Rice processing digital twinning and quality control management system for full-chain quality tracing

Abstract

The invention discloses a rice processing digital twinning and quality control management system for full-chain quality tracing, and relates to the technical field of rice processing quality control and tracing. Comprising a multi-dimensional perception layer, a quaternary collaborative digital twinborn model layer, a block chain twinborn double-track tracing layer, a historical process-quality database, an intelligent quality control optimization layer driven by unprocessed grain characteristics, a distributed data center and a dynamic authority management and control module. Through block chain double-track tracing and virtual-real verification, a dynamic backtracking quality forming process is supported, the tracing credibility is remarkably improved, grain characteristics drive personalized process generation, virtual debugging is combined to pre-judge the risk, the traditional fixed process and post detection limitation are broken through, the finished product reaching rate is improved, and a quaternary twin model restores the whole process of the quality problem. The cross-link root cause is accurately positioned, the analysis efficiency is improved, the data security is guaranteed through authority control and a distributed architecture, and the method is suitable for enterprises of different scales and the whole rice processing scene.

Description

Technical Field

[0001] This invention relates to the field of rice processing quality control and traceability technology, and in particular to a digital twin and quality control management system for rice processing oriented towards full-chain quality traceability. Background Technology

[0002] As a staple food in my country, the quality and safety of rice processing directly affect consumer health and market order. The entire rice processing chain encompasses multiple stages, including planting, storage, processing, logistics, and sales. Quality is influenced by complex factors, and the rice processing industry currently faces challenges in full-chain quality control and traceability, with significant shortcomings identified in several areas:

[0003] Existing digital twin systems focus on equipment simulation of a single processing stage, but fail to construct a collaborative mapping of "raw grain - equipment - process - quality". Traceability only records static data and cannot reconstruct the dynamic process of quality formation, leading to difficulties in root cause analysis.

[0004] Traditional blockchain traceability only uploads physically collected data, lacking two-way verification between virtual and physical data. This makes it prone to data tampering or collection errors that lead to traceability distortion, and it cannot verify the consistency between the data and the actual processing process.

[0005] Existing processing technologies mostly rely on fixed parameters or manual experience adjustments, failing to generate personalized solutions based on the real-time characteristics of the raw grains (such as moisture, variety, and impurities). Furthermore, quality control depends on post-processing testing, making it impossible to predict quality risks during the processing in advance.

[0006] Existing systems rely solely on data statistics to pinpoint root causes, failing to reconstruct the entire process of quality issues and making it difficult to distinguish whether the problem originates in planting, processing, warehousing, or logistics. Furthermore, optimization solutions lack dynamic validation. Therefore, this invention proposes a digital twin and quality control management system for rice processing, oriented towards end-to-end quality traceability, to address the shortcomings of existing technologies. Summary of the Invention

[0007] To address the aforementioned issues, the present invention aims to provide a digital twin and quality control management system for rice processing with end-to-end quality traceability. Through dual-track traceability and virtual-to-real verification using blockchain, the system enhances traceability credibility. The characteristics of raw grains drive personalized processes, and the system predicts quality risks, breaking through traditional limitations. The four-element twin model recreates the entire process of quality problems and accurately locates the root cause. Access control and a distributed architecture ensure security, making the system adaptable to all rice processing scenarios and enterprises of different sizes.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] The rice processing digital twin and quality control management system for full-chain quality traceability includes a multi-dimensional perception layer, a four-element collaborative digital twin model layer, a blockchain twin dual-track traceability layer, a historical process-quality database, an intelligent quality control optimization layer driven by raw grain characteristics, a distributed data center, and a dynamic permission management module.

[0010] The multi-dimensional sensing layer is deployed throughout the entire rice processing chain. It collects multi-dimensional sensing data in real time through a fusion acquisition network to form a standardized sensing dataset for the entire chain.

[0011] The four-element collaborative digital twin model layer is used to construct a four-element collaborative model that includes a material twin, an equipment twin, a process twin, and a quality twin.

