A method and system for human participation process monitoring and traceability

Abstract

The application belongs to the technical field of product traceability management, and discloses a human participation process monitoring and traceability method and system. Sensing collection nodes are arranged for four core processing procedures of cleaning, steaming, drying and processing. Hyperspectral detection is carried out at the entrances and exits of each procedure to obtain saponin component estimation and water content data. Process parameters in the procedure execution process are synchronously collected to form single-procedure quality change records. Based on multi-batch historical processing data, quality influence coefficients between adjacent procedures are calculated and fitted to determine the attenuation law and amplification threshold of the deviation of the previous process to the subsequent process, and a cross-procedure cascading quality mapping model is constructed. In the processing process, the current procedure data is substituted into the model to pre-act the product saponin retention rate and quality grade, and the parameter adjustment scheme is output according to the deviation degree. For batches that cannot be adjusted through subsequent procedures, a pause instruction is automatically triggered and the digital identity is locked.

Description

Technical Field

[0001] This invention belongs to the field of product traceability management technology, specifically a method and system for monitoring and tracing the ginseng processing. Background Technology

[0002] Ginseng, a high-value economic crop with both medicinal and edible properties, undergoes multiple processing steps, including washing, sorting, steaming, drying, and processing. The process parameters of each step directly determine the retention rate and conversion efficiency of active ingredients such as ginsenosides. Currently, digital traceability methods have been gradually introduced into the ginseng processing field, but existing technical solutions still have the following technical problems:

[0003] Existing traceability systems primarily collect procedural information such as process time, operators, and equipment numbers, but fail to collect and analyze core process parameters such as steaming temperature curves, drying gradients, and moisture content changes at a fine-grained level. They also fail to establish a quantitative mapping relationship between parameter changes and the transformation of active ingredients such as ginsenosides Rb1, Rg1, and rare saponins Rg3. Consequently, they cannot reflect the quality of process execution, making it difficult for consumers and regulators to determine the true medicinal value of products through traceability information.

[0004] After steaming, drying and other processes, the appearance of different ginseng varieties tends to be similar. The existing traceability system only relies on the upstream origin label data for continuity and lacks a secondary characteristic verification mechanism at the processing end. Downstream, it is impossible to reverse verify based on the characteristics of the processed product, which makes it easy for the variety to be counterfeited. There is a break in the identity trust in the traceability chain at the processing stage.

[0005] The ginseng processing industry chain involves multiple independent entities such as ginseng farmers, primary processing points, advanced processing plants, and brand owners. Existing blockchain traceability solutions only require each entity to upload data to the blockchain for evidence storage, without establishing a cross-verification mechanism between upstream and downstream entities. It is difficult to identify data tampering by a single entity, and thus cannot truly solve the problem of business trust.

[0006] Existing traceability solutions only have the ability to hold people accountable after the fact. They cannot provide real-time warnings and corrections for deviations in process parameters during the processing. When parameters are abnormal, they cannot intervene immediately. Usually, the quality is not found to be substandard until the entire batch is processed and quality inspection is carried out, resulting in raw material loss, production capacity waste and commercial default risks. Summary of the Invention

[0007] The purpose of this invention is to provide a method and system for monitoring and tracing the ginseng processing process, so as to solve one or more problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring and tracing the ginseng processing process, comprising the following specific steps:

[0009] Furthermore, a unique digital identity is generated for each batch of ginseng. During the fresh ginseng storage process, multi-dimensional feature vectors such as ginseng surface texture, rhizome morphology, and epidermal spectral characteristics are extracted using hyperspectral imaging and microscopic texture acquisition equipment. At the same time, basic attribute data such as place of origin coordinates, cultivation mode, growth years, and harvest time are recorded to generate an initial germplasm feature fingerprint and bind it to the digital identity.

[0010] Based on the characteristic evolution law of different germplasm ginseng processing, a germplasm gradient baseline model is pre-constructed for five core stages: fresh ginseng, after cleaning, after steaming, after drying, and after processing. The characteristic change range and morphological evolution law of each type of germplasm at each processing node are clarified. The index identifier and hash value of the germplasm gradient baseline model and the initial germplasm feature fingerprint are synchronously written into the consortium blockchain identity contract layer.

[0011] Furthermore, sensor acquisition nodes are deployed for the four core processes of cleaning, steaming, drying, and processing. Combined with the initial germplasm characteristic fingerprint, quality data and process parameters are collected in a linked manner. At the entrance of each process, hyperspectral detection is performed to obtain the estimated data of saponin composition and moisture content at the current stage. During the execution of the process, process parameters such as temperature curve, duration, pressure, and ambient humidity are collected in real time. After the process is completed, quality data is detected and output again.

[0012] After the data of a single process is associated and packaged to form a process-level quality change record, the quality influence coefficient of the deviation of the process parameters of the preceding process on the exit quality of the subsequent process is calculated, the quality transmission attenuation and amplification law of the deviation of the process parameters of the preceding process to the subsequent process is established, and a cross-process cascade quality mapping model is formed.

[0013] The cross-process cascaded quality mapping model can correct the benchmark range of process parameters for subsequent processes based on the actual output quality of the preceding processes.

[0014] Furthermore, an automatic cross-validation mechanism for data from multiple upstream and downstream entities is constructed through a consortium blockchain architecture, and the data collected from the process and the quality data generated by the cross-process cascaded quality mapping model are both included in the cross-validation scope.

[0015] After a processing company uploads process data for a certain batch, the system automatically triggers a triple logic check. It matches the export weight and category data of the previous process with the import weight and category data of the current process, and combines this with the preset loss threshold for the corresponding process to complete the consistency verification of business volume values. It compares the range of process parameters and the magnitude of quality changes for that process to verify the consistency between the quality changes and the execution of process parameters with the cross-process cascaded quality mapping model. Finally, it retrieves sampling data of intermediate products from the same batch from a third-party quality inspection node to complete the distribution consistency verification of the process output quality data.

[0016] When at least two of the three verifications pass, the system will push the corresponding batch of data to the consortium blockchain node to complete the consensus on-chain process and generate normal evidence storage certificates; data that fails the verification will be marked as pending confirmation, and the normal on-chain process will not be executed for the time being, and an abnormal alarm will be automatically triggered and pushed to the relevant entities.

