Integrated power transformation engineering project full-process comprehensive management system

Abstract

The invention discloses an integrated power transformation engineering project full-process comprehensive management system, and relates to the technical field of electric power engineering management. The system comprises a unified data center platform, a full-life-cycle management module group, an intelligent analysis and decision support engine and a cooperative work interface. Unified storage and management of multi-source heterogeneous data are realized, all stages of planning, design, purchasing, construction and transfer operation and maintenance are covered, and seamless business connection and data linkage are realized; progress prediction, cost hyper-branched automatic traceability and security risk dynamic early warning are provided; a unified operation interface is provided for all participants; according to the invention, the problems of information island, process disjunction and risk early warning lag in a traditional management mode are effectively solved, the integrated, intelligent and visual management of the whole process of the project is realized, and the project management efficiency, the cost control capability and the safety risk prevention and control level are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of power engineering management technology, and specifically relates to an integrated management system for the entire process of substation engineering projects. Background Technology

[0002] Substation engineering projects are characterized by long construction periods, large investments, complex technologies, numerous stakeholders, and high safety risks. Traditional project management models typically rely on decentralized and independent information systems (such as independent CAD design software, budgeting software, and schedule management software) and extensive manual coordination, leading to poor information flow and the formation of information silos. Furthermore, traditional substation project management models suffer from inconsistent data standards across planning, design, procurement, construction, and operation and maintenance stages, making data sharing and traceability difficult and resulting in frequent discrepancies between drawings and models, and overestimation of costs. Different stakeholders (owners, designers, construction companies, supervisors, etc.) use different tools and platforms at different stages, resulting in high communication costs, inefficient process approvals, and slow change response. Risk assessment for project schedule, cost, safety, and quality relies heavily on post-project statistics and analysis, lacking intelligent means for pre-project prediction and in-process control, and management decisions lack data support. Valuable data, experience, and lessons learned during the project are not effectively accumulated and formed into a knowledge base for reuse in subsequent projects, leading to the recurrence of similar problems.

[0003] Existing project management software, such as traditional PMS systems, although covering some management aspects, mostly focus on a specific stage (such as the construction stage) or are simply information records. They lack the ability to deeply integrate and intelligently manage the entire project lifecycle and all elements, and therefore cannot fundamentally solve the above problems.

[0004] Existing technologies suffer from problems such as data fragmentation and inconsistency, process disconnect and difficulty in collaboration, delayed risk warning, and insufficient knowledge and experience transfer. Summary of the Invention

[0005] (a) Technical problems to be solved To address the problems in related technologies, this invention provides an integrated management system for the entire process of substation engineering projects, thereby overcoming the aforementioned technical problems existing in existing related technologies.

[0006] (II) Technical Solution To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: An integrated management system for the entire process of a substation engineering project is characterized by comprising a unified data hub platform, a full lifecycle management module group, an intelligent analysis and decision support engine, and a collaborative working interface. The unified data hub platform constructs a three-dimensional visual digital foundation through a BIM+GIS fusion model and stores data from the entire process of substation engineering projects in a structured manner; the standardized and interconnected project lifecycle data pool output by the unified data hub platform provides a unique and reliable data source for the integrated substation engineering project full-process comprehensive management system. The entire lifecycle management module group covers planning and operation and maintenance stages, relies on real-time data provided by the unified data hub platform, and is connected through the business process engine. After each module executes business logic, it feeds back the updated business status, triggered process tasks and newly generated business data to the unified data hub platform, forming a business data flow on the unified data hub platform. The intelligent analysis and decision support engine obtains real-time and historical data from the business data stream in the unified data hub platform, processes and analyzes it through machine learning models, and outputs actionable progress predictions, cost overrun analysis, and security risk warning data to obtain intelligent analysis results. The intelligent analysis results are pushed back to the unified data hub platform and directly serve the management and collaboration links in the collaborative work interface. The collaborative work interface obtains comprehensive information about substation engineering projects, including pending tasks, approval workflows, early warning messages, analysis reports, and 3D models, from the unified data hub platform, and pushes it to the corresponding users through a unified interface; it also feeds back the user's operation results to the unified data hub platform and business process engine to drive task flow.

[0007] Preferably, the unified data hub platform is used to store and manage all data throughout the entire project process; The unified data hub platform includes a BIM+GIS7 integrated data model, a project full-element database, and a data standardization and interface engine. The BIM+GIS integrated data model is used to integrate Building Information Modeling and Geographic Information System to construct a three-dimensional visual digital foundation that includes geographical environment, equipment assets, and spatial relationships; the project full-element database is used to structure and store project WBS decomposition structure, schedule, cost, contract, quality, safety, and human resources information; the data standardization and interface engine is used to define unified data standards and coding rules, and provide API interfaces for integrating and connecting with external heterogeneous systems to achieve automatic data collection and synchronization.

[0008] Preferably, the full lifecycle management module group includes an integrated planning and design module, an intelligent procurement and supply chain module, a refined construction management module, and a digital handover and operation and maintenance interface module; The integrated planning and design module supports scheme comparison based on historical data and standard library, realizes linkage between design and budget, and automatically triggers budget adjustment warning when design changes. The intelligent procurement and supply chain module works in conjunction with the integrated planning and design module to automatically generate a bill of materials and link it to the supplier management, electronic bidding, contract management and logistics tracking system. The refined construction management module integrates functions such as schedule planning, resource allocation, safety and quality inspection, and construction log reporting, and is associated with the BIM model to realize the visual management of construction progress on the three-dimensional model. Upon project completion, the digital handover and operation and maintenance interface module automatically packages the as-built BIM model, equipment data, and test reports to generate a standard-compliant digital asset package, which is then pushed to the operation and maintenance management system with one click.

