A Multi-Enterprise Carbon Emission Monitoring Method and System for Industrial Parks Based on Blockchain

Abstract

The present invention relates to a method and system for monitoring carbon emissions of multiple enterprises in an industrial park based on blockchain, belonging to the field of carbon emission monitoring, including: applied to the carbon emission management pipeline, including a blockchain retrieval module, a hierarchical tree construction module, and a carbon emission prediction module; obtaining the distribution topology of multiple enterprises in the industrial park and constructing a graph neural network topology of the industrial park; deploying the modules on nodes to configure multiple carbon emission prediction functions, generating a scatter carbon emission prediction model for the industrial park to perform carbon emission monitoring. By predicting and giving early warnings of abnormal carbon emissions of multiple enterprises in the industrial park for a period of time in the future, forward-looking monitoring of carbon emissions in the industrial park is achieved.

Description

Technical Field

[0001] The present invention relates to the field of carbon emission monitoring, and particularly to a method and system for monitoring carbon emissions of multiple enterprises in an industrial park based on blockchain. Background Art

[0002] Currently, carbon emission monitoring methods rely on real-time data collection and post-event data analysis. Although they can achieve real-time monitoring and abnormal alarm of carbon emissions of enterprises in the park, they cannot predict the carbon emission status in a future period of time. This makes it difficult to formulate effective carbon emission control strategies in advance, and can only passively respond to the carbon emission anomalies that have occurred, greatly reducing the timeliness of carbon emission management measures. Therefore, there is a lag in carbon emission monitoring in the prior art, and there is a technical problem that the carbon emission status in the future time domain cannot be predicted. Summary of the Invention

[0003] Aiming at the technical problem that there is a lag in carbon emission monitoring in the prior art and the carbon emission status in the future time domain cannot be predicted, the present invention provides a method and system for monitoring carbon emissions of multiple enterprises in an industrial park based on blockchain to solve this problem.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] In the first aspect, the present invention provides a method for monitoring carbon emissions of multiple enterprises in an industrial park based on blockchain, including: obtaining the distribution topology of multiple enterprises in the industrial park, constructing the industrial park graph neural network topology, where the industrial park graph neural network topology has the same structure as the distribution topology of multiple enterprises in the industrial park, and any enterprise distribution area is abstracted as a node of the industrial park graph neural network topology; deploying a blockchain retrieval module, a hierarchical tree construction module, and a carbon emission prediction module on multiple nodes of the industrial park graph neural network topology, configuring multiple carbon emission prediction functions, and generating an industrial park scatter carbon emission prediction model; constructing an industrial park carbon emission fitting rule, fully connecting it with the output layer of the industrial park scatter carbon emission prediction model, obtaining an industrial park carbon emission prediction model, and performing carbon emission monitoring.

[0006] In the second aspect, the present invention provides a system for monitoring carbon emissions of multiple enterprises in an industrial park based on blockchain, including: a topology construction unit for obtaining the distribution topology of multiple enterprises in the industrial park and constructing the industrial park graph neural network topology, where the industrial park graph neural network topology has the same structure as the distribution topology of multiple enterprises in the industrial park, and any enterprise distribution area is abstracted as a node of the industrial park graph neural network topology; a scatter prediction deployment unit for deploying a blockchain retrieval module, a hierarchical tree construction module, and a carbon emission prediction module on multiple nodes of the industrial park graph neural network topology, configuring multiple carbon emission prediction functions, generating an industrial park scatter carbon emission prediction model, and performing carbon emission monitoring.

[0007] The beneficial effects of the present invention are as follows:

[0008] Obtain the distribution topology of multiple enterprises in the industrial park, construct the graph neural network topology of the industrial park, and map the actual enterprise distribution in the industrial park to the graph neural network, laying a topological foundation for subsequent carbon emission prediction; deploy the blockchain retrieval module, hierarchical tree construction module, and carbon emission prediction module on multiple nodes of the industrial park graph neural network topology, configure multiple carbon emission prediction functions, and generate a scattered carbon emission prediction model for the industrial park. By distributing each functional module on multiple nodes and configuring multiple prediction functions, multi-point prediction of the park's carbon emissions is achieved, improving the flexibility and scalability of the prediction; construct the fitting rule for the carbon emissions of the industrial park, fully connect it to the output layer of the scattered carbon emission prediction model of the industrial park, obtain the carbon emission prediction model of the industrial park, comprehensively consider the carbon emission prediction results of each node for carbon emission monitoring, improve the advance nature, and can timely discover and identify future periods that may have abnormal carbon emissions, providing a warning basis for park managers and achieving the technical effect of forward-looking monitoring of carbon emissions in the industrial park. Description of the Drawings

[0009] Figure 1 It is a schematic flowchart of a method for monitoring carbon emissions of multiple enterprises in an industrial park based on blockchain provided by the present invention;

[0010] Figure 2 It is a schematic structural diagram of a system for monitoring carbon emissions of multiple enterprises in an industrial park based on blockchain provided by the present invention;

[0011] Figure 3 It is a schematic structural diagram of an electronic device provided by the present invention;

[0012] Figure 4 It is a schematic structural diagram of a computer-readable storage medium provided by the present invention.

[0013] In the drawings, the list of components represented by each reference numeral is as follows:

[0014] Topology construction unit 11, scattered point prediction deployment unit 12, output layer full connection unit 13, electronic device 200, memory 210, processor 220, computer program 211, computer-readable storage medium 300. Detailed Embodiments

[0015] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the 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 skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0016] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0017] In the description of the present invention, the term "for example" is used to mean "serving as an example, illustration, or explanation". Any embodiment described as "for example" in the present invention is not necessarily construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are set forth for purposes of explanation. It should be understood that those of ordinary skill in the art can recognize that the present invention can be implemented without the use of these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed in the present invention.

[0018] Embodiment 1:

[0019] As Figure 1 shown, an embodiment of the present invention provides a multi-enterprise carbon emission monitoring method for industrial parks based on blockchain, which is applied to a carbon emission management pipeline. The carbon emission management pipeline includes a blockchain retrieval module, a hierarchical tree construction module, and a carbon emission prediction module.

[0020] Specifically, a multi-enterprise carbon emission monitoring method for industrial parks based on blockchain is applied to a carbon emission management pipeline. The carbon emission management pipeline includes a blockchain retrieval module, a hierarchical tree construction module, and a carbon emission prediction module. Among them, the blockchain retrieval module is used to retrieve and obtain carbon emission-related data of multiple enterprises in the industrial park from the blockchain network, including but not limited to the production scale, production duration, carbon emission monitoring values, etc. of the enterprises, to ensure the reliability and immutability of the data through blockchain technology; the hierarchical tree construction module is used to construct a hierarchical tree model for carbon emission prediction according to the data obtained by the blockchain retrieval module, organize the carbon emission-related data into a tree-like hierarchical structure, and provide a data basis for subsequent carbon emission prediction; the carbon emission prediction module then uses the hierarchical tree obtained by the hierarchical tree construction module to predict the carbon emissions of the industrial park in a future period of time, and provide a basis for realizing the early warning and control of carbon emissions.

[0021] The multi-enterprise carbon emission monitoring method for industrial parks includes:

[0022] S1: Obtain the multi-enterprise distribution topology of the industrial park, and construct the graph neural network topology of the industrial park. Among them, the graph neural network topology of the industrial park has the same structure as the multi-enterprise distribution topology of the industrial park, and any enterprise distribution area is abstracted as a node of the graph neural network topology of the industrial park.

