Water conservancy project list automatic quota, intelligent conversion and set price method

Abstract

The application discloses a kind of water conservancy project list automatic quota, intelligent conversion and group price method, belong to water conservancy engineering intelligent management technical field.The method includes: constructing water conservancy engineering quota knowledge graph and conversion rule base;Intelligent analysis is carried out to bill of quantities, and structured feature is extracted;Quota subgoal matching is carried out using multi-strategy fusion algorithm, and multiple subgoal combination is identified by graph covering algorithm;Conversion demand is automatically identified and conversion coefficient is calculated, and multiple rule superposition and conflict are handled according to priority;Real-time price is obtained from multiple sources, and list comprehensive unit price is calculated;Intelligent verification and optimization are carried out to group price result, and are stored to historical scheme library.The application realizes the automatic quota of water conservancy project list, intelligent conversion and dynamic group price, significantly improves the efficiency and accuracy of cost preparation, solves the problems of low efficiency, easy to make mistakes, multiple subgoal combination and difficulty in application of traditional manual operation, and has good industrial practicability.

Description

Technical Field

[0001] This invention belongs to the field of intelligent management technology for water conservancy projects, and in particular relates to a method for automatic quota application, intelligent conversion and pricing of water conservancy project bills. Background Technology

[0002] Cost management for water conservancy projects is an important component of water conservancy project construction, directly affecting the control of project investment and the rational use of construction funds. Water conservancy projects are characterized by long construction periods, large investment scales, diverse project types, and complex construction conditions, making cost preparation a labor-intensive, highly specialized, and technically demanding task.

[0003] In the traditional process of compiling water conservancy project costs, cost estimators need to manually consult the quota database for each item in the bill of quantities, select the appropriate quota item for application, and perform quota conversion and coefficient adjustments based on project characteristics and construction conditions to finally complete the pricing. However, this traditional method has the following prominent problems: (i) Low efficiency of manual operation. Water conservancy and hydropower projects are characterized by complex design, construction spanning multiple disciplines, high construction difficulty, and numerous components. The cost list is far larger than that of traditional building construction, municipal engineering, and highway engineering, which brings a huge workload to cost preparation. When compiling cost documents for water conservancy and hydropower projects, cost estimators manually compile cost lists item by item and manually assemble prices, which has the disadvantages of being time-consuming, having complex operation steps, and being inefficient.

[0004] (II) Low accuracy in quota application. Hydropower industry quotas are consumption-based quotas, only specifying the unit consumption of labor, materials, and machinery. In pricing, it is necessary to manually select the appropriate quota item from the quota library based on the project characteristics and construction plan of the corresponding bill of quantities, and then place it under the corresponding bill of quantities to achieve quota item application and pricing. Moreover, quota application in the hydropower industry is quite complex, with multiple quota items often applied to a single bill of quantities. Traditional quota application methods rely on manual selection of quota items, which is inefficient. Furthermore, some quota items have conversion factors, making it easy to miss items during manual operation, thus affecting the accuracy of pricing.

[0005] (iii) The pricing results vary from person to person and have poor consistency. For the same quota item, due to differences in the work experience and ability level of the cost estimators, there will be certain differences in the selection of quotas, which will inevitably affect the accuracy and rationality of the project investment preparation.

[0006] (iv) The conversion of quota items is complex. In water conservancy projects, the conversion of quota items involves various types, such as transportation distance conversion, thickness conversion, depth conversion, height conversion, strength conversion, and gradation conversion. Each type of conversion has its own conversion coefficient calculation formula. Cost engineers need to be proficient in various conversion rules and calculate them one by one, which is very easy to make mistakes.

[0007] (v) The problem of inconsistency between the units in the bill of quantities and the quota is prominent. When priced according to the quota items, inconsistencies between the units in the bill of quantities and the units in the quota items are frequently encountered, requiring unit consumption conversion. Currently, there is a lack of automated unit conversion mechanisms, and the process relies entirely on manual judgment and handling.

[0008] (vi) Lack of intelligent pricing schemes tailored to the characteristics of water conservancy projects. Existing technologies include some automatic cost estimation methods, such as using natural language processing to analyze the semantic similarity of keywords in the quota database to obtain similar terms, or using semantic analysis to analyze the bill of quantities to form structured data. However, these methods generally have the following shortcomings: First, they fail to design a complete conversion rule library and automatic conversion mechanism specifically for the special needs of quota conversion in water conservancy projects; second, they lack effective solutions for the problem of combining multiple quota items; third, they fail to automatically incorporate market price information of the project's location into the pricing process; and fourth, they lack mechanisms for verifying and intelligently optimizing the pricing results. Summary of the Invention

[0009] To address the aforementioned issues, this invention proposes an automatic quota application, intelligent conversion, and pricing method for water conservancy project cost estimates. This method aims to solve problems such as low efficiency of manual operation, poor accuracy of quota application, complex conversion processing, and inconsistent pricing results in the existing water conservancy project cost estimate compilation process, thereby achieving intelligent and automated compilation of water conservancy project cost estimates.

