A carbon management strategy generation method, device, equipment and medium

Abstract

The application discloses a carbon management strategy generation method, device, equipment and medium, and relates to the technical field of data processing. Demand content of carbon management of a target park is acquired; the demand content comprises a strategy type, strategy demand and a strategy time; quantitative constraint data is determined from a dynamic knowledge graph based on the strategy type; a target optimization model is called from an optimization model library according to the strategy time; the target optimization model comprises a long-term optimization model or a real-time optimization model; and a target strategy is generated according to the demand content, a park database, the target optimization model and the quantitative constraint condition. By adopting the technical scheme, how to dynamically capture constraint conditions and generate an optimal carbon management strategy is solved; the quantitative constraint condition is captured through the dynamic knowledge graph, and the corresponding model is called in combination with an intelligent agent to solve the optimal strategy in the target park under the constraint condition, so that the whole-process intelligent management of the carbon management system of the target park is realized.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, device, and medium for generating carbon management strategies. Background Technology

[0002] Currently, carbon management technologies for zero-carbon parks have formed two mainstream solutions: one is carbon asset management technology based on knowledge graphs, which can realize the structured processing and accounting of carbon emission data; the other is AI-driven park energy and carbon management system, which can realize energy consumption monitoring, carbon emission prediction and scheduling optimization.

[0003] However, knowledge graphs are only used as accounting databases, without being linked to intelligent decision-making, unable to update policies in real time, and lacking autonomous execution capabilities. They can only perform auxiliary accounting and cannot support full-process intelligent management. AI, on the other hand, is limited to numerical prediction and simple rule triggering, lacking true task planning and autonomous execution capabilities. Furthermore, it lacks a dynamic policy knowledge system, has not integrated energy-carbon-finance data, and cannot achieve global optimization and multi-standard compliance automatic reporting. Overall, it still relies heavily on manual operations, which is a passive and localized optimization management approach.

[0004] Therefore, existing methods for generating carbon management strategies have significant shortcomings in terms of automatic generation and intelligent management. Summary of the Invention

[0005] This invention provides a carbon management strategy generation method, apparatus, equipment, and medium, which solves the problem of how to dynamically capture constraints and generate the optimal carbon management strategy; by capturing and quantifying constraints through dynamic knowledge graphs, and combining intelligent agents to call corresponding models to solve the optimal strategy in the target park under constraints, the invention achieves intelligent management of the entire process of the carbon management system of the target park.

[0006] According to one aspect of the present invention, a method for generating a carbon management strategy is provided, comprising: Obtain the carbon management requirements of the target industrial park; the requirements include strategy type, strategy requirements, and strategy timing. Quantitative constraint data is determined from the dynamic knowledge graph based on the aforementioned strategy type; The target optimization model is retrieved from the optimization model library according to the strategy time; the target optimization model includes a long-term optimization model or a real-time optimization model. The target strategy is generated based on the stated requirements, the park database, the target optimization model, and the quantitative constraints.

[0007] According to another aspect of the present invention, a carbon management strategy generation apparatus is provided, comprising: The acquisition module is used to acquire the carbon management requirements of the target park; the requirements include strategy type, strategy requirements, and strategy time. A determination module is used to determine quantified constraint data from the dynamic knowledge graph based on the strategy type; The calling module is used to call the target optimization model from the optimization model library according to the strategy time; the target optimization model includes a long-term optimization model or a real-time optimization model; The generation module is used to generate a target strategy based on the requirements, the park database, the target optimization model, and the quantitative constraints.

[0008] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the carbon management strategy generation method according to any embodiment of the present invention.

[0009] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the carbon management strategy generation method according to any embodiment of the present invention.

[0010] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the carbon management strategy generation method according to any embodiment of the present invention.

[0011] The technical solution of this invention uses a dynamic knowledge graph to capture and quantify constraints, and combines this with an intelligent agent to call the corresponding model to solve the optimal strategy within the target park under the constraints. This solves the problem of how to dynamically capture constraints and generate the optimal carbon management strategy, thereby achieving intelligent management of the entire process of the carbon management system in the target park.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of a carbon management strategy generation method provided by an embodiment of the present invention; Figure 2 This is a flowchart of a dynamic knowledge graph construction method provided by an embodiment of the present invention; Figure 3 This is a schematic diagram of a carbon management strategy generation device according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device that implements the carbon management strategy generation method of this invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0016] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification, claims and accompanying drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product or device.

[0017] Furthermore, it should be noted that the information collected in the technical solution of this invention is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data all comply with the relevant laws, regulations and standards of relevant countries and regions, necessary confidentiality measures have been taken, and public order and good morals are not violated. Corresponding operation entry points are provided for users to choose to authorize or refuse.