[0009] Preferably, the intelligent analysis and decision support engine includes a schedule prediction model, a cost overrun analysis model, and a safety risk profiling module; and includes the following steps: S31. The progress prediction model analyzes historical project progress data and the actual progress of the current project, uses machine learning algorithms to predict the completion time of the critical path, and automatically issues an early warning when a delay risk is predicted. S32. The cost overrun analysis model monitors cost data such as contract payments and change orders in real time. When the actual cost of a certain sub-item of the project exceeds the budget threshold, the system automatically traces the associated design change orders, contract terms or purchase orders, generates a cost overrun analysis report, and indicates the main reasons. S33. The safety risk profiling module integrates safety violation records, personnel location hotspot maps, and environmental monitoring data to dynamically calculate the risk level of each construction area and highlight it in different colors in the three-dimensional model, sending automatic alarms to personnel entering high-risk areas.

[0010] Preferably, step S31 includes the following steps: S311. Construct a GRU neural network. Based on the characteristics of the substation project schedule data, improve the GRU neural network to obtain an initial GRU schedule prediction neural network. S312. Collect historical substation project progress data; the historical substation project progress data includes the daily planned completion percentage and daily actual completion percentage of each critical path task in the substation project, as well as external factor data; the external factor data includes weather data, resource data, supply chain data, and management data; The historical substation project progress data was cleaned, aligned, and decomposed according to time sequence to obtain optimized historical substation project progress data; S313. Optimize the historical substation project progress data into a training set and a validation set; use the Adam optimizer to iteratively train the initial GRU progress prediction neural network on the training set; after each epoch, calculate the loss of the current initial GRU progress prediction neural network on the validation set, monitor whether the model is overfitting, and stop training when the validation set loss no longer decreases, save the model parameters with the best performance, and obtain the GRU progress prediction neural network. S314. Input real-time project progress and real-time external factor data into the GRU progress prediction neural network to obtain predicted project progress data; if there is a risk of delay in the predicted project progress data, an early warning will be automatically issued.

[0011] Preferably, step S32 includes the following steps: S321. Retrieve the total budgeted cost and actual completion percentage of the task from the database; calculate the earned value of the specified task using the earned value calculation formula; S322. Summarize all relevant actual costs for the specified task from the Finance and Contracts module; Based on the earned value of a specified task and all relevant actual costs of the specified task, the cost deviation and cost deviation rate of the specified task are calculated using the cost deviation formula and the cost deviation rate calculation formula. S323. Set a cost deviation rate threshold; determine whether there is a cost overrun based on the cost deviation threshold and the cost deviation rate of the specified task. If the cost deviation of a specified task is less than or equal to the cost deviation rate threshold, the specified task is deemed to have exceeded its cost, and the source analysis process is automatically triggered. S323 uses data association to retrieve related design changes, related purchase orders, and related contracts and visas within a unified data hub platform, thereby obtaining cost overrun analysis results.

[0012] Preferably, step S33 includes the following steps: S331. Set a baseline value for the number of traffic violations; count the number of traffic violations in a specified area within a specified time; based on the number of traffic violations in the specified area within a specified time and the baseline value for the number of traffic violations, obtain the frequency of traffic violations in the specified area through the first quantification formula. S332. Obtain the real-time number of people and area of ​​a designated area through a personnel positioning system; based on the real-time number of people and area of ​​the designated area, obtain the personnel density of the designated area through a second quantification formula; S333. Acquire meteorological sensor data; derive environmental risk based on the meteorological sensor data and the third quantitative formula. S334. Based on the frequency of violations, population density, and environmental risks in a designated area, obtain the safety risk value of the designated area through a risk calculation formula; S335. Set security level rules; automatically classify the risk level of the specified area based on the security risk value of the specified area and the security level rules.

[0013] Preferably, step S331 includes the following steps: S3111, Improve the single progress completion rate input of the standard GRU neural network to a multivariate input that incorporates external influencing factors; S3112. Add an attention mechanism layer above the output layer of the standard GRU neural network; S3113. Improve the deterministic single-value output of the standard GRU neural network to be able to predict a probability distribution, thereby quantifying the uncertainty of the prediction; S3114. Explicitly remove the periodic components of the data at the front end of the standard GRU neural network, allowing the GRU to learn the nonlinear trends and the impact of sudden events in the residuals.

[0014] Preferably, the integrated substation engineering project full-process comprehensive management system also has a built-in business process engine for defining, executing and monitoring cross-departmental and cross-stage business processes, realizing the self-driving and flow of tasks.

[0015] Preferably, the collaborative work interface provides a unified web and mobile access point for all project participants, enabling task push, online approval, instant messaging, collaborative document editing, and meeting management.

[0016] (III) Beneficial Effects The present invention has the following beneficial effects: This invention breaks down the "information silos" phenomenon in traditional substation project management by constructing a unified data hub platform and a full lifecycle management module group. It achieves data connectivity and business collaboration throughout the entire process from planning, design, procurement, construction to handover and operation. The system utilizes unified data standards and interface engines to ensure the consistency and traceability of multi-source heterogeneous data, significantly reducing problems such as "inconsistent drawings and models" and "overestimation of budgets" caused by data fragmentation. At the same time, the collaborative work interface greatly improves the communication efficiency among all participants and the automation level of the approval process, effectively reducing project management costs and the risk of time delays.

[0017] The intelligent analysis and decision support engine built into this invention is based on machine learning algorithms, enabling real-time monitoring and proactive early warning of project progress, cost, and safety risks. The progress prediction model can accurately predict critical path delays, the cost overrun analysis model can automatically trace the root causes of overruns, and the safety risk profiling module dynamically assesses the risk level of construction areas, thereby transforming passive management into proactive control and enhancing the scientific and forward-looking nature of project management. It not only improves risk response capabilities but also provides reusable experience and data support for subsequent projects through digital handover and knowledge accumulation, fundamentally solving the problem of insufficient knowledge and experience transfer.

