Multi-source heterogeneous construction cost data penetration alignment method and system

Abstract

The present application relates to a kind of multi-source heterogeneous cost data penetration alignment method and system, the method obtains the cost data of target power grid project in multiple cost-related systems, constructs including total cost node, expense subject node and atomic expense item node cost subject semantic nested tree;According to the expense boundary of each subject node, applicable stage, attribution basis and splittable state generates caliber characteristics, and according to the source characteristics of expense record, expense content characteristics, amount feature and associated basis characteristics generate source data characteristics;Through cross-system feature mapping layer, source data characteristics are mapped to uniform cost expression space, determine the candidate attribution position and final attribution position of expense record;According to final attribution position, generate uniform cost topology amount and output penetration alignment result;The present application can improve multi-source cost data alignment accuracy and the traceability of difference attribution.

Description

Technical Field

[0001] This invention relates to a method and system for penetrating and aligning multi-source heterogeneous valence data, belonging to the field of data processing technology. Background Technology

[0002] Engineering projects generate a large amount of cost data during stages such as investment control, contract execution, material procurement, cost accounting, and financial settlement. Due to the different business objectives at different stages, the relevant data is usually recorded in multiple business systems, and these systems differ in terms of expense coding, expense names, account levels, statistical scope, and amount collection methods.

[0003] In current engineering cost management, cost data from different systems are typically aggregated and compared through methods such as matching cost codes, matching field names, manual compilation, or fixed subject comparison tables. While these methods can achieve a certain degree of data aggregation, they mainly rely on the superficial consistency of fields or human experience, making it difficult to accurately reflect the actual cost boundaries and hierarchical relationships behind different cost records.

[0004] In practical applications, the same expense item may appear in different systems with different names, codes, or granularities; similar or identical expense names may also correspond to different expense ranges. For example, an expense record may correspond to a specific sub-item, cover multiple subordinate expense items, or only be a part of a superior expense account. If matching is based solely on name or code, problems such as misclassification, duplicate collection, omission, or unexplained discrepancies in expense amounts can easily occur.

[0005] Furthermore, existing methods typically focus more on comparing differences in the total project amount or primary expense categories, making it difficult to penetrate the total cost difference layer by layer to more granular expense items. It also makes it difficult to determine whether expense records should be categorized under higher-level, lower-level, or multiple peer-level categories, or temporarily retained under the current category. This means that while cost differences can be identified, it is difficult to accurately pinpoint the specific source and stage of cost formation.

[0006] For power grid projects, the project types are diverse, the construction phases are long, the participating entities are numerous, and the sources of funding are complex. The differences in data standards between stages such as budget estimation, procurement, contracts, settlement, and final accounts are even more pronounced. Related project data also indicates that power grid project cost management suffers from problems such as a high variance rate, reliance on experience in identifying key drivers, and a lack of systematic tracing of underlying causes. Therefore, it is necessary to systematically decompose historical project costs, conduct variance analysis, and construct standard indicators. Summary of the Invention

[0007] To address the problems existing in the prior art, this invention proposes a method and system for penetrating and aligning multi-source heterogeneous valence data.

[0008] The technical solution of the present invention is as follows: On one hand, the present invention provides a method for penetrating and aligning multi-source heterogeneous valence data, comprising the following steps: Obtain cost data for the target power grid project from multiple cost-related systems. The cost data includes cost identification information, cost description information, cost amount information, cost stage information, and related information used to characterize the source or basis of costs. Based on the hierarchical relationship between the cost items of the power grid project, the cost attribution relationship, and the historically confirmed cost details, a semantic nesting tree of cost items is constructed. The semantic nesting tree of cost items includes multiple item nodes, specifically including a total cost node for receiving the total project cost, a cost item node for receiving the sub-items cost, and an atomic cost item node for receiving the minimum attributable cost. Based on the cost boundaries, applicable stages, attribution basis, and splittable status of each subject node, generate the caliber characteristics corresponding to each subject node. Based on the source characteristics, content characteristics, amount characteristics, and correlation basis characteristics of the expense records in each cost-related system, corresponding source data characteristics are generated; The source data features from different cost-related systems are mapped to a unified cost representation space to obtain the unified cost features corresponding to the cost records; Based on the matching relationship between the unified cost features and the caliber features of each subject node, the candidate location of the cost record in the semantic nesting tree of the cost subject is determined; Based on the correspondence between the cost range, detailed support and amount between the cost record and the candidate location, the final location of the cost record in the semantic nesting tree of the cost item is determined. The final location includes the parent node, the next node, multiple sibling nodes or the current node of the candidate location. Generate a unified cost topology amount based on the final location of the cost record, and output the penetration alignment result between the cost record and the semantic nesting tree of the cost item.

[0009] Preferably, the plurality of cost-related systems include at least two of the following: a preliminary budget system, a financial settlement system, an ERP system, a material procurement system, and a contract settlement system; The fee identification information includes at least one of the following: project identifier, fee code, and fee name; The cost description information includes at least one of the following: cost item, cost category, cost content, and the part of the project to which the cost belongs; The associated information includes at least one of the following: contract information, material information, project quantity information, settlement information, final account voucher information, and evidentiary document information.

[0010] Preferably, constructing a semantic nested tree for cost items includes: The main hierarchy of the semantic nesting tree of cost accounts is determined by the preliminary cost account and the final cost account; Link cost items, material categories, contract list items, settlement audit details, and historical confirmed expense details to the corresponding main levels; When the cost names are different in different cost-related systems but the cost boundaries are the same, configure them as aliases for the same subject node; When the cost names are similar but the cost boundaries are inconsistent in different cost-related systems, different subject nodes should be established for each.

[0011] Preferably, the caliber features include cost attribute features, stage attribute features, contract attribute features, material attribute features, quantity attribute features, evidence attribute features, and divisible attribute features; Among them, the cost attribute feature is used to characterize the cost category to which the subject node belongs, the stage attribute feature is used to characterize the cost stage to which the subject node applies, the contract attribute feature is used to characterize whether the subject node is supported by a contract or order, the material attribute feature is used to characterize whether the subject node corresponds to equipment or materials, the quantity attribute feature is used to characterize whether the subject node can be broken down into quantity and unit price, the evidence attribute feature is used to characterize the type of evidence document required for the subject node, and the splittable attribute feature is used to characterize whether the subject node is allowed to continue drilling down to the next level node.

[0012] Preferably, the source data features include source system features, original coding features, cost name features, amount features, project ownership features, contract ownership features, material attribute features, engineering quantity attribute features, stage attribute features, evidence completeness features, and cost coverage features. Among them, the expense coverage feature is used to characterize whether an expense record is a single expense record, a combined expense record, or an expense record to be split.

[0013] Preferably, the source data features from different cost-related systems are mapped to a unified cost representation space, including: Based on the source data characteristics, compensation criteria characteristics, and project context characteristics of the cost records, generate unified cost characteristics corresponding to the cost records; Among them, the caliber compensation feature is used to characterize whether the cost record is a merged record, a split record, whether it lacks contract information, material information, quantity information, evidentiary documents, and whether there is a stage offset; Project context features are used to characterize the project type, voltage level, construction area, construction scale, and technical solution of the target power grid project.