[0010] To achieve the above objectives, the technical solution adopted by this invention is: a method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities, comprising the following steps: Step S1: Construct a knowledge graph of water conservancy engineering quotas and a historical pricing scheme database. The quota knowledge graph stores quota sub-item nodes and the relationship edges between quota sub-item nodes in a graph database format. Step S2: Obtain the bill of quantities for the water conservancy project to be priced, and use natural language processing technology to intelligently parse each item in the bill of quantities, extracting core engineering type terms, construction method terms and technical parameter terms from the project name to form structured data to be priced. Step S3: Based on the quota knowledge graph constructed in Step S1 and the list item features parsed in Step S2, a multi-strategy fusion algorithm is used for intelligent matching of quota sub-items. The multi-strategy fusion algorithm includes semantic similarity matching, knowledge graph reasoning matching, and historical case matching. The comprehensive matching score of candidate quota sub-items or sub-item combinations is calculated by weighted summation or machine learning ranking model, and recommended in descending order of comprehensive matching score. For list items that require multiple quota sub-item combinations, the main quota sub-item and auxiliary quota sub-item are identified by graph covering algorithm, and the minimum quota sub-item set covering all process units is found. Step S4: After selecting a quota item, based on the quota conversion rule library built in step S1, automatically identify the differences between the list item parameters and the standard parameters of the quota item, match the applicable conversion rules and calculate the conversion coefficient. When there are multiple conversion rules, apply the conversion coefficients in sequence according to the preset priority order and handle rule conflicts according to the priority adjudication mechanism. Step S5: After completing the application and conversion of quota items, obtain the current market prices of labor, materials and machinery from multiple price information sources, and perform intelligent price matching through a three-level matching strategy of precise code matching, name semantic matching and specification parameter matching. Calculate the comprehensive unit price of the list items based on the consumption of labor, materials and machinery, conversion coefficients and unit prices of the quota items. Step S6: Perform intelligent verification and optimization on the pricing results generated in Step S5, including verification of the rationality of quota matching, the rationality of conversion coefficients, and the rationality of unit price. When the verification finds that something is unreasonable, generate optimization suggestions and store the final confirmed pricing results in the historical pricing scheme library.

[0011] Furthermore, the attributes of the quota sub-item node include quota number, quota name, quota professional category, quota application scope, labor consumption, material consumption, machinery consumption, quota unit, quota base price, and applicable geographical range; the relationship edges include parent-child relationship, substitution relationship, matching relationship, and constraint relationship; at the same time, a quota conversion rule library is constructed, and each conversion rule in the conversion rule library includes triggering conditions, conversion type, conversion coefficient calculation formula, constraint conditions, and priority used to determine the execution order when multiple conversion rules conflict.

[0012] Furthermore, the conversion rule base mentioned in step S1 includes at least one of the following eight conversion types: distance conversion, thickness conversion, depth conversion, height conversion, strength conversion, gradation conversion, coefficient adjustment conversion, and affirmative conversion; the priority order is: mandatory conversion, conditional conversion, selective conversion, and affirmative conversion.

[0013] Furthermore, the natural language processing technology described in step S2 employs a pre-trained language model based on the Transformer architecture and performs fine-tuning training on a corpus of water conservancy engineering cost; the intelligent parsing also includes: extracting list item parameters using a method combining pattern matching and sequence labeling models, and performing synonym normalization processing on the list item names.

[0014] Furthermore, step S2 also includes calculating the similarity between the items in the list to be processed and the items in the historical pricing scheme library using a hybrid similarity calculation method. The hybrid similarity calculation method includes a weighted sum of the semantic similarity of the item name, the numerical similarity of the item parameters, and the classification similarity of the item type.

[0015] Furthermore, the multi-strategy fusion algorithm in step S3 adopts the random forest algorithm model to fit the list of items to be priced with the list of pricing schemes, and then reassembles the list of items to be priced after forming a matching result; the graph covering algorithm adopts dynamic programming or branch and bound algorithm to find the optimal quota item combination scheme, and the optimization objective is to minimize the total combined price or maximize the comprehensive matching degree.

[0016] Furthermore, the calculation of the conversion factor in step S4 uses the following formula for linear conversion: Conversion factor = base factor + (actual parameter value - standard parameter value) / step size × step size factor; For nonlinear conversion, interpolation calculations are performed using a pre-built parameter-coefficient mapping table; The method further includes: establishing a two-way mapping mode between list item parameters and quota sub-item parameters, so as to complete the conversion in the forward direction and verify the correctness of the conversion in the reverse direction.