[0012] The blockchain twin dual-track traceability layer receives the full-link standardized perception dataset and the simulation data of the four-element collaborative model. Through two-way verification of virtual and real data, the trusted data base is put on the chain in the form of encrypted hash value. At the same time, a unique "twin traceability ID" is assigned to each batch of products.

[0013] The historical process-quality database is used to store historical process-quality correlation data throughout the entire rice processing chain.

[0014] The intelligent quality control optimization layer driven by the characteristics of raw grains is based on the driving data of the material twin and combined with historical process-quality correlation data to generate personalized process parameter schemes to drive the processing equipment to adaptively adjust, and predict quality risks through virtual debugging of the process twin.

[0015] The distributed data center is used to construct a three-tier storage architecture including a real-time database, a historical database, and a twin model database, and to classify and store sensing, simulation, on-chain data, and optimization solutions.

[0016] The dynamic permission management module dynamically allocates system permissions based on user roles and operation scenarios, and is used to synchronize and back up the system operation logs and the operation trajectory of the four-element collaborative model to the blockchain.

[0017] Further improvements are made in the following aspects: When the multi-dimensional perception layer is deployed throughout the entire rice processing chain, specifically in planting fields, raw grain warehouses, processing production lines, packaging lines, logistics links, and terminal sales points, the integrated data acquisition network includes a gene characteristic detection module, a multi-physics field sensor group, a visual dynamic detection unit, and a terminal feedback acquisition module. The gene characteristic detection module collects data on the purity of raw grain varieties and the correlation of disease resistance genes through near-infrared spectroscopy. The multi-physics field sensor group includes soil moisture-fertility sensors, equipment vibration-temperature-pressure sensors, and logistics microenvironment temperature, humidity, vibration, and oxygen concentration sensors. The visual dynamic detection unit uses a high-speed industrial camera combined with machine vision algorithms to collect data on material morphology changes, dynamic evolution of broken rice rate, and packaging integrity in real time. The terminal feedback acquisition module connects with quality testing equipment through QR code questionnaires to collect consumer feedback and terminal sampling data.

[0018] Further improvements are made in that the fusion acquisition network of the multi-dimensional perception layer also includes a rapid detection sensor for pesticide residues in raw grains and a logistics impact sensor. The rapid detection sensor for pesticide residues in raw grains is used to detect the pesticide residue content in raw grains, and the impact sensor is used to record the number of impacts ≥0.5g during the logistics process. The acquired data on pesticide residue content, number of impacts and duration are associated with the quaternary collaborative model.

[0019] A further improvement lies in the following: the collaborative mechanism of the quaternary collaborative digital twin model layer is as follows:

[0020] The material twin dynamically updates material attribute parameters based on the real-time physicochemical properties of the raw grain and outputs a raw grain processing suitability score.

[0021] The equipment twin receives data from a multi-physics sensor array, simulates the wear degree of key components of the equipment and the trend of machining accuracy decay, and outputs the equipment's machining capacity threshold.

[0022] The process twin is based on the raw grain processing adaptability score and equipment processing capacity threshold to simulate the material flow and processing effect under different combinations of process parameters and construct a "parameter-state" coupling matrix.

[0023] The quality twin extracts dynamic data from the material twin, equipment twin, and process twin in real time, and outputs real-time quality scores and potential risk points through a mapping model of "physical parameters-quality indicators".

[0024] A further improvement is that the quaternary collaborative digital twin model layer also includes a model self-calibration unit. The model self-calibration unit periodically collects the deviation values ​​between the actual physical data and the simulation data, and uses a Bayesian optimization algorithm to iteratively update the model parameters to ensure that the model simulation accuracy remains stable within the threshold for a long time.

