A blockchain-based green product certification and traceability system

Abstract

The present invention relates to the technical field of data processing, and discloses a green product certification and traceability system based on blockchain. The system includes a data collection module that collects green attribute data of the entire product life cycle using multi-source perception devices; a blockchain evidence storage module that hierarchically encrypts data based on a hybrid hashing algorithm to generate an immutable ledger record; an intelligent contract audit module that automatically generates green certification results through a multi-rule verification model; a traceability query module that constructs a privacy protection query channel using zero-knowledge proof technology to output a verifiable traceability chain; and a dynamic optimization module that adjusts certification standards according to changes in environmental policies. The system realizes comprehensive collection, secure storage, accurate certification, reliable traceability, and dynamic optimization of standards for green product data, improves the credibility of certification and regulatory efficiency, and promotes the development of the green industry.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a blockchain-based green product certification and traceability system. Background Art

[0002] With the continuous improvement of global environmental awareness, the green product market has developed rapidly. Consumers' demand for green products is increasing day by day. They not only focus on the functions and quality of products, but also pay more attention to the environmental protection attributes of products throughout the entire life cycle, such as production, transportation, and use.

[0003] There are many problems in the traditional product certification and traceability methods. On the one hand, data management is decentralized and lacks credibility. Throughout the life cycle of a product, it involves multiple links such as raw material suppliers, manufacturers, transporters, and sellers. The data of each link is often independently managed by different entities, with inconsistent data formats and various storage methods. This makes the data easily tampered with or lost, and it is difficult to ensure the authenticity and integrity of the data, resulting in the lack of credibility of the certification results.

[0004] On the other hand, the traceability efficiency is low and it is difficult to achieve precise traceability. Most existing traceability systems are based on centralized databases. When the data volume is large, the query speed is slow, and it is difficult to quickly locate the specific information of the product from the vast amount of data. Once a quality problem or environmental protection violation occurs, it is impossible to quickly and accurately trace the source of the problem, bringing great difficulties to supervision and consumer rights protection. Taking the clothing industry as an example, when it is found that a batch of clothing fails to meet the environmental protection standards, due to the complex supply chain information, it is very difficult to quickly determine whether the problem occurs in the raw material procurement link, the production and processing link, or the transportation and storage link.

[0005] In addition, the traditional certification and traceability system is difficult to adapt to the dynamic changes of environmental policies. Environmental policies are constantly updated, and the certification standards for green products are also continuously adjusted. However, the traditional system lacks the ability of automatic adjustment and optimization, and requires manual updates of the certification standards and processes. This is not only inefficient, but also prone to errors and omissions, and it is impossible to accurately certify and supervise green products in a timely and effective manner.

[0006] With the rise of blockchain technology, its characteristics such as decentralization, immutability, and traceability provide new ideas for solving the problems of green product certification and traceability. However, at present, the research and practice of applying blockchain technology to green product certification and traceability are still in the exploration stage. Existing solutions still have deficiencies in aspects such as the comprehensiveness of data collection, the security of encrypted storage, the accuracy of certification and review, and the adaptability to policy changes, and cannot fully meet the needs of the market and supervision. Therefore, developing an efficient, reliable, intelligent blockchain-based green product certification and traceability system that can adapt to policy changes has important practical significance. Summary of the Invention

[0007] The object of the present invention is to provide a blockchain-based green product certification and traceability system to solve the problems raised in the above-mentioned background art.

[0008] To achieve the above object, the present invention provides the following technical solution: A blockchain-based green product certification and traceability system, the system includes:

[0009] Data acquisition module: used to collect green attribute data of the entire product life cycle through multi-source sensing devices;

[0010] Blockchain evidence storage module: hierarchically encrypt the green attribute data based on a hybrid hashing algorithm to generate an immutable distributed ledger record;

[0011] Intelligent contract audit module: input the distributed ledger record into a preset multi-rule verification model to automatically generate a green certification result;

[0012] Traceability query module: use zero-knowledge proof technology to construct a privacy protection query channel and output a verifiable traceability chain of the product's green attributes according to user requests;

[0013] Dynamic optimization module: construct a multi-objective certification parameter optimization model according to environmental policy changes, and adaptively adjust the certification standard using the gradient descent and constraint relaxation algorithms.

