A system and method of water conservancy BIM cost AI agent based on ontology model construction

Abstract

The application relates to a system and method for water conservancy project BIM cost AI intelligent agent based on an ontology model, and belongs to the technical field of water conservancy project cost intelligentization. A structured water conservancy cost knowledge base is built by fusing industry norms, quotas and enterprise historical cases, a four-layer water conservancy BIM cost ontology model is constructed, an AI dynamic reasoning engine is carried to digitize and model business semantics, data relationships and decision logic, and a self-driven cost business operation system is formed; multi-source data collaborative preprocessing is realized through phased BIM model management, accurate calculation and intelligent grouping are completed by binding BIM components and cost parameters based on ontology reasoning, cost information is written back to the BIM model in the form of shared parameters to build BIM5D data assets, and nine scenes from planning estimation to completion settlement are covered to realize whole-life-cycle dynamic management and control. Whole-process intelligentization and data assetization management and control of water conservancy project cost are realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent cost estimation technology for water conservancy projects, specifically to a system and method for constructing an AI intelligent body for BIM cost estimation of water conservancy projects based on an ontology model. Background Technology

[0002] Water conservancy projects are characterized by diverse structural types, complex construction techniques, significant regional differences in cost standards, long construction periods, and large investment scales. Cost estimation and management are core aspects of project investment control. Building Information Modeling (BIM) technology serves the entire process of project investment decision-making, design, bidding, construction, and final acceptance, aiming to provide a platform for collaborative work and information sharing among all project participants. With the deep integration of BIM and artificial intelligence technologies, intelligent cost estimation for water conservancy projects has become an industry trend. However, current mainstream BIM cost estimation applications still have many technical shortcomings, failing to meet the needs of refined, automated, and asset-based cost management in water conservancy projects.

[0003] Specifically, existing BIM cost accounting for water conservancy projects suffers from the following key technical deficiencies: First, research on BIM cost application is lagging and a knowledge system is lacking. Currently, BIM applications in water conservancy projects mainly focus on hydraulic design modeling, construction simulation, and construction management. Research and application of BIM technology in water conservancy project cost estimation are relatively limited. Existing results are mostly limited to application concepts, management framework design, and secondary development of cost estimation software, and a systematic knowledge system and standardized workflow for water conservancy project cost estimation have not yet been formed. Second, data silos are severe and collaborative management is insufficient. Each stage of water conservancy projects uses its own independent models and software systems, resulting in poor information flow among design, cost estimation, and construction stakeholders, leading to duplicate cost data recording. The problems of delayed input and information transmission, data fragmentation, and inefficient collaboration are prominent. Third, the lag in updating quota standards and the difficulty in dynamic cost control make traditional quota-based cost management models ill-suited to the dynamic cost control needs arising from fluctuations in building material prices and engineering changes. Fourth, the low level of intelligence and lack of autonomous decision-making capabilities mean that most existing cost management solutions only achieve basic data statistics and report generation, lacking in-depth modeling and dynamic reasoning capabilities for cost business rules; intelligent applications remain at a superficial auxiliary stage. Fifth, the lack of full lifecycle management capabilities means that existing technologies typically only cover single aspects such as construction drawing budgets and bidding control prices, failing to cover the entire cost management process from planning estimates to final settlement. Furthermore, the model and cost data are independent, making it difficult to effectively write cost information back into the BIM model and forming traceable and reusable cost data assets. Some existing solutions attempt to use AI algorithms to optimize cost accounting or realize BIM model quantity calculation functions, but they generally suffer from technical fragmentation and fail to build a complete technical architecture covering knowledge base, ontology semantic modeling and dynamic reasoning, thus failing to form an intelligent closed loop of "data governance - model building - rule reasoning - autonomous decision-making - automatic execution".

[0004] In summary, existing BIM cost estimation technologies for water conservancy projects face multiple technical bottlenecks, including a lack of knowledge systems, severe data silos, low levels of intelligence, insufficient full lifecycle management capabilities, and the inability to assetize data. There is an urgent need for a systematic solution that can integrate water conservancy cost estimation knowledge bases, ontology semantic models, and AI dynamic reasoning engines to achieve a technological leap from passive control to proactive value management of water conservancy project costs throughout their entire lifecycle. Summary of the Invention

[0005] To address the problems of existing technologies, this invention provides a system and method for constructing an AI intelligent agent for water conservancy engineering BIM cost estimation based on an ontology model.

[0006] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: Firstly, a system for a water conservancy project BIM cost AI intelligent body based on ontology model construction, including a water conservancy cost knowledge base module, an ontology model construction and reasoning module, a BIM data access and collaboration module, an AI intelligent accounting module, a cost verification and optimization module, a dynamic update and model write-back module, a full life cycle cost control module, a human-computer interaction terminal and a data storage module. The water conservancy cost knowledge base module is used to collect water conservancy project cost compilation regulations, quota library, list pricing rules and fee standards based on the design characteristics of water conservancy project BIM cost software components, and to establish standardized water conservancy cost workflow description standards by combining enterprise historical cost case data, forming a structured and iterative water conservancy cost exclusive knowledge base. The ontology model construction and reasoning module is used to construct a water conservancy project BIM cost ontology model based on the water conservancy cost knowledge base and enterprise structured cost data. The model includes an entity layer, attribute layer, relationship layer, and rule layer. It is equipped with an AI dynamic reasoning engine. The dynamic reasoning engine is based on the business thresholds set in the rule layer and the mapping logic established in the relationship layer. Through business rules and dynamic reasoning mechanisms, it digitizes and models the water conservancy cost business semantics, data association relationships, and intelligent decision-making logic, realizing the integrated fusion of data governance, model construction, business rules, and action instructions, so as to autonomously drive cost business decision-making and automated operation. The BIM data access and collaboration module is used to access and parse mainstream water conservancy BIM model files. It adopts the "one model, multiple uses" and phased model management methods to realize the creation, sharing and multi-party collaborative management of BIM model information, and performs structured preprocessing on BIM component data. The AI ​​intelligent accounting module, based on the technical route of BIM, data, collaboration and AI, has a built-in semantic matching AI model, intelligent pricing engine and automatic quantity statistics model, which is used to realize BIM digital measurement, intelligent pricing and pricing and automatic generation of standardized results reports. The cost verification and optimization module is used to perform multi-dimensional compliance verification, anomaly identification and automatic error correction on the quantity calculation, pricing and fee calculation results based on the standards of the rule layer of the ontology model and the cases in the knowledge base, so as to optimize the pricing accuracy and reduce the workload of manual verification. The dynamic update and model write-back module is used to respond in real time to engineering design changes, building material price fluctuations and quota specification adjustments, and dynamically iterate the cost results; at the same time, it writes the comprehensive unit price and total cost information back to the BIM model in the form of shared parameters, so that BIM components carry cost information and construct BIM5D cost data assets. The full life cycle cost management module is used to cover nine major cost business scenarios, including planning budget, project suggestion estimate, feasibility study estimate, preliminary design estimate, construction drawing budget, bidding control price, bid price, construction process settlement and final settlement, so as to realize the full life cycle cost value management of water conservancy projects. The human-computer interaction terminal is used to realize the functions of knowledge base configuration, ontology rule editing, model import, parameter configuration, result display, manual review and data export. The data storage module is used to store knowledge base data, ontology model data, BIM model data, cost accounting results, verification logs, and full lifecycle cost data assets.

[0007] In one specific implementation of the first aspect, the entity layer of the ontology model includes four core entities: engineering type entities, BIM component entities, cost resource entities, and cost business scenario entities. The engineering type entities cover all categories of water conservancy projects, including dams, canals, pumping stations, dikes, and sluices. The BIM component entities include water conservancy-specific components such as earthwork, concrete, steel reinforcement, seepage prevention, slope protection, and electromechanical equipment. The cost resource entities include labor quotas, material quotas, machinery quotas, fee standards, tax rate parameters, and real-time building material prices. The cost business scenario entities correspond to the nine full life cycle cost control stages.

[0008] In one specific implementation of the first aspect, the relational layer of the ontology model includes component-work quantity association, component-bill quota matching relationship, quota-price mapping relationship, project type-accounting rule adaptation relationship and business scenario-pricing parameter association relationship, which are used to realize multi-dimensional data structured bidirectional binding to support AI dynamic reasoning and decision-making.

[0009] In one specific implementation of the first aspect, the AI ​​intelligent accounting module can autonomously execute a full set of repetitive water conservancy cost calculation tasks, replacing manual quantity calculation, pricing, summarization, and report generation operations, thereby achieving accurate and rapid calculation of engineering quantities and improving the accuracy of matching quotas and fees.

[0010] In one specific implementation of the first aspect, the full life cycle cost control module drives cost management to upgrade from the traditional price determination and control model to a proactive value management model covering the entire life cycle of the project.