[0014] Preferably, determining the candidate location of the expense record in the semantic nesting tree of the cost item includes: Calculate the degree of matching between the uniform cost characteristics of the cost records and the caliber characteristics of each item node; Based on the matching degree and the matching threshold corresponding to each subject node, the candidate location of the expense record is determined; When the candidate location is not determined, mark the fee record as a record to be verified and output the reason for the mismatch; When a candidate location is determined, that candidate location is used as the initial location. When multiple candidate locations are identified, the final location is determined based on the scope of expenses, detailed supporting documents, and corresponding amounts.

[0015] Preferably, determining the final location of the expense record in the semantic nesting tree of cost items includes: When the cost coverage of a cost record exceeds the cost range of the corresponding subject node of the candidate location, the cost record will be assigned to the next higher level node of the candidate location. When the expense record has contract details, material details, project quantity details, settlement details or evidence documents, and the details can correspond to the next level node of the candidate location, the expense record will be assigned to the corresponding next level node. When expense records cover multiple peer nodes at the same time, the splitting ratio is determined based on the contract amount, material amount, project amount, settlement amount or historical expense structure, and the expense records are split into multiple peer nodes. When a fee record lacks the details or supporting documentation required for further breakdown, the fee record is retained at the current node, and an un-segmented amount is generated.

[0016] Preferably, after generating a unified cost topology amount, it also includes: Perform consistency checks on the parent node amount, the next level node amount, and the unsubdivided amount in the semantic nesting tree of cost items; When the deviation between the amount of the parent node and the amount of the next level node and the undivided amount exceeds a preset threshold, a verification exception record is output. In a unified cost expression space, compare the amount differences of the same subject node under different cost-related systems or different cost stages, and generate a difference location table based on the amount differences; The discrepancy location table includes project identifier, original cost information, unified subject node, node level, amount of each cost stage, node difference amount, unspecified amount, supporting documents, and review status.

[0017] On the other hand, the present invention also provides a multi-source heterogeneous valence data penetration alignment system, comprising: The data acquisition module is used to acquire cost data of the target power grid project in multiple cost-related systems. The cost data includes cost identification information, cost description information, cost amount information, cost stage information, and related information used to characterize the source or basis of costs. The nested tree construction module is used to construct a semantic nested tree of cost items based on the hierarchical relationship, cost attribution relationship, and historically confirmed cost details among the cost items of the power grid project. The semantic nested tree of cost items includes multiple item nodes, which are hierarchically divided into at least a total cost node, a cost item node, and an atomic cost item node. The total cost node is used to cover the total project cost, the cost item node is used to cover the sub-items cost, and the atomic cost item node is used to cover the minimum attributable cost. The caliber feature generation module is used to generate caliber features for each subject node based on the cost boundary, applicable stage, attribution basis, and splittable status of each subject node. The source data feature generation module is used to generate corresponding source data features based on the source features, cost content features, amount features, and correlation basis features of the cost records in various cost-related systems. The cross-system mapping module is used to map the source data features from different cost-related systems to a unified cost expression space through the cross-system feature mapping layer, so as to obtain the unified cost features corresponding to the cost records. The candidate attribution determination module is used to determine the candidate attribution position of the cost record in the semantic nesting tree of the cost item based on the matching relationship between the unified cost feature and the caliber feature of each item node. The final attribution determination module is used to determine the final attribution position of the expense record in the semantic nesting tree of the cost item based on the correspondence between the expense record and the candidate attribution position in terms of expense range, detailed support and amount; wherein, the final attribution position includes the parent node, the next node, multiple sibling nodes or the current node of the candidate attribution position; The topology amount generation module is used to generate a unified cost topology amount based on the final attribution location of the cost record; The results output module is used to output the penetration alignment results between the cost records and the semantic nesting tree of cost items.

[0018] The present invention has the following beneficial effects: 1. This invention constructs a semantic nested tree of cost items and generates caliber features for each item node, enabling the total project cost, sub-item costs, and minimum attributable cost items to be expressed in a unified hierarchy, thereby avoiding mismatch, omission, and duplicate aggregation problems caused by inconsistencies in cost names, codes, calibers, or granularities among different systems.

[0019] 2. This invention maps cost records from different cost-related systems to a unified cost expression space and determines the final location of cost records based on cost scope, detailed support, and amount correspondence. This allows cost records to be accurately categorized, split, or retained, regardless of whether the granularity is too coarse, too fine, or covers multiple cost items simultaneously. This improves the accuracy and traceability of multi-source heterogeneous cost data penetration alignment.

[0020] 3. This invention generates a unified cost topology amount based on the final location of the cost record and outputs the penetration alignment result, enabling cost data from stages such as budget, procurement, contract, settlement and final account to be compared under a unified standard, providing a reliable data foundation for subsequent cost difference location, cost composition analysis, core cost item identification and closed-loop management. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.

[0024] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0025] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0026] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.

[0027] See Figure 1 In some embodiments, a multi-source heterogeneous valence data penetration alignment method is proposed, including the following steps: Obtain cost data for the target power grid project from multiple cost-related systems. The cost data includes cost identification information, cost description information, cost amount information, cost stage information, and related information used to characterize the source or basis of costs. Based on the hierarchical relationship between the cost items of the power grid project, the cost attribution relationship, and the historically confirmed cost details, a semantic nesting tree of cost items is constructed. The semantic nesting tree of cost items includes multiple item nodes, specifically including a total cost node for receiving the total project cost, a cost item node for receiving sub-items of cost, and an atomic cost item node for receiving the minimum attributable cost. The atomic cost item node is an account node that meets the stop splitting condition, which includes: Able to correspond to the contract list record; Alternatively, it can correspond to material procurement orders; Alternatively, it can correspond to specific items in the bill of quantities; Alternatively, it can correspond to the settlement audit details; Alternatively, it can correspond to the expense collection items in the financial settlement vouchers; Alternatively, further breakdown may only result in execution details at the level of suppliers, invoices, or payment records, failing to further explain the differences in cost items.

[0028] Based on the cost boundaries, applicable stages, attribution basis, and splittable status of each subject node, generate the caliber characteristics corresponding to each subject node. Based on the source characteristics, content characteristics, amount characteristics, and correlation basis characteristics of the expense records in each cost-related system, corresponding source data characteristics are generated; The source data features from different cost-related systems are mapped to a unified cost representation space to obtain the unified cost features corresponding to the cost records; Based on the matching relationship between the unified cost features and the caliber features of each subject node, the candidate location of the cost record in the semantic nesting tree of the cost subject is determined; Based on the correspondence between the cost range, detailed support and amount between the cost record and the candidate location, the final location of the cost record in the semantic nesting tree of the cost item is determined. The final location includes the parent node, the next node, multiple sibling nodes or the current node of the candidate location. Generate a unified cost topology amount based on the final location of the cost record, and output the penetration alignment result between the cost record and the semantic nesting tree of the cost item.