[0018] Figure 1 This invention provides a flowchart of a carbon management strategy generation method. This embodiment is applicable to the intelligent management of carbon management strategies within a park. The method can be executed by a carbon management strategy generation device, which can be implemented in hardware and / or software and can be configured in a server. Figure 1 As shown, the method includes: S110. Obtain the carbon management requirements of the target park; the requirements include strategy type, strategy requirements, and strategy time.

[0019] The target parks are those that require long-term or short-term carbon emission management; the requirements are for intelligent carbon management, including the purpose and timing of the carbon management strategy; the strategy type is either a pure cost strategy or a carbon emission dual control redundancy strategy; the strategy requirements include the park's energy and cost needs, such as carbon compliance cost control strategies and developing the optimal comprehensive cost strategy for the park under the dual control of carbon emissions in 2026; the strategy timing refers to the planning time for this carbon management requirement, which can be a short-term management time such as this week or real-time, or a long-term management time such as the next quarter or the next year.

[0020] Specifically, the interactive interface allows users to obtain the carbon management needs of relevant technical personnel for the target park. These needs may include the strategy type, strategy requirements, and strategy timeframe for this carbon management initiative, such as "helping the park develop an energy scheduling cost strategy for next week," "adjusting the park's load in real time to reduce current carbon costs," and "determining a cost-saving carbon dual control strategy for this year."

[0021] S120. Determine quantitative constraint data from dynamic knowledge graphs based on strategy type.

[0022] The dynamic knowledge graph is pre-built based on standard data extracted from multiple source knowledge bases, such as national dual-control policies, local dual-control policies, EU CBAM rules and other policy and regulatory databases; technical standard databases such as IPCC, ISO14064, GHG Protocol and other technical standard databases; market databases such as carbon prices, electricity prices, and subsidy policies; and industry case databases such as emission reduction technologies and carbon asset operation cases. It serves as the dual-control constraints for generating target strategies. The quantitative constraint data are the constraints for generating target strategies, including total constraints such as total carbon emission indicators; intensity constraints such as industrial park intensity indicators; and auxiliary constraints such as power balance constraints, equipment operation constraints, and market transaction constraints.

[0023] Specifically, based on the strategy type, the corresponding quantitative constraint data can be obtained from the dynamic knowledge graph. This can be dual control constraint data for carbon emissions, such as carbon emission standard data, cost constraint data, and equipment constraint data.

[0024] Optionally, quantitative constraint data can be determined from the dynamic knowledge graph based on the strategy type, including: Text extraction is performed on the strategy type to obtain the search terms; Semantic similarity retrieval of dynamic knowledge graphs based on search terms yields constrained data nodes; The node attributes in the constraint data nodes are quantized and transformed to obtain quantized constraint data.

[0025] The search terms are keywords used to retrieve constraint data, representing the time planning scope of the strategy, such as annual planning, minute-level planning, and daily planning, like "industrial park + annual planning + compliance redundancy strategy + dual control threshold" or "logistics park + real-time scheduling + cost-priority strategy + hourly carbon emission threshold"; constraint data nodes are nodes in the knowledge graph corresponding to the search terms, which may include constraint content such as annual total indicators, hourly decomposition thresholds, and compliance redundancy reduction coefficients; node attributes can be the node content corresponding to the constraint data node, including the specific value of carbon emission data and standard thresholds; quantified constraint data is standardized unified data.

[0026] Specifically, text extraction is performed on strategy types to obtain search terms; for example, "industrial park + annual plan + compliance redundancy strategy + dual control threshold". Semantic similarity retrieval is performed through the vector knowledge base layer of the knowledge graph to initially screen out dual control constraint data nodes that match the strategy type. For example, the compliance redundancy strategy type matches the constraint value of "dual control threshold × reduction coefficient (0.9)"; the real-time scheduling strategy type matches "hourly or minute-level time-segmented decomposition threshold"; the green electricity consumption priority strategy type matches "green electricity zero-carbon deduction coefficient + actual dual control threshold after deduction". Entity relationship path reasoning is performed through the knowledge graph layer of the knowledge graph to eliminate... Excluding irrelevant nodes, accurately match constraint data that is fully compatible with the current strategy, and quantify the node attributes in the constraint data nodes to obtain quantified constraint data. For example, convert the structured text data retrieved from the knowledge graph into quantitative parameters that the model can recognize, and unify the unit and time granularity. For example, convert "the total annual carbon emission target of the park is 10,000 tons" into "the hourly threshold of the day-ahead planning = 10,000 / (365 × 24) ≈ 1.14 tons / hour", and convert "compliance redundancy reduction coefficient 0.9" into "compliance redundancy strategy hourly threshold = 1.14 × 0.9 ≈ 1.03 tons / hour".

