Electric power engineering design project management system and method

Abstract

The invention belongs to the technical field of computer-aided project management, and discloses an electric power engineering design project management system and method, and the system comprises a data integration and processing module, a multilayer dynamic knowledge graph data storage library, a graph-based automatic compliance check engine, and a graph neural network-based predictive risk engine. And an active design change impact analysis module. According to the method, a unified and intelligent project data core is constructed, heterogeneous data sources such as design files and law and regulation texts are integrated in a multi-layer dynamic knowledge graph, automatic compliance review is realized by utilizing graph query, data-driven risk prediction is realized by utilizing a graph neural network GNN, and the risk prediction efficiency is improved. And the linkage influence of design change can be simulated and quantitatively evaluated. According to the method, the problems of data islanding, passive compliance examination, subjective risk management, difficulty in change analysis and the like in the prior art are solved, and the management efficiency, accuracy and initiative of the electric power engineering design project can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided project management, and in particular to a power engineering design project management system and method. Background Art

[0002] Currently, project management in the power engineering design industry primarily relies on traditional business process management (BPM) platforms, simple document management systems, or standalone building information modeling (BIM) tools. While these systems have achieved paperless office work and digitized processes to a certain extent, they suffer from several fundamental technical flaws and are no longer able to meet the deep intelligent management requirements of modern, complex power engineering projects.

[0003] Existing systems manage design documents, such as CAD drawings, BIM models, technical specifications, project tasks, and regulatory standards, as opaque data objects. These systems are unable to deeply understand the rich semantics within design documents, nor can they identify and model the complex, multi-dimensional relationships between design components, regulatory constraints, project tasks, team members, and cost budgets.

[0004] For example, while traditional Building Information Modeling (BIM) systems can represent the geometric information and physical relationships of building components, they were not designed for full project lifecycle management, making it difficult to seamlessly integrate them with non-geometric project management information such as task dependencies, resource allocation, budget consumption, and compliance clauses. This fragmented information leads to inefficient cross-domain information retrieval. For example, engineers cannot simply query "all design components affected by a specific regulation and responsible for Team A." Instead, they must manually search and link across multiple independent systems, a cumbersome and error-prone process. Ultimately, this makes it difficult to accumulate and reuse project knowledge.

[0005] Power engineering design must adhere to numerous and complex national, industry, and local standards and regulations. Current practices require compliance review to be a time-consuming, error-prone, and manual process. Domain experts must manually review vast amounts of design drawings and technical documentation, comparing design parameters against regulatory requirements. This process is not only inefficient but also prone to errors and omissions due to human oversight. Discovery of non-compliance issues later in the project or during construction can lead to costly design rework and construction delays.

[0006] Although some automated rule-checking (ARC) systems already exist, most of them are “black box” tools that are independent of the project management process. Their rule bases are usually hard-coded, difficult to update, and difficult to adapt to frequent changes in regulations. They are also unable to conduct linkage analysis with other elements in the entire project life cycle, such as cost and schedule.

[0007] Current risk management relies heavily on project managers' personal experience and qualitative assessments, lacking objective, quantitative data support. Automated, data-driven forecasting mechanisms are lacking for potential risks arising from complex dependencies within projects. While some modern project management tools, such as Wrike and Asana, have begun to incorporate AI for risk prediction, their predictive capabilities are limited by the data dimensions used, primarily relying on project management metadata such as task completion rates, overdue deadlines, and start and end dates.

[0008] These systems can identify the "symptoms" of project delays, but they can't delve into their "root causes." This is especially true when risks stem from underlying technical or regulatory issues, such as a key piece of equipment not meeting the latest environmental standards, or a seemingly unrelated upstream design change indirectly leading to downstream construction risks. This presents a critical gap in existing technology.

[0009] Design changes are inevitable in complex engineering projects. A single, local modification often ripples through the design, triggering a series of chain reactions—known as change propagation—on other related components, project schedules, cost budgets, and even regulatory compliance. Analyzing these chain reactions is a complex, time-consuming, and manual task with incomplete assessments. This prevents decision makers from foreseeing all potential consequences, leading to suboptimal or even incorrect decisions.