[0029] In a specific embodiment, taking a new 110kV substation construction project as an example, the cost items recorded in the preliminary cost estimate system include construction costs, installation costs, equipment purchase costs, and other costs; the cost items recorded in the financial settlement system include construction costs, equipment costs, deferred investment, and other apportioned costs; the ERP system records the actual costs incurred under the project cost center; the materials procurement system records purchase orders for main transformers, switchgear, cables, and protection automation equipment; and the contract settlement system records civil construction contracts, electrical installation contracts, equipment purchase contracts, design contracts, supervision contracts, and related change orders. Because the cost names, codes, and granularities in the above systems are different, matching only by field names may result in duplicate data collection of equipment costs, omissions in data collection of installation costs, or other costs being mistakenly included in equipment costs.

[0030] This embodiment first obtains the cost data of the target power grid project from multiple business systems. These multiple business systems include a preliminary cost estimation system, a financial settlement system, an ERP system, a materials procurement system, and a contract settlement system. Each cost data entry includes at least the following: project number, project name, project type, voltage level, construction unit, project area, cost code, cost name, cost amount, stage to which the cost belongs, time of cost occurrence, source system, contract number, supplier name, material code, quantity, unit price, approval status, settlement status, settlement voucher number, and evidence document identifier.

[0031] The cost data in the preliminary budget system is used to represent the cost control amount during the preliminary budget stage; the cost data in the financial settlement system is used to represent the final financial confirmation amount of the project; the cost data in the ERP system is used to represent the cost accumulation amount during the project execution process; the cost data in the material procurement system is used to represent the equipment and material procurement amount; and the cost data in the contract settlement system is used to represent the contract, change, approval and settlement amount.

[0032] After acquiring the cost data, basic standardization processing is performed on the multi-source cost data. Standardization processing includes unifying project identifiers, monetary units, time fields, contract numbers, supplier names, material names, and cleaning up expense names.

[0033] Specifically, if project numbers are consistent across different systems, project associations are directly established. If project numbers are missing or different systems use different project numbers, candidate project associations are generated based on project name, construction unit, project region, project type, voltage level, construction time, and contract number, and these candidate associations are submitted for review. If the review confirms that the projects belong to the same project, the project identifiers from different systems are unified to the same target project identifier. For data with different monetary units, they are uniformly converted to ten thousand yuan or yuan. System prefixes, parenthetical descriptions, abbreviations, version numbers, spaces, and special symbols in fee names are cleaned up, but the final fee attribution is not directly determined based on the cleaned fee names.

[0034] After completing basic standardization, a semantic nested tree for cost items is constructed. This semantic nested tree expresses the hierarchical relationship between the total project cost, cost items, and atomic cost items. The semantic nested tree includes a root node, first-level cost nodes, second-level cost nodes, third-level cost nodes, and atomic cost item nodes.

[0035] The root node is the total cost node for the target project. First-level cost nodes can include building construction cost nodes, installation cost nodes, equipment purchase cost nodes, other cost nodes, land acquisition and demolition cost nodes, compensation for damaged crops cost nodes, construction measures cost nodes, and design change cost nodes. Second-level and third-level cost nodes continue to expand downwards according to project type. For example, the equipment purchase cost node can include main transformer equipment nodes, switchgear nodes, cable equipment nodes, communication equipment nodes, and protection automation equipment nodes; the building construction cost node can include substation civil engineering nodes, foundation engineering nodes, road engineering nodes, cable trench engineering nodes, and temporary works nodes; other cost nodes can include survey and design cost nodes, supervision cost nodes, construction management cost nodes, and testing and inspection cost nodes, etc.

[0036] The semantic nesting tree of cost items can be represented as: ; in: Represents a semantic nested tree of cost items; Represents a set of subject nodes; Represents the set of parent-child connections between nodes; This represents the set of rules for attributing fees; Represents a set of node levels; This indicates the set of conditions for stopping the splitting; When constructing the semantic nesting tree of cost items, the main hierarchy is first determined by the preliminary cost items in the preliminary cost system and the final cost items in the financial settlement system. Since the preliminary cost system typically reflects the planned control scope of engineering construction investment, and the financial settlement system typically reflects the final financial recognition scope, the two together form the main trunk of the cost item semantic nesting tree. Then, cost items from the ERP system, material classifications from the material procurement system, contract list items from the contract settlement system, and historically confirmed expense details are linked to the corresponding main hierarchy.

[0037] For example, if the budget estimation system has an "Equipment Purchase Cost" item, and the materials procurement system has "Main Transformer Purchase Orders," "10kV Switchgear Purchase Orders," and "Power Cable Purchase Orders," then the materials corresponding to these purchase orders will be categorized and linked under the "Equipment Purchase Cost" node. Similarly, if the contract settlement system has a "Civil Engineering Contract List," which includes items such as foundation excavation, concrete pouring, cable trench construction, and station area roads, then these contract list items will be linked under the corresponding nodes for "Construction Engineering Costs" or "Construction Measures Costs."

[0038] When expense names differ across different business systems but have consistent expense boundaries, they should be configured as aliases for the same account node. For example, "Cable Equipment Fee," "Power Cable Fee," and "Cable Procurement Fee," all pointing to cable procurement amounts at their expense boundaries, can be used as aliases for the "Cable Equipment" node. When expense names are similar across different business systems but have inconsistent expense boundaries, they should not be merged into the same node. For example, both "Equipment Installation Fee" and "Equipment Purchase Fee" contain "equipment" in their names, but the former belongs to installation costs while the latter belongs to procurement costs; separate nodes should be created for each, and they should be distinguished by direct measurement.

[0039] When expense names differ across different business systems but have consistent expense boundaries, they should be configured as aliases for the same account node. For example, "Cable Equipment Fee," "Power Cable Fee," and "Cable Procurement Fee," all pointing to cable procurement amounts at their expense boundaries, can be used as aliases for the "Cable Equipment" node. When expense names are similar across different business systems but have inconsistent expense boundaries, they should not be merged into the same node. For example, both "Equipment Installation Fee" and "Equipment Purchase Fee" contain "equipment" in their names, but the former belongs to installation costs while the latter belongs to procurement costs; separate nodes should be created for each, and they should be distinguished by direct measurement.

[0040] Generate a radial quantity for each account node in the semantic nesting tree of cost accounts. The radial quantity is used to describe the cost boundary of the account node, enabling cost records from different systems to be matched based on the cost substance.

[0041] The radial parameters of each subject node include project type characteristics, cost attribute characteristics, stage attribute characteristics, contract attribute characteristics, material attribute characteristics, project quantity attribute characteristics, responsible entity attribute characteristics, evidence attribute characteristics, and divisible attribute characteristics.