[0017] Furthermore, a quality pre-simulation and pre-intervention system is built based on the cross-process cascade quality mapping model. The processing process is controlled by combining the real-time on-chain process data. During the processing, the system collects the process parameters and quality data of the current process in real time, substitutes them into the cross-process cascade quality mapping model, and performs pre-simulation calculations on the saponin retention rate and quality grade of the finished product after the subsequent processes are completed.

[0018] If the simulation results show that the quality indicators of the finished product will be lower than the preset qualified threshold, the system will push the warning information to the on-site operation terminal and output the current process parameter adjustment plan and the parameter benchmark correction suggestions for subsequent processes according to the direction of deviation.

[0019] If the pre-simulation results show that the quality indicators of the finished product are lower than the preset qualified threshold and cannot be restored to the qualified range through subsequent process parameter callbacks, the system will automatically trigger a process pause command and lock the corresponding batch digital identity, prohibiting the current batch of products from flowing into the next process. All pre-simulation records, early warning events and handling operations will be simultaneously uploaded to the blockchain for evidence storage.

[0020] Furthermore, the system extracts indicators such as germplasm grade, process compliance, active ingredient retention rate, pre-process intervention records, and quality stability, and generates finished product quality value scores and corresponding grades through a weighted scoring model;

[0021] The hyperspectral feature data of the finished product is retrieved and matched with the feature interval of the post-processing stage in the germplasm gradient baseline model to confirm the consistency of germplasm identity. The verification basis and process data are synchronously associated with the digital identity of the corresponding batch, and the quality value level, identity verification result and finished product feature data are synchronously bound to the digital identity.

[0022] Furthermore, a hierarchical traceability query system is built for terminal application scenarios, and hierarchical traceability information display permissions are configured for different user groups. All displayed data is associated with the corresponding batch's on-chain digital identity. After scanning the code, ordinary consumers can view basic traceability information such as product origin, batch, compliance conclusion of core process, finished product quality grade, and quality inspection report, and obtain the germplasm identity consistency verification conclusion.

[0023] Targeting professional channel partners and regulators, it supports authorized viewing of in-depth data such as process-level quality transfer details, process parameter execution records, and pre-process intervention files. All displayed data is directly retrieved from the consortium blockchain.

[0024] Furthermore, based on the quality and process data accumulated on the blockchain, supply chain credit linkage and business collaborative scheduling are carried out. The credit assessment is based on traceable business data on the blockchain. The system counts indicators such as the timeliness of data uploads by each entity, the pass rate of triple logic verification, batch quality stability, and compliance of abnormal intervention and handling, and automatically generates a comprehensive credit score for processing enterprises and suppliers.

[0025] The comprehensive credit score is mapped to the preset transaction premium coefficient, order priority, and cooperation trust level within the alliance. For suppliers whose batch quality stability indicators meet the preset high-level threshold, the corresponding preset transaction premium coefficient is applied; for entities whose triple logic verification pass rate meets the preset low-level threshold, their order priority is lowered accordingly.

[0026] Simultaneously, through full-chain data aggregation and analysis, the distribution of process duration, parameter fluctuation range, and quality deviation characteristics of each processing link are statistically analyzed, and process parameter adjustment references and supply and demand matching prediction data are output to the main players in the industry chain.

[0027] This invention also provides a ginseng processing monitoring and traceability system for implementing the above-mentioned method, comprising the following modules:

[0028] The identity baseline management module generates a unique digital identity for each batch of ginseng, collects optical characteristics and basic attribute data of fresh ginseng, generates germplasm feature fingerprints, retrieves a pre-built germplasm gradient baseline model, completes the binding of batch feature fingerprints with the baseline, and writes the digital identity, initial germplasm feature fingerprint, index identifier and hash value of the germplasm gradient baseline model into the consortium blockchain identity contract layer.

[0029] The process cascade mapping module connects to the sensor nodes of each process, collects inlet quality data, process parameters and outlet quality data, generates process-level quality change records, calculates the quality influence coefficient between processes, and constructs a cross-process cascade quality mapping model.

[0030] The cross-validation and evidence storage module receives batch process data uploaded by processing enterprises, automatically performs triple verification of business volume value, quality logic, and third-party sampling data, pushes data that meets the verification conditions to the consortium blockchain node, completes consensus and on-chain, and generates normal evidence storage certificates, and marks data that fails the verification as pending confirmation and temporarily does not execute the normal on-chain process.

[0031] The quality pre-simulation intervention module collects process data and substitutes it into the cross-process cascaded quality mapping model to calculate the finished product quality pre-simulation results. It outputs parameter adjustment schemes according to the degree of deviation, triggers a pause command for a specified batch and locks the digital identity, and synchronizes relevant records to the blockchain for evidence storage.

[0032] The finished product grading and verification module extracts data from the entire chain to generate a quality value score and corresponding grade for the finished product, retrieves the characteristic data of the finished product to complete the germplasm identity matching and verification, and binds the results to the corresponding digital identity.

[0033] The hierarchical traceability query module allows for the configuration of corresponding on-chain information display dimensions based on the subject's permissions, supporting product verification and identity verification.

[0034] The credit coordination and scheduling module calculates the credit score of the entity based on on-chain data and matches it with business rules, outputting reference data for adjusting process parameters and supply and demand matching prediction data.

[0035] The beneficial effects of this invention are as follows:

[0036] 1. This invention deploys sensor acquisition nodes for four core processing steps: cleaning, steaming, drying, and processing. Hyperspectral detection is performed at the inlet and outlet of each step to obtain estimated saponin composition and moisture content data. Simultaneously, process parameters such as temperature curves, duration, pressure, and ambient humidity are collected during the process, forming a single-step quality change record. Based on historical processing data from multiple batches, the quality influence coefficient between adjacent steps is calculated, clarifying the attenuation law and amplification threshold of deviations from previous processes transmitted to subsequent processes, and constructing a cross-process cascaded quality mapping model. During processing, the current process data is substituted into the model in real time to pre-simulate the finished product saponin retention rate and quality grade. Parameter adjustment schemes are output according to the degree of deviation. For batches that cannot be corrected through subsequent processes, a pause command is automatically triggered and their digital identity is locked. All pre-simulation and handling records are simultaneously uploaded to the blockchain for evidence storage. This realizes the transformation of processing quality control from post-process inspection to process pre-control, enabling precise identification and prevention of quality risks during the process, reducing the output rate of unqualified batches, reducing raw material loss and capacity waste, and improving the quality stability and process control accuracy of ginseng processing.