[0025] Further improvements are made in the following ways: The two-way verification mechanism of the blockchain twin dual-track traceability layer is as follows: the original data and quality inspection results collected by the multi-dimensional perception layer are encrypted and uploaded to the chain by timestamp, and the simulation data, virtual and real data comparison results, and quality evolution trajectory of the quaternary twin model are encrypted and uploaded to the chain. Then, the two tracks of data are associated through hash values, and the deviation between physical data and simulation data is verified before being uploaded to the chain.

[0026] Further improvements include: when the blockchain twin dual-track traceability layer verifies the deviation between physical data and simulation data, it automatically triggers re-collection and model calibration when the deviation exceeds the limit, and supports calling the digital quaternary collaborative model through the "twin traceability ID".

[0027] Further improvements are made in the following aspects: The raw grain characteristic-driven intelligent quality control optimization layer includes a raw grain characteristic clustering module, a virtual process debugging module, a quality prediction model, and a parameter adaptive execution module. The raw grain characteristic clustering module classifies raw grains into N processing adaptation types based on data such as raw grain moisture, impurities, variety, and hardness using the K-means algorithm. The virtual process debugging module simulates the processing effects of different processing parameters on different types of raw grains through process twins, generating personalized process parameter combinations. The quality prediction model is based on the LSTM-gradient boosting tree hybrid algorithm, which takes raw grain characteristic data, equipment status data, and environmental parameters as input to predict key quality indicators such as broken rice rate, whiteness, and yellowing rate of finished products. The parameter adaptive execution module sends the optimized process parameters to the processing equipment in real time to achieve dynamic adjustment.

[0028] Further improvements include: the distributed data center includes a real-time data processing engine, a twin model parameter library, a blockchain traceability database, and a quality optimization scheme library. The real-time data processing engine adopts a stream computing framework. The twin model parameter library stores the optimal parameters of the model under different grain types and equipment states, and supports automatic invocation and iterative updates. The blockchain traceability database adopts an "on-chain hash + off-chain details" storage mode, storing key index data on-chain and complete perception and simulation data off-chain, supporting fast retrieval and traceability. The quality optimization scheme library records historical quality problems and corresponding optimization schemes, and supports inferring the effect of the scheme based on the quaternary twin model.

[0029] Further improvements include: the system also includes a cross-process quality cause twin simulation module. This module uses blockchain traceability data to call a quaternary twin model to recreate the dynamic scenario of the entire process of quality problems. It also simulates the influence weight of different process parameters on quality indicators through the control variable method, accurately locates the process causing the quality problem, and supports the implementation effect of the simulation optimization scheme in the quaternary twin model, outputting the optimal solution.

[0030] The beneficial effects of this invention are as follows:

[0031] The blockchain twin dual-track traceability layer of this invention encrypts and uploads the simulation data, virtual-real data comparison results, and quality evolution trajectory of the quaternary twin model to the chain. Then, the two tracks of data are associated through hash values. Before being uploaded to the chain, the deviation between physical data and simulation data is verified. It not only records static data, but also supports the backtracking of the dynamic process of quality formation through the twin model, which significantly improves the traceability credibility.

[0032] The intelligent quality control optimization layer driven by the characteristics of raw grains in this invention generates personalized process parameter schemes based on material twin driving data and combined with historical process-quality correlation data to drive the processing equipment to adaptively adjust. It can also predict quality risks through virtual debugging of process twins, enabling parameters to be adjusted in advance. The finished product quality compliance rate can be significantly improved, breaking through the limitations of fixed process + post-event inspection in traditional technology.

[0033] The system of this invention can realize the reconstruction of the entire process of quality problems by utilizing the four-element collaborative digital twin model layer, locate the root cause across links by controlling the variable method, significantly improve the analysis efficiency, and can deduce the effect of optimization schemes, avoiding blind adjustments;

[0034] The system's dynamic permission control module ensures stronger system security and adaptability, while the on-chain operation traceability ensures data security. The distributed data center is adaptable to processing enterprises of different sizes, supporting flexible deployment from small workshops to large factories. It also covers the entire "planting-terminal" chain and adapts to the needs of all rice processing scenarios. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0036] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0037] according to Figure 1 As shown, this embodiment proposes a digital twin and quality control management system for rice processing with full-chain quality traceability, including a multi-dimensional perception layer, a four-element collaborative digital twin model layer, a blockchain twin dual-track traceability layer, a historical process-quality database, an intelligent quality control optimization layer driven by raw grain characteristics, a distributed data center, and a dynamic permission management module.

