Intelligent Feasibility Review and Progress Control Methods and Systems for Power Distribution Network Infrastructure Projects

By integrating multi-source heterogeneous data, constructing a multi-dimensional review indicator system, and setting up a three-level early warning mechanism, combined with BIM technology and IoT monitoring, the problems of data collection lag and low review efficiency in the review and progress control of traditional power distribution network infrastructure projects have been solved. This has enabled intelligent, refined, and collaborative management, improving project management efficiency and quality.

CN123070802APending Publication Date: 2026-11-10GUANGDONG POWER GRID CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611293248.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-25
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Traditional methods for reviewing and managing the progress of power distribution network infrastructure projects suffer from problems such as fragmented data collection, low review efficiency, highly subjective setting of indicator weights, and lagging progress management, making it difficult to achieve intelligent, refined, and collaborative management.

Method used

By integrating data from the power grid, the field, and the external environment through a hybrid review model and human-machine collaboration mechanism, a multi-dimensional review indicator system is constructed. Combined with BIM technology and IoT monitoring, a three-level early warning mechanism is set up to achieve intelligent management and control of the entire process.

Benefits of technology

It significantly improved project management efficiency and quality, achieved objectivity and accuracy in reviews, dynamic and precise progress control, and effectively avoided the risk of project delays.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123070802A_ABST
    Figure CN123070802A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for intelligent feasibility review and progress control of power distribution network infrastructure projects, relating to the field of artificial intelligence. The method includes: constructing a standardized dataset by integrating heterogeneous data from the power grid, site, and external environment; establishing a review indicator system covering five dimensions: technology, economy, environment, safety, and compliance; determining weights using the analytic hierarchy process (AHP) and expert scoring; performing scoring and semantic analysis on power distribution network projects based on a hybrid review model integrating a large language model and a gradient boosting tree to generate preliminary review opinions; initiating human-machine collaborative review for critical projects; developing multi-level construction plans by combining the critical path method and BIM technology; collecting progress data through multiple channels; analyzing the causes of deviations; and establishing a three-level early warning mechanism. The advantages of this invention are: improving review efficiency and accuracy through the hybrid review model and human-machine collaborative mechanism, achieving intelligent management of the entire process of power distribution network infrastructure projects from feasibility review to progress control.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, and in particular to a method and system for intelligent feasibility review and progress control of power distribution network infrastructure projects. Background Technology

[0002] As the fundamental carrier for ensuring reliable power supply and energy transition, power distribution network infrastructure projects increasingly require intelligent transformation in feasibility review and progress management. Traditional reviews rely heavily on manual experience, which leads to bottlenecks such as long review cycles, inconsistent standard implementation, and delayed risk insight when faced with massive amounts of multi-source design and construction data. Furthermore, the geographically dispersed nature of project sites, the large number of participating parties, and the complex external environment make it difficult for traditional management methods to achieve dynamic progress tracking and accurate risk warnings, easily resulting in cost overruns and project delays.

[0003] Current market methods for reviewing and managing the progress of power distribution network infrastructure projects suffer from fragmented data collection. They fail to achieve comprehensive integration and standardized processing of heterogeneous data from multiple sources, including the power grid, the site, and the external environment. Data duplication, missing data, and anomalies are prominent issues, making it difficult to support accurate review and management. Reviews often rely on manual processes or single models, lacking a multi-dimensional and multi-level indicator system. The setting of indicator weights is highly subjective, and the deep integration of semantic analysis and intelligent scoring is not achieved. This results in low review efficiency, large biases, and a lack of human-machine collaboration mechanisms, making it difficult to handle critical or complex projects. Progress management often adopts traditional planning models, failing to integrate BIM technology to optimize construction connections or achieve deep integration with material and manpower systems. Progress data collection is lagging, deviation analysis is superficial, and there is a lack of a scientific three-level early warning and dynamic adjustment mechanism, making it difficult to quickly address progress deviations and easily leading to project delays. Overall, the level of intelligence, refinement, and collaboration is far inferior to the proposed methods. Summary of the Invention

[0004] To improve existing methods and systems, this paper provides a method and system for intelligent feasibility review and progress control of power distribution network infrastructure projects. This method improves the efficiency and accuracy of review through a hybrid review model and human-machine collaboration mechanism. By combining BIM technology, IoT monitoring and a three-level early warning mechanism, it realizes intelligent management and control of the entire process of power distribution network infrastructure projects from feasibility review to progress control.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] Intelligent feasibility review and progress control methods for power distribution network infrastructure projects include:

[0007] Collect and integrate three types of data: power grid side, field side, and external environment. Classify and process the collected heterogeneous data according to type, remove duplicates, fill in missing data, correct anomalies, and obtain a standardized dataset with a unified format.

[0008] A five-dimensional evaluation indicator system is constructed, encompassing technology, economy, environment, safety, and compliance. This system includes secondary and tertiary indicators, and the weights of these indicators are determined using the analytic hierarchy process (AHP) and expert scoring to form the evaluation indicator standards.

[0009] By integrating a large language model and a gradient boosting tree into a hybrid review model, the model is trained with historical project data, performs semantic analysis to match indicators and automatically scores, and iteratively updates the model by comparing the model output with historical human review results.

[0010] Data on power distribution network projects are entered into a hybrid review model for indicator matching and automatic scoring to generate preliminary review opinions. For projects that are critical or have significant shortcomings, human-machine collaborative review is triggered to generate a formal review report.

[0011] Based on historical data and current project conditions, the critical path method is used to divide the baseline plan into five major construction phases. BIM technology is used to simulate and optimize construction connections, and the material and manpower systems are integrated to decompose the baseline plan into multi-level plans.