[0027] Understandably, by accurately retrieving, intelligently matching, and automatically converting quantitative constraint parameters from dynamic knowledge graphs into parameters that the model can recognize, these parameters can be directly integrated into the constraints of the target optimization model when generating target strategies. This becomes a hard constraint for model solving, ensuring that the strategy always complies with the latest policy requirements. This builds a carbon management knowledge base that can be updated in real time, automatically adapts to compliance requirements such as dual control and CBAM, and has the ability to plan tasks and execute autonomously, thus automating complex carbon management tasks.

[0028] S130. Call the target optimization model from the optimization model library according to the strategy time; the target optimization model includes a long-term optimization model or a real-time optimization model.

[0029] Among them, the long-term optimization model can be a mixed-integer linear programming model, used for daily, monthly, annual, and long-term optimization; the real-time optimization model can be a deep reinforcement learning model, used for real-time scheduling, with optimization cycles at the minute or hour level.

[0030] Specifically, based on the optimization cycle in the strategy time, the target optimization model is called from the optimization model library, which can be a long-term optimization model or a real-time optimization model. For requirements such as "generating the optimal cost strategy for energy dispatch in the park next week" and "formulating a monthly carbon compliance cost control strategy", if the strategy time is greater than or equal to 24 hours, then a mixed integer linear programming model is selected. For requirements such as "adjusting the park load at the minute level to reduce real-time carbon compliance costs" and "optimizing energy storage charging and discharging strategies to cope with sudden changes in photovoltaic output", if the strategy time is at the minute or hour level, then a deep reinforcement learning model is selected.

[0031] S140. Generate target strategies based on requirements, park database, target optimization model, and quantitative constraint data.

[0032] The park database contains real-time operational monitoring data for the target park, which may include monitoring data and cost data. The monitoring data includes energy flow data and carbon emission flow data. Energy flow data includes real-time and predicted loads of the park, output forecasts for clean energy sources such as photovoltaics and wind power, operating parameters of equipment such as energy storage and cold and heat sources, and time-of-use electricity prices of the grid. Carbon emission flow data includes real-time and predicted carbon emission factors, carbon emissions per unit energy consumption of each production line and equipment in the park, and zero-carbon deduction coefficients for clean energy. Cost data includes spot / forward carbon prices in the carbon market, carbon quota / CCER trading costs, policy subsidy standards for green electricity / emission reduction projects, and marginal costs of equipment operation and maintenance. The target strategy is a carbon management strategy, which may include, for example, energy storage charging and discharging power curves, grid power purchase plans, flexible load adjustment, and carbon quota trading timing / quantity.

[0033] Specifically, based on the requirements, real-time operational data is retrieved from the park's database and combined with quantitative constraint data as dual-control constraints for the target optimization model. The real-time operational data is then processed to generate target strategies for subsequent carbon management of the park.

[0034] Optionally, if the target optimization model is a long-term optimization model, and the long-term optimization model is a mixed-integer linear programming model; accordingly, a target strategy is generated based on the demand content, the park database, the target optimization model, and the quantitative constraints, including: Retrieve cost and monitoring data from the park's database based on the strategy type; A long-term optimization model is used to generate strategies based on quantitative constraint data, cost data, and monitoring data, resulting in at least one candidate strategy. The target policy is obtained from at least one candidate policy according to the policy requirements.

[0035] Among them, the preferred long-term optimization model is the MILP model; cost data can include energy procurement costs such as photovoltaic / energy storage / grid power purchase in the park, compliance costs corresponding to carbon emissions, policy subsidies for green electricity / emission reduction projects, and carbon asset trading revenue; monitoring data can include the total carbon emissions of the park, photovoltaic power output, energy storage charging and discharging, grid power purchase and other system equipment operation data of the park; candidate strategies are carbon emission management strategies that meet the quantitative constraints.

[0036] Specifically, cost and monitoring data are obtained from the park database based on the strategy type. For example, a pure cost-first strategy focuses solely on the park's overall operating costs, obtaining cost data such as carbon market spot / forward carbon prices, carbon quota / CCER trading costs, and equipment operation and maintenance marginal costs from the park database. A compliance redundancy strategy requires increasing carbon emission dual control redundancy, so it obtains monitoring data from the park database, including real-time / predicted load, output forecasts of clean energy such as photovoltaic / wind power, operating parameters of equipment such as energy storage / cold and heat sources, and time-of-use electricity prices from the grid. A green electricity consumption priority strategy obtains policy subsidy standards for green electricity / emission reduction projects from the park database. A long-term optimization model is used to generate strategies based on quantitative constraint data, cost data, and monitoring data to obtain at least one candidate strategy that meets the quantitative constraint data. Finally, the target strategy is obtained from at least one candidate strategy according to the strategy requirements.