[0014] Preferably, the collection of green attribute data of the entire product life cycle through multi-source sensing devices includes:

[0015] The multi-source sensing devices include a carbon emission monitor, an energy consumption sensor, a logistics tracking tag, and an environmental humidity sensor;

[0016] Align the carbon emission monitor data and the energy consumption sensor data in space and time to construct a product carbon footprint distribution map; perform multi-scale feature decomposition on the logistics tracking tag data and the environmental humidity sensor data to generate a supply chain state feature sequence;

[0017] Construct a two-stream feature fusion network. The first stream uses a three-dimensional convolutional network to extract the spatial correlation features of the carbon footprint distribution map, and the second stream uses a temporal attention network to extract the dynamic evolution features of the supply chain state feature sequence;

[0018] Fuse the spatial correlation features and the dynamic evolution features through a cross-modal graph neural network to generate a comprehensive green attribute dataset including raw material traceability, production process energy consumption, and transportation path optimization.

[0019] Preferably, the hierarchical encryption of the green attribute data based on the hybrid hashing algorithm includes:

[0020] Divide the green attribute data into a basic layer, a process layer, and a result layer according to business types; generate a digest value for the basic layer data using a one-way hash function; use a chained hash structure for the process layer data to ensure data continuity; introduce a Merkle tree structure for the result layer data to achieve batch verification.

[0021] Construct a dynamic key management mechanism, generate differentiated encryption keys according to data levels, and achieve distributed storage through key sharding technology.

[0022] Preferably, the multi-rule verification model adopts a multi-channel parallel decision-making architecture, including:

[0023] Construct a rule graph, where the nodes include national environmental protection standard nodes, industry certification specification nodes, and enterprise-defined rule nodes, and the node attributes include threshold ranges, effective times, and weight coefficients.

[0024] Adopt a two-level attention mechanism. The first level filters associated rules through rule matching degree calculation, and the second level filters expired rules through timeliness verification.

[0025] Integrate the verification results of each channel based on a weighted voting mechanism, and output a green certification report containing compliance scores and anomaly marks.

[0026] Preferably, the zero-knowledge proof technology is realized by combining ring signature and homomorphic encryption, including:

[0027] Encode the product green attribute data into a set of verifiable claims; use the ring signature algorithm to hide user identity information; perform homomorphic encryption processing on the claim set to support logical operation verification in the ciphertext state.

[0028] Preferably, the multi-objective authentication parameter optimization model adopts a decomposition and coordination strategy, including:

[0029] Model the authentication standard parameters as a multi-constraint nonlinear optimization problem, and the decision variables include carbon emission caps, the proportion of recycled materials, and energy consumption efficiency coefficients.

[0030] Initialize the relaxation sub-problem and calculate the approximate Pareto front solution, and use the dynamic sensitivity analysis algorithm to adjust the objective function weights according to policy changes.

[0031] In the decomposition stage, divide the global problem into multiple single-objective sub-problems based on the objective decomposition technology; in the coordination stage, introduce a shadow price mechanism to balance the resource conflicts between sub-problems.

[0032] Preferably, the three-dimensional convolutional network adopts a multi-granularity pooling structure, including:

[0033] Divide the carbon footprint distribution map into multiple levels of grids according to geographical regions, and each level of grid stores the carbon emission intensity and the time cumulative amount.

[0034] In the feature extraction stage, dilated convolutional kernels are adopted to capture cross-region associations, and in the feature compression stage, multi-scale information is retained through spatial pyramid pooling;

[0035] A channel attention mechanism is introduced to adaptively weight the feature maps.

[0036] Preferably, the chained hash structure is implemented through timestamp binding and includes:

[0037] Construct a hash verification index table to record the dual verification information of the location hash and timestamp hash of the data block.

[0038] Preferably, the timeliness verification is implemented through a sliding time window and includes:

[0039] Set the effective time interval and priority label for each rule; construct a time window sliding model to dynamically intercept the currently valid rule subset;

[0040] Adopt an expired rule recycling mechanism to move the rules beyond the time window into the historical rule library and freeze their verification weights.

[0041] Preferably, the dynamic sensitivity analysis algorithm is implemented based on an online learning framework and includes:

[0042] Collect the objective function gradient and constraint relaxation degree in the historical optimization process as training samples;

[0043] Construct a Gaussian process regression model to fit the non-linear relationship between the weight coefficient and the external policy variable;

[0044] Online update the model parameters through the incremental Kalman filter algorithm and adjust the sensitivity threshold of the authentication standard in real time.