[0037] 2. This invention employs a consortium blockchain architecture to establish an automatic cross-verification and evidence storage mechanism for multi-entity data across upstream and downstream processes. It incorporates all on-site data collected during processing and quality data generated by a cross-process cascaded quality mapping model into the verification scope. After processing enterprises upload batch process data, the system automatically triggers a triple logical verification: first, it matches the weight and category data of the previous process's exit with the current process's entry, combining this with the corresponding process's preset loss threshold to verify the consistency of business volume; second, it compares the range of process parameters with the magnitude of quality changes to verify their fit with the quality mapping model's patterns; and third, it retrieves the same batch's sampling data from a third-party quality inspection node to verify the consistency of quality distribution. Only data that passes at least two of these checks can be consensus-uploaded to the blockchain and generate evidence storage certificates. Simultaneously, a germplasm gradient baseline model is pre-constructed, and at the finished product stage, hyperspectral feature matching verifies the consistency of germplasm identity, with the result bound to a unique batch digital identity. This eliminates trust gaps in germplasm identity during processing, enhancing the traceability credibility and anti-counterfeiting capabilities of ginseng products.

[0038] 3. This invention constructs a hierarchical traceability query and industry collaboration system with differentiated on-chain information display dimensions for different participating entities. Ordinary consumers can scan a code to view basic traceability information such as product origin, batch, compliance conclusions of core processes, finished product quality grade, and germplasm verification conclusions. Professional distributors and regulatory authorities, after authorization, can access in-depth data such as process-level quality transfer details, process parameter execution records, and pre-intervention archives. All displayed data is directly retrieved from consortium blockchain nodes. Simultaneously, based on the full-chain data accumulated on the blockchain, a comprehensive credit score is generated by statistically analyzing indicators such as the timeliness of data uploads, triple verification pass rate, and batch quality stability of each entity. This score is matched with preset transaction premiums, order priorities, and other business rules, and outputs process parameter optimization references and supply and demand matching prediction data. Through the credit linkage mechanism, the system compels each entity to standardize its operations and provides data support for the optimal allocation of resources in the industry chain, thereby improving the overall collaborative efficiency and commercial value of the ginseng industry. Attached Figure Description

[0039] Figure 1 This is the overall flowchart of the entire process monitoring and traceability of the inventor's participation in the project;

[0040] Figure 2 This is a flowchart of the triple cross-validation evidence storage sub-process of the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0042] This embodiment uses 5-year-old cultivated ginseng from Fusong, Jilin Province, batch number RS-FS-20250912-007 as the subject. After harvesting, this batch of fresh ginseng was sent to a ginseng processing enterprise for deep processing, including cleaning, steaming, drying, and preparation. (The last sentence appears to be incomplete and possibly refers to a separate, unrelated point.) Figures 1 to 2 The implementation process of this plan will be fully explained.

[0043] During the fresh ginseng warehousing and acceptance process, the system generates a unique digital identity for batch RS-FS-20250912-007. Using a hyperspectral imager and microscopic texture acquisition equipment at a fixed workstation, the system extracts multi-dimensional feature vectors of this batch of ginseng, such as surface texture roughness, rhizome diameter-to-thickness ratio, and epidermal spectral reflectance in the 420-980nm band. Simultaneously, basic attribute data such as origin coordinates, forest-based field cultivation mode, 5-year growth period, and harvest time are entered to generate the initial germplasm feature fingerprint of this batch and bind it to the digital identity.

[0044] The Z-score method is used to normalize and standardize the extracted multi-dimensional feature vectors. A fixed-length germplasm feature fingerprint string is generated through a feature encoding algorithm. At the same time, a key-value correspondence is established between basic attribute data such as place of origin and growth years and the feature fingerprint.

[0045] Using the digital identity of batch RS-FS-20250912-007 as the primary key, the feature fingerprint code and basic attribute data hash value of this batch are bound together. The binding relationship is synchronously written into the local batch management database and the identity contract layer of the ginseng industry alliance chain to ensure that the digital identity of this batch corresponds to a unique and tamper-proof germplasm feature fingerprint.

[0046] The system pre-collects hyperspectral, morphological, and component data of three types of ginseng—cultivated ginseng, forest-grown ginseng, and American ginseng—at five stages: fresh ginseng, after washing, after steaming, after drying, and after processing. Through statistical analysis, it clarifies the characteristic change range and morphological evolution law of each type of ginseng at each processing stage and pre-constructs a germplasm gradient baseline model. This batch of 5-year-old Fusong cultivated ginseng corresponds to the 5-year-old cultivated ginseng sequence in the matching baseline. The index identifier and hash value of the germplasm gradient baseline model, along with the initial germplasm characteristic fingerprint of this batch, are simultaneously written into the consortium blockchain identity contract layer.

[0047] The consortium blockchain's identity contract layer uses a key-value pair structure to store batch data. The batch digital identity (RS-FS-20250912-007) is used as the primary key, and it is associated with three core data types: the initial germplasm characteristic fingerprint hash value, the 5-year-old ginseng germplasm gradient baseline index, and the basic attribute data hash value. The write operation of this batch's identity data requires digital signature confirmation from both the supplier in Fusong and the deep-processing enterprise before the smart contract can be executed. Upon completion of the write operation, a unique identity certificate number for this batch is generated.

[0048] The subsequent data for each process is written to the blockchain in a batch incremental writing mode. The data that passes the verification of each process is packaged and uploaded to the blockchain in the form of a single transaction. The evidence for a single process includes five items: process number, process parameter hash value, quality data hash value, triple verification result, and digital signature of the operating entity.

[0049] Each on-chain transaction generates a corresponding certificate of authenticity. The certificate contains four core pieces of information: batch digital identity, on-chain timestamp, block height, and certificate hash value. Downstream entities can directly complete the certificate of authenticity verification on the consortium blockchain node using the batch digital identity.

[0050] The germplasm gradient baseline model was pre-constructed according to fixed steps. The corresponding baseline interval was directly called for this batch of 5-year-old garden ginseng. Three types of target germplasm were selected: garden ginseng, forest ginseng, and American ginseng. Each type of germplasm was grouped according to different growth years. Sufficient independent samples were collected from each group. The standard process of cleaning, steaming, drying, and processing was completed in sequence. Hyperspectral data, microscopic morphological data, and measured data of saponin components were collected simultaneously at five nodes: fresh ginseng, after cleaning, after steaming, after drying, and after processing.

