A Life Cycle Assessment Method and System Based on Heterogeneous LCI Data

By performing structured processing and global unique identifier reconstruction on heterogeneous LCI data, the problems of conflict and redundancy in data fusion were solved, cross-source hybrid modeling and parallel accounting of multiple methodologies were realized, the availability and compliance adaptability of data were improved, and dual compliance evaluation of the electrical equipment industry was supported.

CN123115248APending Publication Date: 2026-11-17XJ GRP CORP +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610989650.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-03
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing heterogeneous LCI data fusion only performs surface format alignment, which easily leads to problems such as conflicts in universally unique identifiers, data redundancy, process link breaks, and the inability to achieve cross-source hybrid modeling and parallel computation of multiple methodologies.

Method used

By acquiring multi-source heterogeneous LCI data, performing integrity verification of the structured dataset, generating globally unique identifiers, performing synonym stream identification and deduplication, establishing mapping relationships, reconstructing globally unique identifiers of data entities, constructing a fused LCI database, and independently parsing the LCIA methodology to form a standardized dataset and methodology library.

Benefits of technology

It has achieved data format unification, avoided link breakage and redundancy, improved data availability and reliability, supported flexible accounting in multiple compliance scenarios, reduced time and manpower costs, and promoted the large-scale application of life cycle assessment technology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115248A_ABST
    Figure CN123115248A_ABST
Patent Text Reader

Abstract

This invention belongs to the field of lifecycle assessment technology and aims to solve the problems of existing heterogeneous LCI data fusion, which only performs surface format alignment and is prone to conflicts in universal unique identifiers and breaks in process links. The method of this invention includes: obtaining a structured dataset based on multi-source heterogeneous LCI data and performing integrity verification on the structured dataset; generating globally unique identifiers and establishing a first mapping relationship; performing synonym stream identification and deduplication to obtain a synonym stream dataset and a baseline stream dataset, and establishing a second mapping relationship; reconstructing the identifiers of each data entity in the structured dataset using globally unique identifiers to obtain a standardized dataset; constructing a fused LCI database based on the standardized dataset, and constructing a product lifecycle model based on the fused LCI database; performing structured parsing and unified storage of multiple LCIA methodologies to obtain an LCIA methodology library; constructing a lifecycle inventory vector based on the product lifecycle model to obtain the impact assessment accounting results and generate an accounting report.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of life cycle assessment technology, specifically relating to a life cycle assessment method and system based on heterogeneous LCI data. Background Technology

[0002] Life Cycle Assessment (LCA) is a method for summarizing and evaluating the environmental impacts and potential effects of all inputs and outputs throughout the life cycle of a product or service. It serves as a core technical support for carbon footprint accounting, domestic compliance certification, and EU ISO 14067 and CBAM cross-border carbon management. LCA generally consists of four phases: defining the purpose and scope of the analysis, Life Cycle Inventory (LCI) analysis, Life Cycle Impact Assessment (LCIA), and life cycle interpretation. Life Cycle Inventory (LCI) data refers to quantitative data on all inputs (such as raw materials and energy) and outputs (such as waste and emissions) involved in the entire life cycle of a product or service. This data is used to assess the environmental impact of the product or service, helping companies and research institutions make sustainable development decisions.

[0003] Taking electrical equipment industry products as an example, the industry currently faces significant application bottlenecks. Major international LCI databases such as ecoinvent and GaBi lack data specific to China's domestic industrial and electrical equipment sectors, while domestic LCI databases struggle to match international compliance standards and cannot directly support export declarations. Existing heterogeneous LCI data fusion only performs surface format alignment without adhering to underlying data dependencies, easily leading to issues such as Universally Unique Identifier (UUID) conflicts, data redundancy, and process link breaks. Furthermore, the LCIA methodology is strongly bound to the native LCI database, making it impossible to achieve cross-source hybrid modeling and parallel calculation of multiple methodologies. This results in redundant modeling and calculation, weak compliance adaptability, and severely restricts the implementation of dual-compliant LCA in the industry. Summary of the Invention

[0004] (a) Purpose of the invention The purpose of this invention is to provide a lifecycle assessment method and system based on heterogeneous LCI data, aiming to solve the problems of existing heterogeneous LCI data fusion, which only performs surface format alignment, easily leading to conflicts in universal unique identifiers, data redundancy, process link breaks, and the inability to achieve cross-source hybrid modeling and parallel accounting of multiple methodologies.

[0005] (II) Technical Solution To address the above problems, a first aspect of the present invention provides a lifecycle assessment method based on heterogeneous LCI data, the method comprising: Acquire multi-source heterogeneous LCI data, obtain a structured dataset based on the multi-source heterogeneous LCI data, and perform integrity verification on the structured dataset; A globally unique identifier is generated based on the structured dataset after verification, and a first mapping relationship is established between the structured dataset and the globally unique identifier; Synonym stream identification and deduplication are performed on the streaming data in the structured dataset to obtain a synonym stream dataset and a baseline stream dataset, and a second mapping relationship is established between the synonym stream dataset and the baseline stream dataset. Based on the first and second mapping relationships, the identifiers of each data entity in the structured dataset are reconstructed using globally unique identifiers, and the association binding between data entities at all levels is completed to obtain a standardized dataset. A fused LCI database is constructed based on the standardized dataset, and a product lifecycle model is constructed based on the fused LCI database; Multiple LCIA methodologies are acquired, and the LCIA methodologies are structured, parsed, and stored uniformly to obtain an LCIA methodology library. Based on the product lifecycle model, a lifecycle inventory vector is constructed. Based on the lifecycle inventory vector and the LCIA methodology library, the impact assessment accounting results are obtained and an accounting report is generated.

[0006] Preferably, the multi-source heterogeneous LCI data includes domestic LCI datasets and international LCI datasets, and the process of obtaining a structured dataset based on the multi-source heterogeneous LCI data includes: Based on the aforementioned domestic LCI dataset, the original structure was parsed without loss, and all core fields of process, stream, stream attributes, unit group, and data source were extracted. Based on the aforementioned international LCI dataset, the core fields of the internationally common format are mapped and converted into structured data in the platform's standard format; The structured dataset is obtained based on the full set of core fields and the structured data.

[0007] Preferably, the integrity verification of the structured dataset includes: Based on the number of required fields according to platform standards included in the structured dataset and the total number of required fields preset by the platform standards, an integrity check is performed. The formula for the integrity check is: Where n is the number of required fields according to platform standards in the structured dataset, and N is the total number of required fields specified by the platform standards. If the conversion rate is ≥90%, the format conversion is considered successful.

[0008] Preferably, generating a globally unique identifier based on the validated structured dataset includes: Obtain the type code, data source identifier, key business attribute fields, and timestamp field of the data entities in the structured dataset; The type code, data source identifier, key business attribute field, and timestamp field are concatenated to obtain the concatenated string; A hash operation is performed on the concatenated string to obtain a hash digest. The hash digest is mapped to a globally unique identifier in a standard format.