[0037] The strategy generation formula for the long-term optimization model under quantized constraint data is shown below: Where T is the policy time; The energy procurement costs for photovoltaic / energy storage / grid power purchases in the park during time period t; The compliance cost corresponding to carbon emissions during period t; The policy subsidies and carbon asset trading revenue for green electricity / emission reduction projects during period t.

[0038] Based on quantitative constraints, such as total emission constraints, the system ensures that the total carbon emissions of the park are less than or equal to the total carbon emission quota approved by the competent authority within the optimization period; furthermore, the carbon emission intensity per unit output value / per unit energy consumption of the park is less than or equal to the approved intensity quota; monitoring data requires ensuring that the park's photovoltaic output + energy storage charging and discharging + grid power purchases equal the park's real-time total electricity load; equipment operation constraints include upper and lower limits of energy storage SOC, charging and discharging power constraints, upper and lower limits of cold and heat source / generator output, and start-up and shutdown frequency constraints; market transaction constraints include upper and lower limits of grid power purchases, carbon quota / CCER trading quantity constraints, and demand response contract constraints; under the above quantitative constraints, the generated strategy is used as a candidate strategy, and the target strategy is selected according to actual needs.

[0039] Understandably, by constructing a differentiated collaborative optimization model, taking the dual control of total carbon emissions and intensity as the core constraints, and minimizing the overall operating cost of the park as the core objective, while also considering auxiliary constraints such as equipment safety and market trading rules, the model can quickly solve the global optimal solution for the mixed integer / continuous variable problem of day-ahead planning, and output the day-ahead energy dispatch plan, carbon asset trading strategy, and emission reduction input plan.

[0040] Optionally, if the target optimization model is a real-time optimization model, and the real-time optimization model is a deep reinforcement learning model, then a target strategy is generated accordingly based on the requirements, the park database, the target optimization model, and the quantification constraints, including: Based on the strategy type, monitoring and cost data are obtained from the park's database; the monitoring data includes carbon emission monitoring data and equipment monitoring data. Carbon emission standard data and equipment standard data are determined based on quantitative constraint data; The equipment penalty value is determined based on equipment monitoring data and equipment standard data; Carbon emission penalty values ​​are determined based on carbon emission monitoring data and carbon emission standard data. A real-time optimization model is used to generate strategies based on equipment penalty values, carbon emission penalty values, and cost data, resulting in at least one candidate strategy. The target policy is obtained from at least one candidate policy according to the policy requirements.

[0041] The carbon emission monitoring data can include real-time / predicted load of the park and output forecasts of clean energy such as photovoltaic / wind power; equipment monitoring data can include operating parameters of equipment such as energy storage / cold and heat sources; carbon emission standard data can include real-time / predicted carbon emission factors, carbon emissions per unit energy consumption of each production line / equipment in the park, and zero-carbon deduction coefficients for clean energy; equipment standard data can include standard thresholds such as rated power / SOC upper and lower limits of energy storage, output range of cold and heat source units, and flexible load adjustment limits; equipment penalty values ​​are used to ensure that all physical equipment operates within safe limits; and carbon emission penalty values ​​are used to ensure that the cumulative carbon emissions of the park do not exceed the quotas or thresholds stipulated by policy.

[0042] Specifically, carbon emission monitoring data, equipment monitoring data, and cost data are obtained from the park's database based on the strategy type; carbon emission standard data and equipment standard data are determined based on the quantitative constraint data; equipment penalty values ​​are determined based on the equipment monitoring data and equipment standard data; it is determined whether the actual operating data of the equipment exceeds the safety limit. If it does not exceed the limit, the penalty for this item is 0; if it exceeds the limit, the excess amount is multiplied by the corresponding safety weight of the equipment, and a weighted aggregation is performed to obtain the equipment penalty value. For example, if the energy storage charging and discharging power exceeds the rated maximum value, the excess amount is counted; if the energy storage capacity is too low / too high, the unit output is too full / too low, or the load adjustment range is too large, only the excess part is counted, and the values ​​are calculated separately and then added together; the calculation formula is as follows: in, denoted as , where i is the equipment penalty value; i represents the number of standard data items for the equipment; violations can include violations in energy storage power, energy storage SOC, chiller / heater unit output limits, and flexible load regulation limits; energy storage power violations are based on the difference between the charging / discharging power at time t and the rated power in the standard data; energy storage SOC violations are based on the difference between the energy storage SOC at time t and the upper / lower limits in the standard data; chiller / heater unit output limits violations are based on the difference between the chiller / heater unit output limits at time t and the upper / lower limits in the standard data; flexible load regulation limit violations are based on the difference between the flexible load regulation limits at time t.