[0018] This invention constructs an efficient, transparent, and traceable substation engineering project management ecosystem through integrated, intelligent, and visual means. It not only improves the control accuracy and operational efficiency of the entire project process, but also provides solid technical support for the digital transformation of the power industry, resulting in significant economic and social benefits.

[0019] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, the drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the modules of an integrated substation engineering project full-process comprehensive management system according to the present invention. Detailed Implementation

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

[0023] In a provincial power grid company's 220kV Qinghe transmission and transformation project, the traditional management model employed involved: During the planning phase, independent feasibility study software was used to generate estimates; the design institute used CAD and independent cost estimation software to complete design drawings and cost estimates, and conducted drawing reviews with the owner via email and in-person meetings; the procurement department used a separate supply chain management system for bidding and procurement, with material lists manually entered from the design drawings, making them highly prone to errors; during the construction phase, the contractor used professional schedule planning software to manage progress, but progress data was reported weekly in report form, manually entered into the traditional project management system by the owner; and the supervision unit used a mobile app to report on-site quality and... Safety inspection records are stored in a separate cloud database; communication among project stakeholders relies mainly on regular meetings, emails, and instant messaging tools, while process approvals require printed documents for offline signatures; when a basic design change occurs during construction, the design institute sends out revised drawings via email, but the procurement department fails to receive the information in a timely manner, resulting in some ordered steel bars being of the wrong type, causing rework and waste; at the same time, the increased costs caused by the change require several weeks to complete all approvals, severely hindering cost control; the project manager only discovered when the construction team submitted its weekly report that the critical path had been delayed by a week due to severe weather and insufficient resources, and the opportunity for decision-making and corrective action had been missed. This demonstrates that existing technologies suffer from core problems such as data fragmentation, process disconnect, and delayed risk warning. To resolve the above issues, please refer to [link / reference]. Figure 1 This invention discloses an integrated management system for the entire process of substation engineering projects, comprising a unified data hub platform, a full lifecycle management module group, an intelligent analysis and decision support engine, and a collaborative working interface. The unified data hub platform constructs a three-dimensional visual digital foundation through a BIM+GIS fusion model and stores data from the entire process of substation engineering projects in a structured manner; the standardized and interconnected project lifecycle data pool output by the unified data hub platform provides a unique and reliable data source for the integrated substation engineering project full-process comprehensive management system. The entire lifecycle management module group covers planning and operation and maintenance stages, relies on real-time data provided by the unified data hub platform, and is connected through the business process engine. After each module executes business logic, it feeds back the updated business status, triggered process tasks and newly generated business data to the unified data hub platform, forming a business data flow on the unified data hub platform. The intelligent analysis and decision support engine obtains real-time and historical data from the business data stream in the unified data hub platform, processes and analyzes it through machine learning models, and outputs actionable progress predictions, cost overrun analysis, and security risk warning data to obtain intelligent analysis results. The intelligent analysis results are pushed back to the unified data hub platform and directly serve the management and collaboration links in the collaborative work interface. The collaborative work interface obtains comprehensive information about the substation engineering project from the unified data hub platform, including pending tasks, approval workflows, early warning messages, analysis reports, and 3D models, and pushes it to the corresponding users in a unified interface; it also feeds back the user's operation results to the unified data hub platform and business process engine to drive task flow. The aforementioned unified data hub platform is used to store and manage all data throughout the entire project process; The unified data hub platform includes a BIM+GIS integrated data model, a project full-element database, and a data standardization and interface engine. The BIM+GIS fusion data model is used to integrate building information modeling and geographic information system to construct a three-dimensional visual digital foundation that includes geographic environment, equipment assets, and spatial relationships; the project full-element database is used to structure and store project WBS decomposition structure, schedule, cost, contract, quality, safety, and human resources information; the data standardization and interface engine is used to define unified data standards and coding rules, and provide API interfaces for integrating and connecting with external heterogeneous systems to achieve automatic data collection and synchronization. The unified data hub platform in the above embodiments provides a unified platform for the entire process of substation engineering projects, from planning, design, construction to operation and maintenance, by constructing a BIM+GIS 3D digital foundation and a structured full-element database; it defines unified standards and interfaces, completely breaks down information silos between different stages and participants, and realizes automatic collection, centralized management and seamless synchronization of multi-source heterogeneous data, laying a solid data foundation for the visualization, collaboration and intelligent management of the entire project process; The aforementioned full lifecycle management module group includes an integrated planning and design module, an intelligent procurement and supply chain module, a refined construction management module, and a digital handover and operation and maintenance interface module; The integrated planning and design module supports scheme comparison based on historical data and standard library, realizes linkage between design and budget, and automatically triggers budget adjustment warning when design changes. The intelligent procurement and supply chain module works in conjunction with the integrated planning and design module to automatically generate a bill of materials and link it to the supplier management, electronic bidding, contract management and logistics tracking system. The refined construction management module integrates functions such as schedule planning, resource allocation, safety and quality inspection, and construction log reporting, and is associated with the BIM model to realize the visual management of construction progress on the three-dimensional model. Upon project completion, the digital handover and operation and maintenance interface module automatically packages the as-built BIM model, equipment data, and test reports to generate a standard-compliant digital asset package, which is then pushed to the operation and maintenance management system with one click. The full lifecycle management module group in the above embodiments covers all core stages from planning, design, procurement, construction to handover and operation and maintenance; the modules are deeply linked, realizing the seamless connection of the entire process from automatic generation of procurement lists from design models, real-time visualization management of construction progress and BIM models, to one-click generation of digital asset packages for seamless handover and operation and maintenance upon completion; it breaks down the stage barriers in the traditional management model, realizes the automatic flow and closed-loop management of business data, greatly improves the efficiency of cross-departmental collaboration, effectively controls the risk of cost overruns and information loss caused by changes and untimely handover, and provides core business support for the lean management and control of projects; The aforementioned intelligent analysis and decision support engine includes a schedule prediction model, a cost overrun analysis model, and a safety risk profiling module; it includes the following steps: S31. The progress prediction model analyzes historical project progress data and the actual progress of the current project, uses machine learning algorithms to predict the completion time of the critical path, and automatically issues an early warning when a delay risk is predicted. S32. The cost overrun analysis model monitors cost data such as contract payments and change orders in real time. When the actual cost of a certain sub-item of the project exceeds the budget threshold, the system automatically traces the associated design change orders, contract terms or purchase orders, generates a cost overrun analysis report, and indicates the main reasons. S33. The safety risk profiling module integrates safety violation records, personnel location hotspot maps, and environmental monitoring data to dynamically calculate the risk level of each construction area and highlight it in different colors in the three-dimensional model, and sends automatic alarms to personnel entering high-risk areas. The above S31 includes the following steps: S311. Construct a GRU neural network. Based on the characteristics of the substation project schedule data, improve the GRU neural network to obtain an initial GRU schedule prediction neural network. S312. Collect historical substation project progress data; the historical substation project progress data includes the daily planned completion percentage and daily actual completion percentage of each critical path task in the substation project, as well as external factor data; the external factor data includes weather data, resource data, supply chain data, and management data; The historical substation project progress data was cleaned, aligned, and decomposed according to time sequence to obtain optimized historical substation project progress data; S313. Optimize the historical substation project progress data into 80% / 20% training and validation sets; use the Adam optimizer to iteratively train the initial GRU progress prediction neural network on the training set; after each epoch, calculate the loss of the current initial GRU progress prediction neural network on the validation set, monitor whether the model is overfitting, and stop training when the validation set loss no longer decreases, save the model parameters with the best performance, and obtain the GRU progress prediction neural network. S314. Input real-time project progress and real-time external factor data into the GRU progress prediction neural network to obtain predicted project progress data; if there is a risk of delay in the predicted project progress data, an automatic warning will be issued. In practice, the historical substation project progress data in step S312 above needs to be improved in accuracy before being used to train the initial GRU progress prediction neural network, in order to ensure that the model trained has higher accuracy and better adaptability; specifically: By introducing multi-scale data augmentation techniques, virtual project sequences are constructed: direct historical project data is limited and may not meet project requirements; by using a combination of additive noise and scaling, the existing data can be reasonably expanded; for example, for a real task sequence, by adding small-amplitude random perturbations (simulating daily progress fluctuations) and scaling the overall completion rate (such as simulating a more or less efficient construction team), multiple new sequences that are statistically similar but not completely the same can be generated, fundamentally alleviating the data sparsity problem; Knowledge transfer is achieved by employing a pre-training strategy of cross-project transfer learning: each substation project is unique, but there are common underlying patterns, so a two-stage training method is adopted. The two-stage training method first uses non-critical path tasks from all historical projects (the amount of such data is much larger than that of the critical path) to pre-train the GRU network, allowing the model to learn the general pattern of how task progress evolves under the influence of external factors such as weather and resources. Then, the key parameters of this pre-trained model are used as initial values ​​and fine-tuned with critical path task data. This method ensures that the model does not learn from scratch, but is trained with prior knowledge, reducing the dependence on the amount of critical path data. By implementing domain-knowledge-based feature engineering and screening: to avoid the presence of junk data in the input data, the external factor data is deeply processed. Instead of simply inputting the raw data, the features of the raw data are combined to obtain composite features with engineering significance; for example, "daily rainfall" is transformed into "number of consecutive rainy days", and "design changes" are quantified as "the percentage of cost involved