[0045] Compared with the prior art, the beneficial effects of the present invention are:

[0046] The blockchain-based green product certification and traceability system of the present invention has many significant beneficial effects. In terms of data management, the data acquisition module uses multi-source perception devices, such as carbon emission monitors, energy consumption sensors, logistics tracking tags, and environmental humidity sensors, etc., to achieve comprehensive acquisition of the green attribute data of the product throughout its life cycle. This method can obtain product information from multiple dimensions and ensure the integrity and accuracy of the data. For example, in the production process of electronic products, not only can the energy consumption and carbon emissions in the production link be monitored, but also the environmental information during transportation can be grasped through logistics tracking tags, providing rich data support for comprehensively evaluating the green attributes of the product.

[0047] The blockchain evidence storage module uses a hybrid hashing algorithm to encrypt data hierarchically, generating an immutable distributed ledger record. This encrypted storage method greatly improves the security and credibility of data. On the one hand, it encrypts by dividing levels according to business types, and adopts different technologies such as one-way hashing functions, chained hashing structures, and Merkle tree structures for different levels of data characteristics, ensuring the integrity and continuity of data. For example, through timestamp binding in the chained hashing structure, the continuity of data in terms of time and location can be verified, preventing data from being tampered with. On the other hand, the dynamic key management mechanism generates differentiated encryption keys according to data levels and stores them distributively through key sharding technology, further enhancing data confidentiality and effectively preventing data leakage.

[0048] The multi-rule verification model of the smart contract audit module improves the accuracy and fairness of authentication. This model constructs a rule graph, integrates national environmental protection standards, industry certification specifications, and enterprise-defined rules, uses a two-level attention mechanism to screen associated rules and filter expired rules, and integrates verification results based on a weighted voting mechanism to output a green certification report containing compliance scores and anomaly marks. This makes the authentication process more scientific and rigorous, and can more accurately evaluate whether a product meets green standards. For example, in the food industry, by comprehensively considering rules at different levels, it is possible to comprehensively evaluate various links of food from raw material procurement, production and processing to packaging and transportation, avoiding the limitations of single-rule evaluation.

[0049] The traceability query module uses zero-knowledge proof technology to build a privacy protection query channel, which ensures that users can obtain a verifiable traceability chain of product green attributes while protecting user privacy. The combination of ring signature and homomorphic encryption can not only hide user identity information, but also support logical operation verification in ciphertext state. When consumers query product information, they don't have to worry about personal privacy leakage, and at the same time can effectively verify the green attributes of products, enhancing consumers' trust in green products.

[0050] The dynamic optimization module constructs a multi-objective authentication parameter optimization model according to changes in environmental policies, and adaptively adjusts authentication standards using gradient descent and constraint relaxation algorithms. It models authentication standard parameters as a multi-constrained non-linear optimization problem, adjusts the weights of objective functions according to policy changes through a dynamic sensitivity analysis algorithm, and uses a decomposition and coordination strategy to balance resource conflicts between sub-problems. This enables the system to keep up with policy changes in a timely manner, ensuring the timeliness and reasonableness of authentication standards. For example, when the country raises the carbon emission requirements for a certain type of product, the system can automatically adjust authentication parameters and re-evaluate the green attributes of products to ensure that the authentication results comply with the latest policies.

[0051] In addition, each module of the system works in coordination to form a complete and efficient green product certification and traceability system. This not only improves the quality and efficiency of green product certification, but also provides a powerful regulatory tool for government regulatory departments, promoting the healthy development of the green industry. At the same time, it enhances consumers' awareness and trust in green products, driving the prosperity of the green consumer market. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is the working principle diagram of the blockchain-based green product certification and traceability system described in the present invention;

[0053] Figure 2 It is the working diagram of the multi-rule verification model;

[0054] Figure 3 It is the schematic diagram of zero-knowledge proof privacy protection;

[0055] Figure 4 It is the schematic diagram of three-dimensional convolutional network feature extraction. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0057] Please refer to Figures 1 - 4 , the present invention provides a blockchain-based green product certification and traceability system, aiming to achieve effective management and monitoring of the entire life cycle of green products, ensuring that the green attributes of products are true, reliable and traceable. The system mainly consists of the following core modules working in coordination to achieve the goal:

[0058] Data acquisition module: Use multi-source sensing devices to comprehensively collect green attribute data of the entire life cycle of the product. These multi-source sensing devices include carbon emission monitors, energy consumption sensors, logistics tracking tags, and environmental humidity sensors. The carbon emission monitor is used to monitor the carbon emissions of the product in each link, the energy consumption sensor is responsible for recording the energy consumption during the production and transportation of the product, the logistics tracking tag can track the location information of the product in the supply chain in real time, and the environmental humidity sensor monitors the humidity of the environment where the product is located. Through these devices, data related to the green attributes of the product are obtained from multiple dimensions, providing a basis for subsequent analysis.

[0059] Blockchain Evidence Storage Module: The hybrid hashing algorithm is used to perform hierarchical encryption processing on the collected green attribute data, thereby generating an immutable distributed ledger record. The green attribute data is divided into a basic layer, a process layer, and a result layer according to business types. Different hashing techniques are used for processing according to the data characteristics of different levels. At the same time, a dynamic key management mechanism is constructed. Differentiated encryption keys are generated based on the data levels, and the key sharding technology is used to achieve distributed storage, enhancing the security and reliability of data storage.

[0060] Intelligent Contract Audit Module: The distributed ledger record stored on the blockchain is input into a preset multi-rule verification model, which will automatically generate a green certification result. The multi-rule verification model adopts a multi-channel parallel decision-making architecture. By constructing a rule graph, integrating various rules such as national environmental protection standards, industry certification specifications, and enterprise-defined rules, a two-level attention mechanism is used to screen effective rules, and the verification results are integrated based on a weighted voting mechanism. Finally, a green certification report containing compliance scores and anomaly marks is output to objectively and fairly evaluate the green attributes of the product.

[0061] Traceability Query Module: The zero-knowledge proof technology is used to construct a privacy protection query channel. This technology combines ring signature and homomorphic encryption to encode the product green attribute data into a set of verifiable statements. The ring signature algorithm is used to hide the user's identity information, and the statement set is homomorphically encrypted to support logical operation verification in the ciphertext state. When the user issues a query request, the system can output a verifiable traceability chain of the product green attributes, meeting the user's query needs for product green attribute information while protecting the user's privacy.

[0062] Dynamic Optimization Module: A multi-objective certification parameter optimization model is constructed according to changes in environmental policies, and the gradient descent and constraint relaxation algorithms are used to adaptively adjust the certification standards. The certification standard parameters are modeled as a multi-constraint non-linear optimization problem, and the decision variables include the carbon emission cap, the proportion of recycled materials, and the energy consumption efficiency coefficient, etc. By initializing the relaxation sub-problem, calculating the approximate solution of the Pareto front, and using the dynamic sensitivity analysis algorithm to adjust the objective function weights according to policy changes, the global problem is divided into multiple single-objective sub-problems in the decomposition stage, and the shadow price mechanism is introduced in the coordination stage to balance the resource conflicts between sub-problems, enabling the certification standards to be optimized in a timely manner with changes in environmental policies and ensuring the adaptability and effectiveness of the system.

[0063] The implementation of the present invention will be further described below in conjunction with Embodiments 1 to 6.

[0064] Embodiment 1:

[0065] In this embodiment, the specific process of green attribute data collection and processing in the data collection module is further elaborated, especially the content related to the three-dimensional convolutional network.

[0066] First, the carbon emission monitors, energy consumption sensors, logistics tracking tags, and ambient humidity sensors in the multi-source sensing device work together. Carbon emission monitors and energy consumption sensors collect data at different times and locations. To construct an accurate product carbon footprint distribution map, these two types of data need to be aligned in time and space. For example, during the product production process, carbon emission monitors and energy consumption sensors may collect data at different time intervals. Through time synchronization algorithms and spatial mapping methods, carbon emission and energy consumption data are accurately matched to the corresponding production links and geographic locations, thus constructing a carbon footprint distribution map that reflects the carbon emissions and energy consumption of the product at different stages.