[0011] The second aspect is a method for implementing a water conservancy project BIM cost AI intelligent agent based on an ontology model, applied to a system of water conservancy project BIM cost AI intelligent agents based on an ontology model, including the following steps: S1. Establish a standardized water conservancy cost knowledge base and workflow: Combining the design features of water conservancy engineering BIM cost software components, collect industry cost compilation regulations, quota library, list and fee specifications, integrate enterprise historical cost data and cases, establish a unified water conservancy cost workflow description standard, and form a structured and iterative exclusive knowledge base. S2. Construct a dynamic reasoning-enabled water conservancy cost ontology model: Based on the knowledge base and enterprise structured cost data, build a four-layer ontology architecture including entity layer, attribute layer, relationship layer and rule layer, configure refined cost business rules, and load an AI dynamic reasoning engine to digitize and model business semantics, data relationships and decision logic, forming an operating system that autonomously drives cost business. S3. Multi-source BIM data access and collaborative preprocessing: Import multi-stage target project BIM models, adopt the "one model for multiple uses" and phased model management mode, complete the extraction, cleaning and standardization of BIM component data, and realize multi-party data collaborative sharing. S4. AI Intelligent Semantic Reasoning and Digital Quantity Calculation: Relying on the dynamic reasoning capability of the ontology model, it intelligently binds BIM components and cost parameters, and automatically and accurately calculates the quantities of various components. S5. Intelligent pricing and output throughout the entire process: Based on the rules of the ontology model and the standards of the knowledge base, the system automatically completes quota matching, fee calculation and price fitting to achieve intelligent pricing and output, and automatically generates a standardized cost report. S6. Intelligent verification, optimization and autonomous decision-making execution: Verify cost results from multiple dimensions, automatically correct abnormal data, and let the AI ​​model autonomously execute the entire set of cost business to replace manual repetitive operations. S7. Cost data write-back and BIM5D data asset generation: Write the verified cost information into the BIM model so that the model components carry complete cost attributes, forming a traceable and transferable BIM5D cost data asset. S8. Full lifecycle dynamic management and iterative updates: For nine major cost business scenarios, it realizes dynamic cost accounting, change adjustment and value analysis of the entire project process, and responds to industry standards and market price updates in real time. S9. Results Archiving and Data Asset Accumulation: Complete the review and export of cost estimates and the archiving of data throughout the entire process, iteratively optimize the knowledge base and the ontology model, and accumulate standardized cost estimates data assets for the enterprise.

[0012] In one specific implementation of the second aspect, the ontology model construction in step S2 specifically includes: collecting official cost standards and high-quality cost data from enterprises in the water conservancy industry; defining standardized terms, attribute parameters, and business thresholds for each entity; establishing multi-dimensional bidirectional correlation mapping relationships to adapt to the accounting rules for different project types and different cost scenarios; and integrating an AI dynamic reasoning mechanism to achieve automatic rule determination, automatic parameter matching, and automatic business decision-making without human intervention, forming an intelligent decision-making closed loop.

[0013] In one specific implementation of the second aspect, the nine full life-cycle cost scenarios mentioned in step S8 include planning budget, project proposal estimate, feasibility study estimate, preliminary design estimate, construction drawing budget, bidding control price, bid price, construction process settlement and final settlement, covering the entire process of water conservancy project construction.

[0014] In one specific implementation of the second aspect, the cost information mentioned in step S7 includes a comprehensive unit price, a total price, and a detailed cost breakdown, which is written back into the Revit BIM model in the form of shared parameters, so that the BIM components carry full-dimensional cost information.

[0015] In one specific implementation of the second aspect, the "one model for multiple uses" and phased model management mode described in step S3 supports the separate management and collaborative sharing of project feasibility study, preliminary design, construction drawings and construction phase models, ensuring the continuous availability of model data throughout the entire life cycle.

[0016] The beneficial effects of this invention are as follows: 1. This invention proposes and constructs for the first time an intelligent BIM cost architecture for water conservancy projects, consisting of a "knowledge base + ontology model + AI dynamic reasoning." It establishes a structured, iterative cost knowledge base integrating industry compilation regulations, bill of quantities pricing standards, fee standards, and historical case data from enterprises. Combined with a water conservancy cost-specific ontology model comprising entity, attribute, relationship, and rule layers, and integrating an AI dynamic reasoning engine, it digitizes and models business semantics, data relationships, and intelligent decision-making logic, forming an integrated cost business operating system encompassing data governance, model computation, rule judgment, and action commands. This architecture breaks through the limitations of traditional cost systems that only perform basic data statistics and report generation. It deeply integrates cost business rules, engineering semantic relationships, and AI reasoning decision-making to form an intelligent decision-making closed loop, significantly improving the cost system's autonomous decision-making capability and business self-driving level, and realizing a shift from passive price control to proactive value management. Meanwhile, the relational layer of the ontology model establishes a multi-dimensional bidirectional mapping structure of components-work quantity, components-bill quota, quota-price, project type-accounting rules, and business scenario-pricing parameters. This transforms cost knowledge from static storage into a dynamic semantic network that can be reasoned and deduced. The AI ​​dynamic reasoning engine automatically completes component classification, rule matching, and parameter fitting based on the business thresholds preset in the rule layer and the mapping logic in the relational layer. This enables autonomous execution of cost business without continuous human intervention, significantly improving the intelligence level and consistency of cost preparation.

[0017] 2. This invention is based on a technical approach that deeply integrates BIM, data, collaboration, and AI. Through a BIM data access and collaboration module, it supports the efficient access and parsing of multi-source heterogeneous BIM model files. Employing a "one model, multiple uses" mechanism and a phased model management method, it achieves collaborative sharing and version-controllable management of BIM model information from project feasibility studies, preliminary designs, construction drawings, and construction phases. Structured preprocessing extracts, cleans, and standardizes component geometric and construction attributes, providing a unified and accurate data source foundation for subsequent quantity surveying and pricing. Furthermore, leveraging the semantic reasoning capabilities of the ontology model, it intelligently binds BIM components and cost parameters, automatically and accurately completing quantity statistics. Through intelligent pricing and multi-dimensional verification and optimization throughout the entire process, it achieves an automated closed-loop operation from quota matching, fee calculation, price fitting to output, significantly improving the collaborative efficiency and quality of the entire cost estimation chain. Furthermore, this invention responds in real time to design changes, price fluctuations, and specification adjustments through a dynamic update mechanism. By sharing parameters, it writes cost information such as comprehensive unit price and cost details back to the BIM model, enabling model components to carry complete geometric, construction, and cost attributes. This constructs a traceable and transferable BIM5D cost data asset. Simultaneously, it covers nine major business scenarios: planning estimation, estimation, preliminary estimate, budget, bidding control price, bid price, construction process settlement, and final settlement. This achieves integrated management and control of dynamic cost accounting, change adjustment, and value analysis throughout the entire life cycle of water conservancy projects. It effectively solves the common technical problems in traditional cost management, such as data fragmentation, inconsistent rules, poor compilation collaboration, and difficulty in reusing and accumulating results. It enables cost data to flow smoothly and continuously increase in value throughout the entire life cycle, providing a complete, intelligent, and asset-based technical solution for water conservancy project cost management. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall system architecture of the present invention.

[0019] Figure 2 This is a schematic diagram of the four-layer architecture of the ontology model and the AI ​​dynamic reasoning process of the present invention.

[0020] Figure 3 This is a schematic diagram of the overall process of the whole life cycle cost control method of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] like Figures 1 to 3 This paper presents a system and method for constructing an AI intelligent agent for BIM cost estimation in water conservancy projects based on an ontology model.

[0023] I. System Architecture: This invention discloses a water conservancy engineering BIM cost AI intelligent system based on an ontology model. The core architecture relies on a "knowledge base + ontology reasoning + BIM5D collaboration + AI empowerment" framework to achieve intelligent management and control of the entire water conservancy cost process. The system is deployed on an enterprise cloud server using a B / S architecture. Each module interacts with the other via standardized RESTful APIs. The underlying database uses a distributed relational database (such as PostgreSQL 14 and above) to store structured data, supplemented by a graph database (such as Neo4j 5.x) to store the entity relationship network of the ontology model, supporting efficient semantic querying and reasoning.

[0024] 1.1 Implementation Method of the Water Conservancy Cost Knowledge Base Module The construction of the water conservancy cost knowledge base module was carried out according to the following process: (1) Data collection: Collect current national and industry standards such as the "Regulations on the Compilation of Design Estimates (Estimations) for Water Conservancy Projects" (Water Resources Department

[2024] No. 323), "Budget Quotas for Water Conservancy Construction Projects", "Estimated Estimates for Water Conservancy Construction Projects", "Budget Quotas for Water Conservancy Equipment Installation Projects", "Estimated Estimates for Water Conservancy Equipment Installation Projects", "Estimated Estimates for Soil and Water Conservation Projects", and "Hourly Rate Quotas for Water Conservancy Construction Machinery", as well as local water conservancy quotas, list pricing specifications, and fee collection documents of various provinces and cities. At the same time, collect data on high-quality water conservancy cost cases completed by enterprises in the past ten years (no less than 100 items), covering all water conservancy projects such as hub projects, water diversion projects, and river projects.

[0025] (2) Knowledge Structuring: The collected specifications, quotas, and case data are structured according to a unified metadata model. Each knowledge entry must contain at least the following fields: knowledge ID, knowledge type (specification / quota / list / fee / case), applicable project type, applicable cost stage, geographical attribute, effective time range, specific rule content, and associated reference source. The structured knowledge is stored in the knowledge base in JSON format, and a full-text search engine is established to achieve semantic-level fast retrieval.

[0026] (3) Workflow standardization: Combining the design features of the BIM cost software components for water conservancy projects, a standardized cost workflow description standard covering the entire chain of "model import, component naming, cost attribute improvement, engineering quantity extraction, quota matching, fee calculation, price fitting, result output, and data write-back" is formulated, modeled and stored in BPMN 2.0 format, and called and executed by the AI ​​intelligent accounting module.

[0027] Inputs to the knowledge base module include: industry standard documents, historical cost estimation case data of enterprises, and design feature documents of BIM cost estimation software components.

[0028] The knowledge base module outputs: a structured, searchable, and iterative water conservancy cost-specific knowledge base, and a standardized workflow model.