[0017] Furthermore, the price information sources mentioned in step S5 include at least two of the following: government-published information prices, market prices from professional platforms, historical procurement prices of enterprises, real-time market inquiry data, and user-defined prices; the comprehensive unit price is calculated according to the following formula: P = Σ(C i ×K i ×P i )+ M; Where P is the comprehensive unit price of the list item, and C i K represents the consumption of personnel, materials, and machinery for the i-th quota item. i P is the conversion factor for the i-th quota item. i Let M be the unit price of the i-th type of resource, M be the fee amount, and n be the number of quota items.

[0018] Furthermore, the unit price rationality verification in step S6 includes: comparing the calculated unit price of the list item with the unit price of similar items in history, calculating the deviation rate, triggering an alarm and displaying comparison reference data when the deviation rate exceeds a preset threshold, the preset threshold can be automatically set by the system based on historical data statistical analysis (e.g., 30%), or it can be manually configured by the user according to the importance level of the item; the optimization suggestions include recommended alternative quota items, adjusted conversion coefficients, and suggested unit price correction values.

[0019] Furthermore, the method also includes: an AI agent based on a large language model, in which the user interacts with the AI ​​agent in natural language, and the AI ​​agent autonomously calls the functional modules of steps S1 to S6 to perform the cost preparation task and feeds back the execution results to the user in natural language.

[0020] The beneficial effects of adopting this technical solution are: (I) Significantly Improves the Efficiency of Water Conservancy Project Cost Estimation. This invention liberates cost estimators from tedious manual operations through automatic quota application, intelligent conversion, and dynamic pricing. In traditional manual pricing methods, cost estimators need to review quotas item by item, manually calculate conversion factors, and manually enter price data. The pricing work for a medium-sized water conservancy project often takes several weeks. After adopting this invention, the pricing time can be shortened by more than 80%, significantly improving work efficiency.

[0021] (II) Significantly improves the accuracy and consistency of quota application. This invention avoids inconsistencies in quota selection caused by differences in the experience of cost estimators through a multi-strategy fusion quota matching algorithm and knowledge graph reasoning. The system recommends quotas according to unified matching rules and algorithms, ensuring that the same or similar list items obtain consistent quota application results, thereby improving the standardization level of cost documents.

[0022] (III) Achieving Automation and Accuracy in Quota Conversion. This invention constructs a complete quota conversion rule library, which can automatically identify conversion needs, match conversion rules, calculate conversion coefficients, and correctly handle the superposition and conflict of multiple rules, freeing cost engineers from complex conversion calculations and avoiding calculation errors and omissions in manual conversion. The unique bidirectional mapping mode of quota parameters further ensures the correctness and traceability of conversion results.

[0023] (iv) Achieving dynamic and intelligent pricing results. This invention ensures the timeliness and accuracy of price information used in pricing by dynamically acquiring and intelligently matching multi-source price information. At the same time, through the intelligent verification and optimization mechanism of pricing results, it can automatically identify unreasonable pricing results and provide optimization suggestions, forming a closed-loop intelligent processing flow of "matching-conversion-pricing-verification-optimization".

[0024] (v) Possesses continuous learning and self-optimization capabilities. This invention, by constructing a historical pricing scheme database and recording user confirmation and modification behaviors, enables the system to continuously learn from the professional experience and preferences of cost estimators, constantly optimize the matching model and recommendation effect, and become more accurate with use.

[0025] (vi) Provides the ability to compare multiple options. This invention can not only recommend the optimal quota scheme, but also provide multiple candidate schemes and their comparison reports, giving cost engineers greater decision-making flexibility and helping to optimize project costs while meeting technical specifications.

[0026] (vii) It innovatively solves the problem of combining multiple quota items in water conservancy projects. There are many situations in water conservancy projects where multiple quota items need to be combined to fully cover the work content of the list. This invention can automatically identify the main quota and auxiliary quota through process decomposition and graph coverage algorithm, forming a complete quota item combination scheme, filling the gap of existing technology in complex application scenarios of water conservancy projects. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the automatic quota application, intelligent conversion and pricing method for water conservancy project bill of quantities according to the present invention. Figure 2 This is a schematic diagram of the quota knowledge graph structure in an embodiment of the present invention; Figure 3 This is a flowchart of the multi-strategy fusion quota matching algorithm in an embodiment of the present invention; Figure 4 This is a flowchart of the intelligent quota conversion rule processing in an embodiment of the present invention; Figure 5 This is a flowchart of the intelligent verification process for pricing results in an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described below with reference to the accompanying drawings.