in the cumulative change orders within this week". Furthermore, a feature screening method based on mutual information is used to identify and eliminate noise factors with extremely low correlation to schedule deviations, ensuring that the data input to the model is highly purified and engineering-related, thereby improving the model's learning efficiency and accuracy. The weather data represents the daily rainfall, wind speed, and temperature corresponding to the project date, and can be obtained from the meteorological department interface; the resource data represents the ratio of the actual number of workers on site to the planned number of workers each day; the supply chain data represents the arrival status (arrived / not yet arrived) of key equipment and materials; the management data represents the issuance date of design change notices and records of delays in the approval process. In specific implementation, the optimized historical substation project progress data obtained in step S312 above is achieved through the following steps: using interpolation to fill in a few missing actual progress data, or using 0 to represent resources not in place; aligning all data by task and date to ensure that the input vector for each time step is complete; performing STL decomposition on the actual completion percentage sequence of each task to separate the trend component reflecting the long-term progress of the project, the seasonal component reflecting the weekly / quarterly construction rhythm (such as slow progress on weekends, slow progress during the rainy season), and the residual component containing random events and special effects after removing the trend and seasonal components; In practical implementation, before inputting real-time project progress and real-time external factor data into the GRU progress prediction neural network in step S314 above, a complete data acquisition and processing system needs to be constructed to ensure automatic, accurate, and real-time data acquisition; and to avoid prediction errors due to data distortion; specifically: A multi-source data automatic acquisition and fusion layer based on the Internet of Things and edge intelligence is constructed to provide real-time, high-fidelity input data for the GRU progress prediction neural network. The acquisition and fusion layer, as the front-end perception system of the model, can solve the problems of delay and distortion caused by manual data entry in noisy construction site environments. It includes a physical perception layer and a data parsing layer. At the physical sensing layer, various intelligent sensing devices are deployed. For daily actual completion percentages, manual reporting by foremen is no longer relied upon; instead, high-definition PTZ cameras are installed in key construction areas (such as the main control building and support structure areas), combined with daily scheduled drone patrols to automatically collect on-site image data. For resource data, smart safety helmets or positioning beacons are provided to construction workers to track the actual number of personnel in each work area in real time. IoT modules are installed on key machinery to monitor their operating time and status. For weather data, miniature weather stations are deployed directly at the construction site to obtain localized, precise microclimate data. For supply chain data, UWB or RFID tags are affixed to important equipment; when these arrive on site, the gateway automatically identifies and updates the arrival status. At the data parsing layer, edge computing and computer vision algorithms are introduced. On the edge computing server at the construction site, computer vision algorithms are run to analyze the transmitted image data in real time, automatically identify and quantify the completion status of engineering entities, such as: "Based on today's image recognition, the main transformer foundation formwork erection has been completed by about 80%", and this result is automatically converted into a structured daily actual completion percentage. This process does not require manual intervention and realizes end-to-end automation from "image" to "progress data". All multi-source data obtained through automatic sensing and identification is timestamped and formatted by the IoT platform and then pushed to the unified data hub platform of this system in real time. The GRU progress prediction neural network directly obtains cleaned and fused high-frequency (up to hourly) input feature vectors from this platform, thereby driving accurate and real-time prediction. The above embodiments, targeting substation project schedule management, creatively propose a schedule prediction method based on a multi-dimensional improved GRU neural network. By introducing external influencing factors that integrate weather, resource, supply chain, and management data as multivariate inputs, and integrating attention mechanisms, probability distribution outputs, and STL time series decomposition technology, an intelligent prediction model capable of accurately quantifying uncertainty is constructed. This model can dynamically perceive and learn the complex impact of various internal and external factors on construction progress, not only predicting the completion time of the critical path with high accuracy but also providing confidence intervals for the prediction results, achieving a leap from passive statistics to proactive early warning. This greatly enhances the ability to proactively perceive and refine schedule risks, providing project managers with a scientific basis for decision-making and effectively ensuring the timely progress of the project. The above S32 includes the following steps: S321. Retrieve the total budget cost P of the task from the database. V and actual completion percentage P C ; Calculate the earned value E of a specified task using the earned value calculation formula. V The earned value calculation formula is as follows: E V=P V ×P C ; S322. Summarize all relevant actual costs for a specified task from the Finance and Contracts module. C The actual costs include labor, materials, and equipment costs; Earned Value E based on specified tasks V And the actual costs A related to the specified task C The cost deviation C for a specified task is calculated using the cost deviation formula and the cost deviation rate calculation formula. V and cost deviation rate C VR The calculation formula is as follows: C V =E V -A C C VR =(C V / E V )×100%; S323, Set the cost deviation rate threshold T C For example, -10%; based on the cost deviation threshold T C Combined with the cost deviation rate C of the specified task VR Determine if the budget has been exceeded; When the cost deviation C of the specified task is... VR ≤ Cost Deviation Rate Threshold T C When the deviation rate is less than -10%, the cost of the specified task is determined to be over budget, and the source analysis process is automatically triggered. S323 uses data association relationships to search for related design changes, related purchase orders, and related contracts and approvals within a unified data hub platform to obtain cost overrun analysis results; the related design changes involve searching for all design change orders associated with the task's WBS code and calculating the cumulative cost increase ΔC caused by these change orders. d The associated purchase orders involve retrieving all purchase orders associated with the bill of materials for the specified task, comparing the order amount with the budget amount, identifying purchase items where (order unit price - budget unit price) × quantity > 0, and calculating their total excess cost ΔC. p The related contracts and visas refer to retrieving on-site visas or off-contract work approval forms related to the task, and calculating the additional cost ΔC incurred. s ; The cost overrun analysis results specifically include overrun tasks, overrun overview, root cause breakdown, and related documents; the overrun tasks are the task names and codes; the overrun overview is C... V With C VR Numerical values; the root cause decomposition includes design changes leading to cost overruns ΔC. d Yuan, the overspending rate is (ΔC) d / |C V |)×100%;The purchase price exceeded the standard, resulting in a cost overrun ΔC p Yuan, overspending percentage (ΔC) p / |C V |)×100%;On-site visa application resulted in cost overrun ΔC s Yuan, overspending percentage (ΔC) s / |C V |)×100%; The associated documents include a list of specific change order numbers and purchase order numbers that led to the overspending, allowing managers to quickly locate the problem; The cost overrun analysis model in the above embodiments is based on earned value management and monitors task cost performance in real time. Once an overrun is detected, the system automatically traces related design changes, purchase orders, and on-site approvals through a unified data hub platform, accurately quantifies the contribution of various factors to the overrun, and generates an analysis report containing specific root causes and related documents. This achieves an automated and intelligent leap from "discovering overruns" to "locating the root causes," greatly improving the accuracy and response speed of cost control, providing managers with direct and actionable decision-making basis, and effectively curbing the risk of cost runaway. The above S33 includes the following steps: S331. Set a baseline value N for the number of traffic violation incidents. b ; Count the number N of traffic violations in a specified area within a specified time period. v Based on the number N of traffic violations in a specified area within a specified time period. v And the baseline value N for the number of traffic violations b The frequency of traffic violations F in the specified area is obtained through the first quantization formula. V The first quantization formula is as follows: F V =min(100,(N v / N b ()×100); where N b This represents a baseline value for the number of traffic violations (e.g., 5 times). Exceeding this value is considered a perfect score (extremely high risk). The aforementioned violations refer to actions that violate established safety procedures, as recorded by the system or confirmed by safety management personnel, within a designated area and statistical period. Examples include not wearing a safety helmet, not wearing a safety belt while working at height, and unauthorized hot work. The baseline value for the number of violations is used for risk assessment, determined based on the average number of violations per unit area per unit time period calculated from historical project data, and comprehensively combined with industry standards and project safety management objectives. For example, N... b Setting it to 5 times means that when the number of violations in a region reaches this value, the region is considered to be in a state of extremely high violation frequency risk. S332. Obtain the real-time number of people N in a designated area through a personnel positioning system. pand area S a Based on the real-time number of people N in a specified area p and area S a The population density F of the specified area is obtained through the second quantification formula. P The second quantification formula is as follows: D=N p / S a D is transformed into a population density fraction F using a piecewise function. P For example, D < 0.1 people / m² gets 20 points, 0.1 ≤ D < 0.5 gets 50 points, and D ≥ 0.5 gets 100 points; the values ​​of 0.1 and 0.5 are based on historical experience. S333, Acquire meteorological sensor data (such as wind speed W) S Rainfall R F The environmental risk F is obtained by combining meteorological sensor data with the third quantitative formula. E The third quantification formula is as follows: Wind speed W S <10m / s gets 0 points, 10m / s≤W S <15m / s gets 60 points, W S ≥15m / s gets 100 points; Rainfall R F >50mm / h gets 100 points; F E Take the highest score among all environmental factors; S334, Violation frequency F based on a specified area V Personnel density F P Environmental risks F E The safety risk value R for a specified area is obtained through a risk calculation formula; the risk calculation formula, R=F1×F V +F2×F P +F3×F E ; Where F1, F2, and F3 represent the frequency of traffic violations, F V Personnel density F P Environmental risks F E The weights F1, F2, and F3 are set according to a comprehensive subjective and objective weighting method that combines the analytic hierarchy process (AHP) and the entropy weight method. First, the AHP is used to determine the subjective weight ranking among the factors. Then, based on historical project data, the entropy weight method is used to obtain the amount of objective information carried by each indicator data. Finally, the subjective and objective weights are weighted and integrated to ensure that the weight allocation is both in line with the management priorities and can dynamically respond to the actual data characteristics. S335. Set security level rules; automatically classify the risk level of the specified area based on the security risk value R of the specified area and the security level rules; the security level rules are as follows: low risk (green), 0≤R<35; medium risk (yellow), 35≤R<70; high risk (red), R≥70. The safety risk profiling module in the above embodiments constructs a dynamic quantitative assessment model by integrating multi-source data such as violation records, personnel location, and environmental monitoring. This model can calculate the comprehensive risk value of each construction area in real time and intuitively present the risk level in a three-dimensional model using green, yellow, and red colors. At the same time, it automatically sends alarms to personnel entering high-risk areas. This realizes a fundamental shift in safety management from post-event statistics to pre-event warnings and from passive response to proactive intervention, significantly improving the inherent safety level and refined management capabilities of construction sites.

