System and Method for Determining Product Carbon Footprint Compliance and Deriving Optimal Reduction Pathways in Multi-Carbon Export Regulation Environment

Abstract

The present invention relates to a system and method for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment. More specifically, it relates to a system and method for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment that utilizes an ontology-based knowledge graph to determine compliance with multiple jurisdictional regulations and derive an optimal emission reduction path in a fragmented environment where multiple carbon export regulations coexist.

Description

Technology Field

[0001] The present invention relates to a system and method for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment. More specifically, it relates to a system and method for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment that utilizes an ontology-based knowledge graph to determine compliance with multiple jurisdictional regulations and derive an optimal emission reduction path in a fragmented environment where multiple carbon export regulations coexist. Background Technology

[0002] As global responses to climate change accelerate, countries are introducing various forms of carbon export regulations to protect the competitiveness of their domestic industries and enforce greenhouse gas reductions. Leading with the European Union's Carbon Border Adjustment Mechanism (CBAM), differing carbon regulations are set to be implemented in parallel across multiple jurisdictions, including the UK's independent CBAM and the US Clean Competition Act (CCA). This multi-regulatory environment creates significant complexity for export-oriented manufacturers in calculating and managing their Product Carbon Footprint.

[0003] Conventional carbon emission calculation systems are either fragmentary spreadsheet-based models that calculate emissions based on a single country or standard (e.g., ISO 14067 or PAS 2050), or adopt hardcoded rule engine methods that rely on simple relational databases. However, these conventional technologies have fundamental limitations in dynamically reflecting fragmented regulatory requirements, such as calculation boundaries, energy source attribution rules, and Product Category Rules (PCRs), which are defined differently by jurisdiction. Whenever regulations are newly established or updated, the source code of the entire system must be modified, resulting in massive maintenance costs and legal and financial risks associated with non-compliance.

[0004] In particular, conventional systems fail to systematically identify the causes when conflicting determination results are obtained from two or more regulations. When one regulation recognizes a specific renewable energy certificate as a reduction achievement while another does not, conventional technology merely outputs the difference in final emission figures and fails to track the fundamental conflict in attribution rules to provide a solution. Furthermore, in practical environments where missing supply chain raw material data frequently occurs, conventional systems return errors and halt calculations, severely undermining the continuity of the entire supply chain analysis.

[0005] Furthermore, conventional technology has limitations in that it evaluates the effects of various reduction measures that companies can take to decrease carbon emissions (e.g., process improvements, energy source switching, raw material replacement) on each regulation by simply summing them up, even though these effects appear non-linear. It fails to identify in advance regulatory deviations—where a specific reduction measure acts favorably under one regulation but disadvantageously under another—and lacks a multi-objective optimization process that combines measures that cannot be physically applied simultaneously or are semantically complementary. Consequently, companies face a critical difficulty in making optimal decisions that satisfy all regulations simultaneously, even while incurring excessive costs.

[0006] Therefore, to overcome the limitations of conventional technologies, there is an urgent need to develop an intelligent product carbon footprint management system that introduces a multi-layered ontology-based knowledge graph capable of flexibly accommodating regulatory domain knowledge, analyzes the causes of regulatory conflicts through semantic inference, provides robust alternative paths in the event of missing data, and derives mathematically verified optimal reduction paths by detecting complex interactions between reduction measures. The problem to be solved

[0007] The present invention has been devised to improve upon the aforementioned problems, and aims to provide a system that automatically determines compliance with product carbon footprints by integrating carbon export regulations of various countries based on a multilayer ontology structure.

[0008] The present invention aims to provide a determination method for calculating compliance based on carbon raw material specifications by utilizing a common abstraction layer regarding supply chains and emission ranges and individual regulatory knowledge graphs.

[0009] The present invention aims to provide a reverse inference method for deriving raw material substitution conditions or energy source conversion conditions by tracing a knowledge graph in reverse. means of solving the problem

[0010] To solve the aforementioned problem, the present invention comprises: a domain-independent shared abstraction layer defining universal concepts of supply chain, products, and raw materials; a carbon export regulation domain layer inheriting from the domain-independent shared abstraction layer and defining an emission range classification system and a carbon raw material specification model; an ontology repository managing a regulation-specific sub-knowledge graph in which product category rule and energy source attribution rule instances of individual carbon export regulations are constructed as sub-instances of the carbon export regulation domain layer; a C-BOM input processing unit that receives input of a carbon raw material specification of a target product based on the item classification code of the target product and the carbon raw material specification model, and identifies a set of multiple carbon export regulations applied to the target product based on the item classification code and the carbon raw material specification; a graph query processing unit that queries a PCR instance corresponding to the item classification code in the regulation-specific sub-knowledge graph for each regulation within the identified set of multiple carbon export regulations; and a method for determining the attribution range of the carbon raw material specification according to the output range boundary defined by the queried PCR instance, and determining the intrinsic carbon amount and compliance under each regulation, wherein the energy source attribution rule applied in the determination A product carbon footprint management system comprising: an inference engine that records instances as inference paths; an actionable combination derivation unit that detects interactions and derives an actionable combination set by calculating the reduction effect for each regulation within the identified set of multiple carbon export regulations for each reduction measure candidate within the set of reduction measure candidates, which includes multiple reduction measure candidates applicable to the carbon raw material specification; and a result output unit that generates and outputs an integrated result including the determined compliance status and the derived set of actionable combinations.