[0010] Existing change impact analysis methods are usually isolated and only focus on the product structure itself, but fail to cover the entire project ecosystem. They are unable to answer complex questions such as "Replacing model A transformer with model B will have what specific impact on the progress of downstream tasks, compliance with relevant regulations, and total project costs?"

[0011] In summary, the fundamental bottleneck of existing technology lies in the lack of an intelligent data core that can express and connect the entire project ecosystem—from the physical attributes of a transformer, a regulatory constraint, to a task deadline, and a budget amount—in a unified, machine-understandable format. Therefore, a new technical solution is urgently needed to overcome these shortcomings. Summary of the Invention

[0012] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a power engineering design project management system and method. By constructing a unified project knowledge model and utilizing artificial intelligence technology, it realizes automated review of project compliance, proactive prediction of potential risks, and quantitative analysis of the impact of design changes, so as to solve the problems raised in the above-mentioned background technology.

[0013] To achieve the above object, the present invention adopts the following technical solutions: A power engineering design project management system comprises: a data integration and processing module for extracting information from heterogeneous data sources including design documents, regulatory texts and project plans; a multi-layer dynamic knowledge graph data repository for semantically representing project-related entities and their multidimensional relationships in the form of nodes and edges; a graph-based automated compliance checking engine configured to convert design specifications and standards into a graph query language and execute queries on the knowledge graph to automatically verify the compliance of design solutions; a graph neural network-based predictive risk engine configured to predict potential risk events of the project based on structural patterns and features learned from the knowledge graph; and a proactive design change impact analysis module configured to respond to design change requests, simulate changes in the knowledge graph, and quantitatively evaluate the chain reaction effects of the changes on the project.

[0014] Preferably, the multi-layer dynamic knowledge graph data repository includes: a static knowledge layer for storing relatively stable and reusable basic knowledge entities, such as regulatory standards, component templates and material properties; a project instance layer for storing specific entity instances in specific projects, such as specific projects, design documents, component instances and tasks, and establishing associations with corresponding templates in the static knowledge layer; and a dynamic state and risk layer for storing the time-varying state of project entities and the predictive risk information generated by the predictive risk engine.

[0015] Preferably, the data integration and processing module includes a natural language processing submodule for automatically extracting entities, attributes and rules from unstructured regulatory standards documents and storing them in a structured manner into the static knowledge layer.

[0016] Preferably, the system also includes a proactive design change impact analysis module, which is configured to: create a temporary "hypothetical" analysis branch in the knowledge graph in response to a design change request proposed by a user; simulate the design change in the branch, and propagate the change and identify all affected entities by traversing the dependency edges in the knowledge graph; call the compliance check engine and the predictive risk engine to re-evaluate on the "hypothetical" analysis branch to quantify the impact of the change on compliance, risk, cost and schedule.

[0017] Preferably, the predictive risk engine based on graph neural network adopts the relational graph convolutional network R-GCN model.

[0018] The present invention also provides a corresponding power engineering design project management method, comprising the following steps: constructing and dynamically maintaining a multi-layer dynamic knowledge graph, which semantically represents project entities and their relationships by integrating information from heterogeneous data sources such as design documents, regulatory texts, and project plans; automatically and continuously checking the compliance of design schemes by converting design specifications and standards into graph queries and executing the queries on the knowledge graph; using a graph neural network model trained on historical and real-time project knowledge graph data to predict future project risks by analyzing the structural patterns of the graph; and responding to proposed design changes, simulating changes in the knowledge graph and propagating them along dependency relationships to identify all affected entities and quantitatively assess the chain effects of the changes.

[0019] Preferably, the method further includes the following steps: in response to the proposed design change, creating a temporary "what-if" analysis branch in the knowledge graph; simulating the change in the branch and propagating it along the dependencies to identify all affected entities; re-performing compliance checks and risk predictions on the "what-if" analysis branch, and summarizing the impact of the change on cost and schedule to generate a quantitative impact analysis report.