[0043] Carbon emission penalties are determined based on carbon emission monitoring data and carbon emission standard data. Only data up to the current moment within the assessment period is considered to determine whether actual carbon emissions exceed the government's real-time limit. If not, the penalty is 0; if exceeding, the penalty is calculated accordingly. The actual cumulative carbon emissions from the start of the government assessment to the present are automatically calculated based on real-time electricity consumption and the generation of clean energy sources such as solar / wind power. If the government sets an annual / quarterly total carbon emission quota for the park, this is broken down into daily and hourly real-time limits. The calculation formula is shown below: in, This is a carbon emission penalty value; This represents the actual cumulative carbon emissions; Let t be the dynamic threshold for permissible carbon emissions. A real-time optimization model is used to generate strategies based on equipment penalty values, carbon emission penalty values, and cost data, resulting in at least one candidate strategy; the calculation formula is shown below: Where R is a candidate strategy that satisfies the constraint; and This is the penalty coefficient.

[0044] The target policy is obtained from at least one candidate policy according to the policy requirements.

[0045] Understandably, by setting up a deep reinforcement learning model that is suitable for real-time cost strategy adjustments at the minute / hour level, it can quickly output the optimal control strategy for uncertain scenarios such as sudden changes in park load, fluctuations in photovoltaic output, and changes in real-time electricity / carbon prices. This compensates for the lack of adaptability of long-term optimization models in dynamic scenarios. With minimizing real-time comprehensive costs as the core, it adds reward deductions for exceeding carbon emission thresholds and equipment safety constraint penalties to ensure that the strategy is optimal under the premise of compliance.

[0046] This invention, through its embodiments, acquires the carbon management requirements of a target industrial park. These requirements include strategy type, strategy requirements, and strategy timeframe. Based on the strategy type, quantitative constraint data is determined from a dynamic knowledge graph. According to the strategy timeframe, a target optimization model is retrieved from an optimization model library. The target optimization model can be a long-term optimization model or a real-time optimization model. A target strategy is generated based on the requirements, the industrial park database, the target optimization model, and the quantitative constraints. This technical solution addresses how to dynamically capture constraints and generate the optimal carbon management strategy. By capturing quantitative constraints through a dynamic knowledge graph and combining this with an intelligent agent to call the corresponding model, the optimal strategy within the target industrial park can be solved under these constraints, achieving intelligent management of the entire carbon management system for the target industrial park.

[0047] Figure 2 This is a flowchart illustrating a dynamic knowledge graph construction method provided by an embodiment of the present invention. This embodiment supplements the knowledge graph construction method based on the above embodiments. It should be noted that for parts not detailed in this embodiment, please refer to the relevant descriptions in other embodiments, such as... Figure 2 As shown, the method includes: S210. Obtain standard data from at least one data source based on an interface.

[0048] The interface can be an interface that connects to the data source for real-time data acquisition; the data source can include policy and regulation databases, technical standard databases, market databases, and industry case databases; standard data is fixed data in the data source.

[0049] Specifically, standardized data related to carbon management can be obtained from policy and regulation libraries, technical standard libraries, market databases, and industry case libraries through interfaces.

[0050] S220. Perform semantic understanding on the standard data according to the semantic understanding model to obtain at least one key entity and entity attribute.

[0051] The semantic understanding model can be a pre-trained model such as BERT; key entities include the CBAM control scope and green electricity accounting varieties; entity attributes are used to describe the key entity, such as its corresponding numerical value.

[0052] Specifically, an entity-relationship extraction model for the carbon management field is constructed through a semantic understanding model. This model performs deep semantic understanding on standard data and automatically extracts key entities and their corresponding entity attributes.

[0053] Optionally, semantic understanding is performed on the standard data based on the semantic understanding model to obtain at least one key entity and entity attributes, including: Based on the semantic understanding model, perform semantic understanding on the standard data to obtain at least one candidate entity and candidate attribute; Calculate the similarity between candidate entities to obtain entity similarity. Entities with a similarity greater than the entity threshold will be considered as entities to be removed. The entities to be removed are determined based on the data source information to which they belong; Remove the candidate entities to obtain at least one key entity.

[0054] Among them, candidate entities are extracted similar entities; entity similarity is the similarity value between candidate entities; there can be at least one entity to be removed, which are entities with inconsistent data sources under the same entity; data source information includes data source, data release time and data download volume; entities to be removed are entities with low credibility of data source.