[0009] Preferably, the step of performing synonym stream identification and deduplication on the streaming data in the structured dataset to obtain a synonym stream dataset and a baseline stream dataset, and establishing a second mapping relationship between the synonym stream dataset and the baseline stream dataset, includes: The classification field, stream type field, and stream attribute field of the streaming data are standardized. Based on the standardized classification, stream type, and stream attribute, a bucketing key is generated. The streaming data is then allocated to the corresponding buckets according to the bucketing key. Within the same bucket, when the classification, stream type, and stream attributes of two streams are consistent, the two streams are determined to be candidate synonym streams. The names of the two streams are semantically similar. When the semantic similarity reaches a preset threshold, the two streams are determined to be synonym streams. The reference stream data is retained in the synonym stream, a synonym stream dataset is generated based on all the synonym stream data, a reference stream dataset is generated based on all the reference stream data, and a second mapping relationship is established between the synonym stream dataset and the reference stream dataset.

[0010] Preferably, based on the first and second mapping relationships, the identifiers of each data entity in the structured dataset are reconstructed using globally unique identifiers, and the association binding between data entities at each level is completed to obtain a standardized dataset, including: Based on the first mapping relationship, the original identifier of the unit group entity in the structured dataset is replaced with a globally unique identifier, thereby completing the unification of the identifier of the underlying unit group; Based on the first mapping relationship, the native identifier of the flow attribute entity in the structured dataset is replaced with a globally unique identifier, and the unit group reference associated with the flow attribute is updated to the globally unique identifier of the unit group, thus completing the association and binding between the flow attribute and the unit group; Based on the first mapping relationship, the native identifier of the flow entity in the structured dataset is replaced with a globally unique identifier. Based on the second mapping relationship, the references of synonymous flows are uniformly pointed to the identifier of the corresponding base flow. Based on the first mapping relationship, the flow attribute reference associated with the base flow is updated to the globally unique identifier of the flow attribute, thus completing the association and binding between the flow and the flow attribute. Obtain the unit process entity, replace the input and output streams of the unit process entity with the identifiers of the corresponding reference streams based on the second mapping relationship, update the identifiers of the input and output streams as the corresponding reference streams with the globally unique identifiers of the reference streams based on the first mapping relationship, complete the association and binding between the unit process entity and the stream, and obtain the standardized dataset with consistent association across all levels.

[0011] Preferably, the step of constructing a fused LCI database based on the standardized dataset and constructing a product lifecycle model based on the fused LCI database includes: The full standardized dataset with completed identifier reconstruction and association binding is stored in a unified storage structure, with the globally unique identifier as the unique primary key, and multi-dimensional joint indexes are established according to the dataset type to form a fused LCI database that supports mixed references across different original data sources; Set core parameters for product modeling, including product functional units, lifecycle system boundaries and target compliance scenarios. Based on the core parameters, filter eligible unit processes and flow data from the fused LCI database to lock the reference output product flow of the target product. By mixing and referencing unit processes and stream data from different original data sources, and based on the unit processes corresponding to the reference output product stream, a product lifecycle topology is built through the relationship between input streams, output streams and upstream and downstream processes. Based on the entire path of the lifecycle topology, input flow nodes without corresponding upstream supply processes are identified, and the optimal upstream unit process is matched from the fused LCI database to complete the supply chain path, thus obtaining the product lifecycle model.

[0012] Preferably, the step of acquiring multiple LCIA methodologies, performing structured parsing and unified storage on the multiple LCIA methodologies to obtain an LCIA methodology library includes: We acquire multiple LCIA methodologies corresponding to different compliance scenarios, perform structured analysis on each LCIA methodology, and extract the impact type, environmental flow definition, characteristic factor parameters, and compliance applicability scope of each LCIA methodology. The impact categories, basic flows, and version numbers of each LCIA methodology are stored in a unified storage structure and preloaded into memory; Based on the unified storage structure and the data structure preloaded into memory, the LCIA methodology library is obtained.

[0013] Preferably, the step of constructing a lifecycle inventory vector based on the product lifecycle model, obtaining impact assessment accounting results based on the lifecycle inventory vector and the LCIA methodology library, and generating an accounting report includes: Based on the product lifecycle model, the environmental flow exchange volume of each process is summarized to construct a lifecycle inventory vector; Based on the lifecycle inventory vector, feature factors are matched for the selected LCIA methodologies in the LCIA methodology library to construct feature factor matrices corresponding to the various LCIA methodologies. A multi-threaded parallel approach is adopted to perform matrix multiplication operations simultaneously, resulting in a vector of calculation results from various LCIA methodologies. Based on the calculated result vector, an impact category integrity check is performed and an accounting report is generated.

[0014] According to another aspect of the present invention, a lifecycle assessment system based on heterogeneous LCI data is provided, the system comprising: The data access module is used to acquire multi-source heterogeneous LCI data, obtain a structured dataset based on the multi-source heterogeneous LCI data, and perform integrity verification on the structured dataset. The global identifier management module is used to generate a globally unique identifier based on the structured dataset after verification, and to establish a first mapping relationship between the structured dataset and the globally unique identifier. The streaming data deduplication module is used to perform synonym stream identification and deduplication on the streaming data in the structured dataset to obtain a synonym stream dataset and a baseline stream dataset, and to establish a second mapping relationship between the synonym stream dataset and the baseline stream dataset. The identifier reconstruction and association module is used to reconstruct the globally unique identifiers of each data entity in the structured dataset based on the first mapping relationship and the second mapping relationship, and to complete the association binding between data entities at all levels to obtain a standardized dataset. The database management and product modeling module is used to build a fused LCI database based on the standardized dataset and to build a product lifecycle model based on the fused LCI database. The methodology management module is used to acquire various LCIA methodologies, perform structured parsing and unified storage of these LCIA methodologies, and obtain an LCIA methodology library. The accounting and reporting module is used to construct a life cycle inventory vector based on the product life cycle model, obtain the impact assessment accounting results based on the life cycle inventory vector and the LCIA methodology library, and generate an accounting report.

[0015] (III) Beneficial Effects The above-described technical solution of the present invention has the following beneficial technical effects: 1. By constructing a globally unified identifier system, combined with synonym stream deduplication and hierarchical association binding, deep standardization is achieved from the data identifier layer and entity association layer. This ensures that the format of the merged data is uniform, and its internal reference relationships and hierarchical dependencies are completely consistent, avoiding problems such as broken links and invalid references, and improving the availability and reliability of the merged LCI data.

[0016] 2. The methodology is parsed independently and stored in a structured manner. It relies on a globally unified stream identifier to adapt the methodology to any LCI data, breaking the dependency of the methodology on the native database. Users can freely choose the combination of methodologies according to different compliance needs, without having to build a product model or separately collect statistical list data for each methodology. This improves the accounting flexibility in multiple compliance scenarios and can simultaneously meet the methodology adaptation needs of different scenarios such as domestic certification and EU declaration.