[0024] The above S311 includes the following steps: S3111. The single progress completion rate input of the standard GRU neural network is improved to a multivariate input that incorporates external influencing factors. The improvement here is located after the input layer and before the GRU layer, which is equivalent to providing the model with richer contextual information. S3112. Add an attention mechanism layer above the output layer of the standard GRU neural network; the improvement here is located after the GRU layer and before the output layer; this makes the model prediction no longer treat all historical data equally, but can focus on key events and patterns. S3113. Improve the deterministic single-value output of the standard GRU neural network to predict a probability distribution, thereby quantifying the uncertainty of the prediction; the improvement here is located in the final output layer of the network; through the above improvement, not only can we obtain the prediction of "expected delay of 3 days", but also the confidence interval of "90% certainty that the delay is between 1 and 5 days", providing a deeper insight for risk management. S3114. Explicitly remove the periodic components of the data at the front end of the standard GRU neural network, allowing GRU to focus more on learning the nonlinear trends and impacts of sudden events in the residuals. This improvement is located at the data preprocessing stage and architecture design level of the entire neural network model. This improvement enables GRU to only process the more difficult-to-describe fluctuations after removing strong periodicity, thus improving the model's learning efficiency and accuracy. In specific implementation, step S3111 above is as follows: at each time step t, the input of the model is no longer just the historical progress value X. t Instead, it is an extended feature vector [X] t W t H t ,R t ]; The X tThis represents the time-series data of the task itself (such as daily planned completion percentage, actual completion percentage); the W t This represents the human resource input factor (such as the ratio of the number of workers on site that day to the planned number); the H... t This represents the human resource input factor (such as the ratio of the number of workers on site that day to the planned number); the R... t This indicates the resource availability factor (e.g., whether key materials have arrived; yes = 1, no = 0). Before being input into the GRU layer, a fully connected embedding layer is used to map the heterogeneous features (numerical and categorical) in the extended feature vector to a unified, dense vector space, making it more suitable for GRU processing. In specific implementation, step S3112 above is as follows: the standard GRU outputs a hidden state sequence containing information from all time steps; the model automatically identifies and focuses on the most critical historical time points for predicting future progress; the attention mechanism calculates a weight α for each historical time step i. i This weight represents the importance of the information at that time step to the current prediction; for example, the model will learn to assign higher attention weights to "the days last week when work was suspended due to heavy rain"; and the hidden states of all time steps will be weighted by α. i The summation yields a "context vector," which encapsulates the most crucial historical information and is then fed into the output layer for prediction. In specific implementation, step S3113 above specifically involves: modifying the output layer so that it no longer outputs only a predicted completion date, but instead outputs a Gaussian distribution parameter: the Gaussian distribution parameter is obtained through a Gaussian function; in this application, the mean μ in the Gaussian function represents the predicted completion time, and the variance σ² represents the prediction confidence level; the larger the variance, the higher the uncertainty (e.g., because there have been multiple unforeseen delays recently); during training, the loss function is changed to maximize the log-likelihood of the observed data under this prediction distribution; In specific implementation, step S3114 is as follows: In the data preprocessing stage, the STL (Seasonal-Trend Decomposition using Loess) method is used to decompose the original progress series into trend components, seasonal components, and residual components; the residual components and seasonal components are fed into the improved GRU model as the main inputs, and the trend components are fitted by linear regression; the final prediction result is the superposition of the GRU's prediction of the residuals, the extrapolation of the seasonal components, and the prediction of the trend components. The aforementioned integrated management system for the entire process of substation engineering projects also has a built-in business process engine, which is used to define, execute and monitor cross-departmental and cross-stage business processes to achieve self-driven and streamlined tasks. The aforementioned collaborative work interface provides all project participants with a unified web and mobile access point, enabling task push, online approval, instant messaging, collaborative document editing, and meeting management; The above embodiments drive the automatic flow of cross-departmental tasks through a built-in business process engine, and provide a unified operation entry point for all participants through a collaborative work interface, realizing the online and standardized approval, communication and collaboration, which greatly improves process efficiency and cross-organizational collaboration capabilities.