[0038] The multi-dimensional perception layer is deployed throughout the entire rice processing chain. It collects multi-dimensional perception data in real time through a fusion acquisition network, forming a standardized perception dataset for the entire chain. Specifically, when deployed throughout the rice processing chain, the multi-dimensional perception layer is located in planting fields, raw grain warehouses, processing production lines (rice hullers, rice milling machines, color sorters, etc.), packaging lines, logistics links (logistics vehicles), and terminal sales points. The fusion acquisition network includes a gene characteristic detection module, a multi-physics field sensor group, a visual dynamic detection unit, and a terminal feedback acquisition module. The gene characteristic detection module uses near-infrared spectroscopy to collect data on the purity of raw grain varieties and the correlation of disease resistance genes. The multi-physics field sensor group includes soil moisture-fertility sensors, equipment vibration-temperature-pressure sensors, and logistics microenvironment temperature, humidity, vibration, and oxygen concentration sensors. The visual dynamic detection unit uses a high-speed industrial camera combined with machine vision algorithms to collect data on material morphology changes, dynamic evolution of broken rice rate, and packaging integrity in real time. The terminal feedback acquisition module connects with quality testing equipment through QR code questionnaires to collect consumer feedback and terminal sampling data. The fusion acquisition network of the multi-dimensional perception layer also includes a rapid detection sensor for pesticide residues in raw grains and a logistics impact sensor. The rapid detection sensor for pesticide residues in raw grains is used to detect the pesticide residue content in raw grains, and the impact sensor is used to record the number and duration of impacts ≥0.5g during the logistics process. The acquired data on pesticide residue content, number of impacts, and duration are associated with the quaternary collaborative model.

[0039] The four-element collaborative digital twin model layer is used to construct a four-element collaborative model that includes a material twin, an equipment twin, a process twin, and a quality twin.

[0040] The collaborative mechanism of the four-element collaborative digital twin model layer is as follows:

[0041] The material twin dynamically updates material attribute parameters (such as hardness and water absorption rate) based on the real-time physicochemical properties (moisture, impurities, and varietal purity) of the raw grain, and outputs a raw grain processing suitability score (0-100 points).

[0042] The equipment twin receives data from a multi-physics sensor array (vibration-temperature-pressure data), simulates the wear degree and processing accuracy decay trend of key equipment components (rice hulling-milling-color sorting-polishing equipment), and outputs the equipment processing capacity threshold (such as the maximum compatible rice milling pressure).

[0043] The process twin is based on the raw grain processing adaptability score and equipment processing capacity threshold to simulate the material flow and processing effect under different combinations of process parameters (simulating the entire process flow of rice hulling-milling-color sorting-polishing) and construct a "parameter-state" coupling matrix.

[0044] The quality twin extracts dynamic data from the material twin, equipment twin, and process twin in real time, and outputs real-time quality scores and potential risk points through a mapping model of "physical parameters-quality indicators".

[0045] The quaternary collaborative digital twin model layer also includes a model self-calibration unit. The model self-calibration unit periodically collects the deviation values ​​between the actual physical data and the simulation data, and uses a Bayesian optimization algorithm to iteratively update the model parameters to ensure that the model simulation accuracy remains stable within the threshold (±2.5%) for a long time.