[0012] By acquiring construction progress-related data through IoT terminals, mobile apps, and supervision systems, calculating the progress deviation rate by comparing it with the baseline plan, and analyzing the causes of deviations by combining historical and environmental data, a deviation analysis report is generated.

[0013] Based on the results of the schedule deviation analysis, a three-level early warning threshold is set, and corresponding rectification and adjustment measures are initiated for different degrees of deviation. For major deviations, a special meeting is organized to re-optimize the schedule control benchmark plan and push the results in real time.

[0014] Preferably, the process of collecting and integrating data from the power grid side, the field side, and the external environment, and classifying and processing the collected heterogeneous data according to type, removing duplicates, filling in missing data, correcting anomalies, and obtaining a standardized dataset in a unified format specifically includes:

[0015] Establish a multi-source data acquisition terminal for power distribution network infrastructure projects to integrate three types of core data: power grid side, field side, and external environment.

[0016] The collected multi-source heterogeneous data were standardized and preprocessed. The structured data was removed by key-value matching. Missing values ​​were filled by the mean of similar items. Outliers were identified and corrected.

[0017] Semi-structured data uses a method of associating data from the same source to sort out hierarchical relationships, extract core information, and transform it into a structured format;

[0018] Unstructured data is processed using streaming data sampling and feature extraction methods to transform on-site images and documents into identifiable structured data. All preprocessed data is then standardized in format and unit to form a standardized dataset.

[0019] Preferably, the assessment indicator system comprises five dimensions: technology, economy, environment, safety, and compliance, with secondary and tertiary specific indicators. The weights of these indicators are determined using the analytic hierarchy process (AHP) and expert scoring, resulting in the assessment indicator standards that specifically include:

[0020] Based on the needs of power distribution network infrastructure projects and industry standards, a review indicator system covering five dimensions—technical feasibility, economic feasibility, environmental feasibility, safety feasibility, and compliance feasibility—is constructed, with specific secondary and tertiary indicators under each dimension.

[0021] The analytic hierarchy process (AHP) combined with expert scoring was used to assign weights to each indicator. Experts in power distribution network planning, construction, operation and maintenance, and economic evaluation were organized to revise the indicator weights through multiple rounds of demonstration.

[0022] The normative clauses are transformed into indicator verification rules and integrated into the indicator system to obtain the pass threshold and scoring criteria for each indicator, thus forming the review indicator standards.

[0023] Preferably, the hybrid review model that integrates a large language model and a gradient boosting tree, trains the model using historical project data, performs semantic analysis to match metrics and automatically scores, and iteratively updates the model by comparing the model output with historical human review results, specifically including:

[0024] A hybrid review model that integrates a large language model and a gradient boosting tree is adopted, and the model is trained based on historical data of similar power distribution network infrastructure projects.

[0025] Based on the semantic analysis capabilities of the large language model, key information is extracted from the project application materials, matched with the review indicator system, and automatically scored for each indicator through a gradient boosting tree model.

[0026] During training, the model's output review results are compared with historical human review results in real time to calculate the deviation value. Based on the deviation value, the indicator weights and model judgment logic are adjusted. New project review data and expert feedback are regularly incorporated to iteratively update the model.

[0027] Preferably, the step of inputting distribution network project data into a hybrid review model for indicator matching and automatic scoring to generate preliminary review opinions, and triggering human-machine collaborative review for projects with critical or significant shortcomings to form a formal review report specifically includes:

[0028] The application materials for power distribution network infrastructure projects are entered into the hybrid review model. The model extracts the standardized data of the projects and matches them one by one with the constructed feasibility review index system. Each third-level index is automatically scored, and the comprehensive review score is calculated based on the index weight.

[0029] For projects with a comprehensive review score higher than the passing line, a preliminary review opinion is generated. For projects with a comprehensive score at the critical value, a human-machine collaborative review is triggered, the model review results are corrected, and a formal review report is generated.

[0030] For projects with an overall score below the passing mark, a review disapproval opinion will be generated, specifying the reasons for disapproval.

[0031] Preferably, the baseline plan is divided into five construction phases using the critical path method based on historical data and current project conditions. BIM technology is used to simulate and optimize construction coordination, and the material and manpower systems are integrated. Specifically, the decomposition of the baseline plan into multi-level plans includes:

[0032] For distribution network infrastructure projects that have passed the feasibility review, a baseline schedule control plan is constructed using the critical path method, based on historical progress data of similar projects, the scale, complexity, and construction conditions of the current project.

[0033] The baseline plan clearly divides the project into five core construction phases: preliminary preparation, civil construction, equipment procurement and installation, commissioning and testing, and final acceptance. Each phase is further detailed with specific task lists, timelines, responsible parties, and resource allocation plans.

[0034] By using BIM technology to visualize and simulate the construction process, we can optimize the connection between construction procedures, identify potential bottlenecks in the construction process, and set reasonable buffer times for each task.

[0035] The baseline plan is integrated with the materials management system and the manpower scheduling system, and the baseline plan is broken down into weekly plans and monthly plans to form a multi-level progress control system.

[0036] Preferably, the step of acquiring construction progress-related data through IoT terminals, mobile apps, and supervision systems, calculating the progress deviation rate by comparing it with the baseline plan, analyzing the causes of deviations by combining historical and environmental data, and forming a deviation analysis report specifically includes:

[0037] Through IoT terminals at the construction site, the real-time data collection includes the operating status of construction equipment, the attendance of construction personnel, the quantity and usage of materials, and the progress of each construction process.