[0029] 1.2 Implementation of Ontology Model Construction and Reasoning Module The ontology model construction and reasoning module is the core innovation of this invention, and its construction process is as follows: (1) Entity layer definition: Based on knowledge base data, four core entity classes are defined: Engineering Type Entities: Covering all categories of water conservancy projects, including dams (gravity dams, arch dams, earth-rock dams), canals (open channels, culverts), pumping stations, dikes, and sluice gates. Each Engineering Type entity includes the following attributes: Engineering Category Code (e.g., "DAM-001"), Engineering Name, Typical Feature Parameter Set, and Applicable Quota Version.

[0030] BIM Components include earthwork components (excavation, backfill), concrete components (subbase, lining, dam), reinforced concrete components, seepage control components (curtain grouting, cutoff wall), slope protection components (dry masonry, mortar-grouted masonry, gabion cages), and mechanical and electrical equipment components (pumps, gates, hoists), among other water conservancy-specific components. Each component entity includes attributes such as: component type code, geometric parameters (volume / area / length / weight), material properties, construction process label, project location, and cost attributes.

[0031] Cost Resource Entity: Includes labor quotas (classified by regional wage categories, from Category 1 to Category 6), material quotas (including real-time building material prices), machinery quotas (hourly rate), fee standards (including rate parameters such as safety and civilized construction fees, night construction surcharges, and temporary facility fees), and tax rate parameters (general VAT calculation / simplified VAT calculation).

[0032] Cost Scenario entities correspond to nine scenarios: planning budget, project proposal estimate, feasibility study estimate, preliminary design estimate, construction drawing budget, tender control price, bid price, construction progress settlement, and final settlement. Each scenario entity includes the following attributes: scenario code, applicable stage, pricing accuracy requirements (e.g., ±20% accuracy in the estimation stage, ±10% accuracy in the preliminary estimate stage, and ±5% accuracy in the budget stage), and a list of mandatory cost items.

[0033] (2) Attribute layer definition: Define a standardized set of attributes for each entity class. Taking BIM component entity as an example, its attributes include: component ID (GUID), component name, building function / location, geometric dimensions (length × width × height), and volume (m). 3 ), area (m 2 ), length (m), weight (t), concrete grade, steel reinforcement grade, construction method, and code of the relevant sub-project. The range of attribute values ​​and business thresholds (e.g., concrete member volume ≥ 0.1m³) are also specified. 3 (e.g., steel bars with a diameter ≥ 6mm) are explicitly defined in the attribute layer.

[0034] (3) Relationship layer establishment: Establish five types of core relationships: Component-Quantity Association: Associates BIM component entities with corresponding quantity calculation rules. For example, "concrete lining component" is associated with the rule "volume measurement of entities calculated according to design drawings".

[0035] Component-Bill of Quantities (BOQ) matching relationship: Associate BIM component entities with corresponding BBOQ items and quota sub-items. For example, “C25 concrete lining” is associated with the BBOQ item “concrete lining (code 500107)” and the quota sub-item “concrete lining pouring (quota number 40073)”.

[0036] Quota-price mapping relationship: Map quota items to corresponding base prices, labor costs, material costs, and machinery costs, and associate them with real-time price data sources.

[0037] Project type - accounting rule adaptation relationship: Different project types (hub / water diversion / river channel, etc.) are adapted to different pricing rules and fee standards.

[0038] Business Scenarios - Pricing Parameter Relationships: Different cost scenarios (estimation / preliminary estimate / budget / settlement) are associated with different sets of pricing parameters and accuracy requirements.

[0039] The above relationships are expressed in the form of OWL (Web Ontology Language) axioms, stored in a graph database, and support SPARQL queries and semantic reasoning.

[0040] Graph database schema definition: The entity-relationship network of the ontology model is stored in the Neo4j graph database as an attribute graph model. Node types include EngineeringType, BIMComponent, CostResource, CostScenario, QuotaItem, Price, etc., and node attributes include fields such as name, code, description, valid_from, and valid_to. Relationship types include HAS_COMPONENT, MATCHES_QUOTA, MAPPED_TO_PRICE, ADAPTS_TO_RULE, BELONGS_TO_SCENARIO, etc., and relation attributes include fields such as confidence (value 0~1) and priority (value 1~10). The graph database supports the Cypher query language for semantic retrieval and association reasoning.

[0041] (4) Rule layer configuration: The water conservancy cost management business rules are formalized into SWRL (Semantic Web Rule Language) rules. A typical rule example is as follows: Rule 1 (Component-Bill of Quantities Matching Rule): If the BIM component type is "Concrete Lining" and the concrete grade is ≥C25 and the thickness is ≥200mm, then match the bill of quantities item "Concrete Lining (500107)" and the quota item "Concrete Lining Pouring (40073)".

[0042] Rule 2 (Fee Calculation Rule): If the project type is "hub project", the safety and civilized construction fee shall be calculated at 2.5% of the investment in the first to fourth building and installation projects (excluding safety and civilized construction fees and other temporary projects); if it is "river project", the rate shall be reduced to 2.0%.

[0043] Rule 3 (Precision Threshold Rule): If the cost scenario is "estimated", the allowable deviation range for the comprehensive unit price is ±20%; if it is "preliminary estimate", the deviation range is ±10%; if it is "budget" or "settlement", the deviation range must be controlled within ±5%.

[0044] Rule 4 (Price Update Rule): If the market price of a material fluctuates by more than ±5% (relative to the current fixed base price), the automatic price update mechanism will be triggered.

[0045] Rule conflict resolution strategy: When multiple SWRL rules are triggered simultaneously and produce conflicting results, the system adopts a priority conflict resolution strategy: each rule has a preset priority level (1~10, the higher the value, the higher the priority), and higher priority rules are executed first; when rules of the same priority conflict, the most recently effective strategy (based on the rule's most recent modification timestamp) or a manually preset rule chain order is used for execution. All rule triggering and conflict resolution processes are recorded in the inference log for manual review and traceability.

[0046] (5) Integration of AI Dynamic Reasoning Engine: Based on the above ontology model (entity + attribute + relation + rule), integrate a semantic matching model based on the Transformer architecture and a deterministic reasoning engine based on a rule engine (such as Drools 8.x). The reasoning process is as follows: Input: BIM component data (geometric attributes + construction attributes) + cost scenario identifier + current time node.

[0047] Reasoning steps: ① The semantic matching model calculates the semantic similarity between the BIM component name / type and the ontology entity (similarity threshold ≥ 0.85) to complete the automatic classification of components; ② The rule engine triggers the corresponding SWRL rules based on the classification results and scene identifiers to complete the quota matching and fee parameter determination; ③ The price mapping module queries the real-time price data source based on the quota-price relationship to complete the price fitting.

[0048] Output: Complete cost inference results (including matched list of quotas, fee parameters, fitted prices, and inference confidence).

[0049] Semantic matching AI model architecture details: The semantic matching AI model adopts a lightweight Transformer encoder architecture (such as BERT-mini or RoBERTa-small), with the number of model parameters controlled between 100 million and 1 billion (ideally around 800M parameters), including a 6-layer Transformer encoder, 12 attention heads, and a hidden layer dimension of 384. The model input is a concatenated sequence of BIM component name description text (maximum length 64 tokens) and ontology entity name text (maximum length 32 tokens), and the output is the similarity score (between 0 and 1) corresponding to the [CLS] tag. The model training uses a contrastive learning objective function (InfoNCE loss), with a temperature coefficient τ=0.07 and a batch size of 64. The training data comes from samples with labeled component-quota matching relationships in the enterprise's historical cost data. After data augmentation (synonym replacement, random deletion), the size is no less than 10,000 labeled samples, which are divided into training, validation, and test sets in an 8:1:1 ratio.

[0050] Inputs to the ontology model building module: structured data from the water conservancy cost knowledge base (including standards, quotas, and cases) and historical cost data of enterprises (structured cost records of no less than 100 projects).

[0051] The output of the ontology model building module includes a four-layer architecture water conservancy engineering BIM cost ontology model (stored in OWL+SWRL format in a graph database and rule base), and a trained and fine-tuned semantic matching AI model.

[0052] 1.3 Implementation Method of BIM Data Access and Collaboration Module The BIM data access and collaboration module is responsible for the access, parsing, and preprocessing of multi-source BIM model data. (1) Model Import: Supports importing and parsing mainstream BIM model formats such as Revit (.rvt, 2018 and above) and IFC (.ifc, IFC4 and above). Model files are read using Revit API 2024 and above and the IFC parser (based on xBIMToolkit), extracting component geometric information (volume, area, length, location coordinates) and attribute information (component type, material, construction stage, floor information). The maximum size of a single model file is 2GB.

[0053] (2) “One Model, Multiple Uses” Management: A phased model management mechanism is adopted, and BIM models for the project feasibility study, preliminary design, construction drawings, and construction stages are archived and managed separately. Models at different stages are linked through a unique project identifier (Project GUID, using UUID v4 format) to ensure data traceability throughout the entire process. Model version management adopts an incremental storage strategy, storing only the differences compared to the previous version to reduce storage overhead. The differences are described in JSON Patch format.

[0054] (3) Data preprocessing: The extracted BIM component data is cleaned and standardized, including: ① Standardization of component names (e.g., unifying "C25 concrete wall" to "concrete wall-C25"); ② Unit standardization (all volumes are standardized to m). 3 The area is uniformly set to m. 2 The preprocessed data undergoes the following format conversions: ① Length is uniformly set to m; ② Missing attribute completion (completed based on component type and contextual reasoning, with inference confidence < 0.7 marked as requiring manual confirmation); ③ Coordinate normalization (unified to the project coordinate system, with the channel starting point or dam centerline as the origin).

