes, the hash value of the data record is recalculated and the subsequent hash pointers are updated, thereby maintaining the integrity and immutability of the traceability main chain.

[0025] Furthermore, based on the aforementioned traceability main chain, a batch processing task is initiated to perform multi-dimensional correlation analysis on the standardized traceability dataset, thereby mining key control point data in the slaughtering production process, including:

[0026] Read the complete data link of the tracing main chain in the big data distributed storage cluster;

[0027] Traverse each node in the data link and extract the process parameter data and quality inspection data;

[0028] The process parameter data is compared with the preset process standard range to identify abnormal parameter nodes that deviate from the standard.

[0029] The quality inspection data is compared with the normal quality baseline of the same batch in history to identify abnormal parameter nodes that deviate from the standard.

[0030] The nodes of the abnormal parameters and the nodes of the quality fluctuations are overlaid on the time axis, and the nodes in the intersection are the key control point data.

[0031] Furthermore, using the aforementioned key control point data, matching and verification are performed in a pre-set risk rule base to generate quality and safety audit logs for the corresponding batch of livestock, including:

[0032] Each control point in the key control point data is broken down into specific detection indicators, measured values, and occurrence times;

[0033] The detection indicators and measured values ​​are used as query conditions to search in the risk rule base, which stores the safety thresholds and risk level definitions of various detection indicators.

[0034] When the measured value exceeds the safety threshold, the corresponding risk level and handling suggestions are obtained from the risk rule base.

[0035] Combine the occurrence time, detection indicators, measured values, risk level, and handling recommendations to form an audit item;

[0036] All audit entries under the same batch traceability code are summarized in chronological order to generate the quality and safety audit log.

[0037] Furthermore, the quality and safety audit logs are merged with the corresponding standardized traceability dataset to generate an enhanced traceability profile, including:

[0038] Using the batch traceability code as the association key, the standardized traceability dataset and the quality and safety audit log are located in the big data distributed storage cluster, respectively.

[0039] Each audit entry in the quality and safety audit log is attached as an extended attribute to the corresponding original data record in the standardized traceability dataset;

[0040] The data records with audit entries are re-encoded to generate enhanced data objects containing quality risk labels;

[0041] All enhanced data objects are arranged in the order of the main traceability chain and encapsulated into the enhanced traceability file.

[0042] Furthermore, the enhanced traceability file is written to the blockchain evidence storage node, including:

[0043] The enhanced traceability archive is serialized and converted into a byte stream format suitable for network transmission;

[0044] Calculate a holistic digital digest for the data packet in the byte stream format, the digital digest being an aggregation of the hash values ​​of all enhanced data objects within the data packet;

[0045] Construct a new block that includes the digital digest, timestamp, operator identification, and operation type;

[0046] The newly added block is broadcast to all consensus nodes in the blockchain evidence storage node network, and the consensus nodes verify and record the newly added block.

[0047] Once the newly added block is successfully written into the blockchain ledger, the corresponding block height and transaction hash are written back to the metadata of the enhanced traceability archive.

[0048] Furthermore, it also includes:

[0049] When a query request for a specific batch traceability code is received, the enhanced traceability file is retrieved from the blockchain evidence storage node and parsed into a visualized flow path diagram, which is then displayed to the querying party. Specifically, this includes:

[0050] Parse the query request and extract the specific batch traceability code and query authorization certificate carried therein;

[0051] Using the specific batch traceability code, query the corresponding block height and transaction hash in the blockchain ledger;

[0052] Based on the retrieved block height and transaction hash, the original enhanced traceability file is located from the blockchain evidence storage node;

[0053] The enhanced traceability archive is subjected to integrity verification. This is done by recalculating its digital digest and comparing it with the digital digest recorded in the block. Once it is confirmed that the data has not been tampered with, the archive content is returned.

[0054] Furthermore, the enhanced traceability archive is parsed into a visualized flow path diagram, including:

[0055] Read the main traceability chain data in the enhanced traceability file to identify each node in the process from breeding, quarantine, slaughter, processing to logistics;

[0056] Extract the occurrence time, geographical location, operators, and quality risk tags corresponding to each stage;

[0057] Arrange the nodes in chronological order in a two-dimensional coordinate system, use connecting lines to indicate the flow direction, and use different colors or shapes to distinguish different quality risk levels;

[0058] Detailed operational information is overlaid and displayed at each stage node, forming the visualized flow path diagram.

[0059] Furthermore, it also includes: a model adaptation module, used to continuously perform incremental learning on the historical standardized source dataset in the big data distributed storage cluster during system operation, including:

[0060] Newly added standardized traceability datasets are periodically extracted from the big data distributed storage cluster as incremental training samples;

[0061] Analyze the novel data features and anomaly patterns reflected in the incremental training samples, and update the rule definitions in the risk rule base;

[0062] The updated risk rule base is redeployed into the batch processing task for the next cycle of key control point data mining, enabling the system to adapt to changes in the slaughtering process.

[0063] Compared with the prior art, the beneficial effects of the present invention are:

[0064] A standardized traceability dataset, with unified coding and a globally unique batch traceability code for each record, is imported into a big data distributed storage cluster. A cross-stage chained index relationship is established based on the temporal relationship of data generation, forming the main traceability chain. The big data distributed storage cluster can stably handle the massive amounts of multi-stage data generated throughout the entire slaughtering process, achieving large-scale data storage and reducing data storage failures and loss. The globally unique batch traceability code uniquely identifies the traceability data of each batch of livestock, enabling precise correlation of data at each stage and avoiding data association deviations caused by coding confusion. The cross-stage chained index relationship connects the traceability data of each slaughtering operation in chronological order, breaking down data silos and achieving continuous traceability of livestock from entry to slaughter completion. This improves the efficiency and accuracy of traceability information retrieval and reduces traceability anomalies caused by breaks in the traceability chain.