[0038] The construction progress can be obtained through a mobile app, and the supervision and acceptance records can be obtained by connecting to the supervision system.

[0039] The collected real-time progress data is compared with the progress control baseline plan in real time to calculate the progress deviation rate of each task. Combined with historical data and environmental data collected on site, the causes of deviations are analyzed and a progress deviation analysis report is generated.

[0040] Preferably, the method involves setting a three-level early warning threshold based on the schedule deviation analysis results, initiating corresponding rectification and adjustment measures for different degrees of deviation, organizing special meetings to re-optimize the schedule control baseline plan for major deviations, and pushing specific details in real time, including:

[0041] Based on the schedule deviation analysis results, a three-level early warning threshold is set. When the schedule deviation rate is within 5%, a level one early warning is issued without adjusting the baseline plan.

[0042] When the schedule deviation rate is between 5% and 10%, a level-two warning will be issued, the schedule adjustment plan will be activated, and the subsequent construction rhythm of the task will be adjusted.

[0043] When the schedule deviation rate exceeds 10% or affects the core tasks on the critical path, a Level 3 warning is issued, a special meeting is held to obtain the impact of the deviation on the overall project duration, and the schedule control baseline plan is re-optimized.

[0044] Warning information is pushed out via SMS and mobile APP, clearly defining the rectification deadline and division of responsibilities. After the rectification is completed, deviation analysis is carried out again.

[0045] Furthermore, a smart feasibility review and progress control system for power distribution network infrastructure projects is proposed, including:

[0046] Data integration and standardization module: Integrates heterogeneous data from multiple sources, including power grid, field, and external environment, and generates a unified standard project dataset through cleaning, deduplication, imputation, and format conversion;

[0047] Review Indicators and Model Management Module: Constructs a multi-level review indicator system covering technology, economy, environment, safety and compliance. The weights are determined based on the analytic hierarchy process and expert experience. It also integrates a large language model and a gradient boosting tree hybrid model to perform semantic understanding, indicator matching and automatic scoring of project data.

[0048] Collaborative review and report generation module: Receives project materials and calls the hybrid review model to conduct preliminary review, generates review opinions, triggers manual intervention review process for critical projects, and generates formal review reports;

[0049] Schedule planning and decomposition module: The critical path method is used to develop a baseline plan that includes five major construction phases. BIM is used for construction simulation and optimization, and the plan is decomposed into multi-level tasks.

[0050] Progress monitoring and deviation analysis module: Collects progress data in real time, compares it with the baseline plan to calculate the deviation rate, analyzes the causes of deviations by combining historical and environmental data, and generates a progress deviation analysis report;

[0051] Early warning response and control module: A three-level early warning mechanism is set up based on the degree of deviation. Adjustment measures are triggered for different levels. Special meetings are organized to re-optimize the baseline plan for major deviations, and early warning information is pushed in real time.

[0052] Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

[0053] Compared with the prior art, the advantages of the present invention are:

[0054] This system achieves intelligent, standardized, and refined management throughout the entire process, significantly improving project management efficiency and quality. By integrating and standardizing multi-source heterogeneous data, it addresses the pain points of data disorder and inconsistency, providing reliable data support for review and control. It constructs a multi-dimensional, multi-level review indicator system, combining a hybrid intelligent model with a human-machine collaborative mode, which improves the objectivity and accuracy of reviews while reducing manual workload, achieving high-efficiency reviews. It optimizes the schedule using the critical path method and BIM technology, and monitors progress in real time using IoT and other means. A three-level early warning mechanism promptly addresses deviations, achieving dynamic and precise schedule control, effectively mitigating the risk of project delays. Furthermore, through model iteration and plan optimization, it continuously improves the adaptability of management, providing strong support for the feasibility control and efficient advancement of power distribution network infrastructure projects. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the method proposed in this invention;

[0056] Figure 2 This is a schematic diagram of the heterogeneous data acquisition and standardized preprocessing proposed in this invention;

[0057] Figure 3 This is a schematic diagram illustrating the construction of the feasibility assessment index system proposed in this invention;

[0058] Figure 4 This is a schematic diagram illustrating the training and optimization of the review model proposed in this invention;

[0059] Figure 5 This is a schematic diagram illustrating the project feasibility review process proposed in this invention.

[0060] Figure 6 This is a schematic diagram illustrating the generation of the progress control baseline plan proposed in this invention;

[0061] Figure 7 This is a schematic diagram of the progress data acquisition and deviation analysis proposed in this invention;

[0062] Figure 8 This is a schematic diagram of the progress deviation early warning and adjustment proposed in this invention. Detailed Implementation

[0063] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0064] The intelligent feasibility review and progress control system for power distribution network infrastructure projects includes:

[0065] Data integration and standardization module: Integrates heterogeneous data from multiple sources, including power grid, field, and external environment, and generates a unified standard project dataset through cleaning, deduplication, imputation, and format conversion;

[0066] Review Indicators and Model Management Module: Constructs a multi-level review indicator system covering technology, economy, environment, safety and compliance. The weights are determined based on the analytic hierarchy process and expert experience. It also integrates a large language model and a gradient boosting tree hybrid model to perform semantic understanding, indicator matching and automatic scoring of project data.

[0067] Collaborative review and report generation module: Receives project materials and calls the hybrid review model to conduct preliminary review, generates review opinions, triggers manual intervention review process for critical projects, and generates formal review reports;

[0068] Schedule planning and decomposition module: The critical path method is used to develop a baseline plan that includes five major construction phases. BIM is used for construction simulation and optimization, and the plan is decomposed into multi-level tasks.