[0042] The node is defined by several key characteristics: Project Type: This describes whether the node applies to the main grid, distribution network, technical upgrade, or a specific project category. Cost Attribute: This describes whether the node falls under categories such as construction, installation, equipment, materials, other costs, land acquisition and demolition, compensation for damaged crops, design changes, or construction measures. Stage Attribute: This describes when the node typically appears in the budget, procurement, contract, settlement, or final account stages. Contract Attribute: This describes whether the node is supported by contracts, orders, or settlement documents. Material Attribute: This describes whether the node corresponds to equipment, materials, spare parts, or non-material costs. Quantity Attribute: This describes whether the node can be broken down into quantities and unit prices. Responsible Entity Attribute: This describes whether the node typically involves entities such as construction, design, engineering, materials, finance, or local coordination. Evidence Attribute: This describes what type of evidence should support the node. Divisibility Attribute: This describes whether the node allows further drilling down to sub-nodes.

[0043] When generating specific radial quantities, the above attributes can be converted into structured codes. For example, for the "Cable Equipment Cost" node, its project type characteristic can be applicable to both main grid and distribution network; the cost attribute characteristic is equipment or material category; the stage attribute characteristics include budget stage, procurement stage, contract stage, settlement stage, and final settlement stage; the contract attribute characteristic is that it can be supported by procurement contracts and orders; the material attribute characteristic is that there is a material code; the project quantity attribute characteristic can be expressed by cable length and unit price; the evidence attribute characteristic is purchase orders, contract lists, delivery and acceptance records, and settlement audit records; and the splittable attribute characteristic is that it can be drilled down to the atomic cost items of different specifications and models of cables.

[0044] For example, for the "Qingpei compensation fee" node, its fee attribute characteristics are Qingpei category; the stage attribute characteristics mainly include the preliminary estimate, settlement and final account stages; the contract attribute characteristics may be weak; the material attribute characteristics are empty or weak; the engineering quantity attribute characteristics may be manifested as the compensation area, the number of compensation households or the number of compensation items; the evidence attribute characteristics are local coordination records, compensation agreements, on-site visas and payment vouchers; the divisibility attribute characteristics depend on whether it is possible to continue drilling down by compensation item or compensation area.

[0045] Next, heterogeneous cost records from each business system are obtained, and a source system feature vector is generated for each heterogeneous cost record. Heterogeneous cost records refer to original cost records from different systems that have not yet been incorporated into the unified cost item semantic nesting tree.

[0046] The source system feature vectors for each heterogeneous cost record include source system features, original cost code features, original cost name features, amount features, project attribution features, contract attribution features, material or equipment attribute features, quantity and unit price features, cost occurrence stage features, evidence completeness features, and cost coverage features.

[0047] The cost coverage feature is used to determine whether a cost record is a single cost record, a consolidated cost record, or a cost record to be split. For example, a "main transformer purchase order" in a materials procurement system is usually a single cost record that directly corresponds to the main transformer equipment node; a "settlement amount for electrical installation contracts" in a contract settlement system may cover multiple cost nodes such as primary equipment installation, secondary equipment installation, and cable laying, and therefore is a consolidated cost record; a "collected amount of engineering construction costs" in a financial settlement system may consolidate multiple contracts or multiple cost categories, and therefore is a cost record to be split.

[0048] A cross-system feature mapping layer is constructed. This layer converts feature vectors from different systems into unified cost features. Instead of simply replacing field names, this layer combines source system features, compensation criteria features, and project context features to identify the appropriate location of cost records within the unified cost item semantic nesting tree.

[0049] Cross-system mapping can be represented as: ; in: Indicates the first Unified cost characteristics after mapping heterogeneous cost records; Representation and source system The corresponding cross-system mapping function; Indicates the first The source system for heterogeneous cost records; Indicates the first The source system feature vector of each heterogeneous cost record; Indicates the first Supplementary features to the scope of heterogeneous expense records; Indicates the first The project context characteristics of the project to which each heterogeneous cost record belongs; Indicates source system The corresponding system feature mapping parameters; Indicates source system Corresponding caliber compensation parameters; Indicates source system The corresponding context mapping parameters; Indicates source system The corresponding bias parameters; This represents the normalized mapping function; The caliber compensation feature indicates the deviation of the cost record relative to the unified cost caliber, including whether it is a merged record, a split record, whether contract information, material information, quantity information, supporting documentation, and whether there is a stage offset. The project context feature indicates the project type, voltage level, construction area, construction scale, and main technical solutions. Since the same cost name may be classified differently in different types of projects, introducing the project context feature can reduce mismatches. For example, "foundation engineering cost" may be classified as substation construction cost in substation engineering, and as pole foundation engineering cost in line engineering; the two should not be classified into the same atomic cost item solely based on name.

[0050] The parameters of the cross-system feature mapping layer can be determined through initial rule samples, historical confirmation samples, and review feedback samples. Initial rule samples are derived from the expense item comparison table, historical manual mapping tables, and preset business rules; historical confirmation samples are derived from expense records whose attribution has been confirmed through settlement audits, financial final accounts, or special reviews; review feedback samples are derived from records to be reviewed output by this method, where business personnel confirm the true attribution node, and then write the confirmation result back to form a sample. Through this method, the cross-system feature mapping layer can be continuously modified based on the initial rules, gradually adapting to different systems, project types, and expense definitions.

[0051] After obtaining the unified cost characteristics, the caliber matching degree between heterogeneous cost records and each subject node in the semantic nesting tree of cost subjects is calculated. The caliber matching degree is used to determine whether a cost record should be assigned to a certain subject node.

[0052] The caliber matching degree can be expressed as: ; in: Indicates the first Heterogeneous cost record and the first The degree of consistency in definitions between subject nodes; Indicates the first The radial quantity of each subject node; Indicates the length of the vector; Indicates the granularity level correction factor; Indicates the correction factor for the completeness of evidence; This indicates the adjustment factor for the cost range; This represents the phase consistency correction coefficient.

[0053] In practical applications, if the granularity of the expense record is consistent with the granularity of the account node, the granularity level correction coefficient is higher; if the expense record is significantly coarser or finer than the account node, the granularity level correction coefficient is lower. If the expense record has evidence such as contracts, purchase orders, bills of quantities, settlement audit records, or final settlement vouchers, the evidence completeness correction coefficient is higher; if the evidence is missing, the evidence completeness correction coefficient is lower. If the expense range of the expense record exceeds the node boundary, the expense range correction coefficient is lower. If the occurrence stage of the expense record is consistent with the stage in which the node usually occurs, the stage consistency correction coefficient is higher.

[0054] Determine the set of candidate nodes based on the caliber matching degree: ; in: Indicates the first The set of candidate home nodes corresponding to each heterogeneous cost record; This represents any subject node in the semantic nesting tree of cost items; Indicates the first Matching thresholds for each subject node; When the candidate attribution node set is empty, the expense record is marked as a record to be verified, and the reason for the mismatch is output. Reasons for mismatch may include an unidentifiable expense name, expense scope exceeding the established tree range, missing evidence, uncertain project attribution, or inconsistent stages. When the candidate attribution node set contains only one node, that node is used as the initial attribution node. When the candidate attribution node set contains multiple nodes, granular backtracking processing begins.