[0051] Reflectance values ​​of multiple characteristic wavelengths within a specified band were extracted from hyperspectral data to form spectral feature vectors. Three quantitative features—rhizome diameter-to-thickness ratio, epidermal texture entropy, and cross-sectional porosity—were extracted from microscopic morphological data. The mass percentages of total saponins and ginsenosides Rb1, Rg1, and Rg3 were extracted from the component data. Grouped by germplasm and growth age, statistical fitting was performed on various features at each processing node to obtain the feature mean and 95% confidence interval, which served as the feature baseline for the corresponding processing node of that germplasm.

[0052] All germplasm five-stage baseline data are stored in the form of a two-dimensional array. The array index corresponds to germplasm type, growth years, processing node and characteristic dimension. After inputting the corresponding germplasm attributes for this batch, the corresponding characteristic baseline interval of the five stages can be automatically matched.

[0053] For the four core processes of cleaning, steaming, drying and processing in this batch, sensor acquisition nodes are deployed at each workstation. Combined with the initial germplasm characteristic fingerprint of this batch, quality data and process parameters are collected in a linked manner. At the entrance of each process, high-spectral rapid detection is first performed to obtain the estimated total saponin content and moisture content at the current stage.

[0054] During the process, process parameters such as temperature curve, runtime, equipment pressure, and ambient humidity are collected in real time. After the process is completed, an outlet hyperspectral test is performed again to obtain output quality data.

[0055] The sensing nodes for each process are divided into three categories: entry detection unit, process acquisition unit, and exit detection unit. In this batch, the entry and exit detection units of each process use fixed-station visible and near-infrared hyperspectral imaging equipment, which, together with the flat feeding of the conveyor line, realizes the scanning of each piece of ginseng in the whole batch, and the field of view of the acquisition covers the entire surface area of ​​the whole batch of 500kg ginseng in the flat state.

[0056] The process acquisition unit is equipped with corresponding sensors according to the process type. The cleaning process is equipped with water pressure, flow rate and water temperature sensors; the steaming process is equipped with temperature, pressure and timing modules; the drying process is equipped with integrated temperature and humidity sensor, wind speed sensor and weighing module; and the processing process is equipped with stir-frying temperature sensor, stir-frying rate sensor and timing module.

[0057] All process parameters are continuously collected and stored at 1-second intervals. Inlet and outlet quality inspection adopts a combination of full-batch scanning and sampling inspection. Full-batch scanning obtains the overall average data of the batch, and single samples are simultaneously sampled proportionally to test the content of monomeric saponins. Before starting data collection for each process of this batch, the digital identity RS-FS-20250912-007 is read first, and the corresponding initial germplasm characteristic fingerprint is automatically retrieved as a benchmark reference. All collected data are automatically associated with the digital identity of this batch to complete labeling and packaging.

[0058] The entry, process, and exit data of a single process are linked and packaged to form a process-level quality change record for this process. With total saponin retention rate and moisture content as the core quality indicators, the partial least squares regression method is used to fit and calculate the quality impact coefficient of the deviation of the process parameters of the preceding process on the exit quality of the subsequent process by using historical processing data of 300 batches of the same germplasm over the past two years. The attenuation coefficient and amplification threshold of the deviation of the preceding process to the subsequent process are clarified, the quality transfer law is established, and a cross-process cascade quality mapping model is formed.

[0059] The cross-process cascaded quality mapping model is a chain multivariate regression model. For the process flow of cleaning, steaming, drying and processing in this batch, the model input and output settings follow a fixed logic. The single-level input of the model contains three types of data: the actual exit quality value of the previous process, the real-time process parameter vector of the previous process, and the baseline process parameters of the subsequent process to be predicted.

[0060] The quality impact coefficient was obtained by fitting historical batch data. Using continuous batch data of four processes as samples, the total saponin retention rate at the exit of each process was the dependent variable, and the deviation of the process parameters of the preceding process was the independent variable. The quality impact coefficient corresponding to each process parameter was obtained by partial least squares regression.

[0061] Cross-process cascade quality prediction formula:

[0062]

[0063] This represents the predicted export quality value for the (i+1)th process, with the core quantitative indicator being the total saponin retention rate.

[0064] This represents the baseline exit quality value of the (i+1)th process, corresponding to the standard quality value of the germplasm in the gradient baseline model.

[0065] This represents the total number of core process parameters for the i-th process step;

[0066] This represents the quality influence coefficient corresponding to the k-th process parameter in the i-th process, characterizing the impact of each unit deviation of this parameter on the outlet quality of the subsequent process; for example, in this embodiment, the quality influence coefficient of the peak steaming temperature is 0.82% / ℃.

[0067] This represents the actual deviation of the k-th process parameter in the i-th process, i.e., the difference between the actual operating value of the parameter and the reference parameter value.

[0068] Among the parameters, a 1°C deviation in the peak steaming temperature corresponds to a 0.82% deviation in the saponin retention rate at the drying stage, exhibiting the highest influence coefficient. A 1°C / h deviation in the drying heating rate corresponds to a 0.57% deviation in the saponin retention rate at the processing stage, with the second highest influence coefficient. The model output includes two types of data: the expected outlet quality value for subsequent processes and the corrected baseline range of process parameters for subsequent processes.

[0069] The cascaded transfer logic is to calculate iteratively according to the process flow. The actual output quality of each process is used as the input benchmark for the next process. If the quality of the previous process deviates from the baseline, the process parameter range of the next process is corrected in reverse through the quality influence coefficient to achieve dynamic compensation for quality deviation.

[0070] By constructing an automatic cross-verification and evidence storage mechanism for data from multiple upstream and downstream entities through the ginseng industry alliance chain architecture, the on-site data collected at each process stage of this batch and the predicted quality data generated by the cross-process cascaded quality mapping model are all included in the cross-verification scope. When the processing enterprise uploads the cleaning process data of batch RS-FS-20250912-007, the system automatically triggers triple logic verification.

[0071] The first item matches the outbound weight and category data of the Fusong supplier corresponding to the previous process (fresh ginseng warehousing) with the inbound weight and category data of the cleaning process, and combines the preset loss threshold of the cleaning process to complete the consistency verification of business volume values.

[0072] The second item compares the range of process parameters and the magnitude of quality changes in the cleaning process to verify the degree of fit between the quality change and the execution of process parameters and the pattern of the cross-process cascade quality mapping model.

[0073] The third step involves retrieving sampling data of intermediate products from the same batch after cleaning from third-party quality inspection nodes to verify the consistency of the distribution of process output quality data.