[0020] Compared with the existing technology, the present invention has the following beneficial effects: the present invention liberates project personnel from a large amount of repetitive and tedious manual work through automated compliance review and design change impact analysis, greatly shortens the design and decision-making cycle, discovers and corrects design defects early, and avoids expensive rework costs in the later stage; the present invention is based on data-driven risk prediction and fully automatic compliance checking, which minimizes errors caused by human negligence and subjective judgment, ensures that the design results strictly comply with various regulations and standards, and significantly improves the overall quality of the project; the present invention transforms the project management model from the traditional "passive response to problems" to "active prediction and risk avoidance", enabling project managers to foresee and respond to potential challenges in advance, thereby improving the controllability and success rate of the project; the present invention provides project managers with quantitative and explainable decision-making basis, whether it is approving a design change or allocating resources to deal with high-risk tasks, there is solid data support, making the decision-making process more scientific, transparent and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some illustrative embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0022] Figure 1This is a system structure diagram of a power engineering design project management system and method according to the present invention; Figure 2 A flowchart of a power engineering design project management system and method according to the present invention; Figure 3 This is a schematic diagram of the structure of a multi-layer dynamic knowledge graph in a power engineering design project management system and method of the present invention; Figure 4 A schematic diagram of a process for automated compliance checking in a power engineering design project management system and method according to the present invention; Figure 5 A schematic diagram of a process for predictive risk assessment in a power engineering design project management system and method according to the present invention; Figure 6 The present invention provides a flow chart of design change impact analysis in a power engineering design project management system and method. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The terms "module", "engine", "repository" and so on used in this application mainly include computer-related entities, such as but not limited to hardware, firmware, a combination of hardware and software, software or software in execution.

[0024] Example 1 See also Figure 1 This embodiment provides a power engineering design project management system, which mainly builds an integrated and intelligent platform to overcome the limitations described in the background technology.

[0025] The system specifically includes: a data integration and processing module 101, a multi-layer dynamic knowledge graph data repository 102, a graph-based automated compliance checking engine 103, a predictive risk engine based on graph neural networks 104, a proactive design change impact analysis module 105, and a user interface and visualization module 106.

[0026] Data integration and processing module 101, which serves as the input portal of the system and is the data foundation for building the knowledge graph, is responsible for collecting, parsing, cleaning, and converting information from diverse heterogeneous data sources, including but not limited to: Design files, such as AutoCAD DWG files, Revit BIM models (RVT files), various technical specifications and equipment lists, contain rich geometric information and non-geometric attribute information.

[0027] Project management documents, such as MS Project project plans, Excel cost estimates, and Word meeting minutes and reports, define the project's scope, schedule, cost, and communication records.

[0028] Regulatory and standard documents, such as national standards GB, industry specifications DL / T, local regulations, etc., in PDF, Word, or HTML formats, are the cornerstone of project compliance.

[0029] Real-time business data, such as material prices and personnel availability, comes from real-time data streams from the enterprise ERP system, human resources system, or supplier management system.

[0030] The core of this module lies in its intelligent parsing capability, which includes two key sub-modules: The CAD / BIM parser submodule embeds parsers for specific file formats. For example, for DWG files, it can parse DXF data, identify CAD layers, blocks, line types, and text, and extract transformer symbols and their associated text blocks as attributes such as model, rated voltage, and capacity, while also analyzing the spatial relationship between them.

[0031] For BIM models, it is possible to directly extract component objects and their rich parameter information through API, including identity data, constraints, phasing information, etc.

[0032] The Natural Language Processing (NLP) submodule is responsible for deeply analyzing unstructured and semi-structured regulatory standard texts to achieve the transformation from text to knowledge. Its specific processing flow may include: First, the text is preprocessed, including removing non-text content and sentence segmentation; Secondly, we use pre-trained language models based on the Transformer architecture, such as BERT, to perform named entity recognition (NER) to accurately identify key technical entities and constraint concepts such as "transformer," "safety clearance," and "fire resistance rating" in the text. Finally, using relation extraction technology, we identify the relationships and constraints between these entities by matching predefined rule templates or training classification models. For example, from the sentence "The minimum clearance between the main transformer and the wall in a 110kV substation should be no less than 5 meters," we can extract a structured rule triple containing {subject: main transformer, constraint: minimum clearance, object: wall, value: 5, unit: meter}.