[0011] The above-mentioned inference engine is a system that, when the judgment result of a first regulation and the judgment result of a second regulation belonging to the above-mentioned set of multiple carbon export regulations differ, calls a special query to cross-compare the opposing points of the reduction recognition attribute applied to the first regulation and the reduction non-recognition attribute applied to the second regulation, and based on the cross-compare results, extracts a specific energy source that caused the discrepancy in the judgment result as the cause of the result difference and transmits it to the result output unit.

[0012] The above-described inference engine is a system that, when a non-compliance determination result is produced for at least one specific regulation among the identified set of multiple carbon export regulations, traces the violated constraint in reverse using the inference path recorded in the regulation-specific sub-knowledge graph and derives the minimum C-BOM condition for achieving compliance, including at least one of the raw material carbon strength substitution condition, process unit energy consumption reduction condition, and energy source conversion condition, through reverse inference based on the reverse tracing.

[0013] This is a system in which the above-mentioned C-BOM input processing unit, when supply chain verification data for a specific raw material within the received carbon raw material specification is missing, processes the missing state at the level of a constraint language warning to prevent input invalidation, and triggers an alternative path mechanism that automatically assigns the default value of the corresponding regulation defined in the regulation-specific sub-knowledge graph as the attribute of the missing raw material.

[0014] An information storage step in which an ontology repository stores a regulation-specific sub-knowledge graph in which instances of product category rules and energy source attribution rules of individual carbon export regulations are constructed as sub-instances of the carbon export regulation domain layer, wherein the ontology repository has a domain-independent shared abstraction layer that defines universal concepts of supply chain, products, and raw materials, and a carbon export regulation domain layer that inherits from the domain-independent shared abstraction layer and defines an emission range classification system and a carbon raw material specification model; a C-BOM input processing step in which a C-BOM input processing unit receives the item classification code of the target product and the carbon raw material specification of the corresponding target product based on the carbon raw material specification model, and identifies a set of multiple carbon export regulations applicable to the target product based on the item classification code and the carbon raw material specification; a graph query processing step in which a graph query processing unit queries a PCR instance corresponding to the item classification code in the regulation-specific sub-knowledge graph for each regulation within the identified set of multiple carbon export regulations; and an inference engine determines the attribution range of the carbon raw material specification according to the output range boundary defined by the queried PCR instance and determines the implied carbon amount and compliance status under each regulation, A method comprising an inference step that records an instance of the energy source attribution rule applied to the judgment as an inference path, and a result output step in which a result output unit generates and outputs an integrated result including the determined compliance status.

[0015] The above inference step is a method comprising the step of calling a query when the inference engine is different from the judgment result of a first regulation and the judgment result of a second regulation belonging to the identified set of multiple carbon export regulations, cross-comparing the opposing points of the reduction recognition attribute applied to the first regulation and the reduction non-recognition attribute applied to the second regulation, and identifying the cause of the difference in results based on the cross-comparison result.

[0016] The above inference step is a method comprising the step of, when the inference engine produces a non-compliance determination result for at least one specific regulation among the identified set of multiple carbon export regulations, deriving a minimum C-BOM condition for achieving compliance, including at least one of the carbon strength substitution condition of raw materials, the process unit energy consumption reduction condition, and the energy source conversion condition, through reverse inference based on the inference path recorded in the regulation-specific sub-knowledge graph.

[0017] The above C-BOM input processing step is a method comprising the step of the C-BOM input processing unit preventing input invalidation by processing the missing state as a warning level of the constraint language when supply chain evidence data for a specific raw material in the received carbon raw material specification is missing, and triggering an alternative path mechanism that automatically assigns the default value of the corresponding regulation defined in the regulation-specific sub-knowledge graph as an attribute of the missing raw material.

[0018] A method comprising: a matrix construction step in which an actionable combination derivation unit constructs an effect projection matrix by mapping the emission reduction amount calculated under each regulation in the identified set of multiple carbon export regulations to each reduction measure candidate in the set of reduction measure candidates, which includes multiple reduction measure candidates applicable to the carbon raw material specification; and a combination derivation step in which the actionable combination derivation unit analyzes the effect projection matrix to identify effect deviations between regulations and detects interactions between each reduction measure candidate to output an actionable combination set.