[0023] Specifically, first, obtain the multi-enterprise distribution topology of the industrial park. The industrial park consists of multiple enterprises, and the geographical distribution and mutual relationship of these enterprises within the park can be abstracted into a topological structure, that is, the multi-enterprise distribution topology of the industrial park, which reflects the spatial location relationship and business interaction relationship among the enterprises in the industrial park, and can comprehensively understand the internal composition structure of the industrial park and the mutual influence among enterprises. After obtaining the multi-enterprise distribution topology of the industrial park, according to the multi-enterprise distribution topology of the industrial park, construct a corresponding graph neural network topology of the industrial park. Specifically, abstract the area where each enterprise in the industrial park is located as a node in the graph neural network topology, and the relationship between multiple enterprise areas corresponds to the edge between nodes. For example, if two enterprises are adjacent geographically or have frequent business cooperation, a connecting edge can be established between the nodes corresponding to these two enterprises, thus obtaining a graph neural network topology structure corresponding to the multi-enterprise distribution topology structure of the industrial park.

[0024] By constructing the graph neural network topology, it is beneficial to the construction and deployment of the carbon emission prediction model, and at the same time can well represent the topological relationship among enterprises. It not only considers the geographical location factors of enterprises, but also considers the business relevance among enterprises, enabling the prediction to more comprehensively and accurately depict the mutual influence among enterprises. At the same time, the distributed characteristics of the graph neural network are compatible with the decentralized characteristics of the blockchain. Each node can correspond to a computing node in the blockchain network, which is convenient for deploying the prediction model into the blockchain network, improving scalability and robustness.

[0025] S2: Deploy the blockchain retrieval module, the hierarchical tree construction module, and the carbon emission prediction module to multiple nodes of the graph neural network topology of the industrial park, configure multiple carbon emission prediction functions, and generate a scatter carbon emission prediction model for the industrial park.

[0026] Specifically, each node in the graph neural network topology of the industrial park corresponds to the area where an enterprise is located. By deploying a blockchain retrieval module on these nodes, the distributed storage and retrieval of data related to the carbon emissions of the enterprise are realized, ensuring the security, reliability, and immutability of the data. At the same time, a hierarchical tree construction module and a carbon emission prediction module are also deployed on each node. The hierarchical tree construction module is used to organize the local carbon emission data of the node into a tree-like hierarchical structure, providing data support for carbon emission prediction; the carbon emission prediction module predicts the carbon emissions of the enterprise corresponding to the node based on the data in the hierarchical tree. By configuring an independent carbon emission prediction function on each node, a distributed scatter carbon emission prediction model for the industrial park can be obtained, making full use of the structural characteristics of the graph neural network topology to realize the distributed calculation and prediction of the carbon emissions of the entire industrial park. Among them, each node can select a suitable machine learning algorithm and model parameters according to its own computing power and data situation, so as to achieve more accurate and efficient prediction.

[0027] By deploying a blockchain retrieval module, a hierarchical tree construction module, and a carbon emission prediction module on multiple nodes of the graph neural network topology of the industrial park, and configuring the corresponding carbon emission prediction function, a distributed scatter carbon emission prediction model for the industrial park is constructed, which not only considers the needs of data security and distributed computing, but also can make full use of the topological structure characteristics of the graph neural network, laying a solid foundation for realizing the accurate monitoring and early warning of the carbon emissions of the industrial park.

[0028] S3: Construct the carbon emission fitting rules for the industrial park, fully connect with the output layer of the scatter carbon emission prediction model of the industrial park, and obtain the industrial park carbon emission prediction model to perform carbon emission monitoring.

[0029] Specifically, the scatter carbon emission prediction model of the industrial park can obtain the predicted values of the carbon emissions of each enterprise in the park through distributed computing on each node of the graph neural network topology. However, these predicted values are relatively independent and scattered, and further integration and correction are required to obtain the predicted value of the total carbon emissions of the entire industrial park. For this reason, carbon emission fitting rules are introduced. These rules comprehensively consider the mutual influence and restriction relationships among enterprises in the industrial park, as well as the overall carbon emission control objectives of the park, and comprehensively process the scatter prediction results of each scatter carbon emission prediction model of the industrial park, so as to obtain the industrial park carbon emission prediction model.

[0030] When constructing the carbon emission fitting rules, fully absorb the knowledge and experience of domain experts, design corresponding parameters and thresholds to balance the prediction accuracy and calculation efficiency. At the same time, consider the interpretability and adaptability of the rules to ensure that the prediction results can provide effective guidance and reference for park managers and decision-makers. Subsequently, fully connect the carbon emission fitting rules with the output layer of the scatter carbon emission prediction model of each industrial park to obtain the industrial park carbon emission prediction model.

[0031] By constructing the industrial park carbon emission prediction model, make full use of the distributed computing power of the graph neural network to obtain the carbon emission prediction values of each enterprise in the park, and integrate and optimize the prediction results through the carbon emission fitting rules to improve the accuracy and reliability of the prediction, laying a model foundation for the precise monitoring, early warning and control of industrial park carbon emissions.

[0032] Furthermore, the embodiments of the present application further include the steps:

[0033] S4: Obtain the production scales and production durations of multiple enterprises in a specified future time zone of a multi-enterprise distribution area.

[0034] Specifically, in order to predict the carbon emissions of an industrial park, first understand the production plans and arrangements of each enterprise in the industrial park in a future period of time. For example, it can be obtained by interacting with the park management department and the enterprise management system, or estimated by analyzing historical data and other methods. When obtaining the data, clearly specify the future time zone for prediction, that is, which future period of time the carbon emissions need to be predicted. For example, select a time span such as the next month, quarter or year.

[0035] For each enterprise in the industrial park, obtain the production scale and production duration data in the specified future time zone to obtain multiple enterprise production scales and multiple enterprise production durations. Among them, the production scale data reflects the quantity or output value of products planned to be produced by the enterprise, and the production duration data reflects the length of time the enterprise plans to invest in production.

[0036] By obtaining the production scales and production durations of multiple enterprises in the industrial park in the specified future time zone, it provides key input information for subsequent carbon emission predictions and a data basis for the refined monitoring of industrial park carbon emissions.

[0037] S5: Input the multiple enterprise production scales and the multiple enterprise production durations into multiple nodes of the industrial park carbon emission prediction model to obtain the industrial park carbon emission prediction value.

[0038] Specifically, after obtaining the production scale and production duration data of each enterprise in the industrial park, these data are input into the carbon emission prediction model of the industrial park to predict carbon emissions. Since the constructed carbon emission prediction model of the industrial park is a distributed model based on graph neural network, the data is input to multiple nodes of the model, and through the calculations between the nodes, the predicted value of the carbon emissions of the entire park, i.e., the predicted value of the carbon emissions of the industrial park, is obtained.

[0039] First, the production scale data and production duration data of each enterprise are correspondingly input to the node representing the enterprise in the carbon emission prediction model of the industrial park. Second, on each node, using the pre-configured carbon emission prediction function, combined with the input production scale and production duration of the enterprise, the predicted value of the carbon emissions of the enterprise represented by the node is obtained. Each node can perform calculations simultaneously to improve the prediction efficiency. Then, through the communication and data exchange between the nodes, the predicted values of the carbon emissions calculated by each node are transmitted and aggregated, and integrated and corrected using the constructed carbon emission fitting rule to obtain the predicted value of the total carbon emissions of the entire industrial park in the specified future time zone, i.e., the predicted value of the carbon emissions of the industrial park.