[0074] The consortium blockchain sets up four types of consensus participating nodes: production supply nodes, processing enterprise nodes, third-party quality inspection nodes, and regulatory nodes. The data corresponding to the triple verification is retrieved from the on-chain evidence stored by the corresponding nodes. The business volume value verification data is taken from the outgoing inventory certificate of the production supply node and the incoming inventory certificate of the processing node. The quality logic verification calls the cross-process cascaded quality mapping smart contract deployed on the chain for automatic calculation. The third-party sampling inspection data is taken from the sampling inspection evidence of the same batch uploaded by the third-party quality inspection node. All verification processes are automatically executed in the on-chain contract, and the entire verification process is traceable.

[0075] The judgment rule for the consistency verification of business volume values ​​is that the difference between the weight at the entry of the current process and the weight at the exit of the previous process is deemed to be passed if the percentage does not exceed the preset loss threshold of the corresponding process.

[0076] The loss thresholds for the cleaning process, steaming process, drying process, and processing process were set at 3%, 8%, 40%, and 2%, respectively. Each threshold corresponds to the moisture content variation range of the 5-year-old cultivated ginseng germplasm gradient baseline.

[0077] The judgment rule for the quality logic fit verification is set as follows: when the deviation between the measured quality change and the predicted value of the cross-process cascade quality mapping model does not exceed ±5%, it is judged as passing. The judgment rule for the third-party sampling consistency verification is set as follows: when the deviation between the average saponin content and moisture content of the third-party sampling sample and the average batch uploaded by the enterprise does not exceed ±7%, it is judged as passing.

[0078] When a single item fails the inspection in this batch, an abnormality detail sheet is automatically generated, indicating the deviation item and deviation range, and pushed to the quality control personnel of the processing enterprise for manual review.

[0079] The data for this batch of cleaning processes has been verified, and all three verifications—business volume value verification, quality logic verification, and third-party sampling verification—have passed. The system pushes the data for this process to the consortium blockchain node to complete consensus and upload it to the blockchain, and generates corresponding evidence certificates.

[0080] If fewer than two items pass the verification, the corresponding data will be marked as pending confirmation, and an anomaly alarm will be automatically triggered and pushed to the processing enterprise and the supplier. For batch data marked as pending confirmation, the corresponding processing entity shall submit an anomaly explanation and supplementary supporting data within 24 hours, and the system will automatically trigger a second verification.

[0081] Data that passes the second verification is pushed to the consortium blockchain to complete consensus and be uploaded to the chain normally. Data that fails the second verification is permanently marked as abnormal and questionable. After being marked as questionable, the data in this process is stored in the on-chain abnormal data area, clearly distinguishing it from normal evidence data. This process does not block the normal on-chain flow of data in subsequent processes of the batch, and the abnormal record is simultaneously included in the credit scoring statistics of the corresponding entity.

[0082] A quality prediction and pre-intervention system is built based on a cross-process cascaded quality mapping model, and dynamic control of the processing is achieved by combining real-time on-chain process data. During the execution of the steaming process in this batch, the system collects process parameters such as current steaming temperature and duration, as well as inlet quality data in real time, and substitutes them into the cross-process cascaded quality mapping model to perform pre-calculation of the total saponin retention rate and quality grade of the finished product after the subsequent drying and processing processes.

[0083] The execution logic of this batch of pre-simulation calculations is as follows: first, extract the inlet quality data and real-time process parameters collected in the steaming process as the current input of the cross-process cascaded quality mapping model. For the subsequent unexecuted drying and processing processes, the benchmark process parameters corresponding to 5-year-old cultivated ginseng in the cross-process cascaded quality mapping model are called as the initial input values ​​by default.

[0084] The model is called sequentially according to the process from steaming to drying and then to processing, calculating step by step until the estimated total saponin retention rate and 95% confidence fluctuation range of the finished product at the processing stage are output. The estimated saponin retention rate is then substituted into the weighted scoring model for finished product quality to obtain the preliminary quality grade.

[0085] The threshold for the retention rate of total saponins in the finished product is preset to be ≥75%. If the pre-tested value is lower than this threshold, the parameter adjustment range of the current steaming process and subsequent drying and processing processes is calculated by reverse iterative derivation with the finished product qualification line as the target, and the parameter adjustment scheme that can be directly executed is output.

[0086] The parameter adjustment range is calculated based on the quality influence coefficients of each process parameter in the cross-process cascade quality mapping model. First, the difference between the estimated total saponin retention rate of the finished product and the 75% qualified threshold is calculated. Then, according to the process cascade transmission relationship, the quality compensation amount is decomposed to each relevant process. Finally, combined with the quality influence coefficient of the corresponding process parameter, the specific parameter adjustment value is calculated.

[0087] For example, the estimated total saponin retention rate of this batch of finished products is 72%, with a difference of 3%. Combining the cascade transmission relationship of steaming, drying, and processing, the quality compensation amount is decomposed in reverse. This corresponds to an increase of 2.5℃ in the peak temperature of the current steaming process. At the same time, a baseline correction suggestion is given to reduce the heating rate of the subsequent drying process by 1.5℃ / h. After the compensation is added, the saponin retention rate of the finished product can be restored to above 75%.

[0088] The adjustment direction follows the principle of quality compensation. When the saponin retention rate is low, the corresponding positive compensation parameters such as increasing the peak steaming temperature and extending the drying low temperature section are adjusted. At the same time, the upper limit of each process parameter is preset, with the peak steaming temperature not exceeding 102℃ and the maximum drying temperature not exceeding 60℃, to ensure that the adjusted parameters do not exceed the safe process range corresponding to 5-year-old garden ginseng.

[0089] Pre-test deviations are categorized into two levels based on the difference between the estimated total saponin retention rate of the finished product and the 75% acceptable threshold. Minor deviations, corresponding to a difference within 2%, are handled by the system which only pushes parameter adjustment plans for on-site personnel to confirm and implement.

[0090] Severe deviations correspond to differences exceeding 2% but still have room for correction. In addition to pushing parameter adjustment plans for the current process, the system will simultaneously provide suggestions for correcting the process parameters of subsequent processes and push strong early warning information to the on-site management terminal. The process can only continue after the management personnel confirm the adjustment plan.