[0065] Batch processing tasks are initiated based on the traceability main chain to perform multi-dimensional correlation analysis on standardized traceability datasets. This uncovers key control point data in the slaughtering process, which is then matched and verified against a pre-defined risk rule base to generate quality and safety audit logs. These logs are then merged with the corresponding standardized traceability datasets to create enhanced traceability profiles, which are then written to blockchain storage nodes. Batch processing enables efficient analysis of massive amounts of traceability data, while multi-dimensional correlation analysis accurately uncovers key control information in the production process, improving the accuracy of risk verification and reducing the subjective bias and one-sidedness of traditional manual audits. The enhanced traceability profiles integrate traceability data and audit information, achieving synchronous correlation between data and audit results, clearly presenting the quality and safety status of each batch of livestock. The writing to blockchain storage nodes ensures the immutability and traceability of the traceability profiles, reducing data tampering and loss, and guaranteeing the authenticity of traceability data and audit results. Attached Figure Description

[0066] Figure 1 This is a sequence diagram of a big data-based full-process traceability management system for slaughtering, as described in this invention.

[0067] Figure 2 A flowchart for establishing a chained index relationship to form the main traceability chain;

[0068] Figure 3 A correlation analysis diagram of abnormal parameters and comprehensive risk levels at key control points throughout the slaughtering process;

[0069] Figure 4 Analysis chart of the number of blockchain-based evidence storage and verification nodes for enhanced traceability of the entire slaughter process;

[0070] Figure 5 A heat map showing the risk level distribution and adaptive rule update analysis for the entire slaughtering process traceability. Detailed Implementation

[0071] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0072] See Figure 1 The data acquisition module obtains raw data sets for the entire life cycle of each batch of livestock through IoT acquisition terminals deployed at various operational stages of the slaughtering production line. The data processing module cleans and standardizes the raw data sets to generate a standardized traceability dataset with unified coding. Each record in the standardized traceability dataset carries a globally unique batch traceability code. The storage and indexing module imports the standardized traceability dataset into a big data distributed storage cluster and establishes a cross-stage chain index relationship according to the temporal relationship of data generation, forming the traceability main chain. Based on the traceability main chain, the analysis and calculation module initiates a batch processing calculation task to perform multi-dimensional correlation analysis on the standardized traceability dataset, mining key control point data in the slaughtering production process. The audit and verification module uses the key control point data to perform matching and verification in a pre-set risk rule base, generating quality and safety audit logs for the corresponding batch of livestock. The archive and evidence storage module merges the quality and safety audit logs with the corresponding standardized traceability dataset to generate an enhanced traceability archive, and writes the enhanced traceability archive to the blockchain evidence storage node.

[0073] In one embodiment of the present invention, the data processing module receives a set of raw data covering the entire lifecycle acquired by the data acquisition module. This set includes livestock breeding records, entry quarantine data, slaughtering and processing data, and cold chain logistics data. The livestock breeding records involve the original records of individual livestock at the farm; the entry quarantine data are the original records generated during the quarantine process when livestock enter the slaughterhouse; the slaughtering and processing data are the original records generated during the processing stage on the slaughtering line; and the cold chain logistics data are the original records generated by sensors during the transportation of the processed products. The data processing module performs deduplication and logical verification on the ear tag numbers, immunization records, and feed feeding records in the livestock breeding records. For example, if the livestock breeding records record "Livestock number A001, immunization record shows that it has been vaccinated with foot-and-mouth disease vaccine type A, but the feed feeding record also contains drug B, which is prohibited from being used with vaccine type A," the data processing module will remove this contradictory record. The format of animal quarantine certificate numbers, clinical examination results, and laboratory test reports in the entry quarantine data is standardized. In the entry quarantine data, animal quarantine certificate numbers may exist in various forms such as "Quarantine No. (2026) 001" and "2026001". The data processing module will uniformly convert them into the standard structure of "JY2026001". The unstructured text in the clinical examination results and laboratory test reports, such as "No obvious abnormalities were found on the body surface, and the mental state is acceptable", will be converted into a structured combination of fields, such as "Body surface condition: normal; mental state: good". The data processing module performs unit normalization on knife disinfection records, carcass weighing data, and by-product classification information in the slaughtering and processing data. For example, the original disinfectant concentration data in the knife disinfection records might be recorded in the form of "5%" or "50g / L," which will be uniformly converted to the standard unit "g / L." Similarly, the carcass weighing data might be in the unit of "jin" or "kilogram," which will be uniformly converted to "kilogram." The module also adds the operator's employee ID information, linking each slaughtering and processing data record to a specific operator. Furthermore, the module synchronizes the temperature and humidity sensor readings with the geographic location coordinates in the cold chain logistics data. Readings from different brands of temperature and humidity sensors may have different time bases, and location coordinates from vehicle GPS also have their own independent timestamps. Using Coordinated Universal Time (UTC) as the reference, the data processing module aligns all sensor readings with the coordinate information to generate a transportation environment trajectory sequence with precise timestamps.

[0074] In practice, a globally unique batch traceability code is generated for each data record after the above data cleaning and standardization transformation. The generation of the batch traceability code follows the formula:

[0075]

[0076] in: This indicates the generated batch traceability code. It is a fixed-digit origin code used to identify the region or farm from which livestock originated. It is a fixed-digit slaughterhouse code used to identify the factory that performs slaughtering and processing. It is the production date, in the format "YYYYMMDD". It is a system-generated random serial number of sufficient length to avoid duplicate codes for different batches from the same day, the same place of origin, and the same slaughterhouse. After generating the batch traceability code, the data processing module attaches the batch traceability code to the corresponding processed data record, and combines all data records with the same batch traceability code to form a standardized traceability dataset.