[0046] The blockchain twin dual-track traceability layer receives standardized perception datasets from the entire chain and simulation data from the quaternary collaborative model. Through bidirectional verification of virtual and physical data, the trusted data base is uploaded to the chain in the form of encrypted hash values. At the same time, a unique "twin traceability ID" is assigned to each batch of products. The bidirectional verification mechanism of the blockchain twin dual-track traceability layer is as follows: the original data and quality inspection results collected by the multi-dimensional perception layer are encrypted and uploaded to the chain by timestamp. The simulation data of the quaternary twin model, the comparison results of virtual and physical data, and the quality evolution trajectory are encrypted and uploaded to the chain. Then, the two tracks of data are associated through hash values. Before being uploaded to the chain, the deviation between physical data and simulation data is verified. When the deviation exceeds the limit (difference threshold ≤ ±3%), re-collection and model calibration are automatically triggered. It also supports calling the digital quaternary collaborative model through the "twin traceability ID".

[0047] The historical process-quality database is used to store historical process-quality correlation data throughout the entire rice processing chain.

[0048] The intelligent quality control optimization layer driven by raw grain characteristics is based on the driving data of the material twin and combined with historical process-quality correlation data to generate personalized process parameter schemes to drive the processing equipment to adaptively adjust, and to predict quality risks through virtual debugging of the process twin. Specifically, the intelligent quality control optimization layer driven by raw grain characteristics includes a raw grain characteristic clustering module, a virtual process debugging module, a quality prediction model, and a parameter adaptive execution module. The raw grain characteristic clustering module classifies raw grains into N processing adaptation types based on data such as raw grain moisture, impurities, variety, and hardness using the K-means algorithm. The virtual process debugging module simulates the processing effects of different processing parameters on different types of raw grains through the process twin to generate personalized process parameter combinations. The quality prediction model is based on the LSTM-gradient boosting tree hybrid algorithm, which takes raw grain characteristic data, equipment status data, and environmental parameters as input to predict key quality indicators such as broken rice rate, whiteness, and yellowing rate of finished product. The parameter adaptive execution module sends the optimized process parameters to the processing equipment in real time to achieve dynamic adjustment.

[0049] The distributed data center is used to construct a three-tiered storage architecture including a real-time database, a historical database, and a twin model database, categorizing and storing sensing, simulation, on-chain data, and optimization solutions. Specifically, the distributed data center includes a real-time data processing engine, a twin model parameter library, a blockchain traceability database, and a quality optimization solution library. The real-time data processing engine adopts a stream computing framework. The twin model parameter library stores the optimal parameters of the model under different grain types and equipment states, and supports automatic invocation and iterative updates. The blockchain traceability database adopts an "on-chain hash + off-chain details" storage mode, storing key index data on-chain and complete sensing and simulation data off-chain, supporting fast retrieval and traceability. The quality optimization solution library records historical quality problems and corresponding optimization solutions, and supports inferring the effect of solutions based on the quaternary twin model.

[0050] The dynamic permission management module dynamically allocates system permissions based on user roles and operation scenarios, and is used to synchronize and back up the system operation logs and the operation trajectory of the four-element collaborative model to the blockchain.

[0051] The system also includes a cross-process quality cause twin simulation module. This module uses blockchain traceability data to call a quaternary twin model to recreate the dynamic scenario of the entire process of quality problems. It simulates the influence weight of different parameters (soil fertility, storage temperature and humidity, processing pressure, and logistics vibration) on quality indicators through the control variable method, accurately locates the cause of quality problems, and supports the implementation effect of the simulation optimization scheme in the quaternary twin model, outputting the optimal solution.