[0055] Specifically, the standard data is semantically understood using a semantic understanding model to obtain at least one candidate entity and candidate attribute; similarity is calculated for the candidate entities to obtain entity similarity; entities with similarity greater than a threshold are designated as entities to be removed; the data source information of the entities to be removed is determined, and calculations are performed based on the data source, data release time, and data download volume; a data source score is determined based on a data source-data authority mapping table; a data release time score is determined based on the data release time and a time threshold; a data accuracy score is determined based on the data download volume and a download volume threshold; the data source score, release time score, and data accuracy score are weighted and aggregated to obtain a credibility score for each entity to be removed; the entity with the highest credibility score is retained, and the remaining entities are removed; and the entities to be removed from the candidate entities are eliminated to obtain at least one key entity.

[0056] Understandably, by constructing an automatic multi-source information conflict detection algorithm, inconsistent information from multiple sources regarding the same entity / attribute can be automatically identified, thereby automatically resolving knowledge conflicts and ensuring the consistency and reliability of knowledge.

[0057] S230. Classify key entities based on entity attributes to obtain entity relationships.

[0058] Among them, the entity relationship is policy constraint-technical path-economic cost. For example, the entity relationship of policy-level entity attributes can be total quantity indicator → park type (industrial / logistics), intensity indicator → industry benchmark, assessment cycle → policy node; the entity relationship of decomposition layer entity attributes is time-based threshold → assessment cycle, decomposition rule → park load characteristics; the entity relationship of adaptation layer entity attributes is constraint details → strategy type, deduction rule → green electricity type, triggering condition → assessment period / non-assessment period.

[0059] Specifically, key entities are classified according to their attributes to obtain the entity relationships corresponding to each key entity. For example, in the policy-level entity attributes, if the key entity can be the total carbon emission index of the park, then its corresponding entity relationship can be the total index → ​​park type.

[0060] S240. Construct entity triples based on entity relationships, entity attributes, and key entities.

[0061] Among them, the entity triple is the association relationship between entity relations, entity attributes, and key entities.

[0062] Specifically, entity relationships, entity attributes, and key entity associations are encapsulated into an entity triple.

[0063] S250. Treat the key entity as a constraint data node and the entity triple as the node attribute of the constraint data node.

[0064] Among them, a data node can be a constraint data node, so that the data content under the constraint data can be quickly found through the knowledge graph.

[0065] Specifically, the key entity is treated as a constraint data node, and the entity triple is treated as the node attribute of the constraint data node, so that the data content under the constraint data can be quickly found through the knowledge graph.

[0066] S260. Associate and integrate at least one constraint data node to construct a dynamic knowledge graph.

[0067] Specifically, at least one constraint data node is associated and integrated to obtain a dynamic knowledge graph under each constraint data node.

[0068] Optionally, after associating and integrating at least one data node to construct a dynamic knowledge graph, the following steps are also included: Retrieve changed data from the data source via an interface; Based on the semantic understanding model, entity extraction is performed on the changing data to obtain changing triples; The similarity between the entity attributes of the variable triples and the entity attributes of the candidate triples in the dynamic knowledge graph is calculated to obtain the candidate similarity. The candidate triples are determined to be updated based on candidate similarity and similarity threshold; Replace the triplet to be updated with the changed triplet to obtain the updated dynamic knowledge graph.

[0069] Among them, changed data refers to the modified data in the data source; changed triples are the triples corresponding to the modified data, which may include the modified entity, modified entity attributes, and modified entity relationships; candidate triples are triples in the dynamic knowledge graph that are similar to key entities; and triples to be updated are triples to be replaced.

[0070] Specifically, the process involves obtaining changed data from a data source via an interface; extracting entities from the changed data using a semantic understanding model to obtain key entities, entity attributes, and entity relationships; constructing changed triples based on the relationships between key entities, entity attributes, and entity relationships; calculating the similarity between the entity attributes of the changed triples and the entity attributes of candidate triples in the dynamic knowledge graph to obtain candidate similarity; selecting candidate triples with similarity greater than a similarity threshold as triples to be updated; and replacing the triples to be updated in the dynamic knowledge graph with the changed triples to obtain the updated dynamic knowledge graph.

[0071] Understandably, the streaming incremental update mechanism enables near real-time iteration of policy and market information, perfectly adapting to the rapid changes in policies such as dual control, and mitigating compliance risks.

[0072] Based on semantic understanding models and graph databases, this invention enables structured, correlated, and real-time management of multi-source heterogeneous information in the field of carbon management, providing interpretable, highly reliable, and timely knowledge support for subsequent intelligent agents to call models and access matching constraint data.