[0069] Progress monitoring and deviation analysis module: Collects progress data in real time, compares it with the baseline plan to calculate the deviation rate, analyzes the causes of deviations by combining historical and environmental data, and generates a progress deviation analysis report;

[0070] Early warning response and control module: A three-level early warning mechanism is set up based on the degree of deviation. Adjustment measures are triggered for different levels. Special meetings are organized to re-optimize the baseline plan for major deviations, and early warning information is pushed in real time.

[0071] Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

[0072] See Figure 1 As shown, the intelligent feasibility review and progress control method for power distribution network infrastructure projects includes:

[0073] Step 1: Collect and integrate three types of data: power grid side, field side, and external environment. Classify and process the collected heterogeneous data according to type, remove duplicates, fill in missing data, correct anomalies, and obtain a standardized dataset in a unified format.

[0074] Step 2: Construct a review indicator system encompassing five dimensions: technology, economy, environment, safety, and compliance. This system includes secondary and tertiary indicators. The weights of these indicators are determined using the analytic hierarchy process (AHP) and expert scoring, thus forming the review indicator standards.

[0075] Step 3: By using a hybrid review model that integrates a large language model and a gradient boosting tree, the model is trained with historical project data, semantic analysis is performed to match indicators and automatic scoring is carried out, and the model is iteratively updated by comparing the model output results with historical human review results.

[0076] Step 4: Input the data of the power distribution network project into the hybrid review model for indicator matching and automatic scoring, generate preliminary review opinions, trigger human-machine collaborative review for critical or major shortcomings, and form a formal review report.

[0077] Step 5: Based on historical data and current project conditions, the critical path method is used to divide the baseline plan into five major construction phases. BIM technology is used to simulate and optimize the construction connection, and the material and manpower systems are integrated to decompose the baseline plan into multi-level plans.

[0078] Step Six: Obtain relevant construction progress data through IoT terminals, mobile apps, and supervision systems; compare the data with the baseline plan to calculate the progress deviation rate; analyze the causes of deviations by combining historical and environmental data; and generate a deviation analysis report.

[0079] Step 7: Based on the schedule deviation analysis results, set three-level early warning thresholds, initiate corresponding rectification and adjustment measures for different deviation levels, organize special meetings to re-optimize the schedule control benchmark plan for major deviations, and push the results in real time.

[0080] See Figure 2 As shown, the process involves collecting and integrating three types of data: data from the power grid side, the field side, and the external environment. The heterogeneous data is categorized and processed according to type, removing duplicates, filling in missing data, correcting anomalies, and obtaining a standardized dataset in a unified format. Specifically, this includes:

[0081] Establish a multi-source data acquisition terminal for power distribution network infrastructure projects to integrate three types of core data: power grid side, field side, and external environment.

[0082] The collected multi-source heterogeneous data were standardized and preprocessed. The structured data was removed by key-value matching. Missing values ​​were filled by the mean of similar items. Outliers were identified and corrected.

[0083] Semi-structured data uses a method of associating data from the same source to sort out hierarchical relationships, extract core information, and transform it into a structured format;

[0084] Unstructured data is processed using streaming data sampling and feature extraction methods to transform on-site images and documents into identifiable structured data. All preprocessed data is then standardized in format and unit to form a standardized dataset.

[0085] Specifically, structured data employs a dual key-value matching method, comparing core key values ​​such as project number and data collection time to remove duplicate data; missing values ​​are categorized by project type and region, and filled using the average of the same indicator for the same type of project, followed by manual sampling verification; outliers are marked with reasonable thresholds set according to industry standards; semi-structured data uses a source data association method to sort out the hierarchical relationship of data, extract core information, and transform it into a structured format; unstructured data uses streaming data sampling combined with feature extraction technology to perform contour recognition and feature annotation on on-site images, and keyword extraction and semantic sorting on document materials, transforming them into identifiable structured data;

[0086] All preprocessed data are uniformly formatted in JSON and use national standard units to form a standardized project database. Distributed database storage is used, and sensitive data is encrypted using the AES encryption algorithm.

[0087] See Figure 3 As shown, a five-dimensional evaluation indicator system is constructed, encompassing technology, economy, environment, safety, and compliance. This system includes secondary and tertiary indicators. The weights of these indicators are determined using the analytic hierarchy process (AHP) and expert scoring, resulting in the evaluation indicator standards that specifically include:

[0088] Based on the needs of power distribution network infrastructure projects and industry standards, a review indicator system covering five dimensions—technical feasibility, economic feasibility, environmental feasibility, safety feasibility, and compliance feasibility—is constructed, with specific secondary and tertiary indicators under each dimension.

[0089] The analytic hierarchy process (AHP) combined with expert scoring was used to assign weights to each indicator. Experts in power distribution network planning, construction, operation and maintenance, and economic evaluation were organized to revise the indicator weights through multiple rounds of demonstration.

[0090] The normative clauses are transformed into indicator verification rules and integrated into the indicator system to obtain the pass threshold and scoring criteria for each indicator, thus forming the review indicator standards.

[0091] Specifically, the technical feasibility indicators focus on the adaptability of distribution network operation, with secondary indicators refined into load carrying capacity and equipment compatibility, and tertiary indicators further clarified into specific quantifiable indicators such as comprehensive voltage qualification rate and line N-1 pass rate. By connecting with the existing power grid monitoring system, the data collection sources and calculation methods of each indicator are clarified.