[0077] In some embodiments, the storage index module imports the standardized traceability dataset into a pre-deployed big data distributed storage cluster, see [reference]. Figure 2 In a big data distributed storage cluster, an independent, physically or logically isolated storage space partition is created for each unique batch traceability code. The storage index module reads the standardized traceability dataset and writes the breeding record data, entry quarantine data, slaughtering and processing data, and cold chain logistics data belonging to the same batch traceability code into the storage space partition created for that batch traceability code in chronological order of data collection. The writing process strictly follows the timestamp order; for example, the earliest breeding record data is written first, followed by the subsequent entry quarantine data, and finally the cold chain logistics data. During the process of writing each data record to the storage space partition, the storage index module calculates a hash value based on the entire content of that data record and uses the hash value of the previously successfully written data record as a "hash pointer to the previous data record," storing it together with the current data record. Through this mechanism, each data record from the breeding record data to the cold chain logistics data is linked together by a hash pointer pointing to its predecessor, forming a logical chain structure, i.e., the traceability main chain.

[0078] In some embodiments, when a data record stored in any link of the traceability main chain needs to be changed due to information correction or other reasons, the storage index module recalculates the new hash value corresponding to the entire content of the changed data record and uses this new hash value to update the hash pointer pointing to this link in the data record of the next link. Due to the change of the hash pointer, the storage index module must continue to recalculate and update the hash pointers in all subsequent link data records, thereby maintaining the integrity and immutability of the entire traceability main chain after data changes. It can be understood that this chain-index relationship maintenance mechanism means that any tampering with data in any historical link will cause all hash pointers from that link to the end of the chain to become invalid, thus enabling tampering to be quickly detected during data verification.

[0079] In one embodiment of the present invention, the analysis and calculation module initiates a batch processing calculation task based on the established traceability main chain. This task performs multi-dimensional correlation analysis on the standardized traceability dataset to mine key control point data. The analysis and calculation module submits a calculation job to the big data distributed storage cluster. This job is designed to read the complete data link of the traceability main chain. The analysis and calculation module locates the storage space partition corresponding to the traceability code of a specific batch in the big data distributed storage cluster and sequentially reads all data records from the beginning of the breeding record data to the end of the cold chain logistics data. These data records, arranged chronologically and linked by hash pointers, constitute a complete data link that can be traversed by the program. The analysis and calculation module traverses each node in the data link, that is, the standardized data record corresponding to each link, and extracts two types of target information from these data records. One type is process parameter data, such as "pre-cooling room temperature," "bleeding time," and "halving speed" in the slaughtering and cutting stage; the other type is quality inspection data, such as "lean meat powder detection value" in the entry quarantine stage and "total bacterial count on the carcass surface" in the slaughtering and cutting stage.

[0080] See Figure 3In the anomaly analysis of key control points in the slaughtering process traceability management system, this chart visually presents the correlation between multi-dimensional detection indicators and comprehensive risk levels. The chart uses the batch traceability code as the horizontal axis, the left vertical axis as the values ​​of each detection indicator, and the right vertical axis as the comprehensive risk level. Six core detection indicators—body temperature, pH value, bacterial count, temperature and humidity, knife disinfection, and carcass weight—are distinguished by different shapes and colors. Horizontal dashed lines mark the safety threshold lines for each indicator, and red broken lines visually reflect the comprehensive risk level of the corresponding batch. The data distribution shows that batch SC20260301004 has a peak comprehensive risk level of 9, with bacterial count, temperature and humidity, and knife disinfection all significantly deviating from the thresholds, forming a high-risk audit entry after matching with the risk rule base. Batch SC20260301002 has a comprehensive risk level of 5, with its bacterial count exceeding the threshold of 200, making it the main source of risk. The detection indicators of the remaining batches are mostly within the safe threshold range, with comprehensive risk levels maintained at a lower level of 2-3. This diagram enables the visualization and linkage between abnormal parameters of key control points and comprehensive risk levels, providing an intuitive basis for quality and safety analysis for the audit and verification module, and supporting the subsequent generation of enhanced traceability files and blockchain evidence storage.

[0081] In practice, the analysis and calculation module compares the extracted process parameter data with the preset process standard range. The preset process standard range is stored in the form of rules, such as "pre-cooling room temperature standard range: 0 degrees Celsius to 4 degrees Celsius". The analysis and calculation module compares the measured value "5 degrees Celsius" of the "pre-cooling room temperature" field in the data record with the standard range, identifies that the measured value deviates from the standard range, and marks the process parameter data node that generated this measured value as an abnormal parameter node. The analysis and calculation module compares the extracted quality inspection data with the historical normal quality baseline for the same batch. This historical normal quality baseline is a reference range calculated by statistically analyzing the numerical distribution of the same test indicator (such as "total colony count on carcass surface") across multiple past normal production batches. For example, the normal quality baseline for "total colony count on carcass surface: not greater than 1000 CFU / g" is used. The analysis and calculation module compares the measured value of "1500 CFU / g" in the "total colony count on carcass surface" field in the data record with the normal quality baseline, identifying any deviations from the normal baseline and marking the quality inspection data node that produced this measured value as a quality fluctuation node. After identifying all abnormal parameter nodes and quality fluctuation nodes, the analysis and calculation module marks these nodes on the same timeline according to the timestamps of their corresponding data records. It then performs an overlap analysis on the timeline of abnormal parameter nodes and quality fluctuation nodes, searching for intersections of nodes that are adjacent or coincident in time.