[0055] Input to the BIM data access module: multi-stage BIM model files (.rvt / .ifc format, a single model file does not exceed 2GB).

[0056] Output of the BIM data access module: Standardized BIM component data table (including fields such as component ID, type, geometric parameters, material properties, construction properties, and stage, output in JSON format).

[0057] 1.4 Implementation Method of AI Intelligent Accounting Module The AI-powered intelligent accounting module, based on the dynamic reasoning capabilities of the ontology model, enables BIM-based digital measurement, intelligent pricing and calculation, and standardized output: (1) Automatic quantity calculation: Based on the "component-quantity association" in the ontology model relationship layer, the calculation rules for the quantity of each BIM component are automatically matched. The calculation logic is as follows: For earthwork components: The excavation volume is calculated based on the component's geometric parameters (excavation boundary, depth, and slope coefficient). The calculation formula is V=(S_upper + S_lower + √(S_upper × S_lower))×H / 3 (frustum formula), where the slope coefficient ranges from 0.33 to 1.00 depending on the soil type (Class I to Class IV). Specifically: the slope coefficient is 0.33~0.50 for Class I soil (soft soil), 0.50~0.67 for Class II soil (ordinary soil), 0.67~0.83 for Class III soil (firm soil), and 0.83~1.00 for Class IV soil (gravelly firm soil).

[0058] For concrete members: Calculations are made directly based on the member's volume parameters, deducting any individual area > 0.3m². 2 The volume of holes (such as gate slots, pipe penetration holes, etc.) is calculated using the formula: V_net = V_total - ΣV_hole (V_hole > 0.3m). 2 ).

[0059] For reinforced concrete members: The theoretical weight is calculated based on the reinforcement information (diameter, spacing, anchorage length, lap length), using the formula W=Σ(L_i×ρ×A_i), where ρ=7850kg / m 3 A_i is the cross-sectional area of ​​the reinforcing bar (π×d_i) 2 / 4), L_i is the total length of the reinforcing bar (including anchorage and lap splice, with the anchorage length taken as 35d~45d and the lap splice length taken as 300mm~600mm, the specific value to be determined according to the concrete grade and seismic resistance level).

[0060] For seepage-proof components: calculate the area or volume based on the design drawing parameters (depth, thickness, length). Curtain grouting is measured by borehole depth (m), and seepage-proof walls are measured by volume (m³). 3 ) Measurement.

[0061] For electromechanical equipment components: measured in "units / sets", the quantity of equipment is directly read from the component attributes.

[0062] The statistical results of the project quantity are summarized by sub-project (such as earthwork, concrete, steel reinforcement, seepage prevention, and mechanical and electrical equipment installation).

[0063] (2) Intelligent pricing: Based on the "component-list quota matching relationship" and "quota-price mapping relationship" in the ontology model relationship layer, the following operations are automatically completed: Quota matching: Based on parameters such as component type, construction process, and material grade, the system retrieves matching quota items from the quota database. The matching algorithm employs a dual verification mechanism of ontology semantic similarity calculation (threshold ≥ 0.85) + rule-based exact matching (SWRL rule triggering). When the semantic similarity score is between 0.75 and 0.85, the system pushes the candidate quota items to the manual review queue; when the score is below 0.75, the system automatically expands the search scope and activates a backup matching strategy (based on keyword Jaccard similarity, threshold ≥ 0.60).

[0064] Fee Calculation: Based on the project type, cost scenario, and region, the corresponding fee standard is retrieved from the rule layer. Fee items include: direct costs (labor costs + material costs + machinery costs), other direct costs (calculated at 3.5%~3.7% of direct costs), indirect costs (calculated at 3.7%~5.7% of direct project costs, with the higher value used for key projects and the lower value used for river projects), corporate profit (calculated at 7% of direct costs + indirect costs), and taxes (value-added tax, generally 9% or 3% for simplified tax calculation).

[0065] Price Fitting: Real-time building material prices are read from the cost resource entity (automatically synchronized daily at 2:00 AM via API connection to the local engineering cost information network), replacing the material prices in the quota base price to generate a dynamic comprehensive unit price. The price data source priority is: ① Prices published on the local engineering cost information network (first choice); ② Information prices from surrounding areas (alternative); ③ Quota base price (last choice).

[0066] (3) Automatic generation of results: Based on the pre-set standardized report templates (Word / Excel format), the bill of quantities, comprehensive unit price analysis table, measure item fee table, regulatory fee and tax table, total price summary table and other contents are automatically filled to generate a complete cost result report.

[0067] The inputs to the AI ​​intelligent accounting module are: standardized BIM component data table + ontology model reasoning results (matched list quotas and fee parameters) + real-time price data.

[0068] Outputs of the AI ​​intelligent accounting module: Detailed engineering quantity statistics table + comprehensive unit price analysis table + cost summary table + standardized cost result report (Word / Excel format).

[0069] 1.5 Implementation Method of Cost Verification and Optimization Module The cost verification and optimization module performs multi-dimensional verification of the accounting results based on the verification thresholds of the ontology model rule layer and industry standards: (1) Quantity accuracy verification: The quantity of work calculated by AI is compared with the historical data of similar works in the knowledge base. An early warning is triggered when the deviation exceeds ±10%. The verification formula is: Deviation rate = |V_AI - V_historical mean| / V_historical mean × 100%. V_historical mean is the statistical mean of no less than 5 samples of similar works in the knowledge base.

[0070] (2) Compliance verification of quota matching: Check whether the selection of quota items conforms to the requirements of project type and construction technology. The verification rules include: ① Whether the quota item is applicable to the current project type (e.g., the quota for earth-rock dam cannot be used for concrete dam); ② Whether the unit of measurement of the quota item is consistent with the unit of measurement of the component (e.g., m). 3 Corresponding to m 3 , t corresponds to t); ③ Whether the work content of the quota sub-item covers all procedures of component construction (e.g., whether the pouring quota includes procedures such as formwork installation and removal, and curing). An early warning will be triggered if any verification rule fails.

[0071] (3) Fee calculation rule verification: Check whether the fee calculation standard matches the project type, cost scenario, and regional category, and whether the fee rate is within a reasonable range. The reasonable ranges for each fee rate are set as follows: safety and civilized construction fee 2.5%±0.5%, other direct costs 3.6%±0.5%, and indirect costs 4.7%±1.0%.

[0072] (4) Price rationality verification: The fitted comprehensive unit price is compared with the historical unit price of similar projects in the knowledge base. If the deviation exceeds ±15%, an early warning is triggered and an alternative price scheme is automatically recommended (refitted from the alternative price data source).

[0073] Automatic error correction mechanism: For projects that trigger warnings, the system automatically performs the following corrective operations based on the pre-defined error correction rules in the rule layer: ① If the quota matching deviation is found (semantic similarity 0.75~0.85), the semantic similarity threshold is lowered (from 0.85 to 0.75) and the search is repeated. The search results are pushed to the manual review queue; ② If the price deviation exceeds the limit (deviation > ±15%), the system automatically switches to the backup price data source for refitting. If the deviation still exceeds the limit, it is marked as requiring manual confirmation; ③ If the fee rate deviates, it is automatically corrected to the median of the reasonable range (e.g., the safety and civilized construction fee is automatically corrected to 2.5%). Error correction operations are recorded in the verification log for manual review reference.

[0074] Input to the cost verification and optimization module: Quantity statistics, pricing, and cost summary results output by the AI ​​intelligent calculation module.

[0075] Output of the cost verification and optimization module: verified cost results + verification warning / error correction logs + optimized pricing accuracy indicators.

[0076] 1.6 Implementation Method of Dynamic Update and Model Write-back Module (1) Dynamic update mechanism: The following data sources are monitored through a scheduled task (executed daily at 2:00 AM): ① Building material prices published on the local engineering cost information network (automatically synchronized via HTTP API or crawled using HTML parsing based on regular expressions); ② Quota specification update announcements published by the Ministry of Water Resources (monitored via RSS subscription); ③ Project design change orders (synchronized via the collaborative platform API). When a price fluctuation exceeding ±5% or a specification update is detected, the cost results are automatically recalculated and iteratively updated. The updated cost results are automatically compared with the previous version to generate a change log.

[0077] (2) Model write-back mechanism: The verified cost information, such as the comprehensive unit price, total price, allocation of provisional costs, and regulatory fees and taxes, is encapsulated as Revit shared parameters and written to the corresponding BIM components in batches using Revit API 2023 and above. The write-back parameters include, but are not limited to, the following 12 items:

[0078] The rewritten BIM model simultaneously possesses geometric information, attribute information, and cost information, forming a standard BIM5D cost data asset.

[0079] Inputs to the dynamic update and model write-back module: verified cost results + real-time price monitoring data + design change information + target BIM model file.

[0080] Output of the dynamic update and model write-back module: BIM5D model (.rvt format) carrying complete cost information + cost iteration update records. 1.7 Implementation Method of Full Life Cycle Cost Control Module The full lifecycle cost management module covers nine major cost business scenarios, and the pricing rules and accuracy requirements for each scenario are as follows: The module automatically switches the corresponding pricing rules and accuracy standards according to the current project stage, realizing full-process value control from investment estimation to final settlement.