[0092] The economic feasibility indicators revolve around investment benefits, with secondary indicators including investment return and cost control, and tertiary indicators including rate of return on investment and investment payback period. The calculation methods for each indicator are clearly defined. For example, the rate of return on investment is calculated as the ratio of the average annual net profit during the project's operating period to the total investment, and the operation and maintenance costs cover all life-cycle expenditures such as equipment maintenance and personnel salaries.

[0093] The environmental, safety, and compliance feasibility indicators are refined into three levels of indicators, which are combined with environmental protection regulations, safety production standards, and government approval requirements. For example, the distributed energy penetration rate in environmental feasibility, the hazard identification compliance rate in safety feasibility, and the land planning compliance in compliance feasibility.

[0094] A combination of the Analytic Hierarchy Process (AHP) and expert scoring was used. The AHP was used to construct a judgment matrix to quantify the relative importance of each dimension and indicator. Then, senior experts in power distribution network planning, construction, operation and maintenance, and economic evaluation were organized to adjust the weights through multiple rounds of anonymous scoring. The expert scoring adopted the 1-9 scale method, and the consistency ratio was calculated after each round of scoring. The technical and economic dimensions were determined to have the highest weight, at 30% and 25% respectively, and the remaining dimensions were reasonably weighted.

[0095] See Figure 4 As shown, a hybrid review model combining a large language model and a gradient boosting tree is used. The model is trained with historical project data, performs semantic analysis to match metrics, and automatically scores. The model is iteratively updated by comparing its output with historical human review results. Specifically, this includes:

[0096] A hybrid review model that integrates a large language model and a gradient boosting tree is adopted, and the model is trained based on historical data of similar power distribution network infrastructure projects.

[0097] Based on the semantic analysis capabilities of the large language model, key information is extracted from the project application materials, matched with the review indicator system, and automatically scored for each indicator through a gradient boosting tree model.

[0098] During training, the model's output review results are compared with historical human review results in real time to calculate the deviation value. Based on the deviation value, the indicator weights and model judgment logic are adjusted. New project review data and expert feedback are regularly incorporated to iteratively update the model.

[0099] Specifically, data from similar power distribution network infrastructure projects that have completed feasibility reviews, acceptance, and have been in operation for at least one year within the past five years were selected. These projects cover different voltage levels, construction scales, and geographical regions. The selected data includes project application materials, expert review opinions, acceptance results, operational data, and rectification records. The data is divided into training, validation, and test sets in a 7:2:1 ratio. The training set is used for model parameter training, the validation set is used for adjusting model hyperparameters, and the test set is used for verifying the final performance of the model. The division process uses random sampling.

[0100] A hybrid architecture integrating a large language model and a gradient boosting tree is adopted. The large language model uses a finely tuned version adapted to the power industry, and imports relevant normative clauses, technical standards and historical review cases related to the review of distribution network infrastructure projects in advance to optimize semantic analysis capabilities. The gradient boosting tree model uses the XGBoost algorithm framework and is customized for the review indicator scoring task.

[0101] The large language model is activated to perform sentence-by-sentence semantic analysis on the project application materials in the training set, accurately extract key information such as project construction scale, equipment parameters, and construction plan, and match them one by one with the constructed review indicator system. The unstructured application materials are transformed into structured indicator data that can be used for scoring. The matched indicator data is then input into the gradient boosting tree model. The model automatically scores each third-level indicator based on the scoring criteria in the historical review data, and then calculates the comprehensive review score by combining the indicator weights.

[0102] Bayesian optimization was used to adjust the model hyperparameters. For large language models, parameters such as the semantic extraction threshold and context window size were optimized. For gradient boosting tree models, parameters such as the learning rate, decision tree depth, and number of leaf nodes were optimized. During the optimization process, the evaluation accuracy of the validation set was used as the core evaluation metric. Through multiple rounds of iterative adjustments, the model's evaluation accuracy on the validation set reached over 97%. At the same time, a 5-fold cross-validation method was used to divide the training set into 5 subsets, which were used as the validation set for training in turn.

[0103] See Figure 5 As shown, data from power distribution network projects is entered into a hybrid review model for indicator matching and automatic scoring, generating preliminary review opinions. For projects at critical junctures or with significant shortcomings, a human-machine collaborative review is triggered, resulting in a formal review report. This report specifically includes:

[0104] The application materials for power distribution network infrastructure projects are entered into the hybrid review model. The model extracts the standardized data of the projects and matches them one by one with the constructed feasibility review index system. Each third-level index is automatically scored, and the comprehensive review score is calculated based on the index weight.

[0105] For projects with a comprehensive review score higher than the passing line, a preliminary review opinion is generated. For projects with a comprehensive score at the critical value, a human-machine collaborative review is triggered, the model review results are corrected, and a formal review report is generated.

[0106] For projects with an overall score below the passing mark, a review disapproval opinion will be generated, specifying the reasons for disapproval.

[0107] Specifically, the model will accurately match the extracted project data with the feasibility review indicator system one by one. For each third-level indicator, it will automatically complete the scoring calculation according to the preset qualification threshold and scoring standard, and record the scoring basis of each indicator. After the scoring is completed, the model will automatically calculate the comprehensive review score of the project according to the weight of each indicator, generate a preliminary scoring report, and clarify the scores of each dimension, the comprehensive score and the ranking.

[0108] The model automatically compares the scores of each level three indicator with the passing threshold, identifies indicators that do not meet the passing standard, and determines them as project shortcomings. For each shortcoming indicator, it automatically matches the corresponding specification requirements, generates targeted problem prompts and actionable improvement suggestions, and links the improvement suggestions to successful rectification cases of similar projects in the past.