[0040] By inputting the production scale and production duration data of each enterprise into multiple nodes of the carbon emission prediction model of the industrial park, and through distributed calculations and data aggregation, the predicted value of the carbon emissions of the industrial park in the specified future time zone is obtained, providing a forward-looking prediction for the carbon emissions of the park.

[0041] S6: When the predicted value of the carbon emissions of the industrial park is greater than or equal to the carbon emission threshold, mark the specified future time zone as an abnormal production identification time zone.

[0042] Specifically, in order to achieve early warning of the carbon emissions of the industrial park, a carbon emission threshold is set as the standard for judging whether abnormal situations occur. First, according to factors such as the carbon emission management target and historical data statistics of the industrial park, the carbon emission threshold is set. This carbon emission threshold is a fixed value of carbon emissions. When the predicted value of the carbon emissions of the industrial park exceeds this threshold, it is considered that abnormal carbon emission situations may occur in the corresponding specified future time zone of the industrial park. Second, compare the predicted value of the carbon emissions of the industrial park in the obtained specified future time zone with the set value of carbon emissions. If the predicted value of the carbon emissions of the industrial park is greater than or equal to the carbon emission threshold, mark this time zone as an abnormal production identification time zone, indicating that in this time zone, the carbon emissions of the industrial park will exceed the normal level and corresponding countermeasures need to be taken.

[0043] Through carbon emission judgment and identification, abnormal carbon emission situations that may occur in the industrial park in a future period can be discovered in advance, and corresponding countermeasures can be formulated accordingly, so as to achieve early warning of the carbon emissions of the park and realize forward-looking control.

[0044] S7: Add the abnormal production identification time zone, industrial park number, and the predicted carbon emissions value of the industrial park to the carbon emissions monitoring results and send them to the carbon emissions monitoring and management terminal for multiple enterprises in the industrial park.

[0045] Specifically, when it is identified that abnormal carbon emissions may occur in a specified future time zone, this early warning information is transmitted to the carbon emissions monitoring and management terminal for multiple enterprises in the industrial park in a timely and accurate manner so that corresponding countermeasures can be taken.

[0046] First, integrate information such as the abnormal production identification time zone, industrial park number, and the obtained predicted carbon emissions value of the industrial park to obtain the carbon emissions monitoring results. Among them, the industrial park number is used to uniquely identify and locate the park, the abnormal production identification time zone indicates the time period when the predicted abnormal carbon emissions occur, and the predicted carbon emissions value of the industrial park gives the carbon emissions result of the park in the abnormal production identification time zone. Subsequently, send the carbon emissions monitoring results to the carbon emissions monitoring and management terminal for multiple enterprises in the industrial park. Among them, the carbon emissions monitoring and management terminal for multiple enterprises in the industrial park is a centralized management platform or system, or can also be multiple endpoints distributed in each enterprise and park management department. In this way, managers can obtain the carbon emissions monitoring results of the park in a timely manner, and based on the carbon emissions monitoring results, take countermeasures in advance, adjust the production and operation plan, so as to control it before the abnormal carbon emissions occur and achieve precise control of the carbon emissions in the park.

[0047] By transmitting the carbon emissions monitoring early warning information to the carbon emissions monitoring and management terminal for multiple enterprises in the industrial park in a timely and accurate manner, forward-looking control of carbon emissions is realized, and the pertinence and effectiveness of carbon emissions management are improved.

[0048] Furthermore, the embodiment of the present application further includes:

[0049] Step S210: Obtain the first node product type of the multiple nodes of the industrial park graph neural network topology;

[0050] Step S220: Initialize the blockchain retrieval module, the hierarchical tree construction module, and the carbon emissions prediction module according to the first node product type to generate a first node carbon emissions prediction function;

[0051] Step S230: Deploy the first node carbon emissions prediction function to the first node.

[0052] In a feasible implementation, first, obtain the product type information of the enterprise corresponding to each node in the graph neural network topology of the industrial park. Since different enterprises in the industrial park produce different types of products, their production processes, energy consumption and carbon emission characteristics are also different, so it is necessary to design and optimize the prediction model according to the product type of each enterprise. By accessing the park management database, enterprise production management system and other channels, obtain the product type data of the enterprise represented by each node. Traverse multiple nodes of the graph neural network topology of the industrial park, take any of them as the first node, and its product type as the first node product type. Then, according to the obtained first node product type, initialize and configure the blockchain retrieval module, hierarchical tree construction module and carbon emission prediction module on the first node. Different types of products have differences in the characteristics, distribution and influencing factors of carbon emission data, so the data retrieval strategy is designed in a targeted manner, the appropriate data hierarchical structure is constructed, and the appropriate machine learning algorithm and model parameters are selected. Through the initialization operation, a carbon emission prediction function suitable for the first node product type is generated, that is, the first node carbon emission prediction function, thereby improving the accuracy and reliability of the prediction.

[0053] Subsequently, the first carbon emission prediction function generated for the first node product type of the first node is deployed to the first node, so that it can calculate and predict the carbon emissions of the enterprise corresponding to the first node in real time based on the local production and operation data and other node data in the blockchain network. Through distributed deployment, the parallel computing capability is used to improve the prediction efficiency; at the same time, different nodes can flexibly adjust and optimize the prediction model according to their own product characteristics and data environment, so as to achieve more refined and differentiated carbon emission management.

[0054] By initializing the prediction module and configuring the function according to the node product type, the carbon emission prediction problem caused by the product differences of enterprises in the industrial park can be effectively solved, and the applicability and accuracy of the prediction results can be improved.

[0055] Furthermore, the embodiment of the present application also includes:

[0056] Step S221: constructing a ternary array label using production scale, production duration and carbon emission monitoring value as retrieval target attributes;

[0057] Step S222: obtaining the training production scale, training production duration and true value of carbon emissions;

[0058] Step S223: Taking the first node product type as a static constraint condition, taking the training production scale and the training production duration as root node dynamic constraints, in the blockchain retrieval module, performing sample growth retrieval on the ternary array label to obtain a carbon emission prediction sample set;

[0059] Step S224: In the hierarchical tree construction module, parse the carbon emission prediction sample set and build a carbon emission prediction hierarchical tree;

[0060] Step S225: In the carbon emission prediction module, recursively aggregate the carbon emission prediction hierarchical tree from the bottom layer to obtain the carbon emission training prediction value;

[0061] Step S226: When the carbon emission training deviation between the carbon emission training prediction value and the carbon emission true value is less than or equal to the training deviation threshold, increment the training convergence frequency by one;

[0062] Step S227: Otherwise, increment the training error frequency by one;

[0063] Step S228: Repeat the training for a certain number of times. When the proportion of the training convergence frequency is greater than or equal to 90%, initialize the blockchain retrieval module, the hierarchical tree construction module, and the carbon emission prediction module to complete, and generate the first node carbon emission prediction function.

[0064] In a preferred embodiment, first, select three attributes closely related to carbon emission prediction, namely production scale, production duration, and carbon emission monitoring value, as retrieval targets, and organize them into a ternary array tag to provide a basis for retrieving relevant data in the blockchain network. At the same time, to train the carbon emission prediction model, obtain a set of known training data, including the true value of the carbon emissions actually monitored under a specific production scale and a specific production duration, as the training production scale, training production duration, and carbon emission true value, to provide a basis for model training. Subsequently, set static constraint conditions according to the first product type of the first node, and use the training production scale and training production duration in the training data as the dynamic constraint conditions of the root node. Use the blockchain retrieval module to perform recursive sample growth retrieval on the ternary array tag, and obtain the historical carbon emission data matching the constraint conditions from the blockchain network to form a carbon emission prediction sample set. Through sample growth retrieval, while ensuring data relevance, expand the quantity and diversity of training samples.

