Thermal power infrastructure project archive map construction method and related device
By employing multimodal dynamic meta-learning alignment and causal perception graph neural network technology, the problems of multimodal data fusion and causal relationship modeling in thermal power infrastructure project archives have been solved, achieving efficient data identification, retrieval, and risk warning, and improving the level of intelligent management of thermal power infrastructure.
Patent Information
- Application Number
- CN202511061183.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-30
- Publication Date
- 2025-11-18
AI Technical Summary
The archives of thermal power infrastructure projects suffer from problems such as low efficiency of multimodal data fusion, insufficient relationship modeling capabilities, and poor model generalization. Traditional methods are difficult to effectively handle multiple types of information and causal relationships.
By employing multimodal dynamic meta-learning alignment and causal perceptive graph neural network technology, a thermal power infrastructure project archive map is constructed by aligning multimodal data features. This includes a dynamic meta-learning alignment algorithm, a causal perceptive graph neural network, and a structural causal model, enabling deep fusion of multimodal data and causal inference.
It achieves deep fusion of multimodal data, improves data recognition accuracy and retrieval efficiency, enhances causal reasoning capabilities, supports cross-project adaptive applications, and provides real-time risk warnings and data security guarantees, in accordance with industry standards.
Smart Images
Figure CN120975201A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of map construction technology, and relates to a method and related apparatus for constructing archive maps of thermal power infrastructure projects. Background Technology
[0002] In thermal power infrastructure projects, archival data exhibits multimodal, highly dynamic, and strongly correlated characteristics. Traditional archival management systems only support structured storage of single-modal data, making it difficult to effectively handle diverse information types such as design drawings, construction videos, business forms, and monitoring data. Existing knowledge graph construction technologies face the following problems when processing thermal power archives:
[0003] 1. Low efficiency of multimodal data fusion: The semantic spaces of different modal data differ greatly, making it difficult for traditional methods to achieve efficient alignment, resulting in information fragmentation. For example, equipment parameters in drawings cannot be automatically correlated with actual construction monitoring data.
[0004] 2. Insufficient relational modeling capabilities: The causal relationships in the construction process of thermal power plants are complex and ever-changing (such as the causal relationship between equipment failure and construction technology), and existing graph neural networks lack the ability to model dynamic causal logic.
[0005] 3. Poor model generalization: The data distribution of thermal power projects varies significantly, and traditional models are difficult to adapt to the feature changes of different projects, requiring a large amount of labeled data for repeated training.
[0006] Against this backdrop, there is an urgent need for an innovative management solution for thermal power infrastructure archives, which can achieve deep integration of archive data, dynamic causal reasoning, and cross-scenario generalization through multimodal dynamic meta-learning alignment and causal perception graph neural network technology. Summary of the Invention
[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and related apparatus for constructing a thermal power infrastructure project archive map. This method and apparatus can construct a thermal power infrastructure project archive map based on multimodal dynamic meta-learning alignment and causal perceptual graph neural network technology.
[0008] To achieve the above objectives, this invention discloses a method for constructing an archive map of thermal power infrastructure projects, comprising:
[0009] Multimodal data features are extracted from thermal power infrastructure project archives, and the multimodal data features are aligned to obtain multimodal aligned feature vectors.
[0010] An archive graph structure is constructed based on the aligned multimodal data features, and causal reasoning is performed.
[0011] Based on the structure of the archive diagram and the results of causal reasoning, an archive diagram of thermal power infrastructure projects is constructed.
[0012] A further improvement of the method for constructing archive maps of thermal power infrastructure projects described in this invention is as follows:
[0013] Furthermore, the process of aligning the multimodal data features to obtain the multimodal aligned feature vector is as follows:
[0014] A dynamic meta-learning alignment algorithm is used to align the features of the multimodal data to obtain the multimodal aligned feature vector.
[0015] Furthermore, the attributes of nodes in the archive graph structure include multimodal aligned feature vectors, and the attributes of edges include relation confidence and timestamps.
[0016] Furthermore, the process of performing causal reasoning is as follows:
[0017] Causal reasoning based on structural causal models.
[0018] Furthermore, the resulting causal model is constructed based on a three-layer causal perceptual graph neural network.
[0019] Furthermore, the loss function of the resulting causal model during the training process is:
[0020]
[0021] Furthermore, the node update formula in the resulting causal model is as follows:
[0022]
[0023] in, Let N(i) be the feature vector of node i at layer l; N(i) be the set of neighboring nodes. For causal attention coefficient, for:
[0024]
[0025] Where a is the learnable attention vector, β ij The causal weight of edge (i,j) is dynamically updated by the causal rule base.