[0017] 3. It covers the complete lifecycle assessment process from raw data access, data fusion and governance, product modeling to impact assessment and report output. Data standards are unified and flow is smooth at each stage, eliminating the need for switching between multiple tools and manual data migration. Addressing the industry pain points of electrical equipment and other sectors facing both domestic carbon accounting requirements and EU cross-border carbon control, the integrated database is compatible with both accurate local data and international compliant data, and the methodology library is adaptable to both domestic and international assessment standards. This allows a single product model to output multiple sets of compliance results, reducing the time and manpower costs of dual compliance assessments and effectively promoting the large-scale application of lifecycle assessment technology in the industry. Attached Figure Description

[0018] Figure 1 This is a flowchart of a lifecycle assessment method based on heterogeneous LCI data provided by the present invention; Figure 2 This is a flowchart of a lifecycle assessment method based on heterogeneous LCI data provided by the present invention; Figure 3 This is a flowchart illustrating the mapping and binding process in a lifecycle assessment method based on heterogeneous LCI data provided by this invention. Figure 4 This is a flowchart illustrating the calculation and analysis process in a lifecycle assessment method based on heterogeneous LCI data provided by this invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0020] Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0021] In the description of this invention, it should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0022] Combination Figures 1 to 4 The first aspect of the present invention provides a lifecycle assessment method based on heterogeneous LCI data, the method comprising: Acquire multi-source heterogeneous LCI data, obtain a structured dataset based on the multi-source heterogeneous LCI data, and perform integrity verification on the structured dataset; A globally unique identifier is generated based on the validated structured dataset, and a first mapping relationship is established between the structured dataset and the globally unique identifier. Synonym stream identification and deduplication are performed on streaming data in structured datasets to obtain synonym stream datasets and baseline stream datasets, and a second mapping relationship is established between the synonym stream dataset and the baseline stream dataset. Based on the first and second mapping relationships, the identifiers of each data entity in the structured dataset are reconstructed with globally unique identifiers, and the association binding between data entities at all levels is completed to obtain a standardized dataset. A fusion LCI database is built based on a standardized dataset, and a product lifecycle model is built based on the fusion LCI database. Acquire multiple LCIA methodologies, perform structured parsing and unified storage of these methodologies, and obtain an LCIA methodology library; A lifecycle inventory vector is constructed based on the product lifecycle model. The impact assessment accounting results are obtained based on the lifecycle inventory vector and the LCIA methodology library, and an accounting report is generated.

[0023] First, LCI data from different sources and in different formats are uniformly processed into structured data and quality verified to resolve heterogeneity at the format level. Second, a globally unique identifier and a two-layer mapping system are used to resolve identifier conflicts and synonym stream redundancy issues between different data sources. Third, a standardized, integrated LCI database is formed through hierarchical identifier reconstruction and association binding, supporting cross-data source hybrid modeling. Finally, the LCIA methodology is decoupled from the native database to form an independent methodology library, completing multi-method accounting and outputting reports based on a unified lifecycle inventory vector. The entire process follows the underlying dependencies of LCI data and the business logic of LCA accounting, achieving a closed-loop process from raw data to compliance reports. This approach constructs a globally unified identifier system, combined with synonym stream deduplication and hierarchical association binding, achieving deep standardization from the data identifier layer and entity association layer. This ensures a unified data format after integration, with completely consistent internal reference relationships and hierarchical dependencies, avoiding issues such as broken links and invalid references, thus improving the usability and reliability of the integrated LCI data. By independently parsing and structurally storing methodologies, and relying on a globally unified flow identifier to adapt methodologies to any LCI data, the dependency of methodologies on native databases is broken. Users can freely choose methodology combinations according to different compliance needs, without having to build separate product models or separately compile list data for each methodology. This improves the accounting flexibility in multiple compliance scenarios and can simultaneously meet the methodological adaptation needs of different scenarios such as domestic certification and EU declaration. It covers the complete life cycle assessment process from raw data access, data fusion and governance, product modeling to impact assessment and report output. Data standards are unified and flow is smooth at each stage, eliminating the need for switching between multiple tools and manual data migration. Addressing the industry pain points of electrical equipment and other industries facing both domestic carbon accounting requirements and EU cross-border carbon control, the integrated database can simultaneously accommodate accurate local data and international compliant data, and the methodology library can simultaneously adapt to domestic and international evaluation standards. This allows a single product model to output multiple sets of compliance results, reducing the time and manpower costs of dual compliance assessments and effectively promoting the large-scale application of life cycle assessment technology in the industry.

[0024] In a preferred embodiment, the multi-source heterogeneous LCI data includes domestic LCI datasets and international LCI datasets. A structured dataset is obtained based on this multi-source heterogeneous LCI data, including: Based on the domestic LCI dataset, the original structure is parsed without loss, and all core fields of process, stream, stream attribute, unit group and data source are extracted. Based on the international LCI dataset, the core fields of the internationally common format are mapped and converted into structured data in the platform's standard format; A structured dataset is obtained based on all core fields and structured data.

[0025] Specifically, for the domestic LCI dataset, a lossless parsing strategy is adopted to directly read its original data structure and fully extract all core fields, including process, stream, stream attributes, unit groups, and data source, without field pruning or information loss, ensuring the integrity of the local data. For the international LCI dataset, a format mapping strategy is adopted, pre-building a field mapping engine between the native format and the platform standard format to convert the core fields of the international data one-to-one into the platform standard format, focusing on ensuring the consistency of data semantics and values. After processing through their respective conversion paths, both types of data are uniformly imported into the structured dataset, providing a unified foundation for subsequent processes. The principle behind this differentiated processing is that the native format of domestic data has stronger compatibility with platform standards, and lossless parsing can maximize the preservation of data value; international data has significant format differences, and the mapping engine can achieve efficient and accurate format conversion.

[0026] This approach ensures that the domestic LCI dataset contains numerous detailed fields and industry-specific attributes adapted to local industrial scenarios. This forms the core foundation for guaranteeing that product accounting results align with actual domestic production conditions. By employing a lossless parsing strategy, all core fields of the domestic dataset are fully preserved. This not only supports basic LCA accounting but also retains extended information such as industry classification, production processes, and regional characteristics. This provides data support for subsequent functions such as data quality grading, compliance scenario adaptation, and intelligent process recommendation, maximizing the application value of local LCI data. International mainstream LCI databases, through long-term accumulation, possess general unit processes covering basic industrial processes, raw material extraction, and transportation globally. They are crucial data sources for supplementing domestic data gaps and meeting international compliance requirements. Through a dedicated format mapping engine, internationally common LCI data can be accurately converted to the platform's standard format without manual data entry and adaptation. This enables batch and efficient access to international data, allowing the system to simultaneously integrate high-quality domestic and international data resources. This avoids the limitations of a single data source; for example, missing raw material extraction data in China can be supplemented by international databases, and missing local electrical equipment manufacturing data in international databases can be supplemented by domestic databases, forming a complementary and integrated data system.

[0027] In a preferred embodiment, integrity verification of the structured dataset includes: Based on the number of required fields according to platform standards included in the structured dataset and the total number of required fields preset by the platform standards, an integrity check is performed. The formula for integrity check is: Where n is the number of required fields according to platform standards in the structured dataset, and N is the total number of required fields specified by the platform standards. If the conversion rate is ≥90%, the format conversion is considered successful.