[0065] Subsequently, in the hierarchical tree construction module, the carbon emission prediction sample set is parsed to build a carbon emission prediction hierarchical tree. The carbon emission prediction sample set is stratified according to the differences in production scale and production duration to construct a multi-level tree structure, obtaining the carbon emission prediction hierarchical tree. Each node in the tree contains a set of sample data with similar attributes, providing organized data support for subsequent prediction model training. Then, in the carbon emission prediction module, the carbon emission prediction hierarchical tree is recursively aggregated from the bottom layer to obtain the carbon emission training prediction value. Starting from the leaf nodes, the sample data of each node is aggregated and calculated layer by layer upwards to obtain the corresponding carbon emission prediction value until the prediction result of the root node, that is, the carbon emission training prediction value, is obtained. When the deviation between the carbon emission training prediction value and the true carbon emission value is less than or equal to the training deviation threshold, it is considered that the deviation between the two is within an acceptable range, and this training is considered to have converged, and the training convergence frequency is incremented by one; if the deviation between the carbon emission training prediction value and the true carbon emission value is greater than the training deviation threshold, this training is marked as an error, and the error frequency is accumulated. Through multiple iterative trainings, repeating the training a certain number of times, when the proportion of the training convergence frequency is greater than or equal to 90%, it is considered that the initialization of the blockchain retrieval module, the hierarchical tree construction module, and the carbon emission prediction module has been completed, forming the first carbon emission prediction function for the first node.

[0066] By adaptively initializing and optimizing the carbon emission prediction function of the first node, the accuracy and generalization ability of the prediction model are improved.

[0067] Furthermore, the embodiments of the present application further include:

[0068] Step S2231: Construct a static constraint condition rule factor: If the sample is of the same product type as the first node, it is satisfied; if the sample is of a different product type from the first node, it is not satisfied;

[0069] Step S2232: Construct a root node dynamic constraint condition rule factor:

[0070] The first rule factor: If the production scale deviation between the sample production scale and the training production scale is less than or equal to the production scale deviation threshold, it is satisfied; if the production scale deviation between the sample production scale and the training production scale is greater than the production scale deviation threshold, it is not satisfied;

[0071] The second rule factor: If the production duration deviation between the sample production duration and the training production duration is less than or equal to the production duration deviation threshold, it is satisfied; if the production duration deviation between the sample production duration and the training production duration is greater than the production duration deviation threshold, it is not satisfied;

[0072] When the first rule factor and the second rule factor are both satisfied, the root node dynamic constraint condition rule factor is considered satisfied; otherwise, the root node dynamic constraint condition rule factor is considered not satisfied.

[0073] In a preferred embodiment, first, a rule factor for determining constraint satisfaction based on whether the sample is of the same product type as the first node is defined as the static constraint condition rule factor. When the sample data to be retrieved is of the same product type as the first node, it is considered to satisfy the static constraint condition; otherwise, it is considered not satisfied. Through the static constraint condition rule factor, samples related to the product type of the first node can be quickly screened out during the blockchain data retrieval process, improving the retrieval efficiency and pertinence. At the same time, a root node dynamic constraint condition rule factor is constructed. Different from the static constraint condition, the dynamic constraint condition needs to consider the deviations of the sample data and the training data in two attributes, production scale and production duration. Therefore, two rule factors are defined, namely the first rule factor and the second rule factor.

[0074] Among them, the first rule factor targets the production scale attribute and compares the production scale deviation between the sample data and the training data. When the deviation is less than or equal to the preset production scale deviation threshold, it is considered to satisfy this rule factor; otherwise, it is considered not satisfied. The second rule factor targets the production duration attribute and compares the production duration deviation between the sample data and the training data. When the deviation is less than or equal to the preset production duration deviation threshold, it is considered to satisfy this rule factor; otherwise, it is considered not satisfied. Only when the first rule factor and the second rule factor are both satisfied, the sample data is considered to satisfy the root node dynamic constraint condition and can be used as an effective retrieval result. Otherwise, if any one of the rule factors is not satisfied, the sample data is considered not to satisfy the dynamic constraint condition and is not included in the retrieval result.

[0075] By constructing the above two types of constraint condition rule factors, the refined control of the blockchain data retrieval process is realized. The static constraint condition ensures the consistency of the retrieved samples and the target node in terms of product type, and the root node dynamic constraint condition further limits the similarity between the sample data and the training data in key attributes. Through the multiple constraint mechanism, the quality and relevance of the retrieved data can be effectively improved, reducing the interference of invalid or noisy data, thereby providing high-quality sample support for the subsequent training of the carbon emission prediction model.

[0076] Furthermore, the embodiments of the present application further include:

[0077] Step S2233: Using the first-node product type as a static constraint condition, and the training production scale and the training production duration as root-node dynamic constraint conditions, retrieve the triple-array tags through the blockchain retrieval module, and collect a first quantity of first-level samples that simultaneously meet the static constraint condition and the root-node dynamic constraint conditions;

[0078] Step S2234: Traverse the first quantity of first-level samples, construct first-level node dynamic constraint conditions, and in combination with the static constraint conditions in the blockchain retrieval module, retrieve the triple-array tags to obtain a first quantity of second-level samples;

[0079] Step S2235: Until traversing the (N - 1)-quantity of (N - 1)-level samples, construct (N - 1)-level node dynamic constraint conditions, and in combination with the static constraint conditions in the blockchain retrieval module, retrieve the triple-array tags to obtain a first quantity of N-level samples;

[0080] Step S2236: Add the first quantity of first-level samples, the first quantity of second-level samples until the first quantity of N-level samples into the carbon emission prediction sample set.

[0081] In a preferred implementation, first, using the first-node product type as a static constraint condition, and the training production scale and the training production duration as root-node dynamic constraint conditions, retrieve the blockchain data to obtain sample data that is the same as the first-node product type and whose production scale and production duration deviation are within the threshold range, and obtain a first quantity of first-level samples as the initial data of the carbon emission prediction sample set. Then, traverse the first quantity of first-level samples, use each first-level sample as a new retrieval starting point, construct first-level node dynamic constraint conditions according to its production scale and production duration values, and retrieve the second-level sample data that meets the conditions again to further expand the carbon emission prediction sample set and obtain a first quantity of second-level samples. Repeat the above process until traversing the (N - 1)-quantity of (N - 1)-level samples, construct (N - 1)-level node dynamic constraint conditions, and in combination with the static constraint conditions, retrieve the triple-array tags in the blockchain retrieval module to obtain a first quantity of N-level samples. As the retrieval process recurs, continuously generate the dynamic constraint conditions of the next-level node based on the previous-level samples and obtain the corresponding next-level samples until the preset N iterations are completed. After that, add the first quantity of first-level samples, the first quantity of second-level samples until the first quantity of N-level samples into the carbon emission prediction sample set in sequence, and merge the sample data of each level obtained through multiple iterative retrievals in order to form the carbon emission prediction sample set for subsequent model training and optimization.