[0026] This invention discloses a system for constructing an archive map of thermal power infrastructure projects, comprising:
[0027] The alignment module is used to extract multimodal data features from thermal power infrastructure project archives, align the multimodal data features, and obtain multimodal aligned feature vectors.
[0028] The first construction module is used to construct the archive graph structure based on the aligned multimodal data features and to perform causal reasoning;
[0029] The second construction module is used to construct an archive map of thermal power infrastructure projects based on the archive map structure and causal reasoning results.
[0030] This invention discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the method for constructing a thermal power infrastructure project archive map.
[0031] This invention discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for constructing a thermal power infrastructure project archive map.
[0032] The present invention has the following beneficial effects:
[0033] The method and related apparatus for constructing archive maps of thermal power infrastructure projects described in this invention construct an archive map structure based on the aligned multimodal data characteristics during specific operation, and perform causal reasoning. Based on the archive map structure and the causal reasoning results, an archive map of thermal power infrastructure projects is constructed. It is highly practical and not only provides the thermal power industry with a technical path that conforms to policy guidance, but also reshapes the archive management paradigm through intelligent means, injecting new momentum into the high-quality development of the power industry. Attached Figure Description
[0034] The accompanying drawings, which form part of this specification, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:
[0035] Figure 1 This is an architecture diagram of the multimodal dynamic meta-learning alignment module for domain knowledge enhancement in this invention;
[0036] Figure 2 This is a flowchart of the method of the present invention;
[0037] Figure 3 This is a diagram illustrating the dynamic response mechanism for engineering changes. Detailed Implementation
[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0039] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.
[0040] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0041] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.
[0042] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.
[0043] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."
[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0045] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.
[0046] Example 1
[0047] refer to Figure 1 , Figure 2 and Figure 3 The method for constructing a thermal power infrastructure project archive map according to the present invention includes the following steps:
[0048] 1) Extract multimodal data features from thermal power infrastructure project archives, align the multimodal data features, and obtain the multimodal aligned feature vector;
[0049] 2) Based on the features of the aligned multimodal data, construct the archive graph structure using the causal perceptual graph neural network module, and perform causal inference;
[0050] 3) Construct an archive map of thermal power infrastructure projects based on the archive map structure and causal reasoning results.
[0051] S4, based on the data security and privacy protection module, implement encrypted storage protection for sensitive data. The specific operation of step 1) is as follows:
[0052] 11) Extract features from multimodal data;
[0053] For text data: A pre-trained model based on RoBERTa for the thermal power industry was used. A dictionary of thermal power terminology (covering standard terms such as those in the "Guidelines for Construction Organization Design of Thermal Power Plants") was introduced into the Masked LanguageModel task, and the model parameters were optimized through self-supervised learning. In the processing of construction logs for a 660MW thermal power project, the entity recognition F1 score reached 95.3%.
[0054] For image data: A hierarchical feature extraction network is constructed. The bottom layer uses an improved Swing Transformer to extract the appearance features of the equipment, and the top layer integrates semantic information of the construction scene through an attention mechanism. In boiler installation image recognition, the accuracy of key component positioning is improved to 92.7%.
[0055] For structured data: Design an LSTM network with a hybrid attention mechanism, which captures the dynamic changes of engineering progress data through time series attention and enhances the expressive power of key parameters (such as steam pressure and coal consumption rate) through feature attention.
[0056] 12) The multimodal data features are aligned using a dynamic meta-learning alignment algorithm to obtain the multimodal aligned feature vector;
[0057] The meta-learning framework in the dynamic meta-learning alignment algorithm employs the MAML-UO (Model-Agnostic Meta-Learning with Uncertainty-aware Optimization) algorithm to construct a two-layer optimization structure.
[0058] Inner layer optimization: For a single thermal power project dataset, the modality-specific encoder parameters θ are quickly adjusted using gradient descent. i Minimize local alignment loss L local .
[0059] Outer layer optimization: On multiple project datasets, update the meta-learner parameters θ based on the meta-gradient and optimize the global alignment loss L. local .