[0073] Figure 3 This is a schematic diagram of a carbon management strategy generation device provided in an embodiment of the present invention. This embodiment is applicable to situations involving intelligent management of carbon management strategies within a park. The carbon management strategy generation device can be implemented in hardware and / or software, and can be configured in a server. Figure 3 As shown, the carbon management strategy generation device 300 includes an acquisition module 310, a determination module 320, a calling module 330, and a generation module 340. Module 310 is used to acquire the carbon management requirements of the target park; the requirements include strategy type, strategy requirements and strategy time. Module 320 is used to determine quantified constraint data from the dynamic knowledge graph based on strategy type; Module 330 is used to call the target optimization model from the optimization model library according to the strategy time; the target optimization model includes a long-term optimization model or a real-time optimization model. The generation module 340 is used to generate target strategies based on the requirements, park database, target optimization model and quantitative constraints.

[0074] This invention, through its embodiments, acquires the carbon management requirements of a target industrial park. These requirements include strategy type, strategy requirements, and strategy timeframe. Based on the strategy type, quantitative constraint data is determined from a dynamic knowledge graph. According to the strategy timeframe, a target optimization model is retrieved from an optimization model library. The target optimization model can be a long-term optimization model or a real-time optimization model. A target strategy is generated based on the requirements, the industrial park database, the target optimization model, and the quantitative constraints. This technical solution addresses how to dynamically capture constraints and generate the optimal carbon management strategy. By capturing quantitative constraints through a dynamic knowledge graph and combining this with an intelligent agent to call the corresponding model, the optimal strategy within the target industrial park can be solved under these constraints, achieving intelligent management of the entire carbon management system for the target industrial park.

[0075] Optionally, the determination module 320 is also used to extract text from the strategy type to obtain search terms; Semantic similarity retrieval of dynamic knowledge graphs based on search terms yields constrained data nodes; The node attributes in the constraint data nodes are quantized and transformed to obtain quantized constraint data.

[0076] Optionally, if the target optimization model is a long-term optimization model, and the long-term optimization model is a mixed-integer linear programming model, then the generation module 340 is also used to obtain cost data and monitoring data from the park database according to the strategy type. A long-term optimization model is used to generate strategies based on quantitative constraint data, cost data, and monitoring data, resulting in at least one candidate strategy. The target policy is obtained from at least one candidate policy according to the policy requirements.

[0077] Optionally, if the target optimization model is a real-time optimization model, and the real-time optimization model is a deep reinforcement learning model, then the generation module 340 is also used to obtain monitoring data and cost data from the park database according to the strategy type; the monitoring data includes carbon emission monitoring data and equipment monitoring data; Carbon emission standard data and equipment standard data are determined based on quantitative constraint data; The equipment penalty value is determined based on equipment monitoring data and equipment standard data; Carbon emission penalty values ​​are determined based on carbon emission monitoring data and carbon emission standard data. A real-time optimization model is used to generate strategies based on equipment penalty values, carbon emission penalty values, and cost data, resulting in at least one candidate strategy. The target policy is obtained from at least one candidate policy according to the policy requirements.

[0078] Optionally, the carbon management strategy generation device 300 also includes a building module for acquiring standard data from at least one data source based on an interface; Based on the semantic understanding model, perform semantic understanding on standard data to obtain at least one key entity and entity attribute; Based on entity attributes, key entities are classified to obtain entity relationships; Construct entity triples based on entity relationships, entity attributes, and key entities; Treat the key entity as a constraint data node, and the entity triple as the node attribute of the constraint data node; By associating and integrating at least one constrained data node, a dynamic knowledge graph can be constructed.

[0079] Optionally, the building module is also used to perform semantic understanding on standard data based on the semantic understanding model to obtain at least one candidate entity and candidate attribute; Calculate the similarity between candidate entities to obtain entity similarity. Entities with a similarity greater than the entity threshold will be considered as entities to be removed. The entities to be removed are determined based on the data source information to which they belong; Remove the candidate entities to obtain at least one key entity.

[0080] Optionally, the building module is also used to retrieve changed data from the data source via an interface; Based on the semantic understanding model, entity extraction is performed on the changing data to obtain changing triples; The similarity between the entity attributes of the variable triples and the entity attributes of the candidate triples in the dynamic knowledge graph is calculated to obtain the candidate similarity. The candidate triples are determined to be updated based on candidate similarity and similarity threshold; Replace the triplet to be updated with the changed triplet to obtain the updated dynamic knowledge graph.

[0081] The carbon management strategy generation device provided in this embodiment of the invention can execute the carbon management strategy generation method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0082] According to embodiments of the present invention, the present invention also provides an electronic device, a readable storage medium, and a computer program product.

[0083] Figure 4 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0084] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0085] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0086] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as carbon management strategy generation methods.

[0087] In some embodiments, the carbon management strategy generation method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the carbon management strategy generation method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the carbon management strategy generation method by any other suitable means (e.g., by means of firmware).