[0019] The above combination derivation step is a method comprising the step of analyzing the effect projection matrix to calculate the reduction effect deviation ratio between each regulation for the same reduction measure candidate, and identifying a specific reduction measure candidate within the set of reduction measure candidates as a regulation deviation measure based on the calculated reduction effect deviation ratio.

[0020] The method further comprises an interaction detection step in which the combination derivation step compares the set of nodes of the carbon raw material specification, which are each affected by the first reduction measure candidate and the second reduction measure candidate within the set of reduction measure candidates, to determine whether there are overlapping physical nodes, and detects and marks hidden complementary relationships by tracing the semantic network between the first reduction measure candidate and the second reduction measure candidate using an attribute chain, and a pruning step in which the measure combinations marked as competitive relationships are excluded from the search space and the measure combinations marked as complementary relationships are classified as priority search targets.

[0021] The method further includes the step of inversely analyzing the energy source attribution rule instances of the regulation-specific sub-knowledge graph corresponding to the specific reduction measure candidate identified as the regulation deviation measure to derive alternative measure candidates that minimize the reduction effect deviation ratio, and updating the effect projection matrix based on the derived alternative measure candidates.

[0022] The method further includes an additivity correction step for a set of feasible combinations remaining within the set of feasible combinations derived after the pruning step, wherein when multiple reduction measures are applied redundantly to a node of the same carbon raw material specification within the remaining feasible combinations, the simple addition of the reduction effects resulting from the redundant application is non-linearly corrected so as not to exceed the physical emission limit of the corresponding node.

[0023] The method further includes a multi-objective optimization step for deriving a ranked path list by applying a multi-objective optimization technique combining a cost function and a reduction effect function to the remaining feasible combinations mentioned above.

[0024] The present invention comprises: an information storage step in which an ontology repository stores a regulation-specific sub-knowledge graph in which instances of product category rules and energy source attribution rules of individual carbon export regulations are constructed as sub-instances of the carbon export regulation domain layer, wherein the ontology repository includes a domain-independent shared abstraction layer that defines universal concepts of supply chain, products, and raw materials, and a carbon export regulation domain layer that inherits from the domain-independent shared abstraction layer and defines an emission range classification system and a carbon raw material specification model; a C-BOM input processing step in which a C-BOM input receives a carbon raw material specification of a target product based on the item classification code of the target product and the carbon raw material specification model, and identifies a set of multiple carbon export regulations applicable to the target product based on the item classification code and the carbon raw material specification; and a matrix construction step in which an executable combination derivation unit constructs an effect projection matrix by mapping the emission reduction amount calculated under each regulation within the identified set of multiple carbon export regulations to each reduction measure candidate within a set of reduction measure candidates including multiple reduction measure candidates applicable to the carbon raw material specification. and the executable combination derivation unit analyzes the effect projection matrix to calculate the reduction effect deviation ratio between each regulation for the same reduction measure candidate, and identifies a specific reduction measure candidate within the set of reduction measure candidates as a regulation deviation measure based on the calculated reduction effect deviation ratio; the method includes a regulation deviation measure identification step.

[0025] The above step of identifying regulatory deviation measures is performed based on a SupplyChainEvidence model specialized for the carbon export regulation domain, inherited from a shared upper abstraction layer built in the above ontology repository, and cross-validates the validity of carbon evidence data submitted by suppliers for specific reduction measure candidates included in the carbon raw material specification against the requirements of the above regulation-specific lower knowledge graph; if the verification result indicates that the reduction effect is not recognized because the evidence requirements demanded by the specific regulation are not met, the corresponding reduction measure candidate is detected as a regulatory deviation measure. Effects of the invention

[0026] According to one embodiment of the present invention, it provides the effect of flexibly integrating and processing fragmented multi-jurisdictional carbon regulatory requirements based on a multi-layered ontology structure, thereby overcoming the limitations of a hardcoded rule engine.

[0027] In addition, according to one embodiment of the present invention, when results between regulations conflict, it has the effect of supporting companies in establishing fundamental response logic by transparently identifying the causes of conflicting attribution rules through semantic queries.

[0028] In addition, according to one embodiment of the present invention, when a result of non-compliance occurs, the logical path of the knowledge graph is traced back to derive specific improvement conditions for raw materials or energy sources, thereby having the effect of improving proactive regulatory responsiveness.

[0029] In addition, according to one embodiment of the present invention, there is a robust system operation effect that ensures subsequent emission calculation by flexibly assigning alternative default values ​​without system interruption, even in practical environments where data is not fully available.

[0030] In addition, according to one embodiment of the present invention, by operating a domain-independent shared abstraction layer and a regulation-specific knowledge graph separately, there is an effect of reducing costs by expanding only the lower instances without completely rebuilding the system, even if regulations from new countries are added in the future.