[0033] In this way, the system can automatically convert human-readable natural language rules into machine-understandable and executable computer instructions, laying a solid foundation for subsequent automated compliance checks.

[0034] The multi-layer dynamic knowledge graph data repository 102 is the core data structure of the present invention. It uses a graph database, such as Neo4j or JanusGraph, as a physical carrier and stores data in a graph structure, i.e., nodes and edges, which fundamentally overcomes the rigidity and data island problems of traditional relational databases.

[0035] All project-related entities, such as equipment, documents, tasks, personnel, and regulations, are represented as "nodes", and the relationships between them, such as "includes", "depends on", "responsible person is", and "constrained by", are represented as "edges" with clear semantics.

[0036] See also Figure 3 , this paper constructs a clear, efficient and scalable knowledge graph by stratifying knowledge according to its change frequency and abstraction level. It specifically includes three logical levels: The static knowledge layer 301, or the rules and template library, stores relatively stable, cross-project reusable basic knowledge and domain ontologies, forming the "worldview" and "rulebook" for power engineering design. Its node types primarily include Regulation, Standard, Component Template, and Material.

[0037] Its edge types primarily describe the inherent relationships between these basic entities, such as "isA" for "is a kind of" and "hasProperty" for "has a property." Data at this layer is automatically parsed and extracted from regulations and standards documents through the aforementioned NLP submodule, or imported from existing structured data sources such as the company's standard parts library and material library.

[0038] The project instance layer 302, or dynamic project snapshot, describes a specific, ongoing project. The nodes in this layer are instantiations of the templates and concepts in the first layer, forming a complete, multi-dimensional digital snapshot of a specific project at a specific point in time.

[0039] Its node types primarily include Project, Design Document, Component Instance, Task, and Person. Its edge types are extremely diverse, allowing it to weave a complex project network. For example, the instanceOf edge represents "is an instance of..." and connects to the first-layer template node; the containsComponent edge represents "contains a component"; the assignedTo edge represents "assigned to"; the dependsOn edge represents "depends on"; and the governedBy edge represents "governed by..." and connects to the first-layer regulation node.

[0040] The data in this layer is generated dynamically. For example, when a designer draws a transformer in CAD, the system creates a component instance node and links it to the corresponding component template in the static knowledge layer through the instanceOf edge.

[0041] The dynamic state and risk layer 303, namely the prediction and insight layer, is a dynamic overlay layer built on top of the first two layers. It is used to capture the time-varying state of the project and the predictive insights generated by the intelligent engine, adding the dimensions of time and probability to the project.

[0042] Its node types primarily include statuses (e.g., "Completed," "In Progress," and "Overdue") and risk events (e.g., "Schedule Delay Risk," "Compliance Violation Risk," and "Cost Overrun Risk." Its edge types are primarily used to associate these statuses and risks with entities at the project instance level. For example, the edge hasStatus represents "has status," the edge hasRisk represents "exists risk," and the edge hasProbability represents "has probability."

[0043] The risk event nodes and their associated hasProbability edges are dynamically generated or updated by the GNN-based predictive risk engine 104 after performing real-time reasoning.

[0044] Table 1 below provides examples of knowledge graph nodes and edges to illustrate their structure and value more specifically.

[0045] Table 1: Example of knowledge graph node and edge definitions The graph-based automated compliance checking engine 103 is responsible for performing compliance reviews of designs. It transforms compliance checks into direct queries on the knowledge graph, achieving a transition from manual review to automated execution. For example, for the rule "the distance between component A and component B must be greater than 2 meters," the engine automatically translates it into a graph query statement, such as SPARQL or Cypher. It locates the nodes representing components A and B in the knowledge graph, obtains their location attributes, calculates the distance in real time, and compares it with the constraint value stored in the corresponding regulatory node.