[0067] Multi-scale feature decomposition is performed on logistics tracking tag data and ambient humidity sensor data. Logistics tracking tags record the product's transportation trajectory within the supply chain, while ambient humidity sensors record changes in humidity during transportation. Using multi-scale analysis methods such as wavelet transforms, this data is decomposed at different time and spatial scales to extract features reflecting changes in supply chain status and generate a supply chain status feature sequence. For example, during the process of transporting a product from its production site to its destination, logistics tracking tag data can reflect information such as transportation speed and dwell time, while ambient humidity sensor data can reflect ambient humidity conditions at different stages of transportation. Multi-scale feature decomposition allows this information to be analyzed at different scales, uncovering more valuable features.

[0068] Then, a two-stream feature fusion network is constructed. The first stream uses a three-dimensional convolutional network to extract the spatial correlation features of the carbon footprint distribution map. The three-dimensional convolutional network uses a multi-granularity pooling structure to divide the carbon footprint distribution map into multi-level grids according to geographical regions. Each level of grid stores the carbon emission intensity and time accumulation. Assuming that the carbon footprint distribution map is divided into Level grid, in In the grid level, carbon emission intensity is expressed as Indicates that the time accumulator is expressed as In the feature extraction stage, dilated convolution kernels are used to capture cross-regional correlations. Dilated convolution kernels can expand the receptive field of the convolution kernel without increasing the number of parameters, thereby better capturing the correlation information of carbon emissions between different regions. In the feature compression stage, multi-scale information is retained through spatial pyramid pooling. Spatial pyramid pooling fuses features of different scales, allowing the network to learn local and global feature information simultaneously. At the same time, a channel attention mechanism is introduced to adaptively weight the feature map. The channel attention mechanism calculates the importance of each channel and assigns different weights to different channels, allowing the network to pay more attention to channel information that is important for green attribute analysis.

[0069] The second stream uses a temporal attention network to extract the dynamic evolution features of the supply chain state feature sequence. The temporal attention network can automatically adjust the degree of attention to features at different times according to the importance of features at different times, so as to better capture the dynamic changes of the supply chain state. Finally, through a cross-modal graph neural network, the spatial correlation features and dynamic evolution features are fused to generate a comprehensive green attribute dataset including raw material traceability, production process energy consumption, and transportation route optimization. The cross-modal graph neural network can effectively fuse data features from different modalities, organically combine the spatial features of the carbon footprint distribution map with the temporal features of the supply chain state feature sequence, and provide comprehensive and accurate data support for subsequent green product certification and traceability.

[0070] Example 2:

[0071] When hierarchically encrypting green attribute data, the data is divided into a basic layer, a process layer, and a result layer according to business types. For the basic layer data, a one-way hash function is used to generate a digest value. The one-way hash function is irreversible. The basic layer data is input into the one-way hash function , and the generated digest value can uniquely identify the basic layer data and cannot restore the original data through the digest value, ensuring the security of the data.

[0072] For the process layer data, a chained hash structure is used to ensure data continuity. The chained hash structure is implemented through timestamp binding, constructing a hash verification index table to record the dual verification information of the location hash and timestamp hash of the data block. Assume that the data blocks are in chronological order as , for the data block , its location hash is , and the timestamp hash is , where is the generation time of the data block . In the hash verification index table, record the [[ID=……]]<000……> information of each data block. When it is necessary to verify the continuity of the data, by comparing the hash values and timestamp hash values of adjacent data blocks, if is calculated based on and , and , it can prove the continuity of the data in terms of time and location, effectively preventing the data from being tampered with or deleted.

[0073] For the result layer data, a Merkle tree structure is introduced to achieve batch verification. The Merkle tree organizes the result layer data into a tree structure, where each leaf node stores the hash value of the data block, and the non-leaf nodes are obtained by calculating the hash values of their child nodes. Through this structure, the integrity of a large amount of data can be quickly verified, improving the verification efficiency.

[0074] Meanwhile, a dynamic key management mechanism is constructed. Differentiated encryption keys are generated according to the data levels. Assume that the base layer key is , the process layer key is , and the result layer key is . These keys are split into multiple fragments through the key sharding technology and stored distributively on different nodes. For example, the key is split into fragments , and are stored separately in different nodes. Only when these nodes cooperate together can the complete key be restored, further enhancing the data security.