[0031] In addition, according to one embodiment of the present invention, heterogeneous rule nodes between regulations are accurately detected through query search, thereby having the effect of identifying key factors that cause deviation in judgment results in an automated manner.

[0032] In addition, according to one embodiment of the present invention, there is an effect of objectively deriving the minimum carbon strength adjustment value and the optimal process energy reduction amount by reverse inferring the violated constraint conditions.

[0033] In addition, according to one embodiment of the present invention, a flawless data correction effect is achieved by actively adjusting the constraint warning level in the event of supply chain data omission and safely injecting default penalty values ​​for each regulation.

[0034] In addition, according to one embodiment of the present invention, there is an effect of maximizing strategic decision-making by exploring a three-dimensional reduction path for simultaneous satisfaction of multiple regulations through an effect projection matrix rather than single regulation optimization.

[0035] In addition, according to one embodiment of the present invention, there is an effect of maintaining the efficiency and consistency of reduction investments by pre-filtering regulatory deviation measures in which specific measures are disadvantageous only to some regulations.

[0036] In addition, according to one embodiment of the present invention, by precisely analyzing physical node overlap and semantic complementary relationships through an ontology attribute chain, there is an effect of preventing conflicting reduction investments and prioritizing the derivation of synergistic combinations.

[0037] In addition, according to one embodiment of the present invention, instead of unconditionally discarding regulatory deviation measures, alternative measure candidates are proposed through reverse analysis, thereby minimizing the loss of reduction opportunities.

[0038] In addition, according to one embodiment of the present invention, there is an effect of mathematically suppressing the contradiction of exceeding physical emission limits that occurs when simply summing the effects of multiple reduction measures, thereby providing a highly reliable calculation result that is realistically achievable.

[0039] In addition, according to one embodiment of the present invention, there is an effect of providing a ranked customized solution that maximizes the greenhouse gas reduction effect relative to the company's available budget cost through a multi-objective optimization function. Brief explanation of the drawing

[0040] FIG. 1 is a diagram schematically showing the relationships between the components of a product carbon footprint management system according to one embodiment of the present invention. FIG. 2 is a diagram schematically showing the relationship between the configuration of an inference engine and an ontology repository, etc., according to one embodiment of the present invention. FIG. 3 is a diagram schematically showing the relationship between an inference engine and a C-BOM input processing unit, etc., according to one embodiment of the present invention. FIG. 4 is a diagram schematically illustrating a carbon footprint compliance determination method according to one embodiment of the present invention. FIGS. 5 to 7 are drawings showing actual applied web pages of a management system according to an embodiment of the present invention. Specific details for implementing the invention

[0041] As illustrated in FIG. 1, a product carbon footprint management system that supports the determination of product carbon footprint compliance and the derivation of an optimal reduction path in a multiple carbon export regulation environment according to one embodiment of the present invention includes an ontology repository (110), a C-BOM input processing unit (120), a graph query processing unit (130), an inference engine (140), an executable combination derivation unit (150), and a result output unit (160).

[0042] The above ontology repository (110) has a domain-independent shared abstraction layer that defines universal concepts of supply chains, products, and raw materials, and stores a carbon export regulation domain layer that inherits from the domain-independent shared abstraction layer and defines an emission range classification system and a carbon raw material specification model. The above ontology repository (110) manages and provides a knowledge model with a multi-layered structure by linking a regulation-specific sub-knowledge graph, which is constructed as a sub-instance of the carbon export regulation domain layer, with instances of product category rules and energy source attribution rules of individual carbon export regulations.

[0043] As illustrated in FIG. 2, the inference engine (140) according to one embodiment of the present invention identifies the cause of the discrepancy between the judgment results of the first regulation and the second regulation belonging to the identified set of multiple carbon export regulations at the semantic level when the judgment results of the first regulation and the second regulation are different.

[0044] The inference engine (140) calls a query to retrieve instances of energy source attribution rules defined in the regulation-specific sub-knowledge graph of the ontology repository (110).

[0045] The inference engine (140) cross-compares the opposing points of the reduction recognition attribute applied to the first regulation and the reduction non-recognition attribute applied to the second regulation through the query. Through this, the inference engine (140) derives a specific energy source that caused the difference in the judgment result, identifies and extracts it as the cause of the result difference to ensure the integrity of the logical basis, and transmits it to the result output unit (160).

[0046] As illustrated in FIG. 3, the inference engine (140) according to one embodiment of the present invention performs reverse inference to resolve the result of non-compliance with the specific regulation when the result of non-compliance is produced.

[0047] The inference engine (140) traces back the inference path recorded in the regulation-specific sub-knowledge graph to identify the violated constraints.

[0048] Based on the above specified constraints, the inference engine (140) derives the minimum C-BOM condition for achieving compliance through reverse inference, which includes at least one of the carbon strength replacement condition of the raw material, the process unit energy consumption reduction condition, and the energy source conversion condition.