[0091] If the pre-test results show that the total saponin retention rate of this batch of finished products is lower than the 75% qualified threshold, and cannot be restored to the qualified range through subsequent drying and processing parameters, the system will automatically trigger a pause command for the steaming process and lock the digital identity of batch RS-FS-20250912-007, prohibiting this batch of products from flowing into the next drying process. All pre-test records, early warning events and handling operations will be simultaneously uploaded to the blockchain for evidence storage.

[0092] Digital identity locking is achieved by calling the state change interface of the consortium blockchain identity contract. After locking, the process flow permission of batch RS-FS-20250912-007 is temporarily closed, prohibiting the batch of products from entering the drying process. The process data and intervention records of the steaming process can still be stored on the blockchain normally, but the submission of entry data applications for the drying process is not supported at the moment.

[0093] The locking operation requires recording the triggering reason, triggering time, and operation subject information and uploading them to the blockchain simultaneously. After the parameters are adjusted and the steaming output quality is retested and meets the standards, the processing enterprise submits an unlocking application. After the quality control department reviews and signs the application, the contract interface is called to unlock the lock and restore the data writing and process flow permissions for this batch.

[0094] After the processing steps of this batch are completed, the system extracts five indicators: germplasm grade, process compliance, active ingredient retention rate, pre-process intervention record, and quality stability. These indicators are then weighted according to preset weights to generate a quality value score for batch RS-FS-20250912-007, and the corresponding quality grades are determined based on preset score ranges.

[0095] A weighted scoring model based on a percentage system is adopted. The weights and scoring rules of each indicator are set in accordance with the ginseng quality evaluation scenario. The germplasm grade has a weight of 25%, which is divided into 4 levels according to germplasm type and growth years, with a corresponding score range of 60 to 100 points. The process compliance has a weight of 20%, which is calculated based on the percentage of time that the process parameters of each process fall within the benchmark range. A 100% percentage yields full marks, and 2 points are deducted for every 5% below the benchmark. The active ingredient retention rate has a weight of 30%, with the total saponin retention rate of the finished product as the core indicator. A retention rate of ≥85% yields full marks, and 2 points are deducted for every percentage point below the benchmark.

[0096] The weighting of pre-process intervention records is 10%, with full marks for no intervention records. Each triggered warning deducts 5 points, and triggering a process suspension deducts 10 points. The weighting of quality stability is 15%, calculated based on the coefficient of variation of ginsenoside content in a single ginseng root from the same batch. A coefficient of variation ≤ 5% earns full marks, and 3 points are deducted for each percentage point increase. This batch of 5-year-old cultivated ginseng germplasm grade scored 85 points, process compliance scored 92 points, active ingredient retention rate scored 82 points, no pre-process intervention records scored 100 points, and quality stability scored 90 points. The weighted average score of the finished product quality is 87.75 points, corresponding to a second-level quality grade.

[0097] Finished Product Quality Value Weighted Scoring Formula:

[0098]

[0099] The overall score for the quality value of the finished product is expressed on a 100-point scale.

[0100] This represents the weight of the germplasm grade index, with a fixed value of 0.25;

[0101] The score represents the individual score of germplasm grade, which is divided into 4 levels according to germplasm type and growth period, with a score range of 60-100 points;

[0102] This indicates the weight of the process compliance indicator, with a fixed value of 0.20;

[0103] The score for process compliance is calculated based on the percentage of time that the process parameters of each process fall within the benchmark range. A 100% percentage earns full marks, and 2 points are deducted for every 5% below the benchmark.

[0104] This indicates the weight of the active ingredient retention rate indicator, with a fixed value of 0.30;

[0105] The score represents the retention rate of active ingredients, with the total saponin retention rate of the finished product as the core indicator. A retention rate of ≥85% earns full marks, and 2 points are deducted for each percentage point decrease.

[0106] This indicates the weight of the pre-intervention record indicators, with a fixed value of 0.10;

[0107] The score represents the individual score for the pre-process intervention record. Full marks are awarded for no intervention record. 5 points are deducted for each warning triggered and 10 points are deducted for each process suspension triggered.

[0108] This represents the weight of the quality stability index, with a fixed value of 0.15.

[0109] The score for quality stability is calculated based on the coefficient of variation of ginsenoside content in a single ginsenoside from the same batch. A full score is awarded if the coefficient of variation is ≤5%, and 3 points are deducted for each percentage point increase.

[0110] The hyperspectral characteristic data of this batch of finished products were retrieved and matched with the characteristic intervals of the post-processing stage in the 5-year-old garden ginseng germplasm gradient baseline model to confirm the consistency of germplasm identity.

[0111] The verification criteria and process data are synchronously linked to the batch digital identity of RS-FS-20250912-007, and the finished product quality grade, identity verification results and finished product feature data are synchronously bound to the digital identity.

[0112] The matching verification process involves sequentially comparing spectral features and morphological features. The spectral feature comparison extracts the reflectance values ​​of 20 characteristic wavelengths of the finished product and compares them one by one with the corresponding intervals of the baseline. If the proportion of characteristic wavelengths falling within the interval is not less than 90%, the spectral matching is considered successful.

[0113] The morphological characteristics are compared and verified. If both indicators, such as the diameter ratio of the rhizome and the entropy value of the epidermal texture, fall within the corresponding range of the baseline, the morphological match is deemed to be successful. If both comparisons are successful, the germplasm identity is finally confirmed to be consistent. If either one fails, the germplasm is marked as questionable and a manual review process is triggered.

[0114] Germplasm identity matching verification formula:

[0115]

[0116] Germplasm identity consistent

[0117] This indicates the success rate of matching the spectral characteristics of the finished product with the germplasm gradient baseline;

[0118] This indicates the number of finished product characteristic wavelengths whose reflectance values ​​fall within the interval corresponding to the germplasm gradient baseline.

[0119] This represents the total number of characteristic wavelengths participating in the comparison; in this embodiment, the value is fixed at 20.

[0120] The symbol represents logical equivalence, indicating that when all conditions on the right are met, the germplasm identity is ultimately determined to be consistent.

[0121] The logical AND symbol represents the requirement that multiple conditions must be met simultaneously.

[0122] This represents the measured value of the diameter ratio of the finished rhizome head;

[0123] This represents the standard confidence interval for the diameter-to-thickness ratio of the rhizome in the post-processing stage within the germplasm gradient baseline model.

[0124] This represents the measured value of the entropy of the finished product's surface texture.

[0125] This represents the standard confidence interval for the epidermal texture entropy value in the post-processing stage within the germplasm gradient baseline model.