[0025] The following is Example 1 of an integrated substation engineering project full-process comprehensive management system based on the present invention; Example 1 is the same 220kV Qinghe transmission and transformation project; At the project's inception, the project was created in a unified data hub platform, and a digital base containing the site's topography and surrounding environment was constructed based on a BIM+GIS integrated data model. Designers directly called the standard component library for 3D design in the integrated planning and design module, with the model and budget linked in real time. After the design was completed, the intelligent procurement and supply chain module automatically extracted an accurate bill of materials from the BIM model and initiated the electronic bidding process. During the construction phase, all parties collaborated through a collaborative interface: the construction team reported progress daily on mobile devices, with data synchronized in real time; the refined construction management module linked actual progress with the BIM model, enabling visualized progress management; quality and safety issues discovered by the supervisor could be uploaded immediately and trigger rectification processes; throughout the process, the intelligent analysis and decision support engine continuously played its role; the progress prediction model, combining current progress, future weather forecasts, and resource availability, predicted a 3-day delay risk in the critical path of the main transformer installation 10 days in advance and issued an alert to the project manager; the project manager adjusted resources in a timely manner to avoid delays; simultaneously, the cost overrun analysis model detected... The cost deviation rate of the civil engineering sub-project has reached -12%. The system automatically traces the source and generates a report, clearly pointing out that the main cause of the cost overrun was a design change and a purchase order that exceeded the budget, providing precise targeting for cost control. The on-site safety risk profiling module detected two high-risk factors in area A: dense population and strong winds. It automatically marked the area in red in the 3D model and sent safety alarms to the handheld terminals of personnel in the area, effectively preventing potential accidents. Upon project completion, the system uses the digital handover and operation and maintenance interface module to generate a digital asset package of the as-built BIM model, all equipment information, and test data with one click, and then transfers it to the operation and maintenance system. The technical effect of this embodiment is that, through full-process integration and intelligent management and control, it has achieved a 5% reduction in project schedule, a cost saving of over 8%, a 60% reduction in security risk events, and a significant improvement in the efficiency and quality of digital asset transfer.