[0082] Traditional retrieval methods only use a fixed reference data as the screening criterion. Although the retrieved sample data is relatively close to the reference data, when the number of target samples is small, the representativeness and reliability of the retrieval results are often insufficient. In the application scenario of industrial park carbon emission data, the historical data volume of each enterprise node is generally small, and it is difficult to obtain sufficient and reliable samples in a single retrieval. Therefore, through the sample growth retrieval method, a recursive iterative sample growth retrieval strategy is adopted, using the sample data obtained in each retrieval as the new constraint condition, continuously expanding the retrieval scope and the number of samples until the preset growth layer number is satisfied, effectively expanding the scale and diversity of the sample set while ensuring the sample correlation, thereby improving the accuracy and reliability of carbon emission prediction.

[0083] Furthermore, the embodiments of the present application further include:

[0084] Step S2241: Construct a sample distance evaluation rule factor:

[0085] Step S2242: Construct a two-dimensional coordinate for sample distance evaluation, where the first-dimensional coordinate of the two-dimensional coordinate for sample distance evaluation is the production scale normalization parameter, and the second-dimensional coordinate of the two-dimensional coordinate for sample distance evaluation is the production duration normalization parameter;

[0086] Step S2243: The Euclidean distance of the two-dimensional coordinate for sample distance evaluation is the sample distance evaluation value;

[0087] Step S2244: Based on the training production scale and the training production duration, construct a root node coordinate based on the two-dimensional coordinate for sample distance evaluation;

[0088] Step S2245: Traverse the carbon emission prediction sample set and construct a sample coordinate set based on the two-dimensional coordinate for sample distance evaluation;

[0089] Step S2246: Sort k neighboring samples from near to far based on the root node coordinate to construct first-level leaf nodes, where 10≥k≥5;

[0090] Step S2247: Traverse the first-level leaf nodes and sort k neighboring samples from near to far respectively to construct second-level leaf nodes;

[0091] Step S2248: Until traversing the M-1 level leaf nodes and sorting k neighboring samples from near to far respectively to construct M-level leaf nodes, where 20≥M≥2;

[0092] Step S2249: Construct the carbon emission prediction hierarchical tree according to the root node, the first-level leaf nodes, the second-level leaf nodes until the M-level leaf nodes.

[0093] In a feasible implementation, first, a two-dimensional coordinate system is defined to measure the similarity between sample data. Among them, the first-dimensional coordinate represents the production scale normalization parameter, and the second-dimensional coordinate represents the production duration normalization parameter, forming a two-dimensional coordinate for sample distance evaluation. Then, the Euclidean distance of the sample in this two-dimensional coordinate system is calculated as the sample distance evaluation value. This evaluation system is a rule factor for sample distance evaluation, which can quantitatively describe the differences in the two key attributes of production scale and production duration of sample data, and provide a distance measurement standard for the subsequent hierarchical tree construction.

[0094] Subsequently, taking the training production scale and training production duration as parameters, the root node coordinates are constructed based on the two-dimensional coordinate for sample distance evaluation. The root node coordinates are directly taken from the production scale and production duration of the training data, representing the position of the prediction target in the attribute space. Then, each sample data in the carbon emission prediction sample set is traversed, and its coordinate value in the two-dimensional coordinate system for sample distance evaluation is calculated to form a sample coordinate set. By mapping the sample data into a unified coordinate system, the similarity and difference between different samples can be intuitively compared. Next, with the root node coordinates as the center, k nearest samples are selected from near to far according to the sample distance evaluation value to construct the first-level leaf nodes of the hierarchical tree. Among them, the value of k is between 5 and 10 to balance the branching factor and computational complexity of the tree. By generating a layer of child nodes most similar to it around the root node, the skeleton structure of the hierarchical tree is initially formed to obtain the first-level leaf nodes. Then, based on the formed first-level leaf nodes, each first-level leaf node is used as a new center to recursively construct the second-level leaf nodes. That is, within the neighborhood of each first-level leaf node, k most similar samples are selected again to form new child nodes, further refining the branching structure of the hierarchical tree, so that the similarity relationship between nodes can be described at a finer granularity. Repeat the above process until the construction of the (M - 1)-level leaf nodes is completed to form the complete structure of the hierarchical tree. Among them, the value range of M is from 2 to 20, which can be adjusted according to the specific number and distribution of samples to balance the depth and width of the tree. After that, the root node, the first-level leaf nodes, the second-level leaf nodes until the M-level leaf nodes are connected according to the parent-child relationship to form a complete carbon emission prediction hierarchical tree. This tree structure takes the training data as the root node and is subdivided level by level until the leaf nodes correspond to a set of sample data most similar to the training data. Each level of the tree represents different similarity granularities. The deeper the level, the higher the similarity between nodes and the more accurate the prediction result.

[0095] By using the sample distance evaluation rule factor and the hierarchical refinement method, the carbon emission prediction sample set is organized into a structured hierarchical tree, enabling the similarity relationship of the sample data to be fully explored and utilized, providing an efficient and interpretable data index and calculation framework for subsequent carbon emission prediction. By performing prediction and aggregation at different levels of the tree, the carbon emissions can be estimated at different granularities, improving the prediction accuracy and robustness.

[0096] Furthermore, the embodiments of the present application further include:

[0097] Step S2251: Obtain the M-level leaf nodes of the j-th leaf node of the (M - 1)-level leaf nodes;

[0098] Step S2252: Statistically calculate the central value of the carbon emission monitoring values of the M-level leaf nodes of the j-th leaf node, and store it in the j-th leaf node;

[0099] Step S2253: Based on the above principle, in the carbon emission prediction module, recursively aggregate the carbon emission prediction hierarchical tree from the bottom layer, and obtain the carbon emission training prediction value.

[0100] In a preferred embodiment, first, obtain the set of M-level leaf nodes corresponding to the j-th leaf node of the (M - 1)-level leaf nodes in the hierarchical tree, denoted as the M-level leaf nodes of the j-th leaf node. Here, M represents the total number of levels of the hierarchical tree, and j represents the node number in the (M - 1)th level. Determine the range of the sub-nodes of each (M - 1)-level node at the bottom layer, providing a starting point for subsequent recursive aggregation. Then, for all nodes in the set of M-level leaf nodes of the j-th leaf node, extract the carbon emission monitoring values of their corresponding sample data, calculate the central tendency of these monitoring values, such as the mean, median, or mode, etc., as the representative carbon emission prediction value of this set of M-level leaf nodes, and store it in the j-th leaf node, thereby obtaining the carbon emission estimation value within a local area on the bottom-layer leaf nodes. Repeat the above process, recursively aggregate the hierarchical tree layer by layer from bottom to top until reaching the root node. In the aggregation process of each layer, taking the nodes of the current layer as a unit, extract the locally calculated carbon emission prediction values of its corresponding lower-layer sub-nodes, and summarize them through weighted average or other integration methods to obtain the carbon emission prediction value of the current node, and recursively pass it upward. When recursively aggregated to the root node, the carbon emission training prediction value of the entire hierarchical tree can be obtained and output as the final prediction result.

[0101] Through the bottom-up recursive aggregation strategy, the structured information of the carbon emission prediction hierarchical tree is fully utilized, realizing the carbon emission prediction process from local to global.

[0102] Furthermore, the embodiments of the present application further include:

[0103] Step S310: Configure the carbon emission fitting threshold;

[0104] Step S320: Delete the carbon emission prediction values that are less than or equal to the carbon emission fitting threshold among the multiple carbon emission prediction values output by the industrial park scatter carbon emission prediction model, and obtain the retained carbon emission prediction values;

[0105] Step S330: Sum up the retained carbon emission prediction values to obtain the carbon emission fitting value.