[0028] Specifically, a list of required fields for the platform standard is pre-defined, clearly defining the fields that different types of LCI data entities must include, serving as a verification benchmark. After data conversion, the number of required fields actually included in each dataset is automatically counted, and the result is used to calculate the format completeness value using the completeness calculation formula. The calculation result is compared with a preset threshold; only when the completeness reaches or exceeds the threshold is the dataset deemed to have passed the format conversion and can proceed to subsequent processing stages; otherwise, it is intercepted and returned for correction. Through a quantitative indicator system, the quality of data access is rigidly controlled, transforming the ambiguous standard of format compliance into a quantifiable and automatically judged numerical standard. This ensures that the data entering the subsequent fusion stage has basic completeness, avoiding errors in subsequent processing or distortion of calculation results due to missing fields.

[0029] This method, using a quantified integrity verification formula, allows for the establishment of clear and unified quality access standards for the data access process. It filters out substandard data with severe field deficiencies at the source, ensuring that data entering subsequent processes possesses basic integrity. This pre-emptive quality control mechanism significantly reduces the cost of subsequent data governance, prevents low-quality data from contaminating the entire fusion database, and guarantees the overall data quality level of the database. Simultaneously, setting reasonable integrity thresholds ensures the integrity of core fields, supporting subsequent normal modeling and calculations, while allowing for a small number of missing non-core fields. This fully preserves usable data resources and maximizes the utilization of multi-source data resources without affecting core usability. It avoids insufficient database coverage due to overly stringent quality requirements, especially for niche unit processes in specific industries, ensuring availability while incorporating them into the fusion database.

[0030] In a preferred embodiment, a globally unique identifier is generated based on the validated structured dataset, including: Obtain the type code, data source identifier, key business attribute fields, and timestamp field of the data entities in the structured dataset; The type code, data source identifier, key business attribute field, and timestamp field are concatenated to obtain the concatenated string; A hash digest is obtained by performing a hash operation on the concatenated string. Globally unique identifiers based on hash digest mapping in a standard format.

[0031] Specifically, firstly, fixed-length type codes are assigned to different types of LCI data entities to distinguish different entity categories such as unit groups, flow attributes, flows, and unit processes. Then, a unique source identifier is assigned to each original data source to mark the original origin of the data. Next, key business attribute fields for each entity are extracted, such as process name, industry classification, flow name, and CAS number, as core elements representing the entity's business characteristics. A standard timestamp is then added to record the time of the first data access. The four types of fields are concatenated sequentially into a complete string, and a fixed-length hash digest is generated using a hash algorithm, ultimately mapping to a globally unique identifier conforming to a standard format. This combination of multi-dimensional features ensures the global uniqueness of the identifier. Furthermore, since the same input will generate the exact same hash result, consistent identifiers are generated when the same data is repeatedly accessed, avoiding duplicate creation. The key business attribute fields include process name, industry classification, and reference product flow; for flow data entities, the key business attribute fields include flow name, flow type, and chemical formula.

[0032] This method, based on multi-dimensional elements such as type encoding, source identifier, business attributes, and timestamps, generates a global identifier that ensures entities from different data sources and of different types have completely unique identifier codes. This thoroughly resolves conflicts caused by incompatibility of native identifiers. Furthermore, the hash mapping generation logic is deterministic; when the characteristic elements of the same entity remain unchanged, the generated global identifier is always consistent. Even if data is repeatedly accessed, no new identifier entries will be generated, ensuring the stability of data identifiers and avoiding database redundancy caused by duplicate data. The global identifier generation elements include entity type and data source information, eliminating the need for additional data detail queries. Entity categories and original data sources can be quickly identified solely through the identifier, facilitating subsequent data management and traceability verification.

[0033] In a preferred embodiment, synonym stream identification and deduplication are performed on the streaming data in the structured dataset to obtain a synonym stream dataset and a baseline stream dataset. A second mapping relationship is then established between the synonym stream dataset and the baseline stream dataset, including: The classification, stream type, and stream attribute fields of the streaming data are standardized. Based on the standardized classification, stream type, and stream attribute, a bucketing key is generated, and the streaming data is allocated to the corresponding bucket according to the bucketing key. Within the same bucket, when two streams have the same classification, stream type, and stream attributes, the two streams are identified as candidate synonyms. The names of the two streams are semantically similar. When the semantic similarity reaches a preset threshold, the two streams are identified as synonyms. The baseline stream data is retained in the synonym stream. A synonym stream dataset is generated based on all the synonym stream data, and a baseline stream dataset is generated based on all the baseline stream data. A second mapping relationship is established between the synonym stream dataset and the baseline stream dataset.

[0034] Furthermore, the process of synonym stream identification and deduplication for streaming datasets within structured datasets also includes: Before performing semantic similarity verification on the names of the streaming data, the names of the streaming data are aligned by Chinese and English translation.

[0035] Specifically, the system first performs standardized preprocessing on the classification, stream type, and stream attribute fields of the streaming data, unifying the format and aligning Chinese and English descriptions. Then, it generates bucketing keys based on the combination of classification, stream type, and stream attributes. Hash bucketing divides the massive streaming data into multiple independent buckets, performing synonym determination only within the same bucket to avoid full pairwise comparisons. Within each bucket, hard rules are used to filter out candidate synonym streams with identical classifications, types, and attributes. Finally, textual semantic similarity is calculated on the names of the candidate streams; streams with similarity reaching a preset threshold are identified as having the same origin and being synonymous. After determination, the baseline stream is retained, and the rest are marked as synonyms and a mapping relationship is established. Bucketing reduces computational scale, hard rules quickly eliminate obviously different streams, and semantic similarity resolves name discrepancies, balancing efficiency and accuracy. It's important to note that the bucketing key is the core of table data distribution. By specifying a column as the bucketing key, the system performs a hash operation on the value of that column and then evenly distributes the data across a specified number of buckets.

[0036] This method, employing a bucketing strategy, divides streaming data into different buckets based on combinations of categories, types, and attributes. Comparisons are performed only within each bucket, reducing the overall number of comparisons by several orders of magnitude. The more buckets there are, the smaller the data volume within each bucket, resulting in a significant improvement in computational efficiency. This divide-and-conquer approach allows synonym stream deduplication to be adapted to processing scenarios with hundreds of thousands or even millions of streaming data points, avoiding performance bottlenecks due to data volume growth and ensuring the efficiency of building large-scale integrated databases. After adopting a hierarchical judgment mechanism, firstly, hard rules based on category, stream type, and stream attributes are used to ensure that the core attributes of candidate streams are completely consistent, eliminating misjudgments due to attribute differences at the source. Then, semantic similarity of names is used to identify homonyms caused by differences in Chinese and English, and differences in expression, accurately capturing homonymous streams from different data sources with different expressions. This combined judgment method reduces the false positive rate and improves the recall rate, identifying as many homonymous streams as possible while ensuring deduplication accuracy, maximizing the elimination of data redundancy. By standardizing and unifying field formats and naming languages ​​through preprocessing, stream names in different formats and languages ​​are converted into a unified standard expression, eliminating superficial format and language differences. This allows subsequent bucketing and similarity determination to focus on the substantive semantics of the stream, ensuring the ability to identify synonym streams across language and format data sources. It enables accurate deduplication and mapping of streaming data from domestic and international databases, laying the foundation for cross-source data fusion.