[0082] In some embodiments, the analysis and calculation module employs a quantified intersection determination formula to identify key control point data. Overlap analysis on the time axis is achieved by calculating temporal proximity; the intersection determination formula is as follows:

[0083]

[0084] in: This represents the set of identified key control points. Represents an abnormal parameter node. This represents a node of quality fluctuation. Indicates abnormal parameter nodes The timestamp of the corresponding data record Indicates quality fluctuation nodes The timestamp of the corresponding data record It is a preset time window threshold used to define the maximum allowed time interval for "overlapping" or "proximity". When any abnormal parameter node... With any mass fluctuation node The absolute value of the timestamp difference is less than or equal to the time window threshold. At that time, the analysis and calculation module determines that these two nodes have an intersection on the time axis, and then identifies this intersection as the abnormal parameter node. and / or quality fluctuation nodes The data represented is identified as key control point data. This formulaic approach transforms "overlap analysis" from a qualitative concept into a calculable and repeatable quantitative operation.

[0085] In some embodiments, the audit verification module receives critical control point (CCP) data mined by the analysis and calculation module, and uses this CCP data to perform matching and verification within a pre-set risk rule base to generate a quality and safety audit log. The audit verification module first decomposes the CCP data, parsing each CCP into three elements: a specific detection indicator, a measured value, and an occurrence time. For example, a control point might be decomposed into the detection indicator "pre-cooling room temperature," the measured value "5 degrees Celsius," and the occurrence time "2026-02-17 03:15:00." The audit verification module uses the decomposed detection indicator and measured value as joint query conditions to search within the pre-set risk rule base. The risk rule base is a structured database or rule engine that stores safety thresholds and risk level definitions for various detection indicators. For example, for the detection indicator "pre-cooling room temperature," the risk rule base might store "Safety threshold upper limit: 4 degrees Celsius; Risk level: Level 1 warning; Handling suggestion: Check the refrigeration equipment and adjust the temperature to 0-4 degrees Celsius." When the measured value of "5 degrees Celsius" exceeds the upper limit of the safety threshold of "4 degrees Celsius", the audit and verification module retrieves the corresponding risk level "Level 1 Warning" and the handling suggestion "Check the refrigeration equipment and adjust the temperature to 0-4 degrees Celsius" from the risk rule base.

[0086] In practice, the audit verification module combines five information items—occurrence time, detection index, measured value, risk level, and handling recommendations—into a structured audit entry. This entry includes fields such as "Timestamp: 2026-02-17 03:15:00," "Indicator: Pre-cooling room temperature," "Value: 5℃," "Level: Level 1 Warning," and "Recommendation: Check refrigeration equipment." The audit verification module scans all key control point data belonging to the same batch of traceability codes, repeating the above process of disassembly, retrieval, acquisition, and combination for each control point to generate multiple audit entries. Finally, the module summarizes and sorts these audit entries according to their chronological order of occurrence, forming a complete, time-based structured document—the quality and safety audit log for that batch of livestock. It can be understood that the generation process of the quality and safety audit log is entirely based on automated matching of data and rules, requiring no manual intervention for risk assessment of each anomaly.

[0087] In one embodiment of the present invention, the document storage module performs a merging operation of the quality and safety audit log and the standardized traceability dataset to generate an enhanced traceability archive. The document storage module receives the standardized traceability dataset from the storage index module and the quality and safety audit log from the audit verification module. The document storage module uses the batch traceability code as the association key. The batch traceability code is a unique identifier that connects all data in the same batch. In the big data distributed storage cluster, the document storage module uses the batch traceability code as a query condition to locate the standardized traceability dataset record and the quality and safety audit log record belonging to the batch traceability code. The document archiving module parses the quality and safety audit logs, reading each audit entry. Each entry includes the occurrence time, detection indicators, measured values, risk level, and handling recommendations. Based on the "occurrence time" and "detection indicator" information in the audit entry, the module searches the standardized traceability dataset for original data records with matching timestamps and related data content. For example, for an audit entry regarding an anomaly in "2026-02-17 03:15:00" and "pre-cooling room temperature," the module will search the slaughtering and segmentation data area of ​​the standardized traceability dataset for records with a timestamp near "2026-02-17 03:15:00" and containing temperature-related fields. After finding a matching original data record, the module appends the entire audit entry as an extended attribute set to that original data record. This appending operation is reflected in the data structure by adding an "audit attribute" field to the original data record object; the value of this field is the complete content of the audit entry.

[0088] In practice, the document archiving module performs secondary encoding on the original data records that have been linked to audit entries. This secondary encoding process reads the original data records and their linked audit entries, and assigns a corresponding quality risk label to each record based on the value of the "risk level" field in the audit entry. For example, a risk level of "Level 1 Warning" corresponds to the label "High Risk," and a risk level of "Attention" corresponds to the label "Low Risk." The encoding process generates a new data object containing all fields from the original data, the newly added "Audit Attribute" field, and the "Quality Risk Label" field. This new data object is called an enhanced data object. The document archiving module iterates through all data records in the standardized traceability dataset that are covered by quality and safety audit logs, generating an enhanced data object for each matching record. For data records not covered by audit logs, i.e., those where no risk was found, the document archiving module generates an enhanced data object with an empty "Audit Attribute" field and a "Quality Risk Label" of "Normal." After the archive storage module obtains all enhanced data objects, it arranges them according to the order of the original data records corresponding to these objects in the traceability main chain, that is, from the breeding archive data, entry quarantine data, slaughter and cutting data to cold chain logistics data. It encapsulates all enhanced data objects into a complete data structure. This data structure, which encapsulates the enhanced information of all links, is the enhanced traceability archive.

[0089] In some embodiments, the document storage module writes the generated enhanced traceability document to the blockchain evidence storage node network. The module first performs serialization processing on the enhanced traceability document, converting this complex data structure into a linear byte stream format suitable for transmission and storage over the network, such as using JSON or Protocol Buffers for serialization encoding. The document storage module then calculates a unique digital digest for the serialized byte stream data packet. The digital digest calculation covers all information within the data packet. The formula is as follows:

[0090]

[0091] in: This represents the final digital digest obtained from the calculation. This represents a standard cryptographic hash function (such as SHA-256). This indicates all data contained within the enhanced source tracing archive data package. An enhanced data object, symbol This indicates a bitwise XOR operation. The archival evidence module performs this operation on each enhanced data object. Calculate the hash value separately Then all The hash values ​​are aggregated through a bitwise XOR operation to obtain a digital digest representing the entire data packet. .