[0046] This approach allows compliance checks to be triggered automatically and in real time with any minor changes to the design, greatly improving the efficiency and accuracy of the review.

[0047] A predictive risk engine 104 based on graph neural networks is the key to achieving proactive management. It uses graph neural networks (GNNs) to learn directly from the complex structure of knowledge graphs.

[0048] In this embodiment, the Relational Graph Convolutional Network (R-GCN) model is preferred. Its core advantage lies in its ability to effectively handle heterogeneous graphs in the present invention's knowledge graph, which contain multiple relationship types, such as dependsOn, assignedTo, governedBy, and so on. Compared to the standard GNN model, R-GCN can more accurately aggregate information from different types of neighboring nodes by learning different weight matrices for each relationship type, thereby providing a deeper understanding of the complex dependencies between projects.

[0049] The engine takes as input a subgraph representing a project, task, or component. By learning patterns from a large number of historical project graphs, it predicts the probability of various risks, such as schedule delays and cost overruns. Because its input is a knowledge graph that includes multi-dimensional information such as technology, management, and regulations, its predictive capabilities far exceed existing technologies that rely solely on surface-level project metadata.

[0050] The proactive design change impact analysis module 105 provides powerful decision support for project decision makers. When a user proposes a change proposal, such as "changing the main transformer model from T1 to T2," the module initiates a transactional "what-if" analysis. It creates a temporary, isolated analysis branch in the knowledge graph, applies the change to this branch, and then uses a graph traversal algorithm to rapidly propagate the change along various dependency edges, such as dependsOn, affects, and governedBy, to identify all affected entities.

[0051] Finally, it will call the compliance engine 103 and the risk engine 104 to re-evaluate this "hypothesis" branch, summarize the impact on cost and progress, and generate a comprehensive quantitative report that clearly shows the pros and cons of the change.

[0052] The user interface and visualization module 106 serves as the interface between the system and the user, providing intuitive dashboards, customizable reports, and an interactive graphical visualization interface. This module allows users to monitor project status, view risk alerts, and explore the impact of design changes, making complex project data intuitive and easy to understand.

[0053] Example 2 See also Figure 2 This embodiment provides a method for managing power engineering design projects, which mainly realizes efficient management of power engineering design projects through a series of automated and intelligent steps. The method specifically includes: Step 1: Build and dynamically maintain a multi-layer dynamic knowledge graph. This is the foundational step. Through the data integration and processing module 101, information is continuously extracted from heterogeneous data sources such as design documents, regulatory texts, and project plans. This information is then populated into the multi-layer dynamic knowledge graph data repository 102, representing project entities and their multidimensional relationships in a semantic manner. This process is ongoing, ensuring that the knowledge graph reflects the latest project status in real time.

[0054] Step 2: Execute automated compliance checks. This step mainly transforms compliance review into a continuous and automated assurance mechanism embedded in the design process. Figure 4 , this step specifically includes: Step 401, rule interpretation and storage, is a prerequisite step. Through the NLP sub-module in the data integration and processing module 101, regulations and standard documents are systematically parsed to convert unstructured text rules such as "...the minimum distance should be..." into structured rule entities in the static knowledge layer 301 of the knowledge graph.

[0055] Step 402: Triggering a check. When the system detects a specific change in the project instance layer 302, such as an update to a design document node or a modification to a component instance node's attributes, the check process is automatically triggered. A full scan can also be configured to occur periodically, such as daily.

[0056] Step 403: Query Generation: After receiving the trigger signal, the graph-based automated compliance checking engine 103 identifies entities related to the change and retrieves all rule nodes that constrain these entities. Each applicable rule is then dynamically translated into a graph query statement.

[0057] Step 404: Graph execution and comparison: the generated graph query statement is executed on the project instance layer 302, and the actual parameters of the design instance are obtained through pattern matching and path traversal, and compared with the standard values ​​in the rules.