[0037] In a preferred embodiment, based on the first and second mapping relationships, the identifiers of each data entity in the structured dataset are reconstructed using globally unique identifiers, and the associations and bindings between data entities at all levels are completed to obtain a standardized dataset, including: Based on the first mapping relationship, the original identifiers of unit group entities in the structured dataset are replaced with globally unique identifiers, thus completing the unification of identifiers for the underlying unit groups; Based on the first mapping relationship, the native identifier of the flow attribute entity in the structured dataset is replaced with a globally unique identifier, and the unit group reference associated with the flow attribute is updated to the globally unique identifier of the unit group, thus completing the association and binding between the flow attribute and the unit group; Based on the first mapping relationship, the native identifier of the flow entity in the structured dataset is replaced with a globally unique identifier. Based on the second mapping relationship, the references of synonymous flows are uniformly pointed to the identifier of the corresponding base flow. Based on the first mapping relationship, the flow attribute references associated with the base flow are updated to the globally unique identifier of the flow attribute, thus completing the association and binding between the flow and the flow attribute. Obtain the unit process entity, replace the input and output streams of the unit process entity with the identifiers of the corresponding baseline streams based on the second mapping relationship, update the identifiers of the input and output streams to the globally unique identifiers of the baseline streams based on the first mapping relationship, complete the association and binding between the unit process entity and the stream, and obtain a standardized dataset with consistent association across all levels.

[0038] Specifically, following the underlying dependencies of LCI data—that is, unit groups are the units of measurement for flow attributes, flow attributes are the attribute definitions of flows, and flows are the input and output elements of processes—identifier reconstruction is performed layer by layer in a bottom-up order. First, the native identifier of the lowest-level unit group is replaced with a global identifier; second, the identifier of the flow attribute is replaced, and the reference of the flow attribute to the unit group is updated to the global identifier simultaneously; then, the identifier of the flow is replaced, and synonymous flows are uniformly pointed to the base flow through the second mapping relationship, while the reference of the flow to the flow attribute is updated to the global identifier; finally, the identifier of the unit process is replaced, and the input and output flows in the process are first replaced with the base flow identifier through the second mapping, and then updated to the global identifier through the first mapping, completing the association and binding at all levels. Lower-level data is the dependency foundation of upper-level data. Fixing the lower-level identifiers first, and then processing the upper-level and updating the references, can avoid reference invalidation and link breakage.

[0039] This method employs a bottom-up reconstruction sequence. When processing each level, all its dependent lower-level data undergoes global identifier replacement, and reference relationships are synchronously updated to the latest global identifier. This ensures that the association references at each level remain valid, completely resolving the common link breakage problem in surface format fusion and guaranteeing the integrity and correctness of the internal relationships within the merged data. For the input and output streams in the process, a second mapping first unifies synonymous streams from different data sources to the base stream, eliminating synonym redundancy. Then, a first mapping converts them to global identifiers, achieving cross-source compatibility. After these two steps, all streams from the same source point to the same global base stream identifier. This two-layer mapping design can simultaneously complete identifier unification and data deduplication in a single reconstruction process, eliminating the need for multiple rounds of processing, improving data governance efficiency, and ensuring that the merged data has neither identifier conflicts nor synonymous stream redundancy.

[0040] In a preferred embodiment, a fused LCI database is constructed based on a standardized dataset, and a product lifecycle model is constructed based on the fused LCI database, including: The full standardized dataset with completed identifier reconstruction and association binding is stored in a unified storage structure, with the globally unique identifier as the unique primary key, and multi-dimensional joint indexes are established according to the dataset type to form a fused LCI database that supports mixed references across different original data sources; Set core parameters for product modeling, including product functional units, lifecycle system boundaries and target compliance scenarios. Based on the core parameters, filter eligible unit processes and flow data from the integrated LCI database to lock the reference output product flow of the target product. By mixing and referencing unit processes and stream data from different original data sources, and based on the unit processes corresponding to the reference output product stream, a product lifecycle topology is built through the relationship between input streams, output streams and upstream and downstream processes. Based on the full path of the lifecycle topology, input flow nodes without corresponding upstream supply processes are identified, and the optimal upstream unit process is matched from the fused LCI database to complete the supply chain path, thus obtaining the product lifecycle model.

[0041] Specifically, during the database construction phase, standardized datasets are stored in a unified storage structure, using a globally unique identifier as the primary key. Multi-dimensional composite indexes are established according to dataset type, including process name, flow name, data source, and industry classification, supporting fast querying and mixed referencing. In the modeling phase, users first configure core parameters such as product functional units, lifecycle system boundaries, and target compliance scenarios. Based on these parameters, the system filters suitable process and flow data from the database, locking in the product's reference output flow as the starting point for modeling. Subsequently, users can mix and match unit processes from different original data sources, connecting upstream and downstream through flow relationships to build a lifecycle topology. The system automatically traverses the entire topology path, identifies input flow nodes lacking upstream supply, and automatically completes the model by matching the optimal upstream process from the database, ultimately forming a complete product lifecycle model.

[0042] This approach, using a globally unique identifier as the primary key and a storage architecture with multi-dimensional composite indexes, supports rapid queries based on multiple dimensions such as name, origin, and industry. Users can quickly obtain matching results when modeling search unit processes. For complex product models containing hundreds of unit processes, the system can also quickly complete data retrieval and association matching without significant delays due to large data volumes. This forms the foundation for supporting the full lifecycle modeling of complex electrical equipment products, improving the smoothness and efficiency of modeling. The hybrid referencing capability allows for the selection of the most suitable data source for different stages within the same product model. For example, domestic LCI data can be used for the local manufacturing stage to ensure that the accounting results match the actual production situation; international database data can be used for general raw materials and transportation stages to ensure methodological adaptability under international compliance scenarios. This allows for a balance between result accuracy and compliance adaptability within a single model, perfectly meeting the business needs of enterprises requiring both domestic and international compliance.

[0043] In a preferred embodiment, multiple LCIA methodologies are acquired, and these methodologies are then structured, parsed, and uniformly stored to obtain an LCIA methodology library, including: We acquire multiple LCIA methodologies corresponding to different compliance scenarios, perform structured analysis on each LCIA methodology, and extract the impact type, environmental flow definition, characteristic factor parameters, and compliance applicability scope of each LCIA methodology. The impact categories, basic flows, and version numbers of each LCIA methodology are stored in a unified storage structure and preloaded into memory; Based on a unified storage structure and data structures preloaded into memory, the LCIA methodology library is obtained.

[0044] Specifically, firstly, multiple LCIA methodologies under different compliance scenarios are collected. Each methodology is then structurally decomposed, extracting core elements such as an impact type list, environment flow definition, feature factor values, compliance scope, and version number. All methodological element data is stored in a unified storage structure, independent of the LCI database. Subsequently, a two-level feature factor in-memory index is constructed, using the global identifier of the basic flow as the first-level key and a combination of methodology number, version number, and impact category number as the second-level key, mapping corresponding feature factor values. The entire index is preloaded into memory, ultimately forming a complete LCIA methodology library. By separating the methodologies from the native LCI database and adapting them to structured storage and global flow identifiers, decoupling the methodologies from the database is achieved. In-memory preloading transforms disk queries into in-memory queries, significantly improving feature factor matching speed.