[0026] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0027] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. An integrated management system for the entire process of a substation engineering project, characterized in that, include: Unified Data Hub Platform: The unified data hub platform constructs a three-dimensional visual digital base through a BIM+GIS fusion model and stores the data of the entire process of substation engineering projects in a structured manner; The standardized and interconnected project lifecycle data pool output by the unified data hub platform provides a unique and reliable data source for the integrated substation engineering project full-process comprehensive management system. Full lifecycle management module group: The full lifecycle management module group covers all operational stages from planning to operation and maintenance, relies on real-time data provided by a unified data hub platform, and is connected through a business process engine; After each module executes its business logic, it reports the updated business status, triggered process tasks, and newly generated business data to the unified data hub platform, forming a business data flow on the unified data hub platform. Intelligent Analysis and Decision Support Engine: The intelligent analysis and decision support engine obtains real-time and historical data from the business data stream in the unified data hub platform, processes and analyzes it through machine learning models, and outputs actionable progress predictions, cost overrun analysis, and security risk warning data to obtain intelligent analysis results; the intelligent analysis results are pushed back to the unified data hub platform and directly serve the management and collaboration links in the collaborative work interface; Collaborative work interface: The collaborative work interface obtains comprehensive information about substation engineering projects, including pending tasks, approval workflows, early warning messages, analysis reports, and 3D models, from the unified data hub platform, and pushes it to the corresponding users in a unified interface; The user's operation results are then fed back to the unified data hub platform and business process engine to drive task flow.

2. The integrated substation engineering project full-process comprehensive management system according to claim 1, characterized in that, The unified data hub platform is used to store and manage all data throughout the entire project process; The unified data hub platform includes a BIM+GIS integrated data model, a project full-element database, and a data standardization and interface engine. The BIM+GIS integrated data model is used to integrate Building Information Modeling and Geographic Information System to construct a three-dimensional visual digital foundation that includes geographical environment, equipment assets, and spatial relationships; the project full-element database is used to structure and store project WBS decomposition structure, schedule, cost, contract, quality, safety, and human resources information; the data standardization and interface engine is used to define unified data standards and coding rules, and provide API interfaces for integrating and connecting with external heterogeneous systems to achieve automatic data collection and synchronization.