[0109] See Figure 6 As shown, based on historical data and current project conditions, the critical path method is used to divide the baseline plan into five major construction phases. BIM technology is used to simulate and optimize construction coordination, and the material and manpower systems are integrated. The baseline plan is decomposed into a multi-level plan, specifically including:

[0110] For distribution network infrastructure projects that have passed the feasibility review, a baseline schedule control plan is constructed using the critical path method, based on historical progress data of similar projects, the scale, complexity, and construction conditions of the current project.

[0111] The baseline plan clearly divides the project into five core construction phases: preliminary preparation, civil construction, equipment procurement and installation, commissioning and testing, and final acceptance. Each phase is further detailed with specific task lists, timelines, responsible parties, and resource allocation plans.

[0112] By using BIM technology to visualize and simulate the construction process, we can optimize the connection between construction procedures, identify potential bottlenecks in the construction process, and set reasonable buffer times for each task.

[0113] The baseline plan is integrated with the materials management system and the manpower scheduling system, and the baseline plan is broken down into weekly plans and monthly plans to form a multi-level progress control system.

[0114] Specifically, the critical path method is used to build a baseline plan framework for schedule control, sort out the tasks of the entire project process, and clearly divide the project into five core construction phases: project preparation, civil construction, equipment procurement and installation, commissioning and testing, and final acceptance. Each phase is broken down into specific executable sub-tasks. By calculating the earliest start time, latest start time, and duration of each task, the critical tasks that affect the overall project duration are identified, the critical path is formed, and the critical tasks are taken as the core focus of schedule control.

[0115] Import project construction drawings, terrain data, and equipment parameters into BIM modeling software to build a 3D visualization construction model, simulate the entire construction process, and identify potential bottlenecks such as poor process connection and construction space conflict through construction simulation. Based on the simulation results, set buffer time for each task. Generally, the buffer time for critical path tasks is set to 10%-15% of the task duration, and for non-critical path tasks, it is set to 5%-10%.

[0116] The baseline plan is interfaced with the power company's material management system and manpower dispatch system to synchronize material procurement progress, equipment arrival status, and construction personnel dispatch information. The baseline plan is decomposed hierarchically, with the overall plan broken down into monthly plans, the monthly plans further refined into weekly plans, and the weekly plans specifying daily construction tasks and objectives, forming a multi-level progress control system of "overall plan - monthly plan - weekly plan - daily plan".

[0117] See Figure 7 As shown, construction progress data is obtained through IoT terminals, mobile apps, and the supervision system. The progress deviation rate is calculated by comparing this data with the baseline plan. The causes of deviations are analyzed by combining historical and environmental data, resulting in a deviation analysis report. This report specifically includes:

[0118] Through IoT terminals at the construction site, the real-time data collection includes the operating status of construction equipment, the attendance of construction personnel, the quantity and usage of materials, and the progress of each construction process.

[0119] The construction progress can be obtained through a mobile app, and the supervision and acceptance records can be obtained by connecting to the supervision system.

[0120] The collected real-time progress data is compared with the progress control baseline plan in real time to calculate the progress deviation rate of each task. Combined with historical data and environmental data collected on site, the causes of deviations are analyzed and a progress deviation analysis report is generated.

[0121] Specifically, the collected real-time progress data is compared with the generated progress control baseline plan in real time. The progress deviation rate of each task is calculated using the formula "(actual progress - planned progress) / planned progress × 100%". Tasks that are behind schedule or ahead of schedule are accurately identified. Deviation tasks are divided into critical path tasks and non-critical path tasks according to their scope of impact, and deviation tasks on the critical path are marked.

[0122] By combining historical schedule deviation data of similar projects in the standardized project database with environmental data collected on site, the causes of deviations are accurately analyzed to identify specific reasons such as material shortages, equipment failures, insufficient personnel, weather effects, or delays in land acquisition and approval. A detailed schedule deviation analysis report is generated to obtain the deviation tasks, the degree of deviation, the scope of impact, and the root cause.

[0123] See Figure 8 As shown, based on the schedule deviation analysis results, a three-level early warning threshold is set, and corresponding rectification and adjustment measures are initiated for different degrees of deviation. For major deviations, a special meeting is organized to re-optimize the schedule control baseline plan, and specific details are pushed out in real time, including:

[0124] Based on the schedule deviation analysis results, a three-level early warning threshold is set. When the schedule deviation rate is within 5%, a level one early warning is issued without adjusting the baseline plan.

[0125] When the schedule deviation rate is between 5% and 10%, a level-two warning will be issued, the schedule adjustment plan will be activated, and the subsequent construction rhythm of the task will be adjusted.

[0126] When the schedule deviation rate exceeds 10% or affects the core tasks on the critical path, a Level 3 warning is issued, a special meeting is held to obtain the impact of the deviation on the overall project duration, and the schedule control baseline plan is re-optimized.

[0127] Warning information is pushed out via SMS and mobile APP, clearly defining the rectification deadline and division of responsibilities. After the rectification is completed, deviation analysis is carried out again.

[0128] Specifically, the first-level warning triggering and handling: When the system detects that the progress deviation rate of a certain task is within 5%, the first-level warning is automatically triggered. The warning information is pushed to the on-site construction manager through the progress control system and mobile APP, which clarifies the warning task, deviation rate and rectification requirements, without the need to adjust the progress control baseline plan; During the handling process, the construction manager reports the rectification measures and progress in real time through the mobile APP. After the rectification is completed, the progress data is verified. The warning is lifted when the deviation rate drops to within the threshold.