[0081] 1.8 Implementation Method of Human-Computer Interaction Terminal and Data Storage Module Human-Computer Interaction Terminal: Developed using the Vue.js 3.x framework, this web-based interactive interface utilizes the Element Plus UI component library. Functions include: ① Knowledge Base Configuration Interface (supports manual entry, batch import, and version management of specifications, quotas, and cases; batch import supports .xlsx / .csv formats); ② Ontology Rule Editing Interface (supports graphical editing and debugging of SWRL rules, providing rule syntax checking); ③ BIM Model Import Interface (supports model file upload, stage configuration, and preprocessing progress viewing; supports breakpoint resume); ④ Cost Parameter Configuration Interface (supports visual configuration of fee standards, price data sources, and accuracy thresholds); ⑤ Results Display Interface (supports online preview, comparative analysis, and one-click export of cost results; export formats support .docx / .xlsx / .pdf); ⑥ Manual Review Interface (supports item-by-item review and confirmation / rejection of early warning projects; supports batch confirmation); ⑦ Data Export Interface (supports batch export of cost results, verification logs, and BIM 5D models; export format supports compressed files (.zip).

[0082] Data storage module: A hybrid storage architecture is adopted—① A relational database PostgreSQL 14 stores structured data of the knowledge base, cost accounting results, verification logs, and full life cycle cost records; ② A graph database Neo4j 5.x stores the entity-relationship network of the ontology model; ③ Object storage MinIO / OSS stores BIM model files (.rvt / .ifc, maximum size 2GB per file), result reports (.docx / .xlsx, maximum size 100MB per file), and operation log archive files; ④ A cache database Redis 7.x stores real-time price data and frequently queried ontology reasoning results to improve response speed (the cache expiration strategy adopts LRU, with an expiration time of 24 hours).

[0083] II. Implementation Examples Example: Intelligent cost management throughout the entire lifecycle of a medium-sized irrigation district renovation project Taking a medium-sized irrigation district renovation water conservancy project as an example, the method of the present invention will be described in detail. The project is located in a province in southern my country, and its main construction contents include: 12.5 km of canal seepage prevention lining, renovation of canal system structures (3 sluice gates, 2 aqueducts, and 1 inverted siphon), renovation of 2 pumping stations, and supporting management facilities, etc. The total investment of the project is about 280 million yuan, and the construction period is 24 months.

[0084] Step S1: Establish a dedicated water conservancy cost knowledge base for irrigation district projects Input: The Ministry of Water Resources' 2024 edition of the "Regulations for the Compilation of Design Estimates (Estimations) for Water Conservancy Projects" and related quotas, the regulations for the compilation of design estimates (estimates) for water conservancy and hydropower projects in the province, and cost data and cases of 15 similar irrigation area renovation projects in the province in the past five years.

[0085] Implementation process: (1) Import the above-mentioned standard documents into the knowledge base module in PDF format, and extract structured knowledge items through OCR (Tesseract 5.x) and natural language processing technology (based on regular expressions + named entity recognition). The extracted standard items include: irrigation area engineering category classification standards, canal lining quota items (including 24 sub-quotas for different lining thicknesses of 100mm / 150mm / 200mm / 250mm and different concrete grades C20 / C25 / C30), sluice gate engineering budget quota (16 items), pump station equipment installation quota (12 items), regional labor wage standards (four types of workers, unit price of XX yuan / man-day), regional building material price system (the local P.O42.5 cement price in July 2026 is XX yuan / t, and medium sand is XX yuan / m). 3 Crushed stone XX yuan / m 3 (HRB400 steel reinforcement, XX yuan / t).

[0086] (2) Enter the historical cost data of 15 similar irrigation district renovation projects in a unified format. Each case data includes: project overview (project type, scale, construction period), cost results at each stage (estimate / preliminary estimate / budget / settlement), and main engineering quantities (earthwork excavation volume XX million m³). 3 Concrete pouring volume: XX million m³ 3 (Including steel reinforcement quantity XX t, equipment XX sets), comprehensive unit price (channel lining XX yuan / m) 3 , Water gate XX yuan / unit, etc.), Cost composition ratio (labor cost XX%, material cost XX%, machinery cost XX%, measures cost XX%, profit XX%, tax XX%).

[0087] (3) Establish a standardized workflow for irrigation area engineering cost: ① Model import and stage configuration, ② Channel / BIM component identification and classification, ③ Earthwork / lining / reinforcement / equipment quantity extraction, ④ Channel engineering quota matching (channel earthwork excavation / backfilling / concrete lining / expansion joint / waterstop, etc., a total of 32 quota items), ⑤ Sluice gate engineering quota matching (16 items), ⑥ Pump station engineering quota matching (12 items), ⑦ Fee calculation (according to irrigation area engineering category, other direct costs 3.5%, indirect costs 4.5%, profit 7%, tax 9%), ⑧ Price fitting (using local real-time building material prices in July 2026), ⑨ Output of results, ⑩ Data write-back.

[0088] Output: A structured knowledge base containing irrigation district engineering-specific specifications, quotas, and case data (containing 156 specification items, 284 quota sub-items, and 15 historical cases), as well as an irrigation district engineering cost standardization workflow model (BPMN 2.0 format).

[0089] Step S2: Construct the BIM cost ontology model for the irrigation area project Input: The irrigation district engineering-specific knowledge base constructed in step S1 + structured cost data of 15 similar irrigation district projects.

[0090] Implementation process: (1) Entity layer definition: Define the specific entity of the irrigation area project based on the knowledge base — Project type entity: Irrigation area project (including five subcategories: canal project, sluice gate project, aqueduct project, inverted siphon project, and pumping station project).

[0091] BIM component entities include 48 categories of components such as channel earthwork excavation, channel earthwork backfilling, channel concrete lining (C20 / C25 / C30), channel expansion joints, channel waterstops, channel slope protection, sluice gate chamber concrete, sluice gate, sluice gate hoist, aqueduct body, inverted siphon, pump station plant, pump unit, and electrical equipment.

[0092] Cost resources include: regional labor quotas (XX yuan / man-day for Class IV labor), material quotas (18 main materials such as P.O42.5 cement, medium sand, crushed stone, HRB400 steel reinforcement, formwork, asphalt, etc.), and machinery quotas (excavator 1m). 3 bulldozer, concrete mixing plant 0.8m 3 The data includes hourly rates for 23 types of machinery such as tower cranes, fee standards (rates for different types of irrigation projects), and real-time building material prices (API connection to the local engineering cost information network).

[0093] Cost estimation business scenario entities: Specific parameter configurations for irrigation district projects corresponding to nine scenarios.

[0094] (2) Attribute layer definition: Define attributes for channel lining components—component type (channel lining), concrete grade (C25), lining thickness (200mm), cross-sectional shape (trapezoidal / rectangular, trapezoidal in this project), design flow rate (XXm³). 3 The system includes 28 attribute fields such as / s, length (total channel length 12.5km, divided into XX sections), volume (automatically extracted from BIM model), steel reinforcement content (automatically extracted from BIM model), and expansion joint spacing (10m / channel).

[0095] (3) Establishment of Relationship Layer: Establish the relationship between channel lining components and engineering quantity calculation rules (calculated by volume according to the dimensions shown in the design drawings, deducting individual areas >0.3m²). 2holes, a total of XX holes including expansion joints, drainage holes, inspection wells, etc. are deducted in this project); establish the matching relationship between C25 channel lining and the bill item "Concrete Lining (500107)" and the quota subitem "Channel Concrete Lining Placement (quota No. 40231)" (confidence level 1.0); establish the matching between quota subitem 40231 and base price (XX yuan / m 3 ), labor cost (XX yuan / m 3 ), material cost (XX yuan / m 3 ), machinery cost (XX yuan / m 3 ); establish the adaptation relationship between irrigation area project types and charging standards (safety and civilized construction fee 2.5%, other direct expenses 3.5%, indirect expenses 4.5%); establish the association between the construction drawing budget scenario and the accuracy requirement of ±5%.

[0096] (4) Rule layer configuration: 6 SWRL rules are compiled—— Rule R-001: If the BIM component type is "channel lining", the concrete grade is ≥C25, and the lining thickness is ≥150mm, then it matches the bill item "Concrete Lining (500107)" and the quota subitem "Channel Concrete Lining Placement (40231)", with a priority of 7.

[0097] Rule R-002: If the BIM component type is "channel lining" and the concrete grade < C25, then it matches the bill item "Concrete Lining (500107)" and the quota subitem "Channel Concrete Lining Placement (40230)" (low-grade version), with a priority of 7.

[0098] Rule R-003: If the BIM component type is "channel earthwork excavation" and the soil category is class III soil, then it matches the quota subitem "Channel Earthwork Excavation (quota No. 10012)", the slope coefficient is 0.67, with a priority of 8.

[0099] Rule R-004: If the cost scenario is "feasibility study estimate", the allowable deviation of the comprehensive unit price is ±20%; if it is "preliminary design estimate", the deviation is ±10%; if it is "construction drawing budget" or "tender control price", the deviation is ±5%, with a priority of 9.

[0100] Rule R-005: If the BIM component type is "sluice gate" and it is a flat gate, it matches the quota subitem "Flat Gate Installation (quota No. 50521)", with a priority of 6.

[0101] Rule R-006: If the BIM component type is "sluice gate" and it is a radial gate, it matches the quota subitem "Radial Gate Installation (quota No. 50522)", with a priority of 6.

[0102] (5) AI Model Training and Fine-tuning: Based on historical cost data of 15 irrigation district projects (including more than 8,500 component-quota matching annotation data), the semantic matching AI model was fine-tuned and trained. The dataset was divided into a training set (6,800 records), a validation set (850 records), and a test set (850 records) in an 8:1:1 ratio. The cross-entropy loss function and Adam optimizer (learning rate 3e-5, β1=0.9, β2=0.999, weight decay 0.01) were used. After training for 10 epochs, the semantic matching accuracy of the model on the test set reached 92.7% (Top-1 accuracy) and the Top-3 accuracy reached 97.3%. An early stopping strategy was adopted during training (training was stopped when the validation set loss did not decrease for 3 consecutive epochs).