[0075] Example 3:

[0076] The multi - rule verification model adopts a multi - channel parallel decision - making architecture. First, a rule graph is constructed. The nodes in the rule graph include national environmental protection standard nodes, industry certification specification nodes, and enterprise - defined rule nodes. Each node has attributes. Among them, the threshold range is used to judge whether the product meets the corresponding rules. For example, in the national environmental protection standard node, the carbon emission threshold for the product is specified as . If the actual carbon emission of the product exceeds , it does not meet the standard; the effective time specifies the valid period of the rule, represented by and indicating the start effective time and end effective time of the rule respectively; the weight coefficient reflects the importance of the rule in the comprehensive evaluation, represented by .

[0077] A two - level attention mechanism is adopted to screen the rules. The first level screens the associated rules through the calculation of rule matching degree. Assume that the green attribute data of the product to be verified is . For each rule node , calculate its matching degree with the data , which can be calculated by methods such as cosine similarity. The rules with a matching degree higher than a certain threshold are screened out and enter the next - level verification. The second level filters out the expired rules through the timeliness verification. The timeliness verification is realized through a sliding time window. An effective time interval and a priority label are set for each rule. A time - window sliding model is constructed to dynamically intercept the current valid rule subset. For example, the current time is , the time - window size is , then the current valid rule subset is the rules that satisfy The rule set. At the same time, an expired rule recycling mechanism is adopted to move the rules that exceed the time window into the historical rule library and freeze their verification weights to avoid interference of expired rules on the authentication results.

[0078] Integrate the verification results of each channel based on the weighted voting mechanism. For the selected valid rules, according to their weight coefficients perform weighted voting. Suppose there are valid rules in total, and the verification result of each rule is ( taking values of passed or not passed), then the comprehensive verification result is calculated as: . Finally, output a green certification report including compliance scores and anomaly marks. The compliance scores are calculated based on the weighted voting results and other relevant indicators, and the anomaly marks are used to indicate in which aspects the product does not meet the rule requirements.

[0079] Embodiment 4:

[0080] This embodiment focuses on elaborating the specific implementation process of the zero-knowledge proof technology in the traceability query module.

[0081] The zero-knowledge proof technology is realized by combining ring signature and homomorphic encryption. First, encode the product green attribute data into a set of verifiable claims.

[0082] Then, use the ring signature algorithm to hide the user identity information. The ring signature algorithm allows the signer to hide their true identity in a set of users. Suppose the set of users is , the signer is , and the ring signature algorithm will generate a signature such that the verifier cannot determine which user in the set the signer is from the signature, but can verify the legality of the signature.

[0083] Next, perform homomorphic encryption processing on the claim set to support logical operation verification in the ciphertext state. The homomorphic encryption algorithm can perform specific operations on the ciphertext without decrypting the data. For example, for two green attribute data and , their ciphertexts are and respectively, and an addition operation can be performed in the ciphertext state, where Represents the addition operation defined by the homomorphic encryption algorithm. When a user issues a query request, the system uses these encryption and signature technologies to obtain relevant data from the blockchain while protecting the user's privacy, and generates a verifiable traceability chain for the green attributes of the product through methods such as ciphertext verification. Users can verify and trace the green attributes of the product based on the information in the traceability chain.

[0084] Example 5:

[0085] This example details the specific implementation process of the multi-objective authentication parameter optimization model and the dynamic sensitivity analysis algorithm in the dynamic optimization module.

[0086] The multi-objective authentication parameter optimization model adopts a decomposition and coordination strategy. First, the authentication standard parameters are modeled as a multi-constraint non-linear optimization problem. The decision variables include the carbon emission cap , the proportion of recycled materials , and the energy consumption efficiency coefficient , etc. Taking the carbon emission cap as an example, it is restricted by various factors such as national environmental protection policies and industry development trends. At the same time, there are also some constraint conditions, such as the restriction of the production process on the energy consumption efficiency coefficient, and the restriction of raw material supply on the proportion of recycled materials.

[0087] Initialize the relaxed sub-problem and calculate the approximate Pareto front solution. In the initialization stage, set some initial relaxation variables. For example, for the constraint of the carbon emission cap, set the relaxation variable , so that the constraint condition becomes . Through some optimization algorithms, such as genetic algorithms, calculate the approximate Pareto front solution under different objective function weights. The approximate Pareto front solution represents the set of optimal solutions that can be achieved when weighing multiple objectives.