[0092] In practice, the document storage module constructs a new block to be uploaded to the blockchain. The data area of ​​the new block contains the following: a calculated digital digest. The block contains the precise timestamp when it was constructed, the identity of the archival storage module or operator performing the storage operation, and the operation type identifying this operation as "archival storage". The archival storage module broadcasts this newly constructed block to all registered consensus nodes in the blockchain storage node network. The consensus nodes in the network verify the new block according to a predetermined consensus algorithm, and the verification content includes a digital digest. The system checks the accuracy of calculations, the validity of timestamps, and the legitimacy of the operator's identity. Once a new block is verified by a sufficient number of consensus nodes, it is officially recorded in the blockchain ledger, becoming an immutable new block. The archival evidence module monitors the blockchain network. When it confirms that a new block has been successfully written to the blockchain ledger, it retrieves the block height (i.e., the block's position number on the chain) and transaction hash (i.e., the unique identifier of this evidence-gathering transaction within the block) from the blockchain network. The archival evidence module then uses the obtained block height and transaction hash information as metadata to write back to the data structure of the initially generated enhanced traceability archive, completing the binding between the archive and the blockchain evidence information.

[0093] In one embodiment of the present invention, the system receives a query request for a specific batch traceability code from an external querying party. The query request is passed through an application programming interface (API) or a web service interface. The payload of the query request must carry the specific batch traceability code and a query authorization credential for verifying the querying party's identity. The query authorization credential can be a digital signature token or authentication information containing an access key. The system parses the query request, extracts the specific batch traceability code and the query authorization credential from the request payload, and sends the query authorization credential to an identity verification service for verification. Only the querying party that passes the verification is authorized to access the data. If the verification fails, the system terminates the query process and returns an authorization error message. After the query authorization credential is verified, the system uses the specific batch traceability code extracted from the request to perform an index query in the blockchain ledger that has been written into the enhanced traceability archive. The system initiates a query transaction to the blockchain evidence storage node network. The query transaction uses the specific batch traceability code as the key to search for and return the metadata associated with that key in the index of the blockchain ledger, namely the block height and transaction hash recorded in the archive metadata when the enhanced traceability archive for that batch was stored.

[0094] See Figure 4In the blockchain-based evidence storage stage of the slaughtering traceability management system, this graph visually presents the distribution and changing trends of the number of verification nodes for each batch of enhanced traceability archives. The graph uses block height as the horizontal axis and the number of verification nodes as the vertical axis, clearly showing the blockchain evidence storage verification status of five batches of enhanced traceability archives through blue broken lines and dots: block height 1001 corresponds to 15 verification nodes, block height 1002 corresponds to 18 verification nodes, block height 1003 corresponds to 16 verification nodes, block height 1004 corresponds to 20 verification nodes, and block height 1005 corresponds to 14 verification nodes. This data directly reflects the degree of consensus participation of the blockchain evidence storage node network in different batches of enhanced traceability archives: the more verification nodes, the more nodes participate in the consensus storage of that batch of archives, and the stronger the guarantee of the immutability and integrity of the stored data; conversely, the participation rate is relatively low. Among them, the batch of files corresponding to block height 1004 achieved the highest level of consensus verification and showed the best security performance. This result can serve as an important basis for subsequent system optimization of the evidence storage consensus mechanism and adjustment of node participation strategy, and also provides the query party with an intuitive reference for the reliability of evidence storage.

[0095] In practice, the system locates the original enhanced traceability archive from the blockchain evidence storage node network based on the retrieved block height and transaction hash. The system sends a request containing the block height and transaction hash to the node storing the target block in the blockchain network. The blockchain node finds the corresponding block based on the block height and searches for a transaction record with a matching transaction hash among all transactions contained in that block. From the data area of ​​that transaction record, it extracts the initially written, serialized enhanced traceability archive byte stream. After obtaining the enhanced traceability archive byte stream, the system does not use it directly but first performs an integrity check. The system recalculates the digital digest of the received enhanced traceability archive data packet in its current state using the same cryptographic hash function and digital digest aggregation algorithm as the archive evidence storage module. From the found blockchain transaction records, the system reads the digital digest originally written and stored on the blockchain by the archive evidence storage module and compares the recalculated digital digest bit-by-bit with the digital digest recorded on the blockchain. When the two digital digests are completely identical, the system confirms that the enhanced traceability file obtained from the blockchain has not been tampered with in any form since it was stored, and the data integrity is guaranteed. The system then returns the file content to the querying party. If the two digital digests are inconsistent, the system determines that the data has been corrupted or tampered with, terminates the query process, and returns a data corruption alarm.

[0096] In some embodiments, the system parses the confirmed and complete enhanced traceability file content into a visualized flow path diagram for intuitive viewing by the querying party. The system reads the main traceability chain data from the enhanced traceability file, which clearly defines the various links and their sequence from livestock breeding, entry quarantine, slaughtering and processing, product segmentation to cold chain logistics. The system extracts specific information corresponding to each link node, including the occurrence time of the link, the geographical coordinates of the operation, the employee number or name of the person performing the operation, and the quality risk label attached to the link data, such as "normal," "low risk," and "high risk." The system internally organizes this extracted information in the form of a link node information table. Refer to Table 1, which shows an exemplary link node information table structure.