[0103] Output: Irrigation area project BIM cost ontology model (OWL format, including 48 entity classes, 156 attribute definitions, 89 relationship definitions, and 6 SWRL rules) + fine-tuned semantic matching AI model (PyTorch2.x format, approximately 800M parameters, model file size approximately 3.2GB).

[0104] Step S3: Multi-source BIM data access and collaborative preprocessing Input: Project feasibility study stage BIM model (.rvt format, approximately 850MB), preliminary design stage BIM model (.rvt format, approximately 1.2GB), construction drawing stage BIM model (.rvt format, approximately 1.8GB).

[0105] Implementation process: (1) Import the three phase model files sequentially through the BIM data access module. A phased model management mechanism is adopted, allocating independent storage space (MinIO bucket) for each phase model and associating them through the project GUID ("IRR-2026-001").

[0106] (2) Data extraction for each stage of the model: Using Revit API 2023, all components (Instance and Symbol) in the model were traversed to extract the geometric parameters (volume, area, length, location coordinates) and attribute parameters (component type, concrete grade, reinforcement information, construction stage, floor / location). A total of 1,250 channel lining components, 186 sluice gate components, 89 pump station components, and 245 other components were extracted, totaling 1,770 components.

[0107] (3) Data cleaning and standardization: ① Unify the component names to the standard names defined by the ontology model (e.g., unify "basic channel section-concrete-C25-200mm" to "channel lining-C25-200"); ② Unify all volume units to m. 3(The original model uses ft) 3 ③ For components with missing reinforcement information (a total of 23), complete them according to the average value of similar components, mark them as "inference completion" after completion, and mark the confidence level in the attribute field; ④ Unify the coordinates of all components to the project coordinate system (with the starting point of the channel as the origin, the X-axis along the channel direction, and the Z-axis as the elevation direction).

[0108] (4) Convert the preprocessed component data into a standardized JSON format according to the attribute definition of the entity layer of the ontology model. Each component record contains 38 fields: component GUID, component name, component type (mapped to ontology entity class), geometric parameters (volume / area / length), material properties (concrete grade / reinforcement grade), construction properties (stage / location / construction method), data source (BIM model ID), preprocessing timestamp, etc.

[0109] Output: A total of 1,770 standardized BIM component data records (JSON format, total data size approximately 45MB) across three phases.

[0110] Step S4: AI Intelligent Semantic Reasoning and Digital Computation Input: Standardized BIM component data output from step S3 + ontology model constructed in step S2 + knowledge base from step S1.

[0111] Implementation process: (1) The system automatically traverses each component record and calls the semantic matching AI model to calculate the semantic similarity between the component name and the ontology entity layer. Taking the channel lining component as an example, among 1,250 channel lining components, the model output similarity score ranges from 0.89 to 0.97, with an average of 0.94 (threshold 0.85). All of them were successfully matched to the "channel lining" entity class, with a matching success rate of 100%.

[0112] (2) Based on the matching results, the rule engine automatically triggers SWRL rule R-001 (C25 concrete lining matching quota item 40231) to complete the automatic binding of components-list-quota. All 1,250 channel lining components are matched to the list item "concrete lining (500107)" and the quota item "channel concrete lining pouring (40231)", with a binding success rate of 100%.

[0113] (3) Automatic quantity calculation: Channel excavation: Based on the excavation boundary and depth parameters extracted from the BIM model (average excavation depth 2.5m, bottom width 3.0m, slope 1:0.67), combined with the soil type (Class III soil, slope coefficient 0.67), the excavation volume was calculated using the frustum formula, totaling 186,000 m³. 3(Calculation process: Channel length 12,500m, trapezoidal cross-sectional area S_upper = (3.0 + 2 × 0.67 × 2.5) × 2.5 = 15.88m²) 2 S_down = 3.0 × 2.5 = 7.50m 2 V = 12,500 × (15.88 + 7.50 + √(15.88 × 7.50)) / 3 = 186,000 m³ 3 ).

[0114] Channel concrete lining: Directly read the volume parameters from the BIM model, deducting expansion joints (one every 10m, each joint 0.02m wide, deducting approximately 3.5m of volume). 3 Drainage holes (Φ50mm, spaced 2m apart, approximately 0.2m³ of volume deducted). 3 The volume of holes (single area > 0.3m²) 2 There are a total of XX holes, minus the volume of XXm³. 3 The net volume is 31,250 m³. 3 (Design thickness 0.20m × channel inner surface area approximately 156,250m²) 2 ).

[0115] Channel reinforcement: Based on the BIM model reinforcement information (Φ12@200 double-layer bidirectional), calculated according to the theoretical weight formula, the total is 1,560t (calculation process: per m 3 The concrete contains approximately 49.92 kg / m of steel reinforcement. 3 ×31,250m 3 ).

[0116] Sluice gate concrete: Gate chamber floor slab (3 gates × 150m) 3 ), gate piers (3 x 80m) 3 ), breast walls (3 x 20m) 3 The total volume of components such as ( ) is 750m³. 3 .

[0117] Sluice gate reinforcement: based on a concrete reinforcement ratio of 120 kg / m 3 The total is calculated to be 90 tons.

[0118] Pumping station equipment: 4 water pumps, 4 motors, 6 gates (2 gates per sluice gate), 6 hoists (2 hoists per sluice gate), totaling 20 sets, measured in "units / sets".

[0119] (4) The quantities of the project are summarized by sub-projects and a detailed statistical table of the quantities of the project is generated (including four sub-projects: canal project, sluice gate project, pumping station project, and management facilities, totaling 56 sub-items).

[0120] Output: Detailed bill of quantities (including 56 sub-items, Excel format), and summary data of quantities for various components (JSON format).

[0121] Step S5: Intelligent pricing and output throughout the entire process Input: Quantity statistics results from step S4 + ontology model inference results (matched bill of quantities and fee parameters) + real-time price data (local engineering cost information in July 2026).

[0122] Implementation process: (1) Quota matching execution: Based on the component-quota binding result in step S4, read the labor, material, and machinery consumption of the corresponding quota sub-item from the quota library. Taking C25 concrete lining of the channel (quota number 40231) as an example: per 10m 3 The concrete production process consumed 3.25 tons of P.O42.5 cement and 4.80 cubic meters of medium sand. 3 7.80m of crushed stone 3 12.5 man-days of labor, 0.8m concrete mixer 3 0.85 machine hours for immersion vibrators and 0.93 machine hours for immersion vibrators.

[0123] (2) Price fitting: Read the current material prices from the real-time price data source - P.O42.5 cement 385 yuan / t, medium sand 165 yuan / m 3 Crushed stone 125 yuan / m 3 Reinforcing steel HRB400 costs 3,850 yuan / t. Replacing the budgeted price in the base price, the dynamic material cost is calculated as follows: cement 3.25 × 385 = 1,251.25 yuan, medium sand 4.80 × 165 = 792.00 yuan, and crushed stone 7.80 × 125 = 975.00 yuan, totaling 3,018.25 yuan / 10m³. 3 Labor costs are calculated based on the regional standard wage of 120 yuan / workday for Category IV workers: 12.5 × 120 = 1,500 yuan / 10m 3 The machinery cost is adjusted based on the fixed hourly rate combined with the current fuel price (diesel 7.5 yuan / L): mixer 0.85 × 78 = 66.30 yuan, vibrator 0.93 × 12 = 11.16 yuan, totaling 77.46 yuan / 10m³. 3 Subtotal of direct costs: 3,018.25 + 1,500 + 77.46 = 4,595.71 yuan / 10m 3 The calculated dynamic comprehensive unit price for C25 concrete lining of the channel is 459.57 yuan / m. 3 .

[0124] (3) Fee calculation: Fee standard according to irrigation area project category - other direct costs are calculated at 3.5% of direct costs: 4,595.71 × 3.5% = 160.85 yuan / 10m3 Indirect costs are calculated at 4.5% of direct project costs (direct costs + other direct costs): (4,595.71 + 160.85) × 4.5% = 214.05 yuan / 10m 3 The company's profit is calculated at 7% of (direct costs + indirect costs): (4,595.71 + 214.05) × 7% = 336.68 yuan / 10m 3 The safety and civilized construction fee is calculated at 2.5% of the investment in the first to fourth construction and installation projects: (4,595.71 + 160.85 + 214.05 + 336.68) × 2.5% = 132.68 yuan / 10m 3 The tax is calculated at 9% of the general VAT rate: (4,595.71 + 160.85 + 214.05 + 336.68 + 132.68) × 9% = 492.83 yuan / 10m 3 Total unit price: 4,595.71 + 160.85 + 214.05 + 336.68 + 132.68 + 492.83 = 5,932.80 yuan / 10m 3 =593.28 yuan / m 3 .

[0125] (4) Cost Summary: Calculate the total cost of each sub-item, the total cost of each section of the project, the total cost of each individual project, and the total project cost in sequence. Channel Engineering: Earthwork excavation 186,000 m³ 3 ×18.50 yuan / m 3 =3.441 million yuan, concrete lining 31,250m 3 ×593.28 yuan / m 3 =18,540,000 yuan, steel reinforcement installation 1,560t × 4,650 yuan / t = 7,254,000 yuan... The total cost of the canal project is 32,658,000 yuan. Total project cost: Canal project 32,658,000 yuan + Sluice gate project 12,845,000 yuan + Pumping station project 9,863,000 yuan + Management facilities 2,634,000 yuan + Independent costs (construction management fee, survey and design fee, supervision fee, etc.) 12,000,000 yuan + Basic contingency reserve (5% of the total of parts one to five) 8,460,000 yuan = Total project cost approximately 278.46 million yuan.