[0126] A hierarchical traceability query system is built for terminal application scenarios. Hierarchical traceability information display permissions are configured for different user groups. All displayed data is associated with the corresponding batch of on-chain digital identity. Permission verification is achieved through the linkage of consortium blockchain identity contract and permission contract. After ordinary consumers scan the traceability code of this batch of products, they call the permission contract to verify the visitor's identity, automatically match public query permissions, and only return traceability fields marked as public in the contract.

[0127] After professional channel providers and regulators complete identity authentication through digital certificates, the permission contract matches the corresponding authorization level and returns the original on-chain data within the corresponding range for this batch. All query requests and returned results generate query evidence records and are synchronously uploaded to the blockchain.

[0128] After scanning the code, ordinary consumers can view basic traceability information for batch RS-FS-20250912-007, including the product's origin, batch number, compliance conclusions of the four core processes, the finished product's secondary quality grade, and third-party quality inspection reports. They can also obtain a conclusion confirming the consistency of germplasm identity. For professional distributors and regulators, authorized access is available to view in-depth data such as process-level quality transfer details, process parameter execution records, and pre-process intervention archives. All displayed data is directly retrieved from the consortium blockchain.

[0129] Based on the quality and process data accumulated on the blockchain, the system links supply chain credit and coordinates business operations. The credit assessment is based on traceable business data on the blockchain. The system counts indicators such as the timeliness of data uploads by each entity, the pass rate of triple logic verification, batch quality stability, and compliance of abnormal intervention and handling, and automatically generates a comprehensive credit score for processing enterprises and suppliers.

[0130] Credit scoring uses a 100-point scale, with four indicators weighted equally. Data upload timeliness is scored based on the percentage of batches that are uploaded within a 2-hour window after the completion of the process. A 100% percentage earns full marks, and a decrease in the percentage results in a corresponding deduction.

[0131] The triple verification pass rate is scored based on the proportion of batches that pass verification out of the total uploaded batches. A pass rate of 100% earns full marks, and a decrease in the pass rate results in a corresponding deduction of marks. Batch quality stability is scored based on the proportion of qualified batches out of the total processed batches. A qualified rate of 100% earns full marks, and a decrease in the qualified rate results in a corresponding deduction of marks.

[0132] The compliance of abnormal intervention and handling is scored based on the percentage of abnormal batches that have completed intervention and handling in accordance with the regulations. A 100% compliance rate earns full marks. Each instance of non-compliance in handling results incurs a deduction of 10 points. The scoring results are updated monthly and simultaneously recorded on the blockchain for evidence storage.

[0133] The comprehensive credit score is mapped to the transaction premium coefficient, order priority, and cooperation trust level preset within the alliance. For high-level suppliers with a batch quality pass rate of ≥98%, a transaction premium coefficient of 3% is applied. For low-level entities with a triple verification pass rate of less than 80%, their order priority is downgraded to level C.

[0134] Simultaneously, through full-chain data aggregation and analysis, the distribution of process duration, parameter fluctuation range, and quality deviation characteristics of each processing link are statistically analyzed, and process parameter adjustment references and supply and demand matching prediction data are output to the main players in the industry chain.

[0135] Supply and demand matching forecasts are generated based on the aggregation of batch data across the entire blockchain. This involves statistically analyzing the ready-to-market fresh ginseng production capacity, germplasm grade, and expected market launch cycle of various production entities, and matching this with the spare capacity of each processing enterprise's processes, raw material demand levels, and production schedules. A supply and demand matching recommendation list is generated by comprehensively ranking the data based on three dimensions: germplasm matching degree, production capacity fit, and entity credit rating.

[0136] The process parameter adjustment reference is generated based on the correlation between parameters and quality statistics of multiple batches of historical processing data of the same material. The common process parameter ranges corresponding to high-scoring batches are extracted and used as optimization references to be output to the corresponding processing entities.

[0137] The ginseng processing monitoring and traceability system corresponding to this embodiment achieves the above-mentioned full-process control through the coordinated operation of multiple functional modules. The identity baseline management module completes the generation of batch digital identity of RS-FS-20250912-007, the extraction of initial germplasm feature fingerprints and the matching of germplasm gradient baseline, and writes the corresponding data into the consortium blockchain identity contract layer.

[0138] The process cascade mapping module connects to the sensor nodes of the four processes in this batch, collects the entry, process, and exit data of the entire process, generates process-level quality change records for this batch, and calls the pre-trained cross-process cascade quality mapping model to complete quality prediction.

[0139] The cross-validation and evidence storage module receives data uploaded from each process in this batch, automatically performs triple verification, pushes the data to the consortium blockchain to complete consensus and put it on the chain, marks the data that fails to pass as pending confirmation and pushes an alarm.

[0140] The quality pre-simulation intervention module collects steaming process data in real time and inputs it into the model to calculate the finished product quality pre-simulation results, outputs parameter adjustment schemes, triggers pause commands for non-returnable batches and locks digital identities, and the finished product grading verification module generates the quality score and grade of the finished product for this batch, completes germplasm identity matching verification and binds the results to the digital identity.

[0141] The hierarchical traceability query module configures corresponding information display dimensions according to three types of permissions: consumers, distributors, and regulators. It supports QR code verification and authorization verification. The credit collaborative scheduling module calculates the monthly credit score of each entity based on the on-chain data of the current batch and historical batches, and outputs process parameter optimization references and supply and demand matching recommendation lists.

[0142] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0143] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring and tracing the processing of ginseng, characterized in that, Includes the following steps: A unique digital identity is generated for each batch of ginseng. Fresh ginseng characteristic data and basic attribute data are collected to generate an initial germplasm characteristic fingerprint, which is then bound to the digital identity. A pre-built germplasm gradient baseline model is retrieved, and the digital identity, initial germplasm characteristic fingerprint, and index identifier and hash value of the germplasm gradient baseline model are written into the consortium blockchain identity contract layer. Sensor acquisition nodes are deployed for core processing steps to collect quality data at the process inlet, process parameters, and process outlet, forming process-level quality change records. The quality influence coefficients between adjacent processes are calculated, and a cross-process cascaded quality mapping model is constructed. By establishing an automatic cross-verification and evidence storage mechanism for upstream and downstream multi-entity data through a consortium blockchain, the batch process data uploaded by the processing enterprise is triple-verified. Data that meets the conditions is pushed to the consortium blockchain node, consensus is achieved, the data is uploaded to the blockchain, and evidence storage certificates are generated. Data that fails the verification is marked as pending confirmation. A quality pre-simulation and pre-intervention system is built based on a cross-process cascaded quality mapping model. The system inputs real-time collected data of the current process, calculates the pre-simulation results of finished product quality, outputs parameter adjustment schemes according to the degree of deviation, triggers a pause command for a specified batch and locks the digital identity, and records the corresponding records are synchronously uploaded to the blockchain for evidence storage. Extract multi-dimensional indicators from the entire processing chain to generate finished product quality value scores and corresponding grades, retrieve finished product characteristic data and germplasm gradient baseline models for matching and verification, and bind the results to digital identities; Build a traceability query system, configure the corresponding on-chain traceability information display dimensions according to the subject's permissions, and support product verification and identity verification; Based on on-chain data, the credit scores of supply chain entities are calculated, matched with preset business cooperation rules, and output references for adjusting process parameters and supply and demand matching prediction data.