[0097] Table 1: Information Table of Link Nodes

[0098] Stage Name Time of occurrence Geographical location Operators Quality risk label Aquaculture 2026-01-15 Farm A coordinates Zhang XX normal Entry quarantine 2026-02-1608:30 Slaughterhouse gate Li XX normal Slaughter and processing 2026-02-1610:15 Slaughterhouse Line 3 Wang XX High risk cold chain logistics 2026-02-1612:00 Logistics Center Zhao XX Low risk

[0099] In its implementation, the system uses data from the process node information table to generate a visualized flow path diagram using a visualization rendering engine. The system defines a two-dimensional coordinate system, with the horizontal axis representing time and the vertical axis representing different operational stages or geographical locations. Each process node in the process node information table is arranged and positioned on the horizontal axis (time axis) of the two-dimensional coordinate system according to its chronological order. These process nodes are then connected sequentially by arrowed lines, with the arrows clearly indicating the path of livestock or products from one stage to the next. Different colors or shapes are used to distinguish different quality risk levels. For example, in the visualized flow path diagram, green circles represent process nodes with a quality risk label of "normal," yellow triangles represent "low-risk" nodes, and red diamonds represent "high-risk" nodes. The system overlays detailed original operational information onto the graphical elements of each stage node. When a user hovers the mouse cursor over or clicks on a stage node, an information box pops up displaying detailed information extracted from the stage node information table, including the time of occurrence, geographical location, operator, quality risk label, and more detailed process data extracted from the enhanced traceability archive. In essence, by parsing the enhanced traceability archive into this visual flow path diagram combining timelines, flow paths, and color coding, complex traceability data is transformed into intuitive graphics, enabling the querying party to quickly grasp the flow process and key risk points throughout the product's entire lifecycle.

[0100] Optionally, when generating a visualized flow path diagram, the system can employ an automatic layout algorithm to determine the precise location of each node in two-dimensional space, thereby reducing visual overlap and intersection. When calculating node positions, the automatic layout algorithm needs to consider the linear arrangement of time sequence and the visual distinction of risk labels. A simplified position calculation function is as follows:

[0101]

[0102] in: Representing the link node In the visualized flow path Figure 2 The final coordinates in the 3D coordinate system Representing the link node The time of occurrence, It is a timestamp Mapped to x-coordinate The monotonically increasing function ensures that nodes with later times are positioned further to the right on the graph. Representing the link node Quality risk labels (which can be quantified into numerical values, such as normal = 0, low risk = 1, high risk = 2). It is a risk label Mapped to ordinate The function is designed to properly separate nodes of different risk levels vertically for easy observation.

[0103] In one embodiment of the present invention, the model adaptation module continuously performs incremental learning on the historical standardized traceability dataset in the big data distributed storage cluster during system operation. The model adaptation module operates independently of the system's batch processing tasks and audit verification module. The model adaptation module has a built-in scheduler that periodically triggers the incremental learning process according to a preset cycle, which can be one day, one week, or one month, depending on the accumulation speed of slaughter production data and analysis needs. When the scheduler triggers the learning process, the model adaptation module initiates a data extraction request to the big data distributed storage cluster. The data extraction request uses a time range as a filtering condition, for example, extracting all newly generated and stored standardized traceability datasets within the time period from "the completion time of the last incremental learning task" to "the trigger time of this task." The model adaptation module extracts this newly added standardized traceability dataset from the big data distributed storage cluster and marks it as incremental training samples. The incremental training samples retain the complete structure of the original data, including data from each stage of the entire lifecycle and globally unique batch traceability codes.

[0104] In its implementation, the model adaptation module analyzes the incremental training samples. The goal of this analysis is to discover novel data features and anomaly patterns reflected in the incremental training samples. The module applies unsupervised learning algorithms for clustering and anomaly detection of the incremental training samples. Unsupervised learning algorithms, such as density-based noise-based spatial clustering or isolated forest algorithms, are used. The model adaptation module first vectorizes the structured fields (such as temperature, weight, and detection index values) in the incremental training samples, converting them into numerical feature vectors suitable for machine learning algorithms. The module then inputs these feature vectors into the unsupervised learning algorithm. The algorithm groups these data points in a high-dimensional feature space, identifying clusters of data points and isolated data points far from any cluster. The module identifies newly formed clusters that have not appeared in historical data cluster centers as novel data features and identifies isolated data points far from all clusters as potential anomaly patterns. For example, records of "new feed additive X" may appear frequently for the first time in the incremental training samples, forming new clusters around the feature vectors related to "feed additive X". This is identified as a new type of data feature. Another example is that several batches of records of "abnormally high slaughter line speed in the early morning" appear in the incremental training samples. These records are isolated in the feature space and are clearly separated from the data points with normal speed distribution. This is identified as an abnormal pattern.

[0105] In practice, the model adaptation module updates the rule definitions in the risk rule base pre-installed in the audit verification module based on the identified new data features and anomaly patterns. The rule definitions in the risk rule base are stored in the form of "condition-action" or "threshold-level". For identified new data features, the model adaptation module adds new rule entries to the risk rule base. For example, when "new feed additive X" is identified as a new data feature, but there is no recorded safety threshold, the model adaptation module can create a new rule in the risk rule base based on the chemical properties of additive X or preliminary correlation analysis. The detection indicator is "residual amount of feed additive X", the initial safety threshold is set to a conservative estimate, the risk level is set to "to be observed", and the rule source is marked as "model adaptation module - new feature". The model adaptation module will set a low initial confidence level for this type of new rule. For identified abnormal patterns, the adaptive model module adjusts the safety thresholds or risk levels of existing rules in the risk rule base. For example, when the adaptive model module identifies a strong statistical correlation between "slaughter cutting speed in the early morning" and "excessive total bacterial count" in multiple batches, it calculates a new, more sensitive speed threshold based on historical and incremental data. The module then updates the rule regarding "cutting speed" in the risk rule base, adjusting the upper limit of the safety threshold from the original V0 to the calculated new threshold V_new, and raising the risk level from "attention" to "warning." The adaptive model module records the basis and confidence level for this update. The update operation of the risk rule base by the adaptive model module is implemented through a version control mechanism. Each update generates a new rule base version and records the update time, content summary, and confidence level.