[0126] (5) Automatic generation of results: Based on the pre-set irrigation area project cost report template, the "Bill of Quantities", "Comprehensive Unit Price Analysis Table" (including 56 sub-items), "Measures Project Fee Table", "Regular Fee and Tax Table", "Total Price Summary Table" and the complete "Irrigation Area Renovation Project Cost Result Report" (Word format, about 86 pages) are automatically generated.

[0127] Output: A complete cost report for the irrigation district renovation project (Word + Excel format) + summary data of costs for each stage.

[0128] Step S6: Intelligent Verification Optimization and Autonomous Decision Execution Input: Cost results from step S5 + historical cost data for 15 similar irrigation district projects from the knowledge base.

[0129] Implementation process: (1) Verification of project quantity: The project quantity of the channel lining (31,250m) 3 ) and the historical average of 15 similar projects (28,600m) 3 Standard deviation 4,200m 3 The deviation rate was +9.3% (<10% threshold) compared to the previous data, and the verification passed.

[0130] (2) Quota matching verification: Check whether all quota items are applicable to irrigation area engineering categories. All 84 quota items corresponding to the 56 sub-projects have passed the verification.

[0131] (3) Fee calculation rule verification: Check whether the fee rate is within a reasonable range - Safety and civilized construction fee 2.5% (reasonable range 2.0%~3.0%) passed; other direct costs 3.5% (reasonable range 3.0%~4.0%) passed; indirect costs 4.5% (reasonable range 3.5%~5.5%) passed; profit 7% (reasonable range 5%~9%) passed.

[0132] (4) Price reasonableness verification: The comprehensive unit price of C25 concrete lining for the canal (593.28 yuan / m) 3 The average cost of similar projects in the past (548.00 yuan / m²) is comparable to that of similar projects in the past. 3 The deviation was +8.26% (<15% threshold), so the verification passed.

[0133] (5) Automatic error correction: During the verification process, a mismatch was found in the sub-item of the sluice gate steel reinforcement quota. The semantic matching AI model matched the sluice gate steel reinforcement to the building construction steel reinforcement quota (quota number 40201, similarity 0.82). The system automatically triggered the error correction mechanism: the semantic similarity threshold was reduced from 0.85 to 0.75 and the search was retried. It was successfully matched to the sub-item of the water conservancy project sluice gate steel reinforcement quota (quota number 50342, similarity 0.79), and the cost of this item was recalculated (90t of sluice gate steel reinforcement, unit price adjusted from 4,350 yuan / t to 4,650 yuan / t, total price adjustment +27,000 yuan).

[0134] (6) Autonomous decision-making and execution: After all verifications are passed, the system automatically confirms the cost results without manual intervention. The whole process takes about 45 minutes (the traditional manual method takes 5 to 7 working days, i.e. 40 to 56 hours).

[0135] Output: Final cost estimate that has passed verification + verification log (including 1 automatic error correction record, with error correction time of 0.3 seconds) + pricing accuracy index (the deviation rate of the comprehensive unit price of 56 sub-projects is all <8%).

[0136] Step S7: Cost Data Write-back and BIM5D Data Asset Generation Input: Final cost estimate from step S6 + BIM model from the construction drawing stage (.rvt format, 1.8GB).

[0137] Implementation process: (1) Encapsulate 12 cost information items, including the comprehensive unit price, total price, labor cost, material cost, machinery cost, allocation of measures cost, allocation of indirect cost, profit, tax, cost stage, compilation date, and data version of each component, into Revit shared parameters (.txt format parameter definition file).

[0138] (2) Using the Revit API 2023, all 1,770 components in the model were batch-traversed. The cost data corresponding to the component GUID was matched, and 12 shared parameters were written to the component properties panel. The batch writing speed was an average of 25 components per second, with a total time of about 71 seconds.

[0139] (3) After the write-back is completed, each channel lining component in the model contains not only geometric information (volume, length, cross-sectional dimensions) and construction information (concrete grade, pouring date), but also complete cost information (comprehensive unit price of 593.28 yuan / m). 3 (e.g., total cost XX yuan, labor cost XX yuan, etc.) to form a BIM5D cost model.

[0140] (4) Export the BIM5D model as .rvt format (1.9GB), store it in MinIO object storage, and generate a data asset registration card (including project name: XX irrigation area renovation project, project GUID: IRR-2026-001, compilation date: 2026-07-01, data version: V1.0, included cost stages: feasibility study estimate / preliminary design estimate / construction drawing budget, number of components: 1,770, number of cost data entries: 21,240).

[0141] Output: BIM 5D model of irrigation district renovation project with complete cost information (.rvt format, 1.9GB) + BIM 5D cost data asset registration card.

[0142] Step S8: Dynamic Management and Iterative Updates Throughout the Entire Lifecycle Input: BIM 5D model + real-time price monitoring data + design change orders during construction.

[0143] Implementation process: (1) Dynamic monitoring during the construction phase: The system automatically synchronizes with the local construction cost information network at 2:00 AM daily (via HTTP API). During the construction period (August 2026 to July 2028, a total of 24 months), the price of P.O42.5 cement experienced two fluctuations—an 8% increase in November 2026 (from RMB 385 / t to RMB 416 / t) and a 5% decrease in May 2027 (from RMB 416 / t to RMB 395 / t). After detecting an 8% increase in cement prices in November 2026 (exceeding the ±5% threshold), the system automatically triggered a price update mechanism to recalculate the comprehensive unit price of major concrete components such as channel C25 concrete lining (from RMB 593.28 / m²). 3 Adjusted to 612.50 yuan / m 3 The price increased by 3.24%, generating a price change notification and sending it to project management for confirmation. In May 2027, after a 5% price drop (to 395 yuan / ton), another update was triggered, and the overall unit price rebounded to 598.20 yuan / m². 3 .

[0144] (2) Response to Design Changes: During construction, a design change occurred—the lining thickness of the channel section from chainage 3+200 to 3+800 (600m in length) was changed from 200mm to 250mm, and the concrete grade was changed from C25 to C30. After the design change order was uploaded through the collaborative platform, the system completed the identification within 5 minutes: automatically located the components involved in the change (60 channel lining components, from chainage 3+200 to 3+800, one section every 10m), and recalculated the quantity of work (concrete volume changed from 600m × 0.20m × XXm). 2 =XXm 3 Change to 600m × 0.25m × XXm 2 =XXm 3 Increase by approximately 600m 3 The system automatically switches to the C30 lining quota item (quota number 40232, base price is about 8% higher than C30 quota) and re-fits the price (C30 concrete comprehensive unit price is about 685 yuan / m). 3 A change cost report was generated (the change increased the cost by approximately 417,000 yuan).

[0145] (3) Phased cost update: During the construction process, the system automatically generates the construction process settlement cost results on a quarterly basis (a total of 8 periods, each period is completed within 5 working days after the end of the quarter). The settlement cost of each period is compared and analyzed with the contract price. The deviation rates of the 8 periods are: +1.2%, -0.8%, +2.1%, +1.5%, -0.5%, +1.8%, +0.3%, and -0.2%, respectively. All are controlled within ±3%, which meets the accuracy requirements of the construction process settlement.

[0146] (4) Final Settlement: After the project is completed in July 2028, the system will automatically generate the final settlement cost result (total settlement price of RMB 292.48 million) based on the final completed BIM model (including all change information, a total of 1,890 components, with 120 additional changed components compared to the construction drawing stage) and the full-process cost data (including 48 months of dynamic price records, 8 price updates, and 3 design change responses). The system will also compare and analyze the costs at each stage: feasibility study estimate RMB 275 million, preliminary design estimate RMB 282 million, construction drawing budget RMB 278.46 million, and final settlement RMB 292.48 million. The deviation rate is controlled within a reasonable range. The system will automatically generate a full life cycle cost control analysis report (including cost comparison charts at each stage, price fluctuation curves, and analysis of change cost composition).

[0147] Output: Full life cycle cost dynamic update record (including 8 price update logs and 3 change response logs) + 8 construction process settlement reports (PDF format) + final settlement report + full life cycle cost control analysis report (Word format, approximately 45 pages).

[0148] Step S9: Archiving Results and Accumulating Data Assets Inputs: Full-process cost estimates, BIM 5D model, operation logs, and verification logs.

[0149] Implementation process: (1) Package and archive all cost results reports (13 reports in total, including estimate / preliminary estimate / budget / settlement / process settlement), BIM5D model files (4-stage models), engineering quantity statistics details (56 sub-items), comprehensive unit price analysis table (56 sub-items), verification log (including 1 error correction record), and operation log (including user operation records and system automatic execution records) into a unified package to generate a project cost data package (total size approximately 3.6GB, .zip format).

[0150] (2) The cost data of this project (including 21,240 cost records for 1,890 components) will be updated synchronously to the enterprise's water conservancy cost knowledge base as typical case data for irrigation district engineering. The newly added case entries include: project overview, cost results at each stage, and main technical and economic indicators (unit cost of channel lining XX yuan / m). 3 The data includes 17 fields such as unit length cost (e.g., XX yuan / m), price fluctuation impact analysis, and change cost percentage.