[0106] In a feasible implementation manner, an industrial park carbon emission fitting rule is constructed. Specifically, according to the carbon emission management objectives and actual needs of the industrial park, a carbon emission fitting threshold is preset as the standard for judging whether the carbon emission prediction value is valid. Subsequently, compare the multiple carbon emission prediction values output by the industrial park scatter carbon emission prediction model with the carbon emission fitting threshold, mark the prediction values less than or equal to the threshold as invalid predictions, and delete them from the multiple carbon emission prediction values to obtain the retained carbon emission prediction values. Through the screening operation, abnormal prediction values that may be caused by factors such as data noise and model errors are filtered out, improving the overall quality of the prediction results. Subsequently, sum up the retained carbon emission prediction values obtained by screening to obtain the carbon emission fitting value of the entire industrial park, representing the overall carbon emission level of the park in a future period.

[0107] A multi-enterprise carbon emission monitoring method for industrial parks based on blockchain provided by an embodiment of the present invention has at least the following technical effects:

[0108] Obtain the multi-enterprise distribution topology of the industrial park, construct the industrial park graph neural network topology, and map the actual enterprise distribution of the industrial park to the graph neural network, laying a topological foundation for subsequent carbon emission prediction; deploy the blockchain retrieval module, hierarchical tree construction module, and carbon emission prediction module on multiple nodes of the industrial park graph neural network topology, configure multiple carbon emission prediction functions, and generate an industrial park scatter carbon emission prediction model. By distributing each functional module on multiple nodes and configuring multiple prediction functions, multi-point prediction of the park's carbon emissions is realized, improving the flexibility and scalability of the prediction; construct an industrial park carbon emission fitting rule, fully connect it to the output layer of the industrial park scatter carbon emission prediction model, obtain an industrial park carbon emission prediction model, comprehensively consider the carbon emission prediction results of each node for carbon emission monitoring, improve the advance nature, can timely discover and identify future periods that may have abnormal carbon emissions, provide a warning basis for park managers, and achieve the technical effect of forward-looking monitoring of industrial park carbon emissions.

[0109] Embodiment 2:

[0110] As Figure 2As shown, based on the same inventive concept as the method for monitoring carbon emissions of multiple enterprises in an industrial park based on blockchain provided in Embodiment 1, the embodiment of the present invention further provides a system for monitoring carbon emissions of multiple enterprises in an industrial park, which is applied to a carbon emission management pipeline. The carbon emission management pipeline includes a blockchain retrieval module, a hierarchical tree construction module, and a carbon emission prediction module, and includes:

[0111] A topology construction unit 11, configured to obtain the distribution topology of multiple enterprises in the industrial park and construct the topology of the industrial park graph neural network. Among them, the topology of the industrial park graph neural network is the same as the structure of the distribution topology of multiple enterprises in the industrial park, and any enterprise distribution area is abstracted as a node of the topology of the industrial park graph neural network;

[0112] A scatter prediction deployment unit 12, configured to deploy the blockchain retrieval module, the hierarchical tree construction module, and the carbon emission prediction module to multiple nodes of the topology of the industrial park graph neural network, configure multiple carbon emission prediction functions, and generate a scatter carbon emission prediction model for the industrial park;

[0113] An output layer fully connected unit 13, configured to construct a carbon emission fitting rule for the industrial park, fully connect with the output layer of the scatter carbon emission prediction model for the industrial park, obtain a carbon emission prediction model for the industrial park, and perform carbon emission monitoring.

[0114] Further, the embodiment of the present application includes:

[0115] A production data acquisition unit, configured to obtain the production scales and production durations of multiple enterprises in a specified future time zone of the distribution area of multiple enterprises;

[0116] A carbon emission prediction calculation unit, configured to input the production scales and production durations of multiple enterprises into multiple nodes of the carbon emission prediction model for the industrial park to obtain a carbon emission prediction value for the industrial park;

[0117] An abnormal identification unit, configured to perform abnormal production identification for the specified future time zone when the carbon emission prediction value for the industrial park is greater than or equal to the carbon emission threshold;

[0118] A monitoring result sending unit, configured to add the abnormal production identification time zone, the industrial park number, and the carbon emission prediction value for the industrial park into the carbon emission monitoring result and send it to the carbon emission monitoring management end of multiple enterprises in the industrial park.

[0119] Further, the scatter prediction deployment unit 12 includes the following execution steps:

[0120] Obtain the first node product type of the multiple nodes of the topology of the industrial park graph neural network;

[0121] Initialize the blockchain retrieval module, the hierarchical tree construction module, and the carbon emission prediction module according to the first node product type to generate a first node carbon emission prediction function;

[0122] Deploy the first node carbon emission prediction function to the first node.

[0123] Furthermore, the scatter prediction deployment unit 12 further includes the following execution steps:

[0124] Construct a ternary array label with production scale, production duration, and carbon emission monitoring value as retrieval target attributes;

[0125] Obtain the training production scale, training production duration, and carbon emission true value;

[0126] Using the first node product type as a static constraint condition and the training production scale and the training production duration as root node dynamic constraint conditions, perform sample growth retrieval on the ternary array label in the blockchain retrieval module to obtain a carbon emission prediction sample set;

[0127] In the hierarchical tree construction module, parse the carbon emission prediction sample set and build a carbon emission prediction hierarchical tree;

[0128] In the carbon emission prediction module, recursively aggregate the carbon emission prediction hierarchical tree from the bottom layer to obtain a carbon emission training prediction value;

[0129] When the carbon emission training deviation between the carbon emission training prediction value and the carbon emission true value is less than or equal to the training deviation threshold, increment the training convergence frequency by one;

[0130] Otherwise, increment the training error frequency by one;

[0131] Repeat the training for a certain number of times. When the proportion of the training convergence frequency is greater than or equal to 90%, the initialization of the blockchain retrieval module, the hierarchical tree construction module, and the carbon emission prediction module is completed, and the first node carbon emission prediction function is generated.

[0132] Furthermore, the scatter prediction deployment unit 12 further includes the following execution steps:

[0133] Construct a static constraint condition rule factor: if the sample is the same as the first node product type, it is satisfied; if the sample is different from the first node product type, it is not satisfied;

[0134] Construct a root node dynamic constraint condition rule factor:

[0135] The first rule factor: If the production scale deviation between the sample production scale and the training production scale is less than or equal to the production scale deviation threshold, it is satisfied; if the production scale deviation between the sample production scale and the training production scale is greater than the production scale deviation threshold, it is not satisfied.

[0136] The second rule factor: If the production duration deviation between the sample production duration and the training production duration is less than or equal to the production duration deviation threshold, it is satisfied; if the production duration deviation between the sample production duration and the training production duration is greater than the production duration deviation threshold, it is not satisfied.

[0137] If both the first rule factor and the second rule factor are satisfied, the root node dynamic constraint condition rule factor is considered satisfied; otherwise, the root node dynamic constraint condition rule factor is considered not satisfied.

[0138] Furthermore, the scatter prediction deployment unit 12 further includes the following execution steps:

[0139] Taking the first node product type as the static constraint condition and the training production scale and the training production duration as the root node dynamic constraint conditions, retrieve the triple array tags through the blockchain retrieval module, and collect the first quantity of first-level samples that simultaneously satisfy the static constraint condition and the root node dynamic constraint conditions.

[0140] Traverse the first quantity of first-level samples, construct the first-level node dynamic constraint conditions, and in combination with the static constraint conditions in the blockchain retrieval module, retrieve the triple array tags to obtain the first quantity of second-level samples.