[0052] The blockchain twin dual-track traceability layer of this invention encrypts and uploads simulation data, virtual-physical data comparison results, and quality evolution trajectory from the quaternary twin model to the blockchain. The two tracks are then linked via hash values. Before uploading, deviations between physical and simulation data are verified. This not only records static data but also supports tracing back the dynamic process of quality formation through the twin model, significantly improving traceability reliability. The raw grain characteristic-driven intelligent quality control optimization layer of this invention generates personalized process parameter schemes based on material twin-based driving data and combined with historical process-quality correlation data. This drives adaptive adjustments of processing equipment and predicts quality risks through virtual debugging using the process twin, enabling advance adjustments. By adjusting parameters, the finished product quality compliance rate can be significantly improved, overcoming the limitations of fixed processes and post-processing inspections in traditional technologies. This invention's system utilizes a four-element collaborative digital twin model layer to reconstruct the entire process of quality problems, locates cross-stage root causes using the controlled variable method, significantly improves analysis efficiency, and can deduce the effects of optimization solutions, avoiding blind adjustments. The system's dynamic permission management module ensures stronger system security and adaptability, while on-chain operation traceability ensures data security. The distributed data center is adaptable to processing enterprises of different sizes, supporting flexible deployment from small workshops to large factories, and covering the entire "planting-terminal" chain, adapting to the needs of all rice processing scenarios.

[0053] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A digital twin and quality control management system for rice processing with end-to-end quality traceability, characterized by: It includes a multi-dimensional perception layer, a four-element collaborative digital twin model layer, a blockchain twin dual-track traceability layer, a historical process-quality database, an intelligent quality control optimization layer driven by raw grain characteristics, a distributed data center, and a dynamic permission management module. The multi-dimensional sensing layer is deployed throughout the entire rice processing chain. It collects multi-dimensional sensing data in real time through a fusion acquisition network to form a standardized sensing dataset for the entire chain. The four-element collaborative digital twin model layer is used to construct a four-element collaborative model that includes a material twin, an equipment twin, a process twin, and a quality twin. The blockchain twin dual-track traceability layer receives the full-link standardized perception dataset and the simulation data of the four-element collaborative model. Through two-way verification of virtual and real data, the trusted data base is put on the chain in the form of encrypted hash value. At the same time, a unique "twin traceability ID" is assigned to each batch of products. The historical process-quality database is used to store historical process-quality correlation data throughout the entire rice processing chain. The intelligent quality control optimization layer driven by the characteristics of raw grains is based on the driving data of the material twin and combined with historical process-quality correlation data to generate personalized process parameter schemes to drive the processing equipment to adaptively adjust, and predict quality risks through virtual debugging of the process twin. The distributed data center is used to construct a three-tier storage architecture including a real-time database, a historical database, and a twin model database, and to classify and store sensing, simulation, on-chain data, and optimization solutions. The dynamic permission management module dynamically allocates system permissions based on user roles and operation scenarios, and is used to synchronize and back up the system operation logs and the operation trajectory of the four-element collaborative model to the blockchain.

2. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 1, characterized in that: When the multi-dimensional perception layer is deployed throughout the entire rice processing chain, it is specifically deployed in planting fields, raw grain warehouses, processing production lines, packaging lines, logistics links, and terminal sales points. The integrated data acquisition network includes a gene characteristic detection module, a multi-physics field sensor group, a visual dynamic detection unit, and a terminal feedback acquisition module. The gene characteristic detection module uses near-infrared spectroscopy technology to collect data on the purity of raw grain varieties and the correlation of disease resistance genes. The multi-physics field sensor group includes soil moisture-fertility sensors, equipment vibration-temperature-pressure sensors, and logistics microenvironment temperature, humidity, vibration, and oxygen concentration sensors. The visual dynamic detection unit uses a high-speed industrial camera combined with machine vision algorithms to collect data on material morphology changes, dynamic evolution of broken rice rate, and packaging integrity in real time. The terminal feedback acquisition module connects with quality testing equipment through QR code questionnaires to collect consumer feedback and terminal sampling data.

3. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 2, characterized in that: The fusion acquisition network of the multi-dimensional perception layer also includes a rapid detection sensor for pesticide residues in raw grains and a logistics impact sensor. The rapid detection sensor for pesticide residues in raw grains is used to detect the pesticide residue content in raw grains, and the impact sensor is used to record the number and duration of impacts ≥0.5g during the logistics process. The acquired data on pesticide residue content, number of impacts, and duration are associated with the quaternary collaborative model.

4. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 3, characterized in that: The collaborative mechanism of the four-element collaborative digital twin model layer is as follows: The material twin dynamically updates material attribute parameters based on the real-time physicochemical properties of the raw grain and outputs a raw grain processing suitability score. The equipment twin receives data from a multi-physics sensor array, simulates the wear degree of key components of the equipment and the trend of machining accuracy decay, and outputs the processing capacity threshold of the equipment. The process twin is based on the raw grain processing adaptability score and equipment processing capacity threshold to simulate the material flow and processing effect under different combinations of process parameters and construct a "parameter-state" coupling matrix. The quality twin extracts dynamic data from the material twin, equipment twin, and process twin in real time, and outputs real-time quality scores and potential risk points through a mapping model of "physical parameters-quality indicators".

5. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 4, characterized in that: The quaternary collaborative digital twin model layer also includes a model self-calibration unit. The model self-calibration unit periodically collects the deviation values ​​between the actual physical data and the simulation data, and uses a Bayesian optimization algorithm to iteratively update the model parameters to ensure that the model simulation accuracy remains stable within the threshold for a long time.

6. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 1, characterized in that: The bidirectional verification mechanism of the blockchain twin dual-track traceability layer is as follows: the original data and quality inspection results collected by the multi-dimensional perception layer are encrypted and uploaded to the chain by timestamp, and the simulation data, virtual and real data comparison results, and quality evolution trajectory of the quaternary twin model are encrypted and uploaded to the chain. Then, the two tracks of data are associated through hash values, and the deviation between physical data and simulation data is verified before being uploaded to the chain.

7. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 6, characterized in that: When the blockchain twin dual-track traceability layer verifies the deviation between physical data and simulation data, it automatically triggers re-collection and model calibration when the deviation exceeds the limit, and supports calling the digital quaternary collaborative model through the "twin traceability ID".

8. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 1, characterized in that: The intelligent quality control optimization layer driven by raw grain characteristics includes a raw grain characteristic clustering module, a virtual process debugging module, a quality prediction model, and a parameter adaptive execution module. The raw grain characteristic clustering module classifies raw grains into N processing adaptation types based on data such as moisture, impurities, variety, and hardness using the K-means algorithm. The virtual process debugging module simulates the processing effects of different processing parameters on different types of raw grains through process twins, generating personalized process parameter combinations. The quality prediction model is based on the LSTM-gradient boosting tree hybrid algorithm, which takes raw grain characteristic data, equipment status data, and environmental parameters as input to predict key quality indicators such as broken rice rate, whiteness, and yellowing rate of finished product. The parameter adaptive execution module sends the optimized process parameters to the processing equipment in real time to achieve dynamic adjustment.

9. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 1, characterized in that: The distributed data center includes a real-time data processing engine, a twin model parameter library, a blockchain traceability database, and a quality optimization scheme library. The real-time data processing engine adopts a stream computing framework. The twin model parameter library stores the optimal parameters of the model under different grain types and equipment states, and supports automatic invocation and iterative updates. The blockchain traceability database adopts an "on-chain hash + off-chain details" storage mode, storing key index data on-chain and complete perception and simulation data off-chain, supporting fast retrieval and traceability. The quality optimization scheme library records historical quality problems and corresponding optimization schemes, and supports the deduction of scheme effects based on the quaternary twin model.

10. The rice processing digital twin and quality control management system for full-chain quality traceability as described in claim 1, characterized in that: The system also includes a cross-process quality cause twin simulation module. This module uses blockchain traceability data to call a quaternary twin model to recreate the dynamic scenario of the entire process of quality problems. It simulates the influence weight of different parameters on quality indicators through the control variable method, accurately locates the cause of quality problems, and supports the implementation effect of the simulation optimization scheme in the quaternary twin model, outputting the optimal solution.

Images