[0106] In some embodiments, after the model adaptation module updates the risk rule base, it redeploys the updated risk rule base to the batch processing task initiated by the analysis and calculation module. Before each execution of critical control point data mining, the batch processing task of the analysis and calculation module actively loads the latest version of the risk rule base from the storage location specified by the model adaptation module. The model adaptation module pushes the new version of the risk rule base file to this specified shared storage location and notifies the analysis and calculation module that the rule base has been updated. When the next batch processing task starts, the analysis and calculation module loads and uses this updated risk rule base. The updated risk rule base includes rules corresponding to new data features learned from the new data, as well as thresholds and levels adjusted for newly discovered abnormal patterns. When performing matching verification, the audit and verification module uses these new rule definitions to determine whether the measured values ​​exceed the safety threshold and matches new risk levels and handling recommendations. This process enables the system to dynamically adjust its risk identification standards and sensitivity using the latest production data feedback. It is understandable that through the closed-loop process of "extracting incremental data - analyzing features and patterns - updating the rule base - redeploying", the system has the ability to adapt to changes in the slaughtering process or new anomalies, without the need for frequent manual revisions of the rule base.

[0107] See Figure 5 In the model adaptation module of the slaughtering full-process traceability management system, this graph uses a heatmap matrix to intuitively quantify the risk level distribution characteristics of different batches at each stage of the entire life cycle, supporting the dynamic updating and iteration of the risk rule base. The graph constructs a two-dimensional matrix with the batch traceability code as the vertical axis and the traceability stage as the horizontal axis. The color bars on the right quantify and map the risk level from dark green to dark red. As can be seen from the data distribution, batch SC20260301004 appears dark red during the quarantine stage, representing the highest risk value in the entire graph. This corresponds to a novel anomaly pattern learned by the model adaptation module and is the core sample triggering a key update to the risk rule base. Batches SC20260301002 and SC20260301005 appear orange-yellow during the quarantine stage, representing medium risk characteristics. Batches SC20260301001 and SC20260301003 exhibit light green risk levels at most stages, belonging to the low-risk range. This heatmap fully reproduces the risk distribution results after incremental learning by the adaptive model module, clearly demonstrating the risk fluctuation patterns in the four stages of breeding, quarantine, slaughter, and logistics, with the quarantine stage being a highly sensitive area for risk identification. Based on the quantitative data from this heatmap, the adaptive model module can accurately identify new anomaly patterns, revise risk thresholds, and redeploy the updated rule base to batch processing tasks, achieving an adaptive closed-loop system for changes in the slaughtering process.

[0108] Optionally, the model adaptation module can employ a time-decay-based incremental learning weight allocation strategy when analyzing incremental training samples and updating rules. This ensures that rule updates respond to new trends without being overly affected by short-term fluctuations. The strategy assigns learning weights to each incremental data point using a weight calculation function, as follows:

[0109]

[0110] in: Indicates assignment to the first The learning weights of each incremental training sample. It is a basic weighting coefficient used to balance the overall impact of incremental learning and historical learning. This indicates the moment when the adaptive module of the model is currently performing incremental learning. Indicates the first The timestamp of the data record corresponding to each incremental training sample. It is a time decay factor greater than zero, symbol This represents the natural constant. This function implies a timestamp. Distance from current learning time The more recent the data, the higher its weight. The larger the timestamp, the greater its impact on rule updates; The further back in time the data is from the present, the greater its weight. The more exponentially decaying the value, the smaller the impact. The adaptive module of the model, during statistical processes such as cluster analysis and threshold calculation, will adjust the weight of each data point accordingly. Weighted calculations are performed to give greater emphasis to new patterns reflected in recent data.

[0111] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A big data-based traceability management system for the entire slaughtering process, characterized in that, include: The data acquisition module acquires raw data sets for the entire life cycle of each batch of livestock through IoT acquisition terminals deployed at various operational stages of the slaughtering production line. The data processing module performs data cleaning and standardization transformation on the raw data set of the entire life cycle to generate a standardized traceability dataset with unified coding. Each record in the standardized traceability dataset carries a globally unique batch traceability code. The storage index module imports the standardized traceability dataset into a big data distributed storage cluster, and establishes a cross-stage chain index relationship according to the temporal relationship of data generation to form the main traceability chain; The analysis and calculation module, based on the traceability main chain, initiates a batch processing calculation task to perform multi-dimensional correlation analysis on the standardized traceability dataset and mine key control point data in the slaughtering production process; The audit and verification module uses the key control point data to perform matching and verification in a pre-set risk rule base and generates a quality and safety audit log for the corresponding batch of livestock. The document storage module merges the quality and safety audit logs with the corresponding standardized traceability dataset to generate an enhanced traceability document, and writes the enhanced traceability document into the blockchain storage node.

2. The big data-based slaughtering process traceability management system as described in claim 1, characterized in that, The original dataset covering the entire lifecycle is cleaned and standardized to generate a standardized traceability dataset with unified coding, including: The complete lifecycle raw data set includes breeding record data, entry quarantine data, slaughtering and processing data, and cold chain logistics data; The ear tag numbers, immunization records, and feed feeding records in the breeding records are deduplicated and logically verified to remove contradictory or missing records. The animal quarantine certificate number, clinical examination results, and laboratory test reports in the entry quarantine data are standardized in format, and unstructured text is converted into structured fields; The knife disinfection records, carcass weighing data, and by-product classification information in the slaughtering and processing data are normalized to a unit, and the operator's employee number information is added. The temperature and humidity sensor readings in the cold chain logistics data are synchronized with the geographical coordinates in time to generate a transportation environment trajectory with timestamps. For each data record after data cleaning and standardization, a globally unique batch traceability code is generated, which includes the place of origin code, slaughterhouse code, production date and random serial number. All data records with attached batch traceability codes are combined into the standardized traceability dataset.