[0029] In Example 1, see Figure 1 As shown, this invention proposes an automatic quota application, intelligent conversion, and pricing method for water conservancy project bills of quantities, including the following steps: Step S1: Construct a knowledge graph of water conservancy engineering quotas and a historical pricing scheme database. The quota knowledge graph stores quota sub-item nodes and the relationship edges between quota sub-item nodes in a graph database format. Step S2: Obtain the bill of quantities for the water conservancy project to be priced, and use natural language processing technology to intelligently parse each item in the bill of quantities, extracting core engineering type terms, construction method terms and technical parameter terms from the project name to form structured data to be priced. Step S3: Based on the quota knowledge graph constructed in Step S1 and the list item features parsed in Step S2, a multi-strategy fusion algorithm is used for intelligent matching of quota sub-items. The multi-strategy fusion algorithm includes semantic similarity matching, knowledge graph reasoning matching, and historical case matching. The comprehensive matching score of candidate quota sub-items or sub-item combinations is calculated by weighted summation or machine learning ranking model, and recommended in descending order of comprehensive matching score. For list items that require multiple quota sub-item combinations, the main quota sub-item and auxiliary quota sub-item are identified by graph covering algorithm, and the minimum quota sub-item set covering all process units is found. Step S4: After selecting a quota item, based on the quota conversion rule library built in step S1, automatically identify the differences between the list item parameters and the standard parameters of the quota item, match the applicable conversion rules and calculate the conversion coefficient. When there are multiple conversion rules, apply the conversion coefficients in sequence according to the preset priority order and handle rule conflicts according to the priority adjudication mechanism. Step S5: After completing the application and conversion of quota items, obtain the current market prices of labor, materials and machinery from multiple price information sources, and perform intelligent price matching through a three-level matching strategy of precise code matching, name semantic matching and specification parameter matching. Calculate the comprehensive unit price of the list items based on the consumption of labor, materials and machinery, conversion coefficients and unit prices of the quota items. Step S6: Perform intelligent verification and optimization on the pricing results generated in Step S5, including verification of the rationality of quota matching, the rationality of conversion coefficients, and the rationality of unit price. When the verification finds that something is unreasonable, generate optimization suggestions and store the final confirmed pricing results in the historical pricing scheme library.

[0030] Example 2: Complete Implementation Process Taking a Large-Scale Water Conservancy Project as an Example This embodiment uses the bill of quantities of a large-scale water conservancy project (including a concrete gravity dam, spillway, water diversion tunnel, metal structure, and electromechanical installation) as an example to describe in detail the implementation steps of the present invention. The bill of quantities contains 1560 items, covering five major categories: earthwork, concrete works, grouting works, metal structure equipment and installation works, and electromechanical equipment and installation works. The following examples use three representative items: earthwork excavation (single process), tunnel concrete lining (multi-process combination), and curtain grouting (requires special conversion).

[0031] The specific implementation of step S1: Construct a knowledge graph of water conservancy project quotas and a historical pricing scheme database.

[0032] (1) Quota data collection and cleaning: All quota items, totaling 16,820, were extracted from the "Budget Quota for Water Conservancy Construction Engineering," the "Budget Quota for Water Conservancy and Hydropower Equipment Installation Engineering," and local supplementary quotas issued by a provincial water resources department. Each quota item includes fields such as quota number, quota name, work content, unit of measurement, labor consumption, material consumption, machinery consumption, other material costs, other machinery costs, quota base price, and applicable geographical scope. The data was cleaned by removing duplicate entries, filling in missing applicable condition fields, and standardizing the unit of measurement (e.g., unifying "100m³" to "m³" as the basic unit).

[0033] (2) Knowledge graph construction: A graph database (such as Neo4j) is used to store the quota sub-item nodes and relation edges. Each quota sub-item is a node, and the node attributes include: quota number (such as "YD-1003"), quota name (such as "Earthwork Excavation - Class III Soil - Open Excavation"), quota professional category (such as "Earthwork Engineering"), quota application scope (such as "General Earthwork Open Excavation"), labor consumption (0.5 man-days / m³), material consumption (0), machinery consumption (0.02 shifts / m³), quota unit ("m³"), quota base price (85 yuan / m³), and applicable geographical scope ("Nationwide").

[0034] Establish four types of relation edges: Parent-child relationship: such as "Earthwork Engineering" node → "Earth Excavation" node → "YD-1003" node, used for hierarchical retrieval.

[0035] Substitution relationship: such as “YD-1003 (old)” node → “YD-1003A (new)” node, which indicates the substitution of old and new quotas.

[0036] Matching relationships: such as "concrete pouring" node → "formwork installation and removal" node, indicating that they are often used in combination.