[0055] Granular backtracking is executed in the following order: "first backtrack, then drill down, then split, and retain if unprocessable." Specifically, it first checks if the expense record exceeds the expense boundary of the current candidate node. If it does, it backtracks to the parent node until a node whose expense boundary covers the expense record is found. Then, it checks if the expense record has contract details, material details, quantity details, settlement details, or supporting documentation. If so, and the child node matching meets the requirements, it drills down to the child node. If the expense record covers multiple sibling nodes, it is allocated to multiple child nodes according to the splitting ratio. If the expense record lacks the necessary evidence for further drill-down or splitting, it is retained at the current node, and an unrecovered retained amount is generated.

[0056] When determining whether backtracking is necessary, the degree of backtracking necessity can be calculated: ; in: Indicates the first The first expense record is in The necessity of uptracing at each node; Indicates the extent to which the cost exceeds the limit; Indicates the degree to which the amount exceeds the limit; Indicates the degree of exceeding the limit in a stage; , , These represent the weights corresponding to the cost range, amount range, and stage attribute, respectively. If the necessity for tracing back exceeds the preset tracing threshold, it indicates that the current node's cost boundary cannot accommodate the cost record, and tracing back to the parent node is necessary. For example, if a contract settlement record is named "Electrical Installation Engineering Settlement," and its amount is significantly larger than the reasonable amount range for a single "Cable Laying" node, and the contract details involve multiple installation sub-items, it should not be directly assigned to the "Cable Laying" node, but should first be traced back to the "Installation Engineering Fee" node or the corresponding contract engineering node.

[0057] When determining whether drilling is feasible, the separability of drilling can be calculated: ; in: Indicates the first The first expense record is in Drill-down separability at each node; Indicates that the contract can be split; This indicates that the supplies can be separated. Indicates that the quantity of work can be broken down; Indicates a splittable settlement tag; This indicates that the evidence can be separated. , , , , These represent the weights of each separable tag; If the drill-down divisibility is greater than the preset drill-down threshold, and the caliber matching degree between the cost record and the child node meets the requirements, then the cost record will continue to drill down from the current node to the corresponding child node. For example, in the materials procurement system, if a record is "Power Cable Procurement Contract", and its contract details further list the amounts for different specifications of cables, then this record can continue to drill down from the "Cable Equipment Fee" node to the "Atomic Cost Items for Different Specification Cables" node.

[0058] When a cost record covers multiple sub-nodes, its allocation ratio to each sub-node is calculated. The allocation ratio is determined by combining the matching degree of the cost scope, contract details, material details, work quantity details, settlement details, and historical cost structure.

[0059] The allocation ratio can be expressed as: ; in: Indicates the first Fee records are assigned to child nodes. The proportion; This represents the child node determined according to the contract details. Reference value for amount; This indicates the sub-nodes determined based on the material details. Reference value for amount; This represents the sub-nodes determined based on the quantity of work and the unit price. Reference value for amount; This indicates the sub-node determined based on the settlement audit details. Reference value for amount; This indicates a child node determined based on the cost structure of similar historical projects. Reference value for amount; , , , , These represent the weights corresponding to contract details, material details, project quantity details, settlement details, and historical cost structure, respectively. Represents a node The set of child nodes; If contract details, material details, quantity details, and settlement details all exist, the weight of the corresponding details is increased. If the above details are missing, the cost structure of similar historical projects can be temporarily used as a supplementary basis, and the splitting result is marked as a low-evidence splitting result. This setting can avoid forcibly splitting costs precisely when evidence is insufficient, while still preserving a traceable path for review.

[0060] The allocation amount for child nodes is obtained based on the allocation ratio: ; in: Indicates the first Fee records are assigned to child nodes. The amount; Indicates the first The original amount of the expense record; After completing the attribution, tracing, drill-down, splitting, and retention of cost records, a unified cost topology amount is generated. For any node in the semantic nesting tree of cost items, the amounts mapped to that node and its descendant nodes are summarized to form a unified amount for that node across different source systems and stages.

[0061] The unified cost topology amount can be expressed as: ; in: Indicates the first Each cost item node in the source system ,stage A uniform amount; Indicates source system In the stage The following is a collection of expense records; Indicates the first The expense record belongs to the first The amount allocation coefficient for each node; If the fee records are complete, they should be included in the node. If the expense record is allocated proportionally to multiple nodes, then the allocation coefficient is 1; if the expense record is allocated proportionally to multiple nodes, then the allocation coefficient is the number of nodes to which the expense record is allocated. The proportion; if the cost record does not belong to the node. If so, the amount allocation coefficient is 0.

[0062] For example, in the materials procurement system, if a "Main Transformer Purchase Order" is fully included in the "Main Transformer Equipment Fee" atomic expense item, then its amount allocation coefficient is 1. In the contract settlement system, if a "Settlement Amount for Electrical Installation Engineering" is split into three sub-nodes—Primary Equipment Installation, Secondary Equipment Installation, and Cable Laying—then its amount allocation coefficients are the corresponding split ratios. In the financial settlement system, a expense record that cannot be further penetrated is retained under the current parent node as an unpenetrated retained amount.

[0063] After generating a unified cost topology amount, a consistency check is performed on the semantic nesting tree of cost items. For any parent node, its amount should be equal to the sum of the amounts of its child nodes plus the sum of the amounts not penetrated and retained, allowing for acceptable deviations caused by rounding, provisional costs, or retention of standards.

[0064] Topology consistency deviation can be expressed as: ; in: Indicates the first Each subject node in the source system ,stage Topological consistency deviations; Represents child nodes In the source system ,stage A uniform amount; Represents a node Reserved amounts that cannot be further penetrated or allocated at this time; This represents the smallest positive number that prevents the denominator from being zero.

[0065] When the topology consistency deviation exceeds the preset consistency threshold, a verification anomaly record is output. The verification anomaly record includes at least the account node, source system, expense stage, parent node amount, sum of child node amounts, amount not penetrated and retained, deviation amount, and an anomaly reason message. Anomaly reason messages may include omission of amount, duplicate aggregation, abnormal splitting ratio, inconsistent expense definitions, insufficient evidence, or incorrect project association.

[0066] Within a unified cost topology space, the amounts for the same item node under different business systems or different cost stages are aligned and compared. For example, the differences between nodes in the budget estimation system and the financial settlement system can be compared, as can the differences in equipment procurement between the budget estimation system and the material procurement system, or the differences in financial aggregation between the contract settlement system and the financial settlement system.

[0067] Node differences can be represented as: ; in: Indicates the first Each subject node in the system With system The difference in amount between them; Indicates the first Each subject node in the system The uniform amount in; Indicates the first Each subject node in the system The uniform amount in the system.