[0045] This approach allows for independent parsing and structured storage of methodologies. Relying on a globally unified stream identification system, methodologies can be adapted to any fused LCI data, completely breaking the binding relationship between the two. Users can freely combine data sources and methodologies according to their evaluation needs. For example, domestic process data can be combined with EU methodologies for calculation to meet the compliance declaration requirements of export products, or it can be combined with domestic methodologies for local certification, improving the flexibility of methodology applications and adapting to more diverse evaluation scenarios. By constructing a two-level in-memory index and preloading it into memory, feature factor queries can be completed directly in memory, achieving constant-time complexity without disk access. For complex product models containing thousands of environmental streams, feature factor matching during multi-method calculations can be completed instantly, significantly shortening calculation time and improving the performance of parallel multi-method calculations. During the structured analysis process, information such as the compliance scope and version number of the methodology is extracted and stored in the methodology library. On the one hand, this allows the system to automatically recommend suitable methodologies based on the compliance scenario selected by the user, avoiding user misselection. On the other hand, it allows the methodology name and version to be clearly marked in the accounting report, meeting the methodology traceability requirements in compliance audits. This not only improves the user experience but also ensures the compliance traceability of the accounting results, directly supporting third-party certification and regulatory declaration scenarios.

[0046] In a preferred embodiment, a lifecycle inventory vector is constructed based on the product lifecycle model. Based on the lifecycle inventory vector and the LCIA methodology library, the impact assessment accounting results are obtained, and an accounting report is generated, including: Based on the product lifecycle model, the environmental flow exchange volume of each process is summarized to construct a lifecycle inventory vector; Based on the lifecycle inventory vector, feature factors are matched for each of the selected LCIA methodologies in the LCIA methodology library, and feature factor matrices corresponding to the various LCIA methodologies are constructed. A multi-threaded parallel approach is used to perform matrix multiplication operations simultaneously, resulting in a vector of calculation results from various LCIA methodologies. Based on the accounting result vector, perform an impact category integrity check and generate an accounting report.

[0047] Specifically, firstly, all unit processes of the product lifecycle model are traversed, and the environmental flow exchange volume of each process is summarized. A standardized lifecycle inventory vector is constructed in a unified format. This vector contains quantitative data of all environmental flows throughout the entire lifecycle and serves as the unified data foundation for accounting. Subsequently, for multiple LCIA methodologies selected by the user, feature factor matrices corresponding to each methodology are constructed in parallel based on pre-loaded feature factor indexes. Multi-threaded matrix multiplication operations are then executed in parallel, with each methodology corresponding to an independent calculation thread, synchronously completing matrix multiplication to obtain their respective result vectors. After accounting is completed, an impact category integrity check is performed to check whether all inventory flows match feature factors. Finally, a standardized accounting report is generated based on the result vectors.

[0048] This approach, employing a single-build inventory vector and reusable multiple methodologies, generates a unified inventory vector by traversing the product model only once. All methodologies then perform calculations based on this vector, eliminating the need for repetitive inventory data collection. When a large number of methodologies are being calculated simultaneously, this significantly reduces repetitive workload in inventory statistics. The reduction in redundant calculations is particularly pronounced for product models with numerous processes and complex structures, with overall calculation efficiency increasing linearly with the number of methodologies. Utilizing multi-threaded parallel matrix multiplication allows multiple computational threads to be launched simultaneously to handle calculations for different methodologies. This fully leverages the server's multi-core computing resources, transforming serial calculations of multiple methodologies into parallel processing. The total calculation time in parallel mode is essentially equivalent to that of a single methodology, and does not increase linearly with the number of methodologies. For scenarios requiring simultaneous output of multiple compliance results, this significantly shortens waiting time and improves the efficiency of evaluation work. Especially in batch product calculation scenarios, it significantly enhances overall processing capacity.

[0049] According to another aspect of the present invention, a lifecycle assessment system based on heterogeneous LCI data is provided, the system comprising: The data access module is used to acquire multi-source heterogeneous LCI data, obtain a structured dataset based on the multi-source heterogeneous LCI data, and perform integrity verification on the structured dataset. The global identifier management module is used to generate globally unique identifiers based on the validated structured dataset and establish the first mapping relationship between the structured dataset and the globally unique identifier. The streaming data deduplication module is used to identify and deduplicate synonym streams in the structured dataset, obtain a synonym stream dataset and a baseline stream dataset, and establish a second mapping relationship between the synonym stream dataset and the baseline stream dataset. The identifier reconstruction and association module is used to reconstruct the globally unique identifiers of each data entity in the structured dataset based on the first and second mapping relationships, and to complete the association and binding between data entities at all levels to obtain a standardized dataset. The database management and product modeling module is used to build a fused LCI database based on standardized datasets and to build a product lifecycle model based on the fused LCI database. The methodology management module is used to acquire various LCIA methodologies, perform structured parsing and unified storage of these methodologies, and obtain an LCIA methodology library. The accounting and reporting module is used to construct a lifecycle inventory vector based on the product lifecycle model, obtain the impact assessment accounting results based on the lifecycle inventory vector and the LCIA methodology library, and generate an accounting report.

[0050] In one specific implementation, a lifecycle assessment method based on heterogeneous LCI data includes the following core steps: S1. Convert all the input multi-source heterogeneous LCI data, regardless of their original format, into the platform's standard format and split them into structured datasets to eliminate format heterogeneity from the bottom up. For mainstream domestic LCI data: lossless parsing of the original dataset, extracting all fields of core nodes such as process, stream, stream attributes, unit group, and source; For international databases, such as ecoinvent and GaBi, we build a native format and platform standard mapping engine to map the core fields of ecoSpold and GaBi formats to standard platform data. Next, a format consistency check is performed according to platform standards. This includes checking for information exchange during the process, reference output streams, and whether the stream data contains CAS numbers and stream attributes. An integrity check is then performed on the converted dataset using the following formula: Where n is the number of platform-standard required fields in the dataset, and N is the total number of required fields as specified by the platform standard, only when If the conversion rate is ≥90%, the format conversion is considered successful, and the process proceeds to the next step.

[0051] S2. Based on the platform standard dataset obtained in step S1, construct a globally unique UUID; Generate a global UUID conforming to the ILCD standard according to the Version 4 UUID specification in RFC4122. The core formula is: in It is a type encoding, a fixed 2-bit encoding, used to distinguish LCI object entities. It is the LCI data source identifier. It extracts key business attribute fields of the current LCI entity to represent business characteristics, such as process name, industry classification, and quantity reference product flow from the unit process set. Use standard timestamps. The SHA-256 hash function outputs a fixed-length hash digest. .) is the ILCD standard UUID mapping function; Establish a mapping table between native UUIDs and global UUIDs. At the same time, it ensures that the same data from the same data source will always generate the same global UUID when accessed repeatedly, and will not create the mapping relationship repeatedly.