[0037] Constraint relationships: such as “In-tunnel construction” node → “In-tunnel lighting and ventilation” node, and marked “must be used simultaneously”.

[0038] Use Cypher statements to import in batches and build an index to improve retrieval speed.

[0039] like Figure 2 As shown, solid lines represent parent-child relationships, green dashed lines represent matching relationships, red dashed lines represent constraint relationships, and yellow solid lines represent substitution relationships.

[0040] (3) Construct a quota conversion rule base: A rule base was constructed containing 312 specific conversion rules across 10 conversion types. Each conversion type includes distance conversion, thickness conversion, depth conversion, height conversion, intensity conversion, gradation conversion, coefficient adjustment conversion, and affirmative conversion. Each rule records the triggering conditions, the conversion coefficient calculation formula, constraints, and the priority used to determine the execution order when multiple conversion rules conflict. The default priority order is: mandatory conversion (e.g., tunnel coefficient) > conditional conversion (e.g., depth, thickness) > selective conversion (e.g., distance, intensity) > affirmative conversion (e.g., expansion coefficient). Example rules are shown in Table 1. Table 1

[0041] (4) Generation of quota feature vector: Each quota item name and project feature is encoded using a fine-tuned pre-trained language model (such as a BERT model based on the Transformer architecture), generating a 768-dimensional feature vector, which is then stored in the Milvus vector database.

[0042] (5) Historical pricing scheme database: We collected pricing data for 2,000 bill of quantities items from completed water conservancy projects. Each data item includes: bill of quantities item ID, bill of quantities name, project characteristics, selected quota sub-item combination (JSON format), conversion factor record, final unit price, actual settlement unit price, project location, and compilation date.

[0043] The specific implementation of step S2: intelligent parsing and standardization of the bill of quantities.

[0044] (1) List input and preliminary analysis: Users upload a bill of quantities in Excel format via a web interface. An example of one row of data is shown in Table 2: Table 2

[0045] The system performs intelligent parsing of the inventory list, using the BERT model to segment and identify named entities in the "project name," extracting the core engineering type term "earthwork excavation," while the construction method term is left blank. The "project features" are then parsed, using a combination of pattern matching and sequence labeling (BiLSTM-CRF) to extract parameters: soil type = "Class III soil," construction method = "open excavation," haulage distance = 2.5 (km), and depth = 6 (m). The system also extracts the unit of measurement "m³" and the quantity "5000."

[0046] (2) Synonym normalization processing: For synonymous expressions in the list name, such as "earthwork excavation" and "earthwork excavation and transportation", they are merged into the standard name "earthwork excavation" through semantic similarity calculation (threshold > 0.9) and assigned the standard code "500101001" in accordance with the specification for pricing of water conservancy engineering quantity list.

[0047] (3) Calculation of mixed similarity: To match subsequent historical cases, calculate the overall similarity between the current list and each item in the historical pricing scheme library: Semantic similarity: The names and features of the two items are concatenated and input into BERT. The cosine similarity is calculated from the [CLS] vector and denoted as Sim_sem.

[0048] Numerical similarity: Euclidean distance is calculated after normalizing numerical parameters such as transport distance, depth, and amount of work: Sim_num = 1 / (1+distance).

[0049] Classification similarity: The Jaccard coefficient is calculated based on the engineering type classification tree (Level 1: earthwork / concrete / grouting, etc.; Level 2: excavation / filling / pouring, etc.).

[0050] Overall score = 0.5 × Sim_sem + 0.3 × Sim_num + 0.2 × Sim_class.

[0051] The similarity score was calculated to be 0.93 with the historical project "Earthwork Excavation of a Reservoir Dam", which will be used as a reference for subsequent matching.

[0052] Specific implementation of step S3: Intelligent matching of quota sub-items and recommendation of multi-sub-item combinations, such as... Figure 3 As shown.

[0053] (1) Semantic similarity matching: The standardized name of the bill of quantities, "Earthwork Excavation - Class III Soil - Open Cut," was semantically matched with all quota sub-item names in the quota knowledge graph, and the top 10 candidates with the highest similarity were selected. Results: YD-1003 (similarity 0.98), YD-1002 (0.85), YD-1004 (0.83), etc.

[0054] (2) Knowledge graph reasoning and matching: In Neo4j, a Cypher query is executed to find applicable sub-items along parent-child relationships, auxiliary sub-items along matching relationships, and inapplicable sub-items along constraint relationships. The query returns the main quota YD-1003, with no matching quota (earthwork excavation is a single process), and the inference path confidence is 0.95.

[0055] (3) Historical case matching: The item with the highest similarity (0.93) was retrieved from the historical pricing scheme database. It used YD-1003 and applied distance and depth conversion.