[0129] Level 2 Early Warning Triggering and Handling: When the progress deviation rate is between 5% and 10%, a Level 2 early warning is triggered, and the warning information is pushed to the construction unit, supervision unit, and project management personnel. When materials are in short supply, an expedited procurement order is sent through the material management system, specifying the delivery deadline and assigning a dedicated person to track it. When personnel are insufficient, backup construction personnel are dispatched through the manpower dispatch system. When equipment fails, the equipment operation and maintenance unit is contacted to allocate alternative equipment to ensure continuous construction. After rectification, the deviation rate is recalculated until the deviation is eliminated and the early warning is lifted.

[0130] Level 3 Early Warning Triggering and Handling: When the schedule deviation rate exceeds 10% or affects the core tasks of the critical path, a Level 3 early warning is triggered, the construction of related processes is suspended, and project management personnel, construction units, supervision units and industry experts are notified through multiple channels. A special meeting is organized to obtain the degree of impact of the deviation on the overall project duration based on the current deviation data, historical deviation handling data of similar projects and on-site environmental data, and generate multiple adjustment plans for the meeting to make decisions.

[0131] Based on the decisions made at the special meeting, the baseline plan for progress control was re-optimized, with a focus on adjusting the time allocation of tasks on the critical path. The delays were made up by compressing the buffer time for non-critical path tasks, optimizing the connection of construction procedures, and increasing resource input.

[0132] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0133] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0134] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent feasibility review and progress control of power distribution network infrastructure projects, characterized in that, include: Collect and integrate three types of data: power grid side, field side, and external environment. Classify and process the collected heterogeneous data according to type, remove duplicates, fill in missing data, correct anomalies, and obtain a standardized dataset with a unified format. A five-dimensional evaluation indicator system is constructed, encompassing technology, economy, environment, safety, and compliance. This system includes secondary and tertiary indicators, and the weights of these indicators are determined using the analytic hierarchy process (AHP) and expert scoring to form the evaluation indicator standards. By integrating a large language model and a gradient boosting tree into a hybrid review model, the model is trained with historical project data, performs semantic analysis to match indicators and automatically scores, and iteratively updates the model by comparing the model output with historical human review results. Data on power distribution network projects are entered into a hybrid review model for indicator matching and automatic scoring to generate preliminary review opinions. For projects that are critical or have significant shortcomings, human-machine collaborative review is triggered to generate a formal review report. Based on historical data and current project conditions, the critical path method is used to divide the baseline plan into five major construction phases. BIM technology is used to simulate and optimize construction connections, and the material and manpower systems are integrated to decompose the baseline plan into multi-level plans. By acquiring construction progress-related data through IoT terminals, mobile apps, and supervision systems, calculating the progress deviation rate by comparing it with the baseline plan, and analyzing the causes of deviations by combining historical and environmental data, a deviation analysis report is generated. Based on the results of the schedule deviation analysis, a three-level early warning threshold is set, and corresponding rectification and adjustment measures are initiated for different degrees of deviation. For major deviations, a special meeting is organized to re-optimize the schedule control benchmark plan and push the results in real time.

2. The method for intelligent feasibility review and progress control of power distribution network infrastructure projects according to claim 1, characterized in that, The process of collecting and integrating data from the power grid side, the field side, and the external environment, and classifying and processing the heterogeneous data according to type to remove duplicates, fill in missing data, correct anomalies, and obtain a standardized dataset in a unified format specifically includes: Establish a multi-source data acquisition terminal for power distribution network infrastructure projects to integrate three types of core data: power grid side, field side, and external environment. The collected multi-source heterogeneous data were standardized and preprocessed. The structured data was removed by key-value matching. Missing values ​​were filled by the mean of similar items. Outliers were identified and corrected. Semi-structured data uses a method of associating data from the same source to sort out hierarchical relationships, extract core information, and transform it into a structured format; Unstructured data is processed using streaming data sampling and feature extraction methods to transform on-site images and documents into identifiable structured data. All preprocessed data is then standardized in format and unit to form a standardized dataset.

3. The method for intelligent feasibility review and progress control of power distribution network infrastructure projects according to claim 1, characterized in that, The assessment indicator system comprises five dimensions: technology, economy, environment, safety, and compliance. It includes secondary and tertiary indicators, and the weights of these indicators are determined using the analytic hierarchy process (AHP) and expert scoring. The specific assessment indicator standards include: Based on the needs of power distribution network infrastructure projects and industry standards, a review indicator system covering five dimensions—technical feasibility, economic feasibility, environmental feasibility, safety feasibility, and compliance feasibility—is constructed, with specific secondary and tertiary indicators under each dimension. The analytic hierarchy process (AHP) combined with expert scoring was used to assign weights to each indicator. Experts in power distribution network planning, construction, operation and maintenance, and economic evaluation were organized to revise the indicator weights through multiple rounds of demonstration. The normative clauses are transformed into indicator verification rules and integrated into the indicator system to obtain the pass threshold and scoring criteria for each indicator, thus forming the review indicator standards.

4. The method for intelligent feasibility review and progress control of power distribution network infrastructure projects according to claim 1, characterized in that, The hybrid review model, which integrates a large language model and a gradient boosting tree, is trained using historical project data. It performs semantic analysis to match metrics and automatically scores the data. The model is then iteratively updated by comparing its output with historical human review results. Specifically, this includes: A hybrid review model that integrates a large language model and a gradient boosting tree is adopted, and the model is trained based on historical data of similar power distribution network infrastructure projects. Based on the semantic analysis capabilities of the large language model, key information is extracted from the project application materials, matched with the review indicator system, and automatically scored for each indicator through a gradient boosting tree model. During training, the model output review results are compared with historical human review results in real time to calculate the deviation value. The indicator weights and model judgment logic are adjusted according to the deviation value. New project review data and expert feedback are regularly incorporated to iteratively update the model.