[0151] (3) Based on the actual cost data of this project, the BIM cost ontology model of the irrigation area project was iteratively optimized - the entity definition of C30 concrete lining was added (the original ontology model only had C20 and C25) and the relationship was added (2 matching relationships and 1 quota-price mapping relationship were added), the rule logic of the change cost of the irrigation area project was supplemented (the SWRL rule R-007 was added: if the design change leads to the increase of concrete grade, the quota item of the corresponding grade will be automatically applied and the price difference will be calculated), and the threshold of the fee parameter was adjusted from a fixed value to a dynamic range (safety and civilized construction fee 2.0%~3.0%, the specific value is automatically determined according to the project scale).

[0152] (4) Generate a standardized water conservancy cost data asset catalog for the enterprise. The data asset number for this project is “ASSET-IRR-2026-001”, which is incorporated into the enterprise's data asset management system. The data card for this project in the enterprise cost data asset catalog contains 12 metadata fields, including: asset number, project name, project type (irrigation area project), total investment of RMB 292 million, cost preparation method (AI fully automatic), preparation time of 45 minutes, quality score (comprehensive verification score of 98.2 points / 100 points), archiving date, data version V1.0, storage path, and access permissions.

[0153] Output: Project cost data package (3.6GB) + updated enterprise water conservancy cost knowledge base (1 new case, 17 fields, approximately 21,240 component-level cost records) + iteratively optimized ontology model (1 new entity, 3 new relationships, 1 new SWRL rule) + enterprise standardized cost data asset catalog (including data asset cards for this project).

[0154] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A system for constructing a BIM cost estimation AI intelligent agent for water conservancy projects based on an ontology model, characterized in that, It includes a water conservancy cost knowledge base module, an ontology model construction and reasoning module, a BIM data access and collaboration module, an AI intelligent accounting module, a cost verification and optimization module, a dynamic update and model write-back module, a full life cycle cost control module, a human-computer interaction terminal and data storage module; The water conservancy cost knowledge base module is used to collect water conservancy project cost compilation regulations, quota library, list pricing rules and fee standards based on the design characteristics of water conservancy project BIM cost software components, and to establish standardized water conservancy cost workflow description standards by combining enterprise historical cost case data, forming a structured and iterative water conservancy cost exclusive knowledge base. The ontology model construction and reasoning module is used to construct a water conservancy project BIM cost ontology model based on the water conservancy cost knowledge base and enterprise structured cost data. The model includes an entity layer, attribute layer, relationship layer, and rule layer. It is equipped with an AI dynamic reasoning engine. The dynamic reasoning engine is based on the business thresholds set in the rule layer and the mapping logic established in the relationship layer. Through business rules and dynamic reasoning mechanisms, it digitizes and models the water conservancy cost business semantics, data association relationships, and intelligent decision-making logic, realizing the integrated fusion of data governance, model construction, business rules, and action instructions, so as to autonomously drive cost business decision-making and automated operation. The BIM data access and collaboration module is used to access and parse mainstream water conservancy BIM model files. It adopts the "one model, multiple uses" and phased model management methods to realize the creation, sharing and multi-party collaborative management of BIM model information, and performs structured preprocessing on BIM component data. The AI ​​intelligent accounting module, based on the technical route of BIM, data, collaboration and AI, has a built-in semantic matching AI model, intelligent pricing engine and automatic quantity statistics model, which is used to realize BIM digital measurement, intelligent pricing and pricing and automatic generation of standardized results reports. The cost verification and optimization module is used to perform multi-dimensional compliance verification, anomaly identification and automatic error correction on the quantity calculation, pricing and fee calculation results based on the standards of the rule layer of the ontology model and the cases in the knowledge base, so as to optimize the pricing accuracy and reduce the workload of manual verification. The dynamic update and model write-back module is used to respond in real time to engineering design changes, building material price fluctuations and quota standard adjustments, and dynamically iterate the cost results; Simultaneously, the comprehensive unit price and total cost information are written back to the BIM model in the form of shared parameters, enabling BIM components to carry cost information and constructing BIM5D cost data assets. The full life cycle cost management module is used to cover nine major cost business scenarios, including planning budget, project suggestion estimate, feasibility study estimate, preliminary design estimate, construction drawing budget, bidding control price, bid price, construction process settlement and final settlement, so as to realize the full life cycle cost value management of water conservancy projects. The human-computer interaction terminal is used to realize the functions of knowledge base configuration, ontology rule editing, model import, parameter configuration, result display, manual review and data export. The data storage module is used to store knowledge base data, ontology model data, BIM model data, cost accounting results, verification logs, and full lifecycle cost data assets.

2. The system according to claim 1, characterized in that, The entity layer of the ontology model includes four core entities: engineering type entities, BIM component entities, cost resource entities, and cost business scenario entities. The engineering type entities cover all water conservancy projects, including key projects, water diversion projects, and river projects. The BIM component entities include water conservancy-specific components such as earthwork, masonry, concrete, formwork, reinforcement, seepage prevention and anchoring, tunnel excavation, and electromechanical equipment. The cost resource entities include labor quotas, material quotas, machinery quotas, fee standards, tax rate parameters, and real-time building material prices. The cost business scenario entities correspond to the nine full life cycle cost control stages.

3. The system according to claim 1, characterized in that, The relational layer of the ontology model includes component-work quantity association, component-bill quota matching relationship, quota-price mapping relationship, project type-accounting rule adaptation relationship, and business scenario-pricing parameter association relationship, which are used to realize multi-dimensional data structured two-way binding to support AI dynamic reasoning and decision-making.

4. The system according to claim 1, characterized in that, The AI ​​intelligent accounting module can autonomously execute a full set of repetitive water conservancy cost calculation tasks, replacing manual quantity calculation, pricing, summarization, and report generation, thereby achieving accurate and rapid quantity calculation and improving the accuracy of matching quotas and fees.

5. The system according to claim 1, characterized in that, The full lifecycle cost control module drives cost management to upgrade from the traditional price determination and control model to a proactive value management model that covers the entire project lifecycle.

6. A method for implementing a BIM cost estimation AI intelligent agent for water conservancy projects based on an ontology model, characterized in that, The system applied to any one of claims 1 to 5 includes the following steps: S1. Establish a standardized water conservancy cost knowledge base and workflow: Combining the design features of water conservancy engineering BIM cost software components, collect industry cost compilation regulations, quota library, list and fee specifications, integrate enterprise historical cost data and cases, establish a unified water conservancy cost workflow description standard, and form a structured and iterative exclusive knowledge base. S2. Construct a dynamic reasoning-enabled water conservancy cost ontology model: Based on the knowledge base and enterprise structured cost data, build a four-layer ontology architecture including entity layer, attribute layer, relationship layer and rule layer, configure refined cost business rules, and load an AI dynamic reasoning engine to digitize and model business semantics, data relationships and decision logic, forming an operating system that autonomously drives cost business. S3. Multi-source BIM data access and collaborative preprocessing: Import multi-stage target project BIM models, adopt the "one model for multiple uses" and phased model management mode, complete the extraction, cleaning and standardization of BIM component data, and realize multi-party data collaborative sharing; S4. AI Intelligent Semantic Reasoning and Digital Quantity Calculation: Relying on the dynamic reasoning capability of the ontology model, it intelligently binds BIM components and cost parameters, and automatically and accurately calculates the quantities of various components. S5. Intelligent pricing and output throughout the entire process: Based on the rules of the ontology model and the standards of the knowledge base, the system automatically completes quota matching, fee calculation and price fitting to achieve intelligent pricing and output, and automatically generates a standardized cost report. S6. Intelligent verification, optimization and autonomous decision-making execution: Verify cost results from multiple dimensions, automatically correct abnormal data, and let the AI ​​model autonomously execute the entire set of cost business to replace manual repetitive operations. S7. Cost data write-back and BIM5D data asset generation: Write the verified cost information into the BIM model so that the model components carry complete cost attributes, forming a traceable and transferable BIM5D cost data asset. S8. Full lifecycle dynamic management and iterative updates: For nine major cost business scenarios, it realizes dynamic cost accounting, change adjustment and value analysis of the entire project process, and responds to industry standards and market price updates in real time. S9. Results Archiving and Data Asset Accumulation: Complete the review and export of cost estimates and the archiving of data throughout the entire process, iteratively optimize the knowledge base and the ontology model, and accumulate standardized cost estimates data assets for the enterprise.

7. The method according to claim 6, characterized in that, Step S2, the ontology model construction, specifically includes: collecting official cost standards and high-quality cost data from enterprises in the water conservancy industry; defining standardized terms, attribute parameters, and business thresholds for each entity; establishing multi-dimensional bidirectional correlation mapping relationships to adapt to the accounting rules for different project types and different cost scenarios; and integrating an AI dynamic reasoning mechanism to achieve automatic rule determination, automatic parameter matching, and automatic business decision-making without human intervention, forming an intelligent decision-making closed loop.

8. The method according to claim 6, characterized in that, The nine full life-cycle cost scenarios mentioned in step S8 include planning budget, project proposal estimate, feasibility study estimate, preliminary design estimate, construction drawing budget, bidding control price, bid price, construction process settlement and final settlement, covering the entire process of water conservancy project construction.

9. The method according to claim 6, characterized in that, The cost information mentioned in step S7 includes the comprehensive unit price, total price, and cost details, which are written back into the Revit BIM model in the form of shared parameters, so that the BIM components carry full-dimensional cost information.

10. The method according to claim 6, characterized in that, The "one model for multiple uses" and phased model management mode described in step S3 support the separate management and collaborative sharing of models in the project feasibility study, preliminary design, construction drawings and construction phases, ensuring the continuous availability of model data throughout the entire life cycle.

Images