[0049] According to one embodiment of the present invention, the C-BOM input processing unit (120) activates an intelligent alternative path mechanism in a data missing situation, verifies supply chain evidence data for a specific raw material within the carbon raw material specification received, and if the supply chain evidence data is missing as a result of the verification, processes the missing state at a warning level of the constraint language to prevent input invalidation.

[0050] Subsequently, the above C-BOM input processing unit (120) triggers an alternative path mechanism that automatically assigns the default value of the corresponding regulation defined in the regulation-specific sub-knowledge graph to the attribute of the missing specific raw material, thereby ensuring uninterrupted judgment operations.

[0051] As illustrated in FIG. 4 (see FIG. 5 to 7), a method for determining product carbon footprint compliance according to one embodiment of the present invention first performs an information storage step of storing multilayer ontology information through the ontology storage (110).

[0052] In the above information storage step, the domain-independent shared abstraction layer and the carbon export regulation domain layer are built and stored in the ontology repository (110) (S110).

[0053] In the subsequent C-BOM input processing step, the item classification code of the target product and the carbon raw material specification are input by the C-BOM input processing unit (120), and based on this, the set of multiple carbon export regulations to be applied is identified (S120).

[0054] In the graph query processing step, PCR instances corresponding to each regulation identified by the graph query processing unit (130) are queried in the regulation-specific sub-knowledge graph, and subsequently, an inference step by the inference engine (140) and a result output step by the result output unit (160) are performed sequentially (S130, S140, S160).

[0055] In the inference step according to one embodiment of the present invention, a query-based analysis function for the difference in judgment results through the inference engine (140) is performed.

[0056] The inference engine (140) cross-searches using the reduction recognition attribute of the first regulation instance and the reduction non-recognition attribute of the second regulation instance as opposing points when the judgment results of the first regulation and the second regulation are different.

[0057] Through this, the inference engine (140) identifies a specific energy source node to which heterogeneous rules between regulations are applied, thereby clearly identifying and deriving the cause of the difference in the result.

[0058] The inference step according to one embodiment of the present invention includes a process in which the inference engine (140) derives an optimal response condition through reverse tracing based on the non-compliance determination result.

[0059] The inference engine (140) utilizes the inference path recorded within the regulation-specific sub-knowledge graph of the specific regulation from which the non-compliance judgment result was produced to conduct reverse inference, thereby tracing back the violated constraint and generating and deriving the minimum C-BOM conditions for achieving compliance, such as the raw material carbon strength substitution conditions described above.

[0060] The C-BOM input processing step according to one embodiment of the present invention includes a process in which the C-BOM input processing unit (120) performs a correction procedure for defect-free calculation when it detects a missing supply chain data for the specific raw material.

[0061] The above C-BOM input processing unit (120) prevents forced termination by processing the data omission status based on the warning attribute of the constraint language.

[0062] Next, the C-BOM input processing unit (120) automatically injects a penalty default value defined by regulation and triggers and executes the alternative path mechanism that assigns it to the missing attribute, thereby ensuring the integrity and stability of subsequent operations.

[0063] In an embodiment of the present invention, the method for deriving an optimal reduction path involves the feasible combination derivation unit (150) performing a matrix construction step to construct an effect projection matrix. In the matrix construction step, the feasible combination derivation unit (150) maps the emission reduction amount calculated under the set of multiple carbon export regulations to each reduction measure candidate within the set of reduction measure candidates, which includes multiple reduction measure candidates applicable to the carbon raw material specification. In the subsequent combination derivation step, the feasible combination derivation unit (150) analyzes the effect projection matrix to identify effect deviations between regulations and performs in-depth detection of interactions between each reduction measure candidate to finally derive and output the set of feasible combinations that are physically and technically coexistable.

[0064] The combination derivation step according to one embodiment of the present invention includes the function of the executable combination derivation unit (150) identifying the regulatory deviation measure.

[0065] The above-mentioned executable combination derivation unit (150) calculates the ratio of reduction effect deviation between the regulations for the same candidate based on the reduction effect for each reduction measure candidate recorded in the effect projection matrix.

[0066] The above-mentioned executable combination derivation unit (150) identifies and marks the above-mentioned specific reduction measure candidate that causes a relatively severe deviation compared to other regulations as the above-mentioned regulation deviation measure based on the above-mentioned calculated reduction effect deviation ratio.

[0067] According to one embodiment of the present invention, the combination derivation step utilizes semantic web technology to deeply analyze the interactions of reduction measures and optimize the search space. In the interaction detection step, the executable combination derivation unit (150) compares the nodes to which the first reduction measure candidate and the second reduction measure candidate are applied within the set of reduction measure candidates to determine whether there exists a set of physically overlapping nodes.