[0068] like For the budget system, In a financial settlement system, this node difference represents the final difference between the preliminary estimate and the final settlement. If... For the budget system, In a materials procurement system, this node difference can be used to determine whether the procurement amount deviates from the budget control amount. If For contract settlement system, For a financial settlement system, this node difference can be used to determine whether there is a discrepancy between the settlement amount and the financial settlement amount.

[0069] Further calculation of node difference attribution degree is used to determine whether the total project cost difference can be accurately attributed to atomic cost items or non-atomic nodes with reasonable reservation explanations in the semantic nesting tree of cost items.

[0070] The node difference relocation degree can be expressed as: ; in: Indicates the first The degree of node difference attribution for each project; Indicates the first The total cost difference for each project; This represents the set of atomic cost nodes in the semantic nesting tree of cost items; Indicates the first The first project The difference in amount between individual atomic cost nodes; This represents the set of non-atomic nodes that have a reserved amount; Indicates the first The first project The amount of the retained difference for each non-atomic node; When the node difference attribution degree is closer to 1, it means that the total project difference can be completely attributed to the specific cost node; when the node difference attribution degree is low, it means that there are still many differences that cannot be explained or penetrated, and further supplementary evidence materials, adjustments to account nodes, or corrections to cross-system mapping parameters are required.

[0071] To aid in subsequent causal analysis, this embodiment also generates difference location markers for nodes where discrepancies occur. These difference location markers are not directly used as the final causal conclusion, but rather to indicate in which business stage or type of caliber shift the discrepancy is more likely to occur.

[0072] The difference attribution marker can be represented as: ; in: Indicates the first The first project The difference in the location markers of each node; Indicates the first The first project The estimated cost for each node; Indicates the first The first project Differences in the amount of work at each node; Indicates the first The first project Estimated project quantity for each node; Indicates the first The first project The unit price difference of each node; Indicates the first The first project The estimated unit price for each node; Indicates stage offset characteristics; Indicates contract offset characteristics; Indicates the characteristics of material offset; Indicates the characteristics of the completeness of evidence; For example, if a significant discrepancy exists between the preliminary cost estimate system and the material procurement system for a certain cable equipment item, while the discrepancies are smaller between the material procurement system, the contract settlement system, and the financial settlement system, then the discrepancy location marker indicates that the discrepancy mainly arose during the procurement stage. If a discrepancy exists between the contract settlement system and the financial settlement system for a certain construction project item, then the discrepancy location marker indicates that there may be a deviation in the financial data collection criteria. If a significant discrepancy exists for a certain expense item, but the quantity of work, unit price, contract details, or supporting documentation are missing, then the discrepancy location marker indicates that supplementary evidence should be provided for that item before determining its cause.

[0073] Furthermore, this embodiment can calculate a penetration alignment quality evaluation value to determine whether the current penetration alignment result meets the requirements of subsequent difference rate analysis, core cost item identification, and cause tracing analysis.

[0074] The penetration alignment quality evaluation value can be expressed as: ; in: This represents the quality evaluation value of the penetration alignment of multi-source heterogeneous valence data; This represents the average value of the project node difference relocation degree; This represents the average value of the topology consistency deviation. This represents the average level of completeness of expense record evidence. This indicates the percentage of expense records that require manual review. This indicates the percentage of the total amount reserved. , , , , These represent the weights corresponding to the degree of discrepancy attribution, topological consistency, completeness of evidence, proportion of manual review, and proportion of retained amount, respectively. When the penetration alignment quality evaluation value reaches the preset quality threshold, it indicates that the current penetration alignment results can be used for subsequent cost difference rate statistics, core cost driver factor identification, and difference cause analysis. When the penetration alignment quality evaluation value does not reach the preset quality threshold, it indicates that the current data contains many cost records that cannot be located, cannot be penetrated, have insufficient evidence, or have inconsistent definitions. Data supplementation, subject node adjustment, or cross-system mapping parameters should be prioritized.

[0075] After completing the above processing, the penetration alignment results are output. The penetration alignment results include the semantic nesting tree of cost items, the radial quantity of each item node, the unified cost characteristics of each heterogeneous cost record, the candidate attribution node set, the final attribution node, the split ratio, the amount retained without penetration, the unified cost topology amount, the topology consistency deviation, the node difference amount, the difference placement mark, the record to be reviewed, and the penetration alignment quality evaluation value.

[0076] Further output the discrepancy location table. The discrepancy location table includes project number, project name, source system, original cost code, original cost name, unified account node, node level, estimated amount, procurement amount, contract amount, settlement amount, final account amount, node discrepancy amount, amount not retained, supporting evidence, discrepancy location marker, and review status.

[0077] For example, in a new 110kV substation construction project, "equipment purchase cost" is a primary cost item in the preliminary cost estimate system. The materials procurement system contains multiple purchase orders for main transformers, switchgear, cables, and protection automation equipment. The contract settlement system contains corresponding purchase contracts and installation contracts. The financial settlement system consolidates some costs into equipment costs. Using the method in this embodiment, the equipment purchase cost is first designated as the parent node, and then the main transformer equipment cost, switchgear cost, cable equipment cost, and protection automation equipment cost are designated as child nodes. Specific purchase orders or contract lists are designated as atomic cost items. Subsequently, a cross-system feature mapping layer maps cost records from different systems to the same cost topology space. If a financial settlement record covers both main transformer equipment cost and switchgear cost, it is determined to be a consolidated cost record through granular backtracking and is then split into corresponding child nodes based on purchase orders, contract details, and historical cost structures. Finally, the system can output a unified amount for each equipment cost node across the preliminary cost estimate, procurement, contract, settlement, and final cost estimates stages, and attribute the differences between the preliminary cost estimate and the final cost estimate to the specific equipment cost node.

[0078] For example, in a power distribution network project, the preliminary cost estimate system categorizes the project cost into the line structure cost and other costs. The contract settlement system includes list items such as cable laying, pole foundations, channel coordination, and compensation for damages, while the financial settlement system contains the combined construction cost. Using the method in this embodiment, the line structure cost can be further broken down into nodes such as cable material costs, cable laying costs, and pole foundation costs. Compensation for damages can be established as a separate node, and the relevant costs can be determined based on site verification, compensation agreements, and settlement audit records to determine whether they should be categorized under the compensation for damages node or the construction measures cost node. Thus, the total project variance is no longer merely a superficial "line project cost exceeding the preliminary estimate," but can be further attributed to specific nodes such as "difference in cable material costs," "difference in pole foundation quantities," or "difference in retained compensation for damages."

[0079] In this embodiment, if data cannot be fully penetrated, it is not forcibly categorized. Instead, it generates a retained amount for non-penetration and a record pending review. For example, if a financial settlement record only shows "Other Engineering Costs" and lacks contract number, material code, engineering quantity information, and supporting documents, this record can be retained under "Other Expenses" or its corresponding parent node and marked as lacking evidence. After supplementing the contract list, approval records, or financial aggregation explanation, it can be re-entered into cross-system feature mapping and granular backtracking processing. This avoids misclassification and ensures that every expense that cannot be penetrated is traceable.