3. The integrated management system for the entire process of a substation engineering project according to claim 1, characterized in that, The full lifecycle management module group includes an integrated planning and design module, an intelligent procurement and supply chain module, a refined construction management module, and a digital handover and operation and maintenance interface module; The integrated planning and design module supports scheme comparison based on historical data and standard library, realizes linkage between design and budget, and automatically triggers budget adjustment warning when design changes. The intelligent procurement and supply chain module works in conjunction with the integrated planning and design module to automatically generate a bill of materials and link it to the supplier management, electronic bidding, contract management and logistics tracking system. The refined construction management module integrates functions such as schedule planning, resource allocation, safety and quality inspection, and construction log reporting, and is associated with the BIM model to realize the visual management of construction progress on the three-dimensional model. Upon project completion, the digital handover and operation and maintenance interface module automatically packages the as-built BIM model, equipment data, and test reports to generate a standard-compliant digital asset package, which is then pushed to the operation and maintenance management system with one click.

4. The integrated substation engineering project full-process comprehensive management system according to claim 1, characterized in that, The system according to claim 1, characterized in that the intelligent analysis and decision support engine includes a schedule prediction model, a cost overrun analysis model, and a safety risk profiling module; and includes the following steps: S31. The progress prediction model analyzes historical project progress data and the actual progress of the current project, uses machine learning algorithms to predict the completion time of the critical path, and automatically issues an early warning when a delay risk is predicted. S32. The cost overrun analysis model monitors cost data such as contract payments and change orders in real time. When the actual cost of a certain sub-item of the project exceeds the budget threshold, the system automatically traces the associated design change orders, contract terms or purchase orders, generates a cost overrun analysis report, and indicates the main reasons. S33. The safety risk profiling module integrates safety violation records, personnel location hotspot maps, and environmental monitoring data to dynamically calculate the risk level of each construction area and highlight it in different colors in the three-dimensional model, and sends automatic alarms to personnel entering high-risk areas.

5. The integrated substation engineering project full-process comprehensive management system according to claim 4, characterized in that, S31 includes the following steps: S311. Construct a GRU neural network. Based on the characteristics of the substation project schedule data, improve the GRU neural network to obtain an initial GRU schedule prediction neural network. S312. Collect historical substation project progress data; the historical substation project progress data includes the daily planned completion percentage and daily actual completion percentage of each critical path task in the substation project, as well as external factor data; the external factor data includes weather data, resource data, supply chain data, and management data; The historical substation project progress data was cleaned, aligned, and decomposed according to time sequence to obtain optimized historical substation project progress data; S313. Optimize the historical substation project progress data into a training set and a validation set; use the Adam optimizer to iteratively train the initial GRU progress prediction neural network on the training set; after each epoch, calculate the loss of the current initial GRU progress prediction neural network on the validation set, monitor whether the model is overfitting, and stop training when the validation set loss no longer decreases, save the model parameters with the best performance, and obtain the GRU progress prediction neural network. S314. Input real-time project progress and real-time external factor data into the GRU progress prediction neural network to obtain predicted project progress data; if there is a risk of delay in the predicted project progress data, an early warning will be automatically issued.

6. The integrated substation engineering project full-process comprehensive management system according to claim 4, characterized in that, S32 includes the following steps: S321. Retrieve the total budgeted cost and actual completion percentage of the task from the database; calculate the earned value of the specified task using the earned value calculation formula; S322. Summarize all relevant actual costs for the specified task from the Finance and Contracts module; Based on the earned value of a specified task and all relevant actual costs of the specified task, the cost deviation and cost deviation rate of the specified task are calculated using the cost deviation formula and the cost deviation rate calculation formula. S323. Set a cost deviation rate threshold; determine whether there is a cost overrun based on the cost deviation threshold and the cost deviation rate of the specified task. If the cost deviation of a specified task is less than or equal to the cost deviation rate threshold, the specified task is deemed to have exceeded its cost, and the source analysis process is automatically triggered. S323 uses data association to retrieve related design changes, related purchase orders, and related contracts and visas within a unified data hub platform, thereby obtaining cost overrun analysis results.

7. The integrated substation engineering project full-process comprehensive management system according to claim 4, characterized in that, S33 includes the following steps: S331. Set a baseline value for the number of traffic violations; count the number of traffic violations in a specified area within a specified time; based on the number of traffic violations in the specified area within a specified time and the baseline value for the number of traffic violations, obtain the frequency of traffic violations in the specified area through the first quantification formula. S332. Obtain the real-time number of people and area of ​​a designated area through a personnel positioning system; based on the real-time number of people and area of ​​the designated area, obtain the personnel density of the designated area through a second quantification formula; S333. Acquire meteorological sensor data; derive environmental risk based on the meteorological sensor data and the third quantitative formula. S334. Based on the frequency of violations, population density, and environmental risks in a designated area, obtain the safety risk value of the designated area through a risk calculation formula; S335. Set security level rules; automatically classify the risk level of the specified area based on the security risk value of the specified area and the security level rules.

8. The integrated substation engineering project full-process comprehensive management system according to claim 4, characterized in that, S331 includes the following steps: S3111, Improve the single progress completion rate input of the standard GRU neural network to a multivariate input that incorporates external influencing factors; S3112. Add an attention mechanism layer above the output layer of the standard GRU neural network; S3113. Improve the deterministic single-value output of the standard GRU neural network to be able to predict a probability distribution, thereby quantifying the uncertainty of the prediction; S3114. Explicitly remove the periodic components of the data at the front end of the standard GRU neural network, allowing the GRU to learn the nonlinear trends and the impact of sudden events in the residuals.

9. The integrated management system for the entire process of a substation engineering project according to claim 1, characterized in that, The integrated substation engineering project full-process comprehensive management system also has a built-in business process engine, which is used to define, execute and monitor cross-departmental and cross-stage business processes to achieve self-driven and streamlined tasks.

10. The integrated substation engineering project full-process comprehensive management system according to claim 1, characterized in that, The collaborative work interface provides a unified web and mobile access point for all project participants, enabling task push, online approval, instant messaging, collaborative document editing, and meeting management.

Images