[0088] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0089] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0090] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0091] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0092] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0093] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product within the cloud computing service system. This addresses the shortcomings of traditional physical hosts and dedicated virtual services, such as high management difficulty and weak business scalability.

[0094] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0095] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for generating a carbon management strategy, characterized in that, include: Obtain the carbon management requirements of the target industrial park; the requirements include strategy type, strategy requirements, and strategy timing. Quantitative constraint data is determined from the dynamic knowledge graph based on the aforementioned strategy type; The target optimization model is retrieved from the optimization model library according to the strategy time; the target optimization model includes a long-term optimization model or a real-time optimization model. The target strategy is generated based on the stated requirements, the park database, the target optimization model, and the quantitative constraints.

2. The method according to claim 1, characterized in that, The step of determining quantified constraint data from the dynamic knowledge graph based on the strategy type includes: Text extraction is performed on the strategy type to obtain search terms; Based on the search terms, semantic similarity retrieval is performed on the dynamic knowledge graph to obtain constrained data nodes; The node attributes in the constraint data nodes are quantized and transformed to obtain quantized constraint data.

3. The method according to claim 1, characterized in that, If the target optimization model is a long-term optimization model, and the long-term optimization model is a mixed-integer linear programming model; accordingly, a target strategy is generated based on the demand content, the park database, the target optimization model, and the quantitative constraints, including: Retrieve cost data and monitoring data from the park database according to the strategy type; A long-term optimization model is used to generate a strategy based on the quantitative constraint data, the cost data, and the monitoring data, resulting in at least one candidate strategy. The target strategy is obtained from at least one candidate strategy according to the strategy requirements.

4. The method according to claim 1, characterized in that, If the target optimization model is a real-time optimization model, and the real-time optimization model is a deep reinforcement learning model, then, accordingly, a target strategy is generated based on the requirements, the park database, the target optimization model, and the quantization constraints, including: According to the strategy type, monitoring data and cost data are obtained from the park's database; the monitoring data includes carbon emission monitoring data and equipment monitoring data. Based on the aforementioned quantitative constraint data, carbon emission standard data and equipment standard data are determined; The equipment penalty value is determined based on the equipment monitoring data and the equipment standard data; The carbon emission penalty value is determined based on the carbon emission monitoring data and the carbon emission standard data. The real-time optimization model is used to generate a strategy based on the equipment penalty value, the carbon emission penalty value, and the cost data, resulting in at least one candidate strategy. The target strategy is obtained from at least one candidate strategy according to the strategy requirements.

5. The method according to any one of claims 1-4, characterized in that, The dynamic knowledge graph is constructed as follows: Standard data is obtained from at least one data source based on an interface; Based on the semantic understanding model, the standard data is semantically understood to obtain at least one key entity and entity attribute; Based on the entity attributes, the key entities are classified to obtain entity relationships; Construct entity triples based on the entity relationships, entity attributes, and key entities; The key entity is used as a constraint data node, and the entity triple is used as the node attribute of the constraint data node. By associating and integrating at least one constrained data node, a dynamic knowledge graph is constructed.

6. The method according to claim 5, characterized in that, Based on the semantic understanding model, the standard data is semantically understood to obtain at least one key entity and entity attributes, including: Based on the semantic understanding model, the standard data is semantically understood to obtain at least one candidate entity and candidate attribute; The similarity of the candidate entities is calculated to obtain the entity similarity. Entities with a similarity greater than the entity threshold are considered as entities to be removed. The entities to be removed are determined based on the data source information to which the entities to be removed belong; The candidate entities are removed to obtain at least one key entity.

7. The method according to claim 5, characterized in that, After associating and integrating at least one constrained data node to construct a dynamic knowledge graph, the process also includes: Change data is obtained from the data source through the interface; Based on the semantic understanding model, entity extraction is performed on the changed data to obtain changed triples; The similarity between the entity attributes of the changed triplet and the entity attributes of the candidate triplet in the dynamic knowledge graph is calculated to obtain the candidate similarity. Based on the candidate similarity and similarity threshold, the triplet to be updated is determined from the candidate triplet; The changed triplet is used to replace the triplet to be updated to obtain the updated dynamic knowledge graph.

8. A carbon management strategy generation device, characterized in that, include: The acquisition module is used to acquire the carbon management requirements of the target park. The requirements include strategy type, strategy requirements, and strategy timing; A determination module is used to determine quantified constraint data from the dynamic knowledge graph based on the strategy type; The calling module is used to call the target optimization model from the optimization model library according to the strategy time; the target optimization model includes a long-term optimization model or a real-time optimization model; The generation module is used to generate a target strategy based on the requirements, the park database, the target optimization model, and the quantitative constraints.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the carbon management strategy generation method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the carbon management strategy generation method according to any one of claims 1-7.

Images