[0080] The method described in this embodiment can be deployed in a cost data analysis platform. The platform may include a data access module, a standardization processing module, a cost item semantic nesting tree construction module, a radial quantity generation module, a source system feature generation module, a cross-system feature mapping module, a granular backtracking module, a unified cost topology amount generation module, a consistency verification module, a difference relocation module, and a result output module.

[0081] The data access module is used to access cost data from the budget estimation system, financial settlement system, ERP system, material procurement system, and contract settlement system. The standardization processing module unifies the processing of project identifiers, monetary units, time fields, contract numbers, supplier names, material names, and expense names. The cost item semantic nesting tree construction module constructs a cost item semantic nesting tree based on budget estimation items, settlement items, ERP cost items, material procurement categories, contract list items, and historically confirmed expense details. The radial quantity generation module generates radial quantities for each item node. The source system feature generation module generates source system feature vectors for each heterogeneous expense record. The cross-system feature mapping module maps source system feature vectors to a unified cost topology space. The granular backtracking module determines whether an expense record should be assigned to a parent node, child node, multiple child nodes, or retained in the current node. The unified cost topology amount generation module generates unified amounts for each item node across different source systems and stages. The consistency verification module verifies the consistency between parent node amounts, child node amounts, and non-penetrated retained amounts. The difference relocation module is used to calculate node differences, node difference relocation degree, and difference relocation markers. The results output module is used to output the penetration alignment results, the difference location table, and the records to be reviewed.

[0082] In other implementations, the semantic nested tree of cost items can be constructed separately according to different project types. For example, the nested tree corresponding to the main network project can highlight cost nodes such as substations, lines, cables, communications, and protection automation; the nested tree corresponding to the distribution network project can highlight cost nodes such as transformer substations, lines, switching stations, ring main units, cable channels, and compensation for damaged or damaged equipment; and the nested tree corresponding to the technical renovation project can highlight cost nodes such as equipment upgrades, system upgrades, protection automation upgrades, communication upgrades, and auxiliary system upgrades. Different nested trees can share some first-level cost nodes, and different atomic cost item stopping conditions can be set according to the project type.

[0083] In other implementations, the cross-system feature mapping layer can be implemented using rule-based mapping, machine learning mapping, or a combination of both. When there are few historical confirmation samples, an initial mapping can be formed using a cost item lookup table and business rules. Once sufficient historical confirmation and review feedback samples have been accumulated, the mapping parameters can be optimized through supervised learning to improve the matching accuracy between cost records and cost item nodes. Regardless of the implementation method, the radial quantity, candidate attribution node, matching result, split ratio, and pending review status should be retained to ensure that the penetration alignment result is interpretable, verifiable, and traceable.

[0084] In other implementations, the alignment results can be further connected to the cost variance analysis module. The variance analysis module can, based on the variance location table, statistically analyze the variance rates between the preliminary and final estimates according to project type, voltage level, construction unit, project area, cost nodes, and business stage, identifying cost nodes and weak points with high variance rates. Since this embodiment has aligned heterogeneous cost data to a unified cost topology space, subsequent variance analysis no longer relies on manual data processing and can directly conduct statistics, tracing, and control from the unified cost node level.

[0085] In some embodiments, a multi-source heterogeneous valence data penetration alignment system is proposed, comprising: The data acquisition module is used to acquire cost data of the target power grid project in multiple cost-related systems. The cost data includes cost identification information, cost description information, cost amount information, cost stage information, and related information used to characterize the source or basis of costs. The nested tree construction module is used to construct a semantic nested tree of cost items based on the hierarchical relationship, cost attribution relationship, and historically confirmed cost details among the cost items of the power grid project. The semantic nested tree of cost items includes multiple item nodes, which are hierarchically divided into at least a total cost node, a cost item node, and an atomic cost item node. The total cost node is used to cover the total project cost, the cost item node is used to cover the sub-items cost, and the atomic cost item node is used to cover the minimum attributable cost. The caliber feature generation module is used to generate caliber features for each subject node based on the cost boundary, applicable stage, attribution basis, and splittable status of each subject node. The source data feature generation module is used to generate corresponding source data features based on the source features, cost content features, amount features, and correlation basis features of the cost records in various cost-related systems. The cross-system mapping module is used to map the source data features from different cost-related systems to a unified cost expression space through the cross-system feature mapping layer, so as to obtain the unified cost features corresponding to the cost records. The candidate attribution determination module is used to determine the candidate attribution position of the cost record in the semantic nesting tree of the cost item based on the matching relationship between the unified cost feature and the caliber feature of each item node. The final attribution determination module is used to determine the final attribution position of the expense record in the semantic nesting tree of the cost item based on the correspondence between the expense record and the candidate attribution position in terms of expense range, detailed support and amount; wherein, the final attribution position includes the parent node, the next node, multiple sibling nodes or the current node of the candidate attribution position; The topology amount generation module is used to generate a unified cost topology amount based on the final attribution location of the cost record; The results output module is used to output the penetration alignment results between the cost records and the semantic nesting tree of cost items.

[0086] In some embodiments, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method as described in any embodiment of the present invention.

[0087] In some embodiments, a computer-readable storage medium is provided on which a computer program is stored, which, when executed by a processor, implements the method as described in any embodiment of the present invention.

[0088] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0089] Those skilled in the art will recognize that the units and algorithm steps described in the embodiments disclosed herein can be implemented using electronic hardware, computer software, or a combination of electronic hardware and software. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0090] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0091] In the several embodiments provided in this application, any function, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0092] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for penetrating and aligning multi-source heterogeneous valence data, characterized in that, Includes the following steps: Obtain cost data for the target power grid project from multiple cost-related systems. The cost data includes cost identification information, cost description information, cost amount information, cost stage information, and related information used to characterize the source or basis of costs. Based on the hierarchical relationship between the cost items of the power grid project, the cost attribution relationship, and the historically confirmed cost details, a semantic nesting tree of cost items is constructed. The semantic nesting tree of cost items includes multiple item nodes, specifically including a total cost node for receiving the total project cost, a cost item node for receiving the sub-items cost, and an atomic cost item node for receiving the minimum attributable cost. Based on the cost boundaries, applicable stages, attribution basis, and splittable status of each subject node, generate the caliber characteristics corresponding to each subject node. Based on the source characteristics, content characteristics, amount characteristics, and correlation basis characteristics of the expense records in each cost-related system, corresponding source data characteristics are generated; The source data features from different cost-related systems are mapped to a unified cost representation space to obtain the unified cost features corresponding to the cost records; Based on the matching relationship between the unified cost features and the caliber features of each subject node, the candidate location of the cost record in the semantic nesting tree of the cost subject is determined; Based on the correspondence between the cost range, detailed support and amount between the cost record and the candidate location, the final location of the cost record in the semantic nesting tree of the cost item is determined. The final location includes the parent node, the next node, multiple sibling nodes or the current node of the candidate location. Generate a unified cost topology amount based on the final location of the cost record, and output the penetration alignment result between the cost record and the semantic nesting tree of the cost item.