[0068] Furthermore, the executable combination derivation unit (150) utilizes an attribute chain, which is an ontology axiom system, to track, detect, and mark hidden complementary relationships between nodes that are semantically connected between the first reduction measure candidate and the second reduction measure candidate, even though there is no physical overlap. In the pruning stage, the executable combination derivation unit (150) prunes and excludes combinations of competitive relationships based on the physical overlap determination in the search space, and classifies the measures marked as complementary relationships as priority search target combinations to maximize optimization computation efficiency.

[0069] A method for deriving an optimal reduction path according to one embodiment of the present invention performs a reverse analysis to replace the identified regulatory deviation measure after identifying the regulatory deviation measure.

[0070] The above-mentioned executable combination derivation unit (150) derives a new alternative measure candidate that can minimize the reduction effect deviation ratio by reverse-examining the energy source attribution rule instance of the regulation-specific sub-knowledge graph corresponding to the above-mentioned specific reduction measure candidate when the above-mentioned specific reduction measure candidate is identified as the above-mentioned regulatory deviation measure.

[0071] Subsequently, the above-mentioned executable combination derivation unit (150) updates the above-mentioned effect projection matrix based on the above-mentioned alternative action candidates.

[0072] A method for deriving an optimal reduction path according to one embodiment of the present invention performs numerical correction considering the physical limit of the reduction effect through an additivity correction step. The feasible combination derivation unit (150) determines cases where multiple reduction measures are applied redundantly to the same node of the carbon raw material specification within feasible combinations remaining in the set of feasible combinations derived after the pruning step.

[0073] The above executable combination derivation unit (150) non-linearly attenuates and corrects the reduction effect value so that the simple sum does not exceed the physical emission limit of the corresponding node, in order to prevent errors that occur when the reduction effect resulting from the overlapping application is simply added.

[0074] A method for deriving an optimal reduction path according to one embodiment of the present invention determines a final optimal path through a multi-objective optimization step. The feasible combination derivation unit (150) applies a multi-objective optimization technique to the remaining feasible combinations and the combinations for which additivity correction has been completed, by combining a cost function associated with the execution of reduction measures and a reduction effect function that is actually manifested.

[0075] Through this, the above-mentioned feasible combination derivation unit (150) derives a ranked list of paths that maximize cost efficiency while achieving simultaneous satisfaction of multiple regulations and reflects this in the above-mentioned integration result.

[0076] Simultaneous response to multiple regulations (EU CBAM and UK CBAM) and interaction detection for steel products

[0077] Referring to FIGS. 5 to 7, this embodiment assumes an environment in which the target product is reinforcing steel (CN code 7214.20) and is simultaneously exported to the European Union (EU) and the United Kingdom (UK).

[0078] First, the C-BOM input processing unit (120) receives a carbon raw material specification containing information on the product's CN code, raw material carbon strength, coke input amount, and electric furnace power consumption (assuming 80% of the Korean grid power and 20% of the virtual PPA).

[0079] Based on the regulation-specific sub-knowledge graph stored in the ontology repository (110), the actionable combination derivation unit (150) constructs an effect projection matrix for four types of applicable reduction measure candidates (low-carbon coke replacement, electric furnace process efficiency improvement, hydrogen reduction steelmaking conversion, virtual PPA additional contract).

[0080] As a result of matrix operations, according to the inference path recorded by the inference engine (140), the fourth measure (virtual PPA) is calculated to have a negligible reduction effect in both regulations. The executable combination derivation unit (150) flags this as an inefficient measure and proposes a replacement with the second measure, and prunes in the search space by identifying, through interaction detection, that the first measure and the third measure are in a competitive relationship sharing a 'coke input node'.

[0081] Finally, the result output unit (160) ranks and outputs the optimal executable candidate combinations.

[0082] Identifying conflicts in energy source attribution rules for battery cell products

[0083] This embodiment deals with a situation in which a lithium-ion battery cell (CN code 8507.60) is simultaneously exposed to EU battery regulations and EU CBAM. Data on the positive electrode material, negative electrode material, and cell assembly power (50% grid, 50% REC based) is transmitted to the system through the C-BOM input processing unit (120).

[0084] The graph query processing unit (130) queries instances of energy source attribution rules (AttributionRule) in the sub-knowledge graph of each regulation.

[0085] When a ‘REC-based power expansion’ measure is input, the inference engine (140) identifies that the EU battery regulation recognizes it but the EU CBAM rejects it, and the feasible combination derivation unit (150) determines that the measure is a ‘regulatory deviation measure’ by confirming that the difference in reduction effect between the two regulations exceeds a threshold.

[0086] Next, the inference engine (140) extracts the difference between the attribution rule instances of the two regulations at the mechanism level through query cross-search, and the result output unit (160) provides this to the user as a warning message to prevent sunk investment.