[0058] Step 405, result analysis and reporting. For cases where the inspection fails, the system creates or updates a "Compliance Violation" risk node in the dynamic status and risk layer 303, and sends an alarm notification to the relevant person in charge through the user interface and visualization module 106. The notification may include details of the violation and recommended modification plans.

[0059] Step 3: Perform predictive risk assessment based on GNN. This step uses the rich structural information of the knowledge graph to achieve in-depth prediction of project risks. Figure 5 , this step specifically includes: Step 501: Graph Feature Engineering. For a target entity requiring risk assessment, such as a Task node, the system first extracts its local subgraph from the knowledge graph. This subgraph, for example, includes its predecessor and successor tasks, responsible individuals, and required equipment. It then extracts rich feature vectors from this subgraph. These features include not only the node's own attributes, such as task duration and budget, but more importantly, the structural information of its neighborhood, such as the type and number of neighboring nodes and the types of connected edges.

[0060] Step 502: GNN model training. This is an offline process. The system uses the enterprise's accumulated historical project knowledge graph data to train the R-GCN model. The training data consists of subgraphs from historical projects and corresponding known outcomes, such as project delays and cost overruns. The model uses backpropagation and gradient descent algorithms to learn to associate specific graph patterns with risk outcomes.

[0061] Step 503: Real-time inference. The trained GNN model is deployed online. When the knowledge graph of the real-time project is updated, the predictive risk engine 104 calls the GNN model to perform forward propagation calculations on the affected entities and their local subgraphs.

[0062] Step 504: Risk Generation and Interpretation. The output of model reasoning is probability values ​​for different risk types. When a risk probability exceeds a preset threshold, the engine creates or updates a Risk_Event node in the dynamic state and risk layer 303. To enhance interpretability, XAI (Explainable Artificial Intelligence) technologies such as SHAP (SHapley Additive Explanations) can be integrated to highlight the key risk factors that contribute most to risk prediction. For example, a delay in a predecessor task is the primary cause of the current task's risk spike, helping project managers understand the root causes of the risk.

[0063] Step 4: Perform proactive design change impact analysis, which provides decision makers with tools to comprehensively, quickly, and quantitatively assess the impact of changes before they are implemented. Figure 6 , this step specifically includes: Step 601 , initiating a change scenario: the user inputs a proposed change through the user interface and visualization module 106 , such as “replace transformer type A with transformer type B”.

[0064] Step 602: Create a "what-if" analysis branch. The system creates a transactional copy of the local subgraph related to the change, the "what-if" branch, in memory or in a temporary database instance. This operation ensures sandboxing of the analysis without affecting the actual project data.

[0065] Step 603: Propagate changes. The system executes the changes proposed by the user in the "hypothesis" branch, such as modifying the instanceOf edge of the component instance node to point to the new component template, and starts a graph traversal algorithm, such as breadth-first search or depth-first search, starting from the change point and propagating along various dependency edges predefined in the knowledge graph, such as dependsOn, affects, etc., to recursively find all directly and indirectly affected nodes.

[0066] Step 604: Reassessment and Quantification. After identifying all affected entities, the proactive design change impact analysis module 105 invokes the compliance checking engine 103 and the predictive risk engine 104 to perform a comprehensive reassessment on the "what-if" branch. Simultaneously, the quantified impact of the change on the total project cost and duration is calculated by summarizing the cost changes of the affected components and the duration changes of the affected tasks.

[0067] Step 605: Generate an impact analysis report. Finally, the system summarizes all analysis results into a comprehensive, intuitive impact analysis report. This report clearly lists the affected equipment, potential new compliance issues, changes in risk probability, and the specific, quantified impact on cost and schedule (for example, "cost increased by 50,000 yuan, construction extended by 3 days"), providing project management with data support for informed decision-making.