[0141] Until traversing the (N - 1)-th quantity of (N - 1)-th level samples, construct the (N - 1)-th level node dynamic constraint conditions, and in combination with the static constraint conditions in the blockchain retrieval module, retrieve the triple array tags to obtain the first quantity of N-level samples.

[0142] Add the first quantity of first-level samples, the first quantity of second-level samples until the first quantity of N-level samples into the carbon emission prediction sample set.

[0143] Furthermore, the scatter prediction deployment unit 12 further includes the following execution steps:

[0144] Construct a sample distance evaluation rule factor:

[0145] Construct a sample distance evaluation two-dimensional coordinate, where the first-dimensional coordinate of the sample distance evaluation two-dimensional coordinate is the production scale normalization parameter, and the second-dimensional coordinate of the sample distance evaluation two-dimensional coordinate is the production duration normalization parameter.

[0146] The Euclidean distance of the sample distance evaluation two-dimensional coordinate is the sample distance evaluation value.

[0147] Based on the training production scale and the training production duration, construct the root node coordinates based on the sample distance evaluation two-dimensional coordinates;

[0148] Traverse the carbon emission prediction sample set, and construct a sample coordinate set based on the sample distance evaluation two-dimensional coordinates;

[0149] Based on the root node coordinates, sort k neighboring samples from near to far to construct the first-level leaf nodes, where 10 ≥ k ≥ 5;

[0150] Traverse the first-level leaf nodes and sort k neighboring samples from near to far respectively to construct the second-level leaf nodes;

[0151] Until traversing the M-1 level leaf nodes and sorting k neighboring samples from near to far respectively to construct the M level leaf nodes, where 20 ≥ M ≥ 2;

[0152] According to the root node, the first-level leaf nodes, the second-level leaf nodes until the M level leaf nodes, construct the carbon emission prediction hierarchical tree.

[0153] Further, the scatter prediction deployment unit 12 further includes the following execution steps:

[0154] Obtain the M-level leaf nodes of the j-th leaf node of the M-1 level leaf nodes;

[0155] Statistically analyze the central value of the carbon emission monitoring values of the M-level leaf nodes of the j-th leaf node and store it in the j-th leaf node;

[0156] Based on the above principle, in the carbon emission prediction module, recursively aggregate the carbon emission prediction hierarchical tree from the bottom layer to obtain the carbon emission training prediction value.

[0157] Further, the output layer fully connected unit 13 includes the following execution steps:

[0158] Configure the carbon emission fitting threshold;

[0159] Delete the carbon emission prediction values that are less than or equal to the carbon emission fitting threshold among the multiple carbon emission prediction values output by the industrial park scatter carbon emission prediction model to obtain the carbon emission prediction value retention value;

[0160] Sum up the carbon emission prediction value retention values to obtain the carbon emission fitting value.

[0161] Embodiment 3:

[0162] Please refer to Figure 3 , Figure 3 which is the schematic diagram of the embodiment of the electronic device provided by the embodiment of the present invention. As Figure 3As shown in the figure, an electronic device 200 provided by an embodiment of the present invention includes a memory 210, a processor 220, and a computer program 211 stored on the memory 210 and executable on the processor 220. When the processor 220 executes the computer program 211, a multi-enterprise carbon emission monitoring method based on blockchain is implemented.

[0163] Embodiment 4:

[0164] Please refer to Figure 4 , Figure 4 which is a schematic diagram of an embodiment of a computer-readable storage medium provided by an embodiment of the present invention. As Figure 4 shown, this embodiment provides a computer-readable storage medium 300, on which a computer program 211 is stored. When the computer program 211 is executed by a processor, a multi-enterprise carbon emission monitoring method based on blockchain is implemented.

[0165] It should be noted that in the above embodiments, the descriptions of each embodiment have their own emphases. For parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0166] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0167] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded computers, or other programmable data processing devices to generate a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices generate a system for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0168] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction system that implements the functions in the processFigure 1 one process or multiple processes and / or boxes Figure 1 the functions specified in one box or multiple boxes.

[0169] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one process or multiple processes and / or boxes Figure 1 one box or multiple boxes.

[0170] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments once they learn the basic creative concept.

[0171] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A blockchain-based method for monitoring carbon emissions of multiple enterprises in an industrial park, characterized in that: Applied to the carbon emission management pipeline, the carbon emission management pipeline includes a blockchain retrieval module, a hierarchical tree construction module and a carbon emission prediction module, including: Obtaining a multi-enterprise distribution topology of an industrial park, and constructing a graph neural network topology of the industrial park, wherein the graph neural network topology of the industrial park has the same structure as the multi-enterprise distribution topology of the industrial park, and any enterprise distribution area is abstracted as a node of the graph neural network topology of the industrial park; Deploy the blockchain retrieval module, the hierarchical tree construction module and the carbon emission prediction module on multiple nodes of the industrial park graph neural network topology, configure multiple carbon emission prediction functions, and generate a scattered carbon emission prediction model for the industrial park; Constructing an industrial park carbon emission fitting rule, fully connecting it with the output layer of the industrial park scattered carbon emission prediction model, and obtaining the industrial park carbon emission prediction model to perform carbon emission monitoring; Wherein, the blockchain retrieval module, the hierarchical tree construction module and the carbon emission prediction module are deployed on multiple nodes of the industrial park graph neural network topology to configure multiple carbon emission prediction functions, including: Obtain a first node product type of the plurality of nodes of the industrial park graph neural network topology; The production scale, production duration and carbon emission monitoring value are used as the retrieval target attributes to construct a ternary array label; Obtain the training production scale, training production duration and true value of carbon emissions; Taking the first node product type as a static constraint condition, taking the training production scale and the training production duration as root node dynamic constraints, in the blockchain retrieval module, performing sample growth retrieval on the ternary array label to obtain a carbon emission prediction sample set; In the hierarchical tree construction module, the carbon emission prediction sample set is parsed based on the sample distance evaluation two-dimensional coordinates, and a carbon emission prediction hierarchical tree is constructed, wherein the first dimension coordinate of the sample distance evaluation two-dimensional coordinates is a production scale normalization parameter, the second dimension coordinate of the sample distance evaluation two-dimensional coordinates is a production time normalization parameter, and the Euclidean distance of the sample distance evaluation two-dimensional coordinates is a sample distance evaluation value; In the carbon emission prediction module, the carbon emission prediction hierarchy tree is recursively aggregated from the bottom layer to obtain a carbon emission training prediction value; When the carbon emission training deviation between the carbon emission training prediction value and the carbon emission true value is less than or equal to the training deviation threshold, the training convergence frequency is increased by one; Otherwise, the training error frequency is increased by one; Repeat the training several times, and when the training convergence frequency is greater than or equal to 90%, the blockchain retrieval module, the hierarchical tree construction module and the carbon emission prediction module are initialized to generate a first node carbon emission prediction function; The first node carbon emission prediction function is deployed on the first node.

2. The method according to claim 1, characterized in that Obtain the industrial park carbon emission prediction model to perform carbon emission monitoring, including: Obtain the production scales and production durations of multiple enterprises in a specified future time zone in a multi-enterprise distribution area; Inputting the production scales of the multiple enterprises and the production durations of the multiple enterprises into multiple nodes of the industrial park carbon emission prediction model to obtain a predicted value of the industrial park carbon emission; When the predicted carbon emission value of the industrial park is greater than or equal to the carbon emission threshold, an abnormal production marking time zone is performed on the designated future time zone; The abnormal production identification time zone, the industrial park number and the industrial park carbon emission forecast value are added to the carbon emission monitoring results and sent to the industrial park multi-enterprise carbon emission monitoring management terminal.