[0088] Adopt the dynamic sensitivity analysis algorithm to adjust the objective function weights according to policy changes. The dynamic sensitivity analysis algorithm is implemented based on an online learning framework. First, collect the objective function gradients and constraint relaxation degrees in the historical optimization process as training samples. Update the model parameters online through the incremental Kalman filter algorithm, and adjust the sensitivity threshold of the authentication standard in real time. The incremental Kalman filter algorithm can continuously update the model parameters according to new sample data, enabling the model to better adapt to policy changes, thereby dynamically adjusting the sensitivity threshold of the authentication standard to ensure the rationality and effectiveness of the authentication standard.

[0089] In the decomposition phase, the global problem is divided into multiple single-objective subproblems using goal decomposition techniques. For example, the optimization problem for green product certification standards can be decomposed into subproblems aimed at minimizing carbon emissions, maximizing the proportion of recycled materials, and improving energy efficiency. In the coordination phase, a shadow price mechanism is introduced to balance resource conflicts among the subproblems. By assigning a shadow price to each subproblem, reflecting the scarcity of resources in different subproblems, the shadow price mechanism effectively allocates resources and achieves the overall optimization goal.

[0090] Example 6:

[0091] This embodiment provides supplementary explanations for other aspects related to the system that are not described in detail in the above embodiments.

[0092] In terms of the system's overall architecture, each module interacts with other modules through efficient interfaces. For example, after the data acquisition module collects green attribute data, it transmits this data to the blockchain evidence storage module through a secure and stable interface for encrypted storage. This data stored in the blockchain evidence storage module can be accessed by the smart contract audit module and the traceability query module. The interface design between these modules ensures the accuracy and timeliness of data transmission, while also implementing security measures such as encrypted transmission to prevent data theft or tampering during transmission.

[0093] Regarding system deployment, a distributed approach can be adopted, with each module deployed on different server nodes to improve system performance and reliability. For example, the blockchain evidence storage module can be deployed on multiple blockchain nodes, leveraging the distributed nature of blockchain to achieve redundant data storage and high availability. The smart contract audit module can be deployed on multiple servers with high computing power, depending on the volume of business, to improve audit efficiency. The traceability query module can be deployed on edge servers close to users to reduce response time for user queries.

[0094] A comprehensive version management mechanism has been established to maintain and update the system. This allows for timely system upgrades when new technologies or policy changes are introduced. For example, when new environmental standards are introduced, the rule map in the smart contract review module can be updated promptly, and the multi-objective authentication parameter optimization model can adjust decision variables and constraints based on the new policy requirements. Furthermore, the system's operational status is monitored in real time, collecting operational data such as processing time and resource consumption of each module to identify potential issues and implement optimizations.

[0095] In addition, the system also considers the compatibility with other relevant systems. For example, in the data collection stage, the data collected by multi-source sensing devices may need to be docked with the enterprise's existing production management system. Therefore, a unified data format and interface specification are designed to ensure the smooth interaction of data. In terms of traceability query, the verifiable traceability chain generated by the system can be docked with the query platform of government regulatory departments, facilitating the supervision and management of green products by regulatory departments. Through these measures, the practicality and scalability of the system in actual applications are ensured, and the needs of green product certification and traceability can be better met.

[0096] It should be noted that in this document, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0097] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A blockchain-based green product certification and traceability system, characterized in that Including: Data acquisition module: used to collect green attribute data of the product throughout its life cycle through multi-source sensing devices; Blockchain evidence storage module: based on a hybrid hashing algorithm, hierarchically encrypt the green attribute data to generate an immutable distributed ledger record; Smart contract audit module: input the distributed ledger record into a preset multi-rule verification model to automatically generate a green certification result; Traceability query module: use zero-knowledge proof technology to construct a privacy protection query channel and output a verifiable traceability chain of the product's green attributes according to user requests; Dynamic optimization module: construct a multi-objective certification parameter optimization model according to changes in environmental policies, and use gradient descent and constraint relaxation algorithms to adaptively adjust the certification standards: The multi-objective certification parameter optimization model adopts a decomposition and coordination strategy, including: Model the certification standard parameters as a multi-constraint non-linear optimization problem, and the decision variables include the carbon emission upper limit, the proportion of recycled materials, and the energy consumption efficiency coefficient; Initialize the relaxed sub-problem and calculate the approximate Pareto front solution, and use the dynamic sensitivity analysis algorithm to adjust the objective function weights according to policy changes; In the decomposition stage, divide the global problem into multiple single-objective sub-problems based on the objective decomposition technology; in the coordination stage, introduce the shadow price mechanism to balance the resource conflicts between sub-problems; The zero-knowledge proof technology is realized by combining ring signature and homomorphic encryption, including: Encode the product green attribute data into a set of verifiable claims; use the ring signature algorithm to hide the user's identity information; perform homomorphic encryption processing on the claim set to support logical operation verification in the ciphertext state.