2. The method for monitoring and tracing the ginseng processing according to claim 1, characterized in that, In the fresh ginseng storage process, the surface texture, rhizome morphology, and epidermal spectral feature vectors of ginseng are extracted using hyperspectral imaging and microscopic texture acquisition equipment; at the same time, basic attribute data including place of origin coordinates, cultivation mode, growth years, and harvest time are entered to generate the initial germplasm feature fingerprint and bind it to the digital identity.

3. The method for monitoring and tracing the ginseng processing according to claim 1, characterized in that, The germplasm gradient baseline model is pre-constructed based on the characteristic evolution law of different germplasm ginseng processing. The model covers five core stages: fresh ginseng, after washing, after steaming, after drying, and after processing, and clarifies the characteristic change range and parameter evolution law of each type of germplasm at each processing node.

4. The method for monitoring and tracing the ginseng processing according to claim 1, characterized in that, The core processing steps consist of four processes: cleaning, steaming, drying, and processing. The sensor acquisition nodes, combined with the initial germplasm characteristic fingerprint, collect quality data and process parameters in a linked manner. The quality influence coefficient is used to characterize the quality transfer law of the deviation of the process parameters of the preceding process on the quality of the output of the subsequent process. The cross-process cascaded quality mapping model can correct the benchmark range of the process parameters of the subsequent process based on the actual output quality of the preceding process.

5. The method for monitoring and tracing the ginseng processing according to claim 1, characterized in that, The triple verification includes business volume value consistency verification, quality logic fit verification, and third-party sampling data distribution consistency verification. When at least two of the triple verifications pass, the corresponding batch of data is pushed to the consortium blockchain node to complete consensus on-chain and generate normal evidence storage certificate. Data that fails the verification is marked as pending confirmation, and the normal on-chain process is not executed for the time being. An abnormal alarm is automatically triggered and pushed to the relevant subject.

6. The method for monitoring and tracing the ginseng processing according to claim 1, characterized in that, The pre-test results of the finished product quality include the saponin retention rate and quality grade of the finished product; when the pre-test results are lower than the preset qualified threshold, an early warning message is pushed to the on-site operation terminal, and the parameter adjustment scheme and parameter benchmark correction suggestions for subsequent processes are output; when the pre-test results cannot be brought back to the qualified range through subsequent processes, a process pause command is automatically triggered and the digital identity of the corresponding batch is locked.

7. The method for monitoring and tracing the ginseng processing according to claim 1, characterized in that, The multi-dimensional indicators used to generate the quality value score of the finished product include germplasm grade, process compliance, active ingredient retention rate, pre-process intervention records, and quality stability. The corresponding scores and grades are generated through a weighted scoring model. The matching verification refers to retrieving the hyperspectral feature data of the finished product and comparing it with the feature interval of the post-processing stage in the germplasm gradient baseline model to confirm the consistency of germplasm identity.

8. The method for monitoring and tracing the ginseng processing according to claim 1, characterized in that, The traceability query system is configured with hierarchical information display permissions for different user groups. All displayed data is associated with the digital identity of the corresponding batch. Ordinary consumers can scan the code to view the product's place of origin, batch, compliance conclusion of core processes, quality grade of the finished product, quality inspection report and germplasm identity verification conclusion. Professional distributors and regulatory authorities can view process-level quality details, process parameter execution records and process pre-intervention records after authorization.

9. The method for monitoring and tracing the ginseng processing according to claim 1, characterized in that, The credit scoring of the supply chain entities is based on the timeliness of data upload, the pass rate of the triple verification, the stability of batch quality, and the compliance of abnormal intervention and handling. The credit scoring of the supply chain entities is mapped to the transaction premium coefficient, order priority, and cooperation trust level preset within the alliance. The supply and demand matching prediction data is generated based on the aggregation and analysis of full-chain process data.

10. A ginseng processing monitoring and traceability system, used to implement the method described in any one of claims 1-9, characterized in that, include: The identity baseline management module generates a unique digital identity for each batch of ginseng, collects fresh ginseng characteristic data and basic attribute data to generate an initial germplasm feature fingerprint, retrieves the germplasm gradient baseline model and completes the binding of the initial germplasm feature fingerprint with the digital identity, and writes the digital identity, the initial germplasm feature fingerprint and the index identifier and hash value of the germplasm gradient baseline model into the consortium blockchain identity contract layer. The process cascade mapping module connects to the sensor nodes of each process, collects process entry quality data, process process parameters and process exit quality data, generates process-level quality change records, calculates the quality influence coefficient between processes, and constructs a cross-process cascade quality mapping model. The cross-validation and evidence storage module receives batch process data and performs triple validation. It pushes qualified data to the consortium blockchain to complete consensus and generate evidence storage certificates, and marks data that fails validation as pending confirmation. The quality pre-simulation intervention module uses real-time collected data from the current process to calculate the pre-simulation results of the finished product quality, outputs parameter adjustment schemes according to the degree of deviation, triggers a pause command for a specified batch and locks the digital identity, and related records are synchronously uploaded to the blockchain for evidence storage. The finished product grading and verification module extracts multi-dimensional indicators from the entire chain to generate a finished product quality value score and corresponding grade, retrieves finished product feature data to complete germplasm identity matching and verification, and binds the result to the digital identity. The hierarchical traceability query module allows for the configuration of on-chain information display dimensions based on the subject's permissions, supporting product verification and identity verification. The credit-based collaborative scheduling module calculates the credit scores of supply chain entities based on on-chain data and matches them with preset business cooperation rules, outputting process parameter adjustment references and supply-demand matching prediction data.

Images