3. The method according to claim 1, characterized in that The first node product type is used as a static constraint condition, and the training production scale and the training production duration are used as root node dynamic constraints, including: Constructing static constraint rule factors: when the sample is of the same type as the first node product, it is satisfied; when the sample is of a different type from the first node product, it is not satisfied; Construct root node dynamic constraint rule factors: First rule factor: when the production scale deviation between the sample production scale and the training production scale is less than or equal to the production scale deviation threshold, it is satisfied; when the production scale deviation between the sample production scale and the training production scale is greater than the production scale deviation threshold, it is not satisfied; Second rule factor: when the production time deviation between the sample production time and the training production time is less than or equal to the production time deviation threshold, the rule is satisfied; when the production time deviation between the sample production time and the training production time is greater than the production time deviation threshold, the rule is not satisfied; If the first rule factor and the second rule factor are satisfied at the same time, the root node dynamic constraint rule factor is deemed to be satisfied; otherwise, the root node dynamic constraint rule factor is deemed not to be satisfied.

4. The method according to claim 1, characterized in that Taking the first node product type as a static constraint condition, taking the training production scale and the training production duration as root node dynamic constraints, in the blockchain retrieval module, performing sample growth retrieval on the ternary array label to obtain a carbon emission prediction sample set, including: Taking the first node product type as a static constraint condition, taking the training production scale and the training production duration as root node dynamic constraints, searching the ternary array label through the blockchain retrieval module, and collecting a first number of first-level samples that simultaneously satisfy the static constraint condition and the root node dynamic constraint condition; Traversing the first number of first-level samples, constructing a dynamic constraint condition for a first-level node, and searching the ternary array label in the blockchain retrieval module in combination with the static constraint condition to obtain a first number of second-level samples; Until the N-1th number of N-1 level samples are traversed, the dynamic constraint conditions of the N-1 level nodes are constructed, and the triple array labels are searched in the blockchain retrieval module in combination with the static constraint conditions to obtain the first number of N level samples; The first number of first-level samples, the first number of second-level samples, and up to the first number of N-level samples are added into the carbon emission prediction sample set.

5. The method according to claim 1, characterized in that In the hierarchical tree construction module, the carbon emission prediction sample set is parsed to build a carbon emission prediction hierarchical tree, including: Construct sample distance evaluation rule factors: Constructing a sample distance evaluation two-dimensional coordinate, wherein the first dimension coordinate of the sample distance evaluation two-dimensional coordinate is a production scale normalization parameter, and the second dimension coordinate of the sample distance evaluation two-dimensional coordinate is a production time normalization parameter; The Euclidean distance of the sample distance evaluation two-dimensional coordinate is the sample distance evaluation value; Based on the training production scale and the training production duration, constructing the root node coordinates through the sample distance evaluation two-dimensional coordinates; Traversing the carbon emission prediction sample set, and constructing a sample coordinate set based on the sample distance evaluation two-dimensional coordinates; Based on the root node coordinates, k neighboring samples are selected from near to far to construct a first-level leaf node, where 10≥k≥5; Traversing the first-level leaf nodes, selecting k neighboring samples from near to far, and constructing second-level leaf nodes; Until the M-1 level leaf nodes are traversed, k neighboring samples are selected from near to far, and M level leaf nodes are constructed, where 20≥M≥2; The carbon emission prediction hierarchical tree is constructed according to the root node, the first-level leaf node, the second-level leaf node and up to the M-level leaf node.

6. The method according to claim 1, characterized in that In the carbon emission prediction module, the carbon emission prediction hierarchy tree is recursively aggregated from the bottom layer to obtain a carbon emission training prediction value, including: Obtaining a set of M-level leaf nodes corresponding to the j-th leaf node of the M-1-level leaf node in the carbon emission prediction hierarchical tree, recorded as the j-th leaf node M-level leaf node; Counting the concentrated value of the carbon emission monitoring values ​​of the j-th leaf node and the M-level leaf nodes, and storing it in the j-th leaf node; Based on the above principle, in the carbon emission prediction module, the carbon emission prediction hierarchy tree is recursively aggregated from the bottom layer to obtain the carbon emission training prediction value.

7. The method according to claim 1, characterized in that Construct the industrial park carbon emission fitting rules, including: Configure carbon emission fitting threshold; Deleting the carbon emission prediction values ​​of the multiple carbon emission prediction values ​​output by the scattered carbon emission prediction model of the industrial park, which are less than or equal to the carbon emission fitting threshold, to obtain a retained carbon emission prediction value; The retained values ​​of the predicted carbon emission values ​​are summed to obtain a fitted carbon emission value.

8. A blockchain-based industrial park multi-enterprise carbon emission monitoring system, characterized in that: Applied to the carbon emission management pipeline, the carbon emission management pipeline includes a blockchain retrieval module, a hierarchical tree construction module and a carbon emission prediction module, including: A topology construction unit, used to obtain a multi-enterprise distribution topology of an industrial park and construct a graph neural network topology of the industrial park, wherein the graph neural network topology of the industrial park has the same structure as the multi-enterprise distribution topology of the industrial park, and any enterprise distribution area is abstracted as a node of the graph neural network topology of the industrial park; A scattered point prediction deployment unit, used to deploy the blockchain retrieval module, the hierarchical tree construction module and the carbon emission prediction module on multiple nodes of the industrial park graph neural network topology, configure multiple carbon emission prediction functions, and generate an industrial park scattered point carbon emission prediction model; The output layer fully connected unit is used to construct the industrial park carbon emission fitting rule, which is fully connected with the output layer of the industrial park scattered carbon emission prediction model to obtain the industrial park carbon emission prediction model to perform carbon emission monitoring; Wherein, the blockchain retrieval module, the hierarchical tree construction module and the carbon emission prediction module are deployed on multiple nodes of the industrial park graph neural network topology to configure multiple carbon emission prediction functions, including: Obtain a first node product type of the plurality of nodes of the industrial park graph neural network topology; The production scale, production duration and carbon emission monitoring value are used as the retrieval target attributes to construct a ternary array label; Obtain the training production scale, training production duration and true value of carbon emissions; Taking the first node product type as a static constraint condition, taking the training production scale and the training production duration as root node dynamic constraints, in the blockchain retrieval module, performing sample growth retrieval on the ternary array label to obtain a carbon emission prediction sample set; In the hierarchical tree construction module, the carbon emission prediction sample set is parsed based on the sample distance evaluation two-dimensional coordinates, and a carbon emission prediction hierarchical tree is constructed, wherein the first dimension coordinate of the sample distance evaluation two-dimensional coordinates is a production scale normalization parameter, the second dimension coordinate of the sample distance evaluation two-dimensional coordinates is a production time normalization parameter, and the Euclidean distance of the sample distance evaluation two-dimensional coordinates is a sample distance evaluation value; In the carbon emission prediction module, the carbon emission prediction hierarchy tree is recursively aggregated from the bottom layer to obtain a carbon emission training prediction value; When the carbon emission training deviation between the carbon emission training prediction value and the carbon emission true value is less than or equal to the training deviation threshold, the training convergence frequency is increased by one; Otherwise, the training error frequency is increased by one; Repeat the training several times, and when the training convergence frequency is greater than or equal to 90%, the blockchain retrieval module, the hierarchical tree construction module and the carbon emission prediction module are initialized to generate a first node carbon emission prediction function; The first node carbon emission prediction function is deployed on the first node.

Images