5. The method for intelligent feasibility review and progress control of power distribution network infrastructure projects according to claim 1, characterized in that, The process of inputting power distribution network project data into a hybrid review model for indicator matching and automatic scoring, generating preliminary review opinions, and triggering human-machine collaborative review for projects with critical or significant shortcomings to form a formal review report specifically includes: The application materials for power distribution network infrastructure projects are entered into the hybrid review model. The model extracts the standardized data of the projects and matches them one by one with the constructed feasibility review index system. Each third-level index is automatically scored, and the comprehensive review score is calculated based on the index weight. For projects with a comprehensive review score higher than the passing line, a preliminary review opinion is generated. For projects with a comprehensive score at the critical value, a human-machine collaborative review is triggered, the model review results are corrected, and a formal review report is generated. For projects with an overall score below the passing mark, a review disapproval opinion will be generated, specifying the reasons for disapproval.

6. The method for intelligent feasibility review and progress control of power distribution network infrastructure projects according to claim 1, characterized in that, Based on historical data and current project conditions, the baseline plan is divided into five construction phases using the critical path method. BIM technology is used to simulate and optimize construction coordination, and the material and manpower systems are integrated. The baseline plan is decomposed into a multi-level plan, specifically including: For distribution network infrastructure projects that have passed the feasibility review, a baseline schedule control plan is constructed using the critical path method, based on historical progress data of similar projects, the scale, complexity, and construction conditions of the current project. The baseline plan clearly divides the project into five core construction phases: preliminary preparation, civil construction, equipment procurement and installation, commissioning and testing, and final acceptance. Each phase is further detailed with specific task lists, timelines, responsible parties, and resource allocation plans. By using BIM technology to visualize and simulate the construction process, we can optimize the connection between construction procedures, identify potential bottlenecks in the construction process, and set reasonable buffer times for each task. The baseline plan is integrated with the materials management system and the manpower scheduling system, and the baseline plan is broken down into weekly plans and monthly plans to form a multi-level progress control system.

7. The method for intelligent feasibility review and progress control of power distribution network infrastructure projects according to claim 1, characterized in that, The process of acquiring construction progress-related data through IoT terminals, mobile apps, and supervision systems, comparing it with the baseline plan to calculate the progress deviation rate, and combining historical and environmental data to analyze the causes of deviations and generate a deviation analysis report specifically includes: Through IoT terminals at the construction site, the real-time data collection includes the operating status of construction equipment, the attendance of construction personnel, the quantity and usage of materials, and the progress of each construction process. The construction progress can be obtained through a mobile app, and the supervision and acceptance records can be obtained by connecting to the supervision system. The collected real-time progress data is compared with the progress control baseline plan in real time to calculate the progress deviation rate of each task. Combined with historical data and environmental data collected on site, the causes of deviations are analyzed and a progress deviation analysis report is generated.

8. The method for intelligent feasibility review and progress control of power distribution network infrastructure projects according to claim 1, characterized in that, The process involves setting three levels of early warning thresholds based on schedule deviation analysis results, initiating corresponding rectification and adjustment measures for different degrees of deviation, organizing special meetings to re-optimize the schedule control baseline plan for major deviations, and pushing specific details in real time, including: Based on the schedule deviation analysis results, a three-level early warning threshold is set. When the schedule deviation rate is within 5%, a level one early warning is issued without adjusting the baseline plan. When the schedule deviation rate is between 5% and 10%, a level-two warning will be issued, the schedule adjustment plan will be activated, and the subsequent construction rhythm of the task will be adjusted. When the schedule deviation rate exceeds 10% or affects the core tasks on the critical path, a Level 3 warning is issued, a special meeting is held to obtain the impact of the deviation on the overall project duration, and the schedule control baseline plan is re-optimized. Warning information is pushed out via SMS and mobile APP, clearly defining the rectification deadline and division of responsibilities. After the rectification is completed, deviation analysis is carried out again.

9. A feasibility intelligent review and progress control system for power distribution network infrastructure projects, used to implement the feasibility intelligent review and progress control method for power distribution network infrastructure projects as described in any one of claims 1-8, characterized in that, include: Data integration and standardization module: Integrates heterogeneous data from multiple sources, including power grid, field, and external environment, and generates a unified standard project dataset through cleaning, deduplication, imputation, and format conversion; Review Indicators and Model Management Module: Constructs a multi-level review indicator system covering technology, economy, environment, safety and compliance. The weights are determined based on the analytic hierarchy process and expert experience. It also integrates a large language model and a gradient boosting tree hybrid model to perform semantic understanding, indicator matching and automatic scoring of project data. Collaborative review and report generation module: Receives project materials and calls the hybrid review model to conduct preliminary review, generates review opinions, triggers manual intervention review process for critical projects, and generates formal review reports; Schedule planning and decomposition module: The critical path method is used to develop a baseline plan that includes five major construction phases. BIM is used for construction simulation and optimization, and the plan is decomposed into multi-level tasks. Progress monitoring and deviation analysis module: Collects progress data in real time, compares it with the baseline plan to calculate the deviation rate, analyzes the causes of deviations by combining historical and environmental data, and generates a progress deviation analysis report; Early warning response and control module: A three-level early warning mechanism is set up based on the degree of deviation. Adjustment measures are triggered for different levels. Special meetings are organized to re-optimize the baseline plan for major deviations, and early warning information is pushed in real time. Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.