3. The big data-based slaughtering process traceability management system as described in claim 2, characterized in that, The standardized traceability dataset is imported into a big data distributed storage cluster, and a cross-stage chained index relationship is established according to the temporal relationship of data generation to form the main traceability chain, including: In the big data distributed storage cluster, an independent storage space partition is created for each batch traceability code; The breeding record data, entry quarantine data, slaughtering and processing data and cold chain logistics data belonging to the same batch traceability code are written into the storage space partition in the order of data collection time. During the data writing process, a hash pointer is automatically generated that points to the data record of the previous stage, thus linking the data records of all stages together. When any data record in any link changes, the hash value of the data record is recalculated and the subsequent hash pointers are updated, thereby maintaining the integrity and immutability of the traceability main chain.

4. The big data-based slaughtering process traceability management system as described in claim 3, characterized in that, Based on the aforementioned traceability main chain, a batch processing task is initiated to perform multi-dimensional correlation analysis on the standardized traceability dataset, thereby mining key control point data in the slaughtering production process, including: Read the complete data link of the tracing main chain in the big data distributed storage cluster; Traverse each node in the data link and extract the process parameter data and quality inspection data; The process parameter data is compared with the preset process standard range to identify abnormal parameter nodes that deviate from the standard. The quality inspection data is compared with the normal quality baseline of the same batch in history to identify abnormal parameter nodes that deviate from the standard. The nodes of the abnormal parameters and the nodes of the quality fluctuations are overlaid on the time axis, and the nodes in the intersection are the key control point data.

5. The big data-based slaughtering process traceability management system as described in claim 4, characterized in that, Using the aforementioned key control point data, a matching and verification process is performed in a pre-set risk rule base to generate a quality and safety audit log for the corresponding batch of livestock, including: Each control point in the key control point data is broken down into specific detection indicators, measured values, and occurrence times; The detection indicators and measured values ​​are used as query conditions to search in the risk rule base, which stores the safety thresholds and risk level definitions of various detection indicators. When the measured value exceeds the safety threshold, the corresponding risk level and handling suggestions are obtained from the risk rule base. Combine the occurrence time, detection indicators, measured values, risk level, and handling recommendations to form an audit item; All audit entries under the same batch traceability code are summarized in chronological order to generate the quality and safety audit log.

6. The big data-based slaughtering process traceability management system as described in claim 5, characterized in that, The quality and safety audit logs are merged with the corresponding standardized traceability dataset to generate an enhanced traceability profile, including: Using the batch traceability code as the association key, the standardized traceability dataset and the quality and safety audit log are located in the big data distributed storage cluster, respectively. Each audit entry in the quality and safety audit log is attached as an extended attribute to the corresponding original data record in the standardized traceability dataset; The data records with audit entries are re-encoded to generate enhanced data objects containing quality risk labels; All enhanced data objects are arranged in the order of the main traceability chain and encapsulated into the enhanced traceability file.

7. The big data-based slaughtering process traceability management system as described in claim 6, characterized in that, Writing the enhanced traceability file to the blockchain evidence storage node includes: The enhanced traceability archive is serialized and converted into a byte stream format suitable for network transmission; Calculate a holistic digital digest for the data packet in the byte stream format, the digital digest being an aggregation of the hash values ​​of all enhanced data objects within the data packet; Construct a new block that includes the digital digest, timestamp, operator identification, and operation type; The newly added block is broadcast to all consensus nodes in the blockchain evidence storage node network, and the consensus nodes verify and record the newly added block. Once the newly added block is successfully written into the blockchain ledger, the corresponding block height and transaction hash are written back to the metadata of the enhanced traceability archive.

8. The big data-based slaughtering process traceability management system as described in claim 7, characterized in that, Also includes: When a query request for a specific batch traceability code is received, the enhanced traceability file is retrieved from the blockchain evidence storage node and parsed into a visualized flow path diagram, which is then displayed to the querying party. Specifically, this includes: Parse the query request and extract the specific batch traceability code and query authorization certificate carried therein; Using the specific batch traceability code, query the corresponding block height and transaction hash in the blockchain ledger; Based on the retrieved block height and transaction hash, the original enhanced traceability file is located from the blockchain evidence storage node; The enhanced traceability archive is subjected to integrity verification. This is done by recalculating its digital digest and comparing it with the digital digest recorded in the block. Once it is confirmed that the data has not been tampered with, the archive content is returned.

9. The big data-based slaughtering process traceability management system as described in claim 8, characterized in that, The enhanced traceability archive is parsed into a visualized flow path diagram, including: Read the main traceability chain data in the enhanced traceability file to identify each node in the process from breeding, quarantine, slaughter, processing to logistics; Extract the occurrence time, geographical location, operators, and quality risk tags corresponding to each stage; Arrange the nodes in chronological order in a two-dimensional coordinate system, use connecting lines to indicate the flow direction, and use different colors or shapes to distinguish different quality risk levels; Detailed operational information is overlaid and displayed at each stage node, forming the visualized flow path diagram.

10. The big data-based slaughtering process traceability management system as described in claim 9, characterized in that, Also includes: The model adaptation module is used to continuously perform incremental learning on the historical standardized source dataset in the big data distributed storage cluster during system operation, including: Newly added standardized traceability datasets are periodically extracted from the big data distributed storage cluster as incremental training samples; Analyze the novel data features and anomaly patterns reflected in the incremental training samples, and update the rule definitions in the risk rule base; The updated risk rule base is redeployed into the batch processing task for the next cycle of key control point data mining, enabling the system to adapt to changes in the slaughtering process.

Images