[0052] S3. Construct stream buckets, standardizing the name, stream type, category, and stream attribute fields of the stream data. Besides formatting, the name field undergoes significant Chinese-English translation and alignment. To improve deduplication performance and avoid pairwise comparisons, all data is bucketed, with judgments performed only within each bucket. The bucket key is defined as follows: in, .) is a hash function. For classification, For stream type, It is a flow attribute; The process involves identifying duplicate flows from the same source within the same bucket. If two flows have the same classification, flow type, and flow attributes, but their names differ only in Chinese and English, they can be identified as candidate synonym flows. Then, the semantic similarity of the standardized name text is checked, and those that reach the threshold are finally identified as synonym flows. Perform deduplication and synonym mapping relationship construction. During deduplication, only the base stream dataset is retained, and a synonym mapping table between synonym stream UUIDs and base stream UUIDs is constructed. .

[0053] S4. Replace the UUID of the underlying dataset. Following the order of unit group, stream attribute, stream and then unit process, replace the unified UUID starting from the lowest unit group to ensure that the UUID of the lower related dataset is fixed when the upper dataset is reconstructed. In hierarchical association binding, the input and output streams in the unit process set are first replaced with the deduplicated base stream UUID, and then based on the mapping table. Replace the new global UUID of the streaming dataset with the new UUID of the streaming attribute dataset; replace the streaming attribute association in the streaming dataset with the new unified UUID of the streaming attribute dataset, and so on.

[0054] S5. Store the full amount of data that has been reconstructed, deduplicated, and hierarchically related into a unified storage structure fused LCI database, using the global UUID as the unique primary key and separate composite indexes according to the dataset type, to achieve the effect of supporting millisecond-level data query and cross-source hybrid reference; When using mixed-reference product modeling, after users define the product functional units, they can seamlessly mix and reference process and stream data from different raw data sources to build the product LCA model. The specific steps are as follows: First, configure the core modeling parameters. Based on the configuration, the system automatically performs pre-screening of data, setting product functional units, lifecycle system boundaries, compliance scenarios, etc., and then filters mix from the unified database. ref =1 (supports mixed references) of unit process and flow data, automatically locks the output product flow of the target product, establishes the mapping relationship between process UUID and output product flow, and serves as the end node of the modeling topology; Here, mix ref =1 is a label used in the database to identify whether a data record supports mixed references. When mix ref When =1, it indicates that the data record can be mixed and referenced with other data records. That is, during LCA modeling, this data record can be combined with other related data records to build a more complete product lifecycle model. ref =1 is used to improve the flexibility and accuracy of LCA modeling, allowing users to select different data records to combine and use according to specific needs, thereby more realistically reflecting the actual environmental impact of the product's life cycle. Within the same product model, process and stream data from different original data sources can be referenced simultaneously, for example: Upstream process of a certain material production: Electronic component production process data converted from domestic LCI data; Data on upstream processes of regional power grid electricity production, converted from ecoinvent data; Data on upstream processes of road freight transport converted from the GaBi database; The system features automatic matching and calibration. Based on global flow node identifiers, the product modeling system automatically completes cross-source flow data unification and semantic matching without manual intervention, achieving a flow matching accuracy of over 95%. For each process flow, the system automatically recommends the three process data that best fit the conditions in that scenario, and labels the data source, compliance and adaptability, and data quality level to help users quickly create models; To construct the product model topology graph, firstly, based on the output unit process and output product flow, connect the input / output flows with the upstream process data using edges to build the product model topology. For example, a raw material flow or energy flow is an input edge, and the output product flow or waste flow is an output edge, thus completely constructing the material and energy consumption flow of the product life cycle. Traverse the entire path from raw material mining to the final product, identify input flow nodes without corresponding upstream supply processes, and automatically match the optimal corresponding process data from the database to ensure the integrity of the life cycle path.

[0055] S6. After processing heterogeneous LCI data in the same way as process and stream data, a unified LCIA methodology library is obtained. Then, multiple versions of LCIA methodologies corresponding to different compliance scenarios are collected and parsed. The impact types, environmental flow definitions, feature factor parameters, and compliance application scope of each methodology are extracted in a structured manner. The impact categories, basic flows, and version numbers of each methodology are stored in a unified storage structure and preloaded into memory for quick retrieval during subsequent LCIA calculations. The memory data structure is defined as follows: in, For basic stream UUID, For methodology ID, For version number, To affect the category ID, The feature factor value is used to achieve precise query of the feature factor with O(1) complexity, and to break the strong binding relationship between the methodology and the original LCI database.

[0056] S7. Based on the product model constructed in step S5, freely select unit processes and flow data across different raw data to complete the full life cycle LCA modeling of electrical equipment products from raw material processing, component manufacturing to assembly. When performing LCIA calculation on the product model, first construct a standardized life cycle inventory vector g, which is an m×1 column vector, where m is the number of environmental flows in the inventory, and save it as a unified standard data structure. The system has a built-in complete methodology rule library for mainstream compliance scenarios. Based on the user's selected compliance scenario, it automatically recommends a suitable methodology combination or a user-selected methodology combination. For example, if IPCC, EF 3.1, and a certain domestic methodology are selected simultaneously, using vector g as the data basis, and through a pre-loaded methodology storage structure, relying on the time complexity of memory indexing (i.e., constant time complexity), the operation time does not change with the amount of data. This enables rapid matching of feature factors and matrix filling, constructing all corresponding feature factor matrices at once. ,in for ×m order matrix m represents the number of impact categories of methodology i, and m represents the number of environmental flows in the list. Multithreaded parallel matrix multiplication, performing n independent matrix multiplications in parallel. Based on the above process, n independent result vectors are obtained simultaneously; After the accounting is completed, an impact category integrity check is performed and an accounting report in a standard format is generated, which includes full-chain traceability information and can be directly submitted to a third-party auditing agency. Next, a multi-methodological comparative analysis is performed to generate a multi-methodological accounting result comparison report, which shows the differences in results of different methodologies for each impact type and analyzes the reasons for the differences.

[0057] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A lifecycle assessment method based on heterogeneous LCI data, characterized in that, The method includes: Acquire multi-source heterogeneous LCI data, obtain a structured dataset based on the multi-source heterogeneous LCI data, and perform integrity verification on the structured dataset; A globally unique identifier is generated based on the structured dataset after verification, and a first mapping relationship is established between the structured dataset and the globally unique identifier; Synonym stream identification and deduplication are performed on the streaming data in the structured dataset to obtain a synonym stream dataset and a baseline stream dataset, and a second mapping relationship is established between the synonym stream dataset and the baseline stream dataset. Based on the first and second mapping relationships, the identifiers of each data entity in the structured dataset are reconstructed using globally unique identifiers, and the association binding between data entities at all levels is completed to obtain a standardized dataset. A fused LCI database is constructed based on the standardized dataset, and a product lifecycle model is constructed based on the fused LCI database; Multiple LCIA methodologies are acquired, and the LCIA methodologies are structured, parsed, and stored uniformly to obtain an LCIA methodology library. Based on the product lifecycle model, a lifecycle inventory vector is constructed. Based on the lifecycle inventory vector and the LCIA methodology library, the impact assessment accounting results are obtained and an accounting report is generated.

2. The method according to claim 1, characterized in that, The multi-source heterogeneous LCI data includes domestic LCI datasets and international LCI datasets. The structured dataset obtained based on the multi-source heterogeneous LCI data includes: Based on the aforementioned domestic LCI dataset, the original structure was parsed without loss, and all core fields of process, stream, stream attributes, unit group, and data source were extracted. Based on the aforementioned international LCI dataset, the core fields of the internationally common format are mapped and converted into structured data in the platform's standard format; The structured dataset is obtained based on the full set of core fields and the structured data.