[0087] Detection of output range boundary discrepancies for aluminum products

[0088] This embodiment is a case where rolled aluminum (CN code 7606.12) is simultaneously subject to EU CBAM and domestic K-ETS Phase 3 emission allowance allocation regulations. When the C-BOM input processing unit (120) processes imported alumina and electrolytic process data,

[0089] The inference engine (140) dynamically determines the scope boundary for each regulation by utilizing PCR data from the ontology repository (110). When applying the 'low-carbon alumina replacement' measure, the inference engine (140) identifies a rule difference in the inference path, such that EU CBAM includes the raw material production stage, whereas K-ETS limits it to the domestic business site boundary.

[0090] The resulting effect deviation is defined as a regulatory deviation mechanism by the feasible combination derivation unit (150), and the system integrates the reverse inference results of the inference engine (140) for non-compliant regulations to propose an optimal reduction path through the result output unit (160) that can simultaneously resolve the compliance gap of all regulations.

[0091] This embodiment assumes a situation in which some supply chain data is missing in an environment where steel products (CN code 7214.20) are subject to multiple carbon export regulations (EU CBAM, UK CBAM, K-ETS) simultaneously.

[0092] When the C-BOM input processing unit (120) detects that supporting data for a specific overseas raw material is missing, it processes this at the level of a warning in the constraint language and automatically assigns the corresponding penalty default value defined in the regulation-specific sub-knowledge graph of the ontology repository (110) to the attribute of the missing raw material to ensure the integrity of subsequent operations.

[0093] Then, when the graph query processing unit (130) queries the PCR instances by regulation based on the completed carbon raw material specification, the inference engine (140) cross-compares the conflicting points of the energy source attribution rules between regulations regarding specific renewable energy certificates, etc., using a semantic query to identify the cause of the discrepancy in the judgment result.

[0094] If a non-compliance determination is made regarding a specific regulation, the inference engine (140) traces the inference path in reverse to calculate the minimum C-BOM conditions for achieving compliance, including the conditions for replacing the raw material carbon strength.

[0095] Subsequently, the feasible combination derivation unit (150) identifies deregulation measures by constructing an effect projection matrix for multiple candidate reduction measures and analyzing the reduction effect deviation of each measure. Furthermore, the feasible combination derivation unit (150) utilizes an ontology attribute chain to determine whether there is overlap between physical nodes affected by each measure, marks measures with overlap as competing relationships to preemptively exclude (Pruning) them from the search space, while marking measures with confirmed synergy without physical overlap as complementary relationships to perform multi-objective optimization. The result output unit (160) outputs an integrated result including the cross-comparison cause, reverse inference condition, and derived optimal reduction path to support corporate decision-making. Explanation of the symbols

[0096] 110: Ontology Repository 120: C-BOM Input Processing Unit 130: Graph Query Processing Unit 140: Inference Engine 150: Executable Combination Derivation Unit 160: Result Output Section