2. The multi-source heterogeneous valence data penetration alignment method according to claim 1, characterized in that, The multiple cost-related systems include at least two of the following: a preliminary budget system, a financial settlement system, an ERP system, a materials procurement system, and a contract settlement system; The fee identification information includes at least one of the following: project identifier, fee code, and fee name; The cost description information includes at least one of the following: cost item, cost category, cost content, and the part of the project to which the cost belongs; The associated information includes at least one of the following: contract information, material information, project quantity information, settlement information, final account voucher information, and evidentiary document information.

3. The multi-source heterogeneous valence data penetration alignment method according to claim 1, characterized in that, Constructing a semantic nested tree for cost items includes: The main hierarchy of the semantic nesting tree of cost accounts is determined by the preliminary cost account and the final cost account; Link cost items, material categories, contract list items, settlement audit details, and historical confirmed expense details to the corresponding main levels; When the cost names are different in different cost-related systems but the cost boundaries are the same, configure them as aliases for the same subject node; When the cost names are similar but the cost boundaries are inconsistent in different cost-related systems, different subject nodes should be established for each.

4. The multi-source heterogeneous valence data penetration alignment method according to claim 1, characterized in that, The caliber characteristics include cost attribute characteristics, stage attribute characteristics, contract attribute characteristics, material attribute characteristics, quantity attribute characteristics, evidence attribute characteristics, and divisibility attribute characteristics; Among them, the cost attribute feature is used to characterize the cost category to which the subject node belongs, the stage attribute feature is used to characterize the cost stage to which the subject node applies, the contract attribute feature is used to characterize whether the subject node is supported by a contract or order, the material attribute feature is used to characterize whether the subject node corresponds to equipment or materials, the quantity attribute feature is used to characterize whether the subject node can be broken down into quantity and unit price, the evidence attribute feature is used to characterize the type of evidence document required for the subject node, and the splittable attribute feature is used to characterize whether the subject node is allowed to continue drilling down to the next level node.

5. The multi-source heterogeneous valence data penetration alignment method according to claim 1, characterized in that, The source data features include source system features, original coding features, cost name features, amount features, project ownership features, contract ownership features, material attribute features, engineering quantity attribute features, stage attribute features, evidence completeness features, and cost coverage features. Among them, the expense coverage feature is used to characterize whether an expense record is a single expense record, a combined expense record, or an expense record to be split.

6. The multi-source heterogeneous valence data penetration alignment method according to claim 1, characterized in that, Mapping source data features from different cost-related systems to a unified cost representation space, including: Based on the source data characteristics, compensation criteria characteristics, and project context characteristics of the cost records, generate unified cost characteristics corresponding to the cost records; Among them, the caliber compensation feature is used to characterize whether the cost record is a merged record, a split record, whether it lacks contract information, material information, quantity information, evidentiary documents, and whether there is a stage offset; Project context features are used to characterize the project type, voltage level, construction area, construction scale, and technical solution of the target power grid project.

7. The multi-source heterogeneous valence data penetration alignment method according to claim 1, characterized in that, Determining the candidate location of expense records in the semantic nesting tree of cost items includes: Calculate the degree of matching between the uniform cost characteristics of the cost records and the caliber characteristics of each item node; Based on the matching degree and the matching threshold corresponding to each subject node, the candidate location of the expense record is determined; When the candidate location is not determined, mark the fee record as a record to be verified and output the reason for the mismatch; When a candidate location is determined, that candidate location is used as the initial location. When multiple candidate locations are identified, the final location is determined based on the scope of expenses, detailed supporting documents, and corresponding amounts.

8. The multi-source heterogeneous valence data penetration alignment method according to claim 1, characterized in that, Determining the final location of expense records within the semantic nesting tree of cost accounts includes: When the cost coverage of a cost record exceeds the cost range of the corresponding subject node of the candidate location, the cost record will be assigned to the next higher level node of the candidate location. When the expense record has contract details, material details, project quantity details, settlement details or evidence documents, and the details can correspond to the next level node of the candidate location, the expense record will be assigned to the corresponding next level node. When expense records cover multiple peer nodes at the same time, the splitting ratio is determined based on the contract amount, material amount, project amount, settlement amount or historical expense structure, and the expense records are split into multiple peer nodes. When a fee record lacks the details or supporting documentation required for further breakdown, the fee record is retained at the current node, and an un-segmented amount is generated.

9. The multi-source heterogeneous valence data penetration alignment method according to claim 1, characterized in that, After generating the unified cost topology amount, it also includes: Perform consistency checks on the parent node amount, the next level node amount, and the unsubdivided amount in the semantic nesting tree of cost items; When the deviation between the amount of the parent node and the amount of the next level node and the undivided amount exceeds a preset threshold, a verification exception record is output. In a unified cost expression space, compare the amount differences of the same subject node under different cost-related systems or different cost stages, and generate a difference location table based on the amount differences; The discrepancy location table includes project identifier, original cost information, unified subject node, node level, amount of each cost stage, node difference amount, unspecified amount, supporting documents, and review status.

10. A multi-source heterogeneous valence data penetration alignment system, characterized in that, include: The data acquisition module is used to acquire cost data of the target power grid project in multiple cost-related systems. The cost data includes cost identification information, cost description information, cost amount information, cost stage information, and related information used to characterize the source or basis of costs. The nested tree construction module is used to construct a semantic nested tree of cost items based on the hierarchical relationship, cost attribution relationship, and historically confirmed cost details among the cost items of the power grid project. The semantic nested tree of cost items includes multiple item nodes, which are hierarchically divided into at least a total cost node, a cost item node, and an atomic cost item node. The total cost node is used to cover the total project cost, the cost item node is used to cover the sub-items cost, and the atomic cost item node is used to cover the minimum attributable cost. The caliber feature generation module is used to generate caliber features for each subject node based on the cost boundary, applicable stage, attribution basis, and splittable status of each subject node. The source data feature generation module is used to generate corresponding source data features based on the source features, cost content features, amount features, and correlation basis features of the cost records in various cost-related systems. The cross-system mapping module is used to map the source data features from different cost-related systems to a unified cost expression space through the cross-system feature mapping layer, so as to obtain the unified cost features corresponding to the cost records. The candidate attribution determination module is used to determine the candidate attribution position of the cost record in the semantic nesting tree of the cost item based on the matching relationship between the unified cost feature and the caliber feature of each item node. The final attribution determination module is used to determine the final attribution position of the expense record in the semantic nesting tree of the cost item based on the correspondence between the expense record and the candidate attribution position in terms of expense range, detailed support and amount; wherein, the final attribution position includes the parent node, the next node, multiple sibling nodes or the current node of the candidate attribution position; The topology amount generation module is used to generate a unified cost topology amount based on the final attribution location of the cost record; The results output module is used to output the penetration alignment results between the cost records and the semantic nesting tree of cost items.

Images