[0068] Finally, it should be understood that 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 spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A power engineering design project management system, characterized in that: include: one or more processors; A memory storing instructions executable by the one or more processors, wherein when the instructions are executed, the processors implement the functions of the following modules: a data integration and processing module for extracting information from the heterogeneous data sources including design documents, regulatory texts, and project plans; A multi-layer dynamic knowledge graph data repository for receiving information extracted by the data integration and processing module and semantically representing project-related entities and their multi-dimensional relationships in the form of nodes and edges; a graph-based automated compliance checking engine configured to convert design specifications and standards into a graph query language and execute queries on the knowledge graph to automatically verify the compliance of design solutions; A predictive risk engine based on a graph neural network, configured to predict potential risk events of a project based on structural patterns and features learned from the knowledge graph; as well as The proactive design change impact analysis module is configured to respond to design change requests, simulate changes in the knowledge graph, and quantitatively evaluate the chain effects of the changes on the project.

2. The power engineering design project management system according to claim 1, characterized in that: The multi-layer dynamic knowledge graph data repository includes: The static knowledge layer is used to store relatively stable and reusable basic knowledge entities, including regulatory standards, component templates, and material properties; A project instance layer, which is used to store specific entity instances in a specific project, including specific projects, design documents, component instances, and tasks, and to establish associations with corresponding templates in the static knowledge layer; and The dynamic state and risk layer is used to store the time-varying state of the project entity and the predictive risk information generated by the predictive risk engine.

3. The power engineering design project management system according to claim 1, characterized in that: The data integration and processing module includes a natural language processing submodule for automatically extracting entities, attributes and rules from unstructured regulatory standard documents and storing them in a structured manner in the static knowledge layer.

4. The power engineering design project management system according to claim 3, characterized in that: The natural language processing submodule uses a pre-trained language model to perform named entity recognition and combines it with relationship extraction technology to convert natural language rules into structured constraints.

5. The power engineering design project management system according to claim 1, characterized in that: The proactive design change impact analysis module is further configured to: In response to a design change request from a user, a temporary "what-if" analysis branch is created in the knowledge graph; Simulate the design change in this branch and propagate the change and identify all affected entities by traversing the dependency edges in the knowledge graph; The compliance checking engine and predictive risk engine are invoked to re-evaluate on the "what-if" analysis branch to quantify the impact of the change on compliance, risk, cost, and schedule.

6. The power engineering design project management system according to claim 1, characterized in that: The graph neural network-based predictive risk engine adopts a relational graph convolutional network (R-GCN) model to process multiple relationship types in the knowledge graph.

7. A method for managing power engineering design projects, characterized in that: The following steps are involved: Through the data integration and processing module, information is continuously extracted from heterogeneous data sources such as design documents, regulatory texts, and project plans to build and dynamically maintain a multi-layer dynamic knowledge graph that semantically represents project entities and their multi-dimensional relationships; Automatically and continuously check the compliance of design solutions by converting design specifications and standards into graph queries and executing the queries on the knowledge graph; Use a graph neural network model trained on historical and real-time project knowledge graph data to predict future project risks by analyzing the structural patterns of the graph; as well as In response to a proposed design change, the change is simulated in the knowledge graph and propagated along dependencies to identify all affected entities and quantitatively assess the chain reaction of the change.

8. A power engineering design project management method according to claim 7, characterized in that: The steps of constructing a multi-layer dynamic knowledge graph include: Build a static knowledge layer to store basic knowledge entities such as regulations and standards, component templates, etc. Constructing a project instance layer for storing specific entity instances in a specific project and associating it with the static knowledge layer; and Build a dynamic status and risk layer to store the time-varying status and predictive risk information of project entities.

9. A power engineering design project management method according to claim 7, characterized in that: The steps for assessing the knock-on effects of changes include: Creating a temporary "what-if" analysis branch in the knowledge graph; Simulate the change in that branch and identify all affected entities; Re-check compliance and risk prediction on the "what-if" analysis branch, and summarize and calculate the impact of the change on cost and schedule to generate a quantitative impact analysis report.

10. A power engineering design project management method according to claim 7, characterized in that: The step of using the graph neural network model to predict risks is to use the relational graph convolutional network R-GCN model to aggregate multiple relational information in the neighborhood of the target node to predict the risk probability.

Images