Claims

Claim 1 A product carbon footprint management system that supports the determination of product carbon footprint compliance and the derivation of optimal reduction paths in an environment of multiple carbon export regulations, comprising: an ontology repository that manages a regulation-specific sub-knowledge graph in which instances of product category rules and energy source attribution rules of individual carbon export regulations are constructed as sub-instances of the carbon export regulation domain layer, and a carbon export regulation domain layer that defines a carbon raw material specification model; a C-BOM input processing unit that receives input of a carbon raw material specification of a target product based on the item classification code of the target product and the carbon raw material specification model, and identifies a set of multiple carbon export regulations applicable to the target product based on the item classification code and the carbon raw material specification; a graph query processing unit that, for each regulation within the identified set of multiple carbon export regulations, queries a PCR instance corresponding to the item classification code in the regulation-specific sub-knowledge graph; and determines the attribution range of the carbon raw material specification according to the calculation range boundary defined by the queried PCR instance, determines the intrinsic carbon amount and compliance under each regulation, and records the energy source attribution rule instance applied to the determination as an inference path. A system for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment, comprising: an inference engine; a feasible combination derivation unit that, for each candidate for a reduction measure within a set of candidates for reduction measures including a plurality of candidates for reduction measures applicable to the carbon raw material specification, calculates the reduction effect for each regulation within the identified set of multiple carbon export regulations to detect interactions and derive a set of feasible combinations; and a result output unit that generates and outputs an integrated result including the determined compliance status and the derived set of feasible combinations. Claim 2 A system for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment, wherein, in the case where the determination result of a first regulation and the determination result of a second regulation belonging to the identified set of multiple carbon export regulations are different, the inference engine calls a query to cross-compare the opposing points of the reduction recognized attribute applied to the first regulation and the reduction non-recognized attribute applied to the second regulation, and based on the cross-compare result, extracts a specific energy source that caused the discrepancy in the determination result as the cause of the result difference and transmits it to the result output unit. Claim 3 A system for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment, wherein, in the case where a non-compliance determination result is produced for at least one specific regulation among the identified set of multiple carbon export regulations, the inference engine traces the violated constraint in reverse using the inference path recorded in the regulation-specific sub-knowledge graph, and derives a minimum C-BOM condition for achieving compliance, including at least one of a raw material carbon strength substitution condition, a process unit energy consumption reduction condition, and an energy source conversion condition, through reverse inference based on reverse tracing. Claim 4 A system for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment, wherein, in claim 1, the C-BOM input processing unit prevents input invalidation by processing the missing state as a warning level of the constraint language when supply chain evidence data for a specific raw material in the input carbon raw material specification is missing, and triggers an alternative path mechanism that automatically assigns the default value of the corresponding regulation defined in the regulation-specific sub-knowledge graph to the attribute of the missing raw material. Claim 5 A method for determining product carbon footprint compliance in a multiple carbon export regulation environment, comprising: an information storage step in which an ontology repository stores a regulation-specific sub-knowledge graph in which instances of product category rules and energy source attribution rules of individual carbon export regulations are constructed as sub-instances of the carbon export regulation domain layer, wherein the ontology repository includes a domain-independent shared abstraction layer that defines universal concepts of supply chain, product, and raw material, and a carbon export regulation domain layer that inherits from the domain-independent shared abstraction layer and defines an emission range classification system and a carbon raw material specification model; a C-BOM input processing step in which a C-BOM input processing unit receives a carbon raw material specification of a target product based on the item classification code of the target product and the carbon raw material specification model, and identifies a set of multiple carbon export regulations applicable to the target product based on the item classification code and the carbon raw material specification; a graph query processing step in which a graph query processing unit queries a PCR instance corresponding to the item classification code in the regulation-specific sub-knowledge graph for each regulation within the identified set of multiple carbon export regulations; and an inference engine, according to the output range boundary defined by the queried PCR instance A method for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment, comprising: an inference step of determining the attribution scope of the carbon raw material specification and determining the intrinsic carbon amount and compliance status under each regulation, wherein the instance of the energy source attribution rule applied to the determination is recorded as an inference path; and a result output step in which a result output unit generates and outputs an integrated result including the determined compliance status. Claim 6 A method for determining and managing product carbon footprint compliance in a multiple carbon export regulation environment, comprising: an information storage step in which an ontology repository stores a regulation-specific sub-knowledge graph in which instances of product category rules and energy source attribution rules of individual carbon export regulations are constructed as sub-instances of the carbon export regulation domain layer, wherein the ontology repository includes a domain-independent shared abstraction layer that defines universal concepts of supply chain, product, and raw material, and a carbon export regulation domain layer that inherits from the domain-independent shared abstraction layer and defines an emission scope classification system and a carbon raw material specification model; a C-BOM input processing step in which a C-BOM input receives the item classification code of the target product and the carbon raw material specification of the target product based on the carbon raw material specification model, and identifies a set of multiple carbon export regulations applicable to the target product based on the item classification code and the carbon raw material specification; and an executable combination derivation unit maps the emission reduction amount calculated under each regulation within the identified set of multiple carbon export regulations to each reduction measure candidate within a set of reduction measure candidates including multiple reduction measure candidates applicable to the carbon raw material specification, thereby determining the effect A method for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment, characterized by comprising: a matrix construction step for constructing a projection matrix; and a regulatory deviation measure identification step in which the executable combination derivation unit analyzes the effect projection matrix to calculate the ratio of reduction effect deviation between each regulation for the same reduction measure candidate, and identifies a specific reduction measure candidate within the set of reduction measure candidates as a regulatory deviation measure based on the calculated ratio of reduction effect deviation. Claim 7 The method for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment according to claim 6, wherein the combination derivation step further comprises an interaction detection step in which the first reduction measure candidate and the second reduction measure candidate within the set of reduction measure candidates each influence a set of nodes of the carbon raw material specification to determine whether there are overlapping physical nodes, and detects and marks hidden complementary relationships by utilizing an attribute chain to track the semantic network between the first reduction measure candidate and the second reduction measure candidate, and a pruning step in which the measure combinations marked as competitive relationships are excluded from the search space and the measure combinations marked as complementary relationships are classified as priority search targets. Claim 8 In claim 6, the step of identifying regulatory deviation measures is performed based on a SupplyChainEvidence model specialized in the carbon export regulation domain, inherited from a shared upper abstraction layer built in the ontology repository; the validity of carbon evidence data submitted by a supplier for a specific reduction measure candidate included in the carbon raw material specification is cross-validated with the requirements of the regulation-specific lower knowledge graph; and if the verification result indicates that the evidence requirements demanded by the specific regulation are not met and the reduction effect is not recognized, the corresponding reduction measure candidate is detected as a regulatory deviation measure. This characterizes a method for determining product carbon footprint compliance and deriving an optimal reduction path in a multi-carbon export regulation environment.

Images