[0056] (4) Multi-strategy fusion sorting: The weights are set as follows: α = 0.4 (semantic), β = 0.4 (knowledge graph), and γ = 0.2 (history). Candidate YD-1003's overall score is 0.4 × 0.98 + 0.4 × 0.95 + 0.2 × 0.93 = 0.954, ranking first. The system recommends YD-1003 and displays the matching criteria.

[0057] (5) Multi-sub-item combination recognition (example of complex project) For the "Tunnel Concrete Lining" bill of quantities (C25 concrete, 40cm thickness, in-tunnel construction, including rebar fabrication and installation), the system decomposes the work content into process units: concrete mixing, concrete transportation (in-tunnel), concrete pouring (lining, 40cm thickness), formwork installation and dismantling, and rebar fabrication and installation. In the knowledge graph, a graph covering algorithm (based on branch and bound, with the optimization objective being the minimum combined total price) is used to find the minimum set of quota items. The algorithm outputs the optimal combination: {"Concrete Mixing - C25", "Concrete Transportation - 200m in-tunnel distance", "Tunnel Concrete Lining Pouring - 40cm Thickness", "Tunnel Formwork Installation and Dismantling", "Rebar Fabrication and Installation"}, a total of 5 quota items, with "Tunnel Concrete Lining Pouring" being the primary item and the others being auxiliary.

[0058] Specific implementation of step S4: intelligent conversion of quotas and automatic calculation of coefficients, such as... Figure 4 As shown.

[0059] Taking earthwork excavation YD-1003 as an example, the standard parameters are: haul distance 1km, depth 4m. The actual bill of quantities parameters are: haul distance 2.5km, depth 6m.

[0060] (1) Automatic identification of conversion requirements: The comparison module identified that: the transport distance exceeded 1.5km, triggering a transport distance conversion; and the depth exceeded 2m, triggering a depth conversion.

[0061] (2) Conversion rule matching: Matched R001 (distance conversion) and R002 (depth conversion) from the rule base.

[0062] (3) Calculation of conversion factor: Linear conversion formula: The distance conversion factor K1 = 1 + (2.5-1) / 0.5 × 0.02 = 1.06; Depth conversion factor K2 = 1 + (6-4) / 1 × 0.01 = 1.02.

[0063] (4) Overlay of multiple rules and conflict handling: Priority order: Mandatory conversion (none) → Conditional conversion (depth priority 2, distance priority 3). If there is no conflict, the conversion is applied according to priority, with a total conversion factor K = K1 × K2 = 1.0812. If a conflict exists (e.g., a rule prohibits the simultaneous use of distance and depth conversion), the system retains the rule with higher priority based on the priority adjudication mechanism and generates a conflict log.

[0064] (5) Verification of conversion results: The preset reasonable range is: individual coefficient 0.85~1.35, total coefficient 0.7~1.5. K=1.0812 is within the range, and the verification is passed.

[0065] (6) Two-way mapping verification: The forward mapping from the list parameters to the quota parameters yields K1=1.06; the reverse mapping: from K1=1.06, the equivalent transport distance is calculated as 1 + (1.06-1) / 0.02×0.5 = 2.5km, which is consistent with the list, verifying the correct conversion.

[0066] The specific implementation of step S5: price acquisition and dynamic pricing.

[0067] (1) Acquisition of multi-source price information: The system is configured with three price sources: the information price (API) from the provincial construction engineering cost management station, the professionally determined price from Guangcai.com (web crawler + API), and the enterprise's internal historical procurement price database. For the labor unit price, the government information price of 150 yuan / man-day is used; for the machinery shift (excavator 1m³), the information price is 950 yuan, the Guangcai.com price is 980 yuan, and the historical price is 920 yuan. The weighted average (weights 0.5, 0.3, 0.2) yields 962 yuan / shift.

[0068] (2) Intelligent price matching: A three-level matching strategy is adopted: precise code matching (e.g., material code "GB01" directly matches the price of steel bars); if no match is found, semantic name matching is performed (e.g., "threaded steel bar HRB400" matches "steel bar" if the similarity is >0.9); otherwise, specification parameter matching is performed (e.g., diameter, strength grade, etc.). The user is prompted to manually input the required information when a match fails.

[0069] (3) Dynamic price updates: The system is set to perform a scheduled task (every Monday at 3:00 AM) to synchronize the latest data from various price sources, and also supports proactive push (updated via API callback when government information prices are released).

[0070] (4) Calculation of comprehensive unit price: According to the formula P = Σ(C i × K i × P i + M calculation: Labor cost: 0.5 man-days / m³ × 150 yuan / man-day = 75 yuan / m³.