2. The green product certification and traceability system according to claim 1, wherein, The collection of the green attribute data of the product throughout its life cycle through multi-source sensing devices includes: The multi-source sensing devices include a carbon emission monitor, an energy consumption sensor, a logistics tracking tag, and an environmental humidity sensor; Perform spatio-temporal alignment on the carbon emission monitor data and the energy consumption sensor data to construct a product carbon footprint distribution map; perform multi-scale feature decomposition on the logistics tracking tag data and the environmental humidity sensor data to generate a supply chain state feature sequence; Construct a two-stream feature fusion network. The first stream uses a three-dimensional convolutional network to extract the spatial correlation features of the carbon footprint distribution map, and the second stream uses a temporal attention network to extract the dynamic evolution features of the supply chain state feature sequence; Fuse the spatial correlation features and the dynamic evolution features through a cross-modal graph neural network to generate a comprehensive green attribute dataset including raw material traceability, production process energy consumption, and transportation path optimization.

3. The green product certification and traceability system according to claim 1, characterized in that, The hierarchical encryption of the green attribute data based on the hybrid hashing algorithm includes: Divide the green attribute data into a basic layer, a process layer, and a result layer according to business types; use a one-way hashing function to generate a digest value for the basic layer data; use a chained hashing structure for the process layer data to ensure data continuity; introduce a Merkle tree structure for the result layer data to achieve batch verification; Construct a dynamic key management mechanism, generate differentiated encryption keys according to data levels, and achieve distributed storage through key sharding technology.

4. The green product certification and traceability system according to claim 1, characterized in that The multi-rule verification model adopts a multi-channel parallel decision-making architecture, including: Construct a rule graph, where the nodes include national environmental protection standard nodes, industry certification specification nodes, and enterprise-defined rule nodes, and the node attributes include threshold range, effective time, and weight coefficient; Adopt a two-level attention mechanism. The first level filters associated rules through rule matching degree calculation, and the second level filters expired rules through timeliness verification; Integrate the verification results of each channel based on a weighted voting mechanism, and output a green certification report containing compliance scores and anomaly marks.

5. The green product certification and traceability system according to claim 2, characterized in that, The three-dimensional convolutional network adopts a multi-granularity pooling structure, including: Divide the carbon footprint distribution map into multiple levels of grids according to geographical regions, and each level of grid stores carbon emission intensity and time accumulation; In the feature extraction stage, dilated convolutional kernels are used to capture cross-regional associations, and in the feature compression stage, spatial pyramid pooling is used to retain multi-scale information; Introduce a channel attention mechanism to adaptively weight the feature maps.

6. The green product certification and traceability system according to claim 3, wherein The chained hash structure is realized through timestamp binding, including: Construct a hash verification index table to record the dual verification information of the location hash and timestamp hash of the data block.

7. The green product certification and traceability system according to claim 4, characterized in that, The timeliness verification is realized through a sliding time window, including: Set the effective time interval and priority label for each rule; construct a time window sliding model to dynamically intercept the currently valid rule subset; Adopt an expired rule recycling mechanism to move the rules beyond the time window into the historical rule library and freeze their verification weights.

8. The green product certification and traceability system according to claim 1, characterized in that, The dynamic sensitivity analysis algorithm is realized based on an online learning framework, including: Collect the objective function gradient and constraint relaxation degree in the historical optimization process as training samples; Construct a Gaussian process regression model to fit the non-linear relationship between the weight coefficient and external policy variables; Online update the model parameters through an incremental Kalman filter algorithm to adjust the sensitivity threshold of the certification standard in real time.

Images