3. The method according to claim 2, characterized in that, The integrity verification of the structured dataset includes: Based on the number of required fields according to platform standards included in the structured dataset and the total number of required fields preset by the platform standards, an integrity check is performed. The formula for the integrity check is: Where n is the number of required fields according to platform standards in the structured dataset, and N is the total number of required fields specified by the platform standards. If the conversion rate is ≥90%, the format conversion is considered successful.

4. The method according to claim 1, characterized in that, The process of generating a globally unique identifier based on the validated structured dataset includes: Obtain the type code, data source identifier, key business attribute fields, and timestamp field of the data entities in the structured dataset; The type code, data source identifier, key business attribute field, and timestamp field are concatenated to obtain the concatenated string; A hash operation is performed on the concatenated string to obtain a hash digest. The hash digest is mapped to a globally unique identifier in a standard format.

5. The method according to claim 1, characterized in that, The process of performing synonym stream identification and deduplication on the streaming data in the structured dataset to obtain a synonym stream dataset and a baseline stream dataset, and establishing a second mapping relationship between the synonym stream dataset and the baseline stream dataset, includes: The classification field, stream type field, and stream attribute field of the streaming data are standardized. Based on the standardized classification, stream type, and stream attribute, a bucketing key is generated. The streaming data is then allocated to the corresponding buckets according to the bucketing key. Within the same bucket, when the classification, stream type, and stream attributes of two streams are consistent, the two streams are determined to be candidate synonym streams. The names of the two streams are semantically similar. When the semantic similarity reaches a preset threshold, the two streams are determined to be synonym streams. The reference stream data is retained in the synonym stream, a synonym stream dataset is generated based on all the synonym stream data, a reference stream dataset is generated based on all the reference stream data, and a second mapping relationship is established between the synonym stream dataset and the reference stream dataset.

6. The method according to claim 1, characterized in that, Based on the first and second mapping relationships, the identifiers of each data entity in the structured dataset are reconstructed using globally unique identifiers, and the association binding between data entities at all levels is completed to obtain a standardized dataset, including: Based on the first mapping relationship, the original identifier of the unit group entity in the structured dataset is replaced with a globally unique identifier, thereby completing the unification of the identifier of the underlying unit group; Based on the first mapping relationship, the native identifier of the flow attribute entity in the structured dataset is replaced with a globally unique identifier, and the unit group reference associated with the flow attribute is updated to the globally unique identifier of the unit group, thus completing the association and binding between the flow attribute and the unit group; Based on the first mapping relationship, the native identifier of the flow entity in the structured dataset is replaced with a globally unique identifier. Based on the second mapping relationship, the references of synonymous flows are uniformly pointed to the identifier of the corresponding base flow. Based on the first mapping relationship, the flow attribute reference associated with the base flow is updated to the globally unique identifier of the flow attribute, thus completing the association and binding between the flow and the flow attribute. Obtain the unit process entity, replace the input and output streams of the unit process entity with the identifiers of the corresponding reference streams based on the second mapping relationship, update the identifiers of the input and output streams as the corresponding reference streams with the globally unique identifiers of the reference streams based on the first mapping relationship, complete the association and binding between the unit process entity and the stream, and obtain the standardized dataset with consistent association across all levels.

7. The method according to claim 6, characterized in that, The construction of a fused LCI database based on the standardized dataset, and the construction of a product lifecycle model based on the fused LCI database, include: The full standardized dataset with completed identifier reconstruction and association binding is stored in a unified storage structure, with the globally unique identifier as the unique primary key, and multi-dimensional joint indexes are established according to the dataset type to form a fused LCI database that supports mixed references across different original data sources; Set core parameters for product modeling, including product functional units, lifecycle system boundaries and target compliance scenarios. Based on the core parameters, filter eligible unit processes and flow data from the fused LCI database to lock the reference output product flow of the target product. By mixing and referencing unit processes and stream data from different original data sources, and based on the unit processes corresponding to the reference output product stream, a product lifecycle topology is built through the relationship between input streams, output streams and upstream and downstream processes. Based on the entire path of the lifecycle topology, input flow nodes without corresponding upstream supply processes are identified, and the optimal upstream unit process is matched from the fused LCI database to complete the supply chain path, thus obtaining the product lifecycle model.

8. The method according to claim 1, characterized in that, The process involves acquiring multiple LCIA methodologies, performing structured parsing and unified storage on these methodologies, resulting in an LCIA methodology library, including: We acquire multiple LCIA methodologies corresponding to different compliance scenarios, perform structured analysis on each LCIA methodology, and extract the impact type, environmental flow definition, characteristic factor parameters, and compliance applicability scope of each LCIA methodology. The impact categories, basic flows, and version numbers of each LCIA methodology are stored in a unified storage structure and preloaded into memory; Based on the unified storage structure and the data structure preloaded into memory, the LCIA methodology library is obtained.

9. The method according to claim 1, characterized in that, The process of constructing a lifecycle inventory vector based on the product lifecycle model, obtaining impact assessment accounting results based on the lifecycle inventory vector and the LCIA methodology library, and generating an accounting report includes: Based on the product lifecycle model, the environmental flow exchange volume of each process is summarized to construct a lifecycle inventory vector; Based on the lifecycle inventory vector, feature factors are matched for the selected LCIA methodologies in the LCIA methodology library to construct feature factor matrices corresponding to the various LCIA methodologies. A multi-threaded parallel approach is adopted to perform matrix multiplication operations simultaneously, resulting in a vector of calculation results from various LCIA methodologies. Based on the calculated result vector, an impact category integrity check is performed and an accounting report is generated.

10. A lifecycle assessment system based on heterogeneous LCI data, characterized in that, The system includes: The data access module is used to acquire multi-source heterogeneous LCI data, obtain a structured dataset based on the multi-source heterogeneous LCI data, and perform integrity verification on the structured dataset. The global identifier management module is used to generate a globally unique identifier based on the structured dataset after verification, and to establish a first mapping relationship between the structured dataset and the globally unique identifier. The streaming data deduplication module is used to perform synonym stream identification and deduplication on the streaming data in the structured dataset to obtain a synonym stream dataset and a baseline stream dataset, and to establish a second mapping relationship between the synonym stream dataset and the baseline stream dataset. The identifier reconstruction and association module is used to reconstruct the globally unique identifiers of each data entity in the structured dataset based on the first mapping relationship and the second mapping relationship, and to complete the association binding between data entities at all levels to obtain a standardized dataset. The database management and product modeling module is used to build a fused LCI database based on the standardized dataset and to build a product lifecycle model based on the fused LCI database. The methodology management module is used to acquire various LCIA methodologies, perform structured parsing and unified storage of these LCIA methodologies, and obtain an LCIA methodology library. The accounting and reporting module is used to construct a life cycle inventory vector based on the product life cycle model, obtain the impact assessment accounting results based on the life cycle inventory vector and the LCIA methodology library, and generate an accounting report.