[0071] Machinery cost: 0.02 machine shifts / m³ × 962 yuan / machine shift = 19.24 yuan / m³.

[0072] Subtotal = 94.24 yuan / m³, multiplied by conversion factor 1.0812, we get 101.92 yuan / m³.

[0073] Fees M: Management fee 8% × 94.24 = 7.54 yuan, profit 5% × (94.24 + 7.54) = 5.09 yuan, tax 9% × (94.24 + 7.54 + 5.09) = 9.62 yuan, total M: 22.25 yuan.

[0074] The comprehensive unit price P = 101.92 + 22.25 = 124.17 yuan / m³.

[0075] Total price of the list = 124.17 × 5000 = 620,850 yuan.

[0076] (5) Comparison of multiple options The system also provides other candidate solutions (such as using YD-1002 and adjusting the conversion), and generates a comparison table for cost estimators to choose the best option.

[0077] Specific implementation of step S6: Intelligent verification and optimization of pricing results, such as... Figure 5 As shown.

[0078] (1) Verification of the reasonableness of quota matching: The YD-1003 application conditions are "open excavation, soil type I-IV", and the soil type III in the inventory list is met; the constraint relationships are checked and there are no conflicts. Verification passed.

[0079] (2) Verification of the rationality of conversion factors: The total conversion factor of 1.0812 is within the preset range (0.85-1.35), and the individual factor is also compliant, so the verification is successful.

[0080] (3) Verification of the reasonableness of the unit price: A total of 45 projects were retrieved from the historical pricing scheme database, specifying "earthwork excavation - Class III soil - open excavation," with a haul distance of 2.0~3.0km and a depth of 5~7m. The average unit price was 118 yuan / m³, and the standard deviation was 12 yuan / m³. The unit price of this scheme is 124.17 yuan, and the deviation rate is (124.17-118) / 118=5.2%, which is less than the preset threshold of 30%, so the verification passed. The system outputs a verification report.

[0081] (4) Intelligent optimization suggestions: If the deviation exceeds 20%, the system will automatically analyze the cause and provide optimization suggestions, such as: "The current unit price of machine shifts is too high. It is recommended to adopt the historical average price of 920 yuan, which can reduce the unit price to 120.5 yuan." (5) Storage and feedback of pricing results: The final confirmed pricing results (list information, quota YD-1003, conversion factor 1.0812, price source, unit price 124.17 yuan, confirmation time, etc.) are stored in the historical pricing scheme library for subsequent model training.

[0082] Example 3: Natural language interaction based on AI agents.

[0083] This embodiment provides an AI agent driven by a large language model. The user enters through a web interface dialog box: "Help me calculate the cost of earthwork excavation for the XX Reservoir dam, with a volume of 5000 cubic meters, Class III soil, and a transportation distance of 2.5 kilometers." An AI agent based on a large language model (such as ChatGLM3-6B, fine-tuned on water conservancy cost dialogue data) is used to automatically perform the following operations: Call the list parser to generate standardized list items; For quota matching, YD-1003 is recommended. Call the conversion function to calculate the distance conversion factor. Call the price list to obtain the current unit price for labor and machinery; Calculate the comprehensive unit price using the group price model; Returning to natural language: "The cost estimate has been completed. The comprehensive unit price for earthwork excavation is 124.17 yuan / m³, with a total price of 620,850 yuan. Detailed pricing breakdown is as follows: ... Is any adjustment required?" The user continued the conversation: "Change the delivery distance to 3 kilometers." The AI ​​agent recalculated and responded. The entire process required no user interaction with the graphical interface.

[0084] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities, characterized in that, Including the following steps: Step S1: Construct a knowledge graph of water conservancy engineering quotas and a historical pricing scheme database. The quota knowledge graph stores quota sub-item nodes and the relationship edges between quota sub-item nodes in a graph database format. Step S2: Obtain the bill of quantities for the water conservancy project to be priced, and use natural language processing technology to intelligently parse each item in the bill of quantities, extracting core engineering type terms, construction method terms and technical parameter terms from the project name to form structured data to be priced. Step S3: Based on the quota knowledge graph constructed in Step S1 and the list item features parsed in Step S2, a multi-strategy fusion algorithm is used for intelligent matching of quota sub-items. The multi-strategy fusion algorithm includes semantic similarity matching, knowledge graph reasoning matching, and historical case matching. The comprehensive matching score of candidate quota sub-items or sub-item combinations is calculated by weighted summation or machine learning ranking model, and recommended in descending order of comprehensive matching score. For list items that require multiple quota sub-item combinations, the main quota sub-item and auxiliary quota sub-item are identified by graph covering algorithm, and the minimum quota sub-item set covering all process units is found. Step S4: After selecting a quota item, based on the quota conversion rule library built in step S1, automatically identify the differences between the list item parameters and the standard parameters of the quota item, match the applicable conversion rules and calculate the conversion coefficient. When there are multiple conversion rules, apply the conversion coefficients in sequence according to the preset priority order and handle rule conflicts according to the priority adjudication mechanism. Step S5: After completing the application and conversion of quota items, obtain the current market prices of labor, materials and machinery from multiple price information sources, and perform intelligent price matching through a three-level matching strategy of precise code matching, name semantic matching and specification parameter matching. Calculate the comprehensive unit price of the list items based on the consumption of labor, materials and machinery, conversion coefficients and unit prices of the quota items. Step S6: Perform intelligent verification and optimization on the pricing results generated in Step S5, including verification of the rationality of quota matching, the rationality of conversion coefficients, and the rationality of unit price. When the verification finds that something is unreasonable, generate optimization suggestions and store the final confirmed pricing results in the historical pricing scheme library.

2. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 1, characterized in that, The attributes of the quota sub-item node include quota number, quota name, quota professional category, quota application scope, labor consumption, material consumption, machinery consumption, quota unit, quota base price, and applicable geographical range; the relationship edges include parent-child relationship, substitution relationship, matching relationship, and constraint relationship; at the same time, a quota conversion rule library is constructed, and each conversion rule in the conversion rule library includes triggering conditions, conversion type, conversion coefficient calculation formula, constraint conditions, and priority used to determine the execution order when multiple conversion rules conflict.

3. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 2, characterized in that, The conversion rule base mentioned in step S1 includes at least one of the following eight conversion types: transport distance conversion, thickness conversion, depth conversion, height conversion, strength conversion, gradation conversion, coefficient adjustment conversion, and affirmative conversion; the priority order is: mandatory conversion, conditional conversion, selective conversion, and affirmative conversion.

4. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 1, characterized in that, The natural language processing technology described in step S2 uses a pre-trained language model based on the Transformer architecture and performs fine-tuning training on the professional corpus of water conservancy engineering cost; the intelligent parsing also includes: using a method based on a combination of pattern matching and sequence labeling models to extract the parameters of the list items, and performing synonym normalization processing on the list item names.

5. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 1 or 4, characterized in that, Step S2 also includes using a hybrid similarity calculation method to calculate the similarity between the items in the list to be processed and the items in the historical pricing scheme library. The hybrid similarity calculation method includes a weighted sum of the semantic similarity of the item name, the numerical similarity of the item parameters, and the classification similarity of the item type.

6. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 1, characterized in that, The multi-strategy fusion algorithm in step S3 uses a random forest algorithm model to fit the list of items to be priced with the list of pricing schemes, and then reassembles the list of items to be priced after forming a matching result. The graph covering algorithm uses dynamic programming or branch and bound algorithm to find the optimal quota item combination scheme, and the optimization objective is to minimize the total combined price or maximize the comprehensive matching degree.

7. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 1, characterized in that, The calculation of the conversion factor in step S4 uses the following formula for linear conversion: Conversion factor = base factor + (actual parameter value - standard parameter value) / step size × step size factor; For nonlinear conversion, interpolation calculations are performed using a pre-built parameter-coefficient mapping table; The method further includes: establishing a two-way mapping mode between list item parameters and quota sub-item parameters, so as to complete the conversion in the forward direction and verify the correctness of the conversion in the reverse direction.

8. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 1, characterized in that, The price information sources mentioned in step S5 include at least two of the following: government-published information prices, market prices from professional platforms, historical procurement prices of enterprises, real-time market inquiry data, and user-defined prices; the comprehensive unit price is calculated according to the following formula: P = Σ(C i ×K i ×P i )+ M; Where P is the comprehensive unit price of the list item, and C i K represents the consumption of personnel, materials, and machinery for the i-th quota item. i P is the conversion factor for the i-th quota item. i Let M be the unit price of the i-th type of resource, M be the fee amount, and n be the number of quota items.

9. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 1, characterized in that, The unit price rationality verification in step S6 includes: comparing the calculated unit price of the list item with the unit price of similar items in history, calculating the deviation rate, triggering an alarm and displaying comparison reference data when the deviation rate exceeds a preset threshold; the optimization suggestions include recommended alternative quota items, adjusted conversion coefficients, and suggested unit price correction values.

10. The method for automatic quota application, intelligent conversion, and pricing of water conservancy project bills of quantities according to claim 1, characterized in that, The method further includes: an AI agent based on a large language model, in which the user interacts with the AI ​​agent in natural language, and the AI ​​agent autonomously calls the functional modules of steps S1 to S6 to perform the cost preparation task and feeds back the execution results to the user in natural language.

Images