[0060] Cross-modal alignment strategy: Design a dynamic temperature regulation contrastive learning algorithm with the following alignment loss function:
[0061]
[0062] in, τ(t) represents the entity feature vectors of text, image, and structured data, respectively; τ(t) is a temperature parameter that dynamically adjusts with training epoch t, employing an adaptive learning rate strategy.
[0063]
[0064] Where τ0 is the initial temperature, and μ and σ are the mean and standard deviation of the historical alignment loss.
[0065] The specific process of step 2) is as follows:
[0066] 21) Construct a structural diagram of thermal power plant infrastructure;
[0067] Entity relationships are constructed: Based on the "Management Measures for the Compliance and Commissioning of Thermal Power Projects", 12 types of entities such as equipment, personnel, and documents are defined, and 8 types of relationships such as "installed at", "caused by", and "associated with" are established to construct a heterogeneous graph structure.
[0068] Graph initialization: Based on the heterogeneous graph structure, a dynamic heterogeneous graph is constructed using the DGL library. Node attributes include multimodal aligned feature vectors, and edge attributes include relation confidence and timestamps.
[0069] 22) Causal reasoning mechanism;
[0070] Constructing causal relationships: Introducing a structural causal model (SCM) and combining it with expert knowledge in the thermal power field to build a causal rule base. For example, establishing a causal chain of "excessive sulfur content in coal → increased load on the desulfurization system → increased equipment failure rate", and quantifying the causal strength through a Bayesian network.
[0071] Constructing a graph neural network architecture: Design a three-layer causal perceptual graph neural network (Causal-GNN), with the node update formula as follows:
[0072]
[0073] in, Let N(i) be the feature vector of node i at layer l; N(i) be the set of neighboring nodes. The causal attention coefficient is calculated using the following formula:
[0074]
[0075] Where a is the learnable attention vector, β ij The causal weight of edge (i,j) is dynamically updated by the causal rule base.
[0076] Causal Intervention Analysis: Causal intervention simulation is implemented based on Do-Calculus theory. During training, the intervention graph structure is randomly selected. By comparing the changes in node features before and after the intervention, the model's causal inference ability is optimized. The loss function is:
[0077]
[0078] The specific process of step 3) is as follows:
[0079] Knowledge graph generation: The multimodal alignment results are fused with the causal reasoning results to generate a knowledge graph of thermal power infrastructure archives, which is stored in the Neo4j graph database.
[0080] The intelligent application features include two parts: multimodal retrieval and risk warning, as detailed below:
[0081] Multimodal retrieval: Supports mixed retrieval of text keywords, image features, and structured parameters, and achieves rapid location of related files through graph traversal, reducing retrieval response time by more than 70%.
[0082] Risk warning: Based on the causal relationship network, the project risks are monitored in real time, such as predicting the risk path of "delayed construction progress → delayed equipment commissioning → postponed production plan", and triggering a warning 7-10 days in advance.
[0083] Step 4) is as follows:
[0084] Sensitive data encryption: The national standard SM4 algorithm is used to encrypt classified drawings and key parameters, and the key lifecycle management complies with the requirements of the "Power Data Security Management Specification".
[0085] This invention has the following characteristics:
[0086] 1. Deep fusion of multimodal data
[0087] This invention successfully achieves semantic-level deep fusion of multi-source heterogeneous data, including text, images, and 3D models, through innovative dynamic meta-learning alignment technology. This method effectively solves the technical challenge of fragmented multimodal data in traditional record management systems, achieving significant results in practical applications: the entity recognition F1 score for text data reaches 95.3%, and the localization accuracy of key components in image data is improved to 92.7%. This high-precision data fusion capability lays a solid foundation for subsequent intelligent analysis and applications.
[0088] 2. Dynamic causal reasoning ability
[0089] This invention creatively combines causal perceptual graph neural networks with structural causal models (SCM) to construct a powerful dynamic causal reasoning system. This system can automatically uncover complex potential causal chains between archival elements, such as typical causal relationships like "excessive sulfur content in coal → increased load on desulfurization system → increased equipment failure rate," and accurately quantify the causal strength using Bayesian networks. This capability significantly improves the accuracy and reliability of engineering risk prediction.
[0090] 3. Cross-project adaptive generalization
[0091] To address the significant data distribution variations in thermal power projects, this invention employs the advanced MAML-UO meta-learning framework and designs a unique two-layer optimization structure. The inner layer optimization focuses on parameter tuning for a single project, while the outer layer optimization performs meta-gradient updates across multiple project datasets. This innovative design enables the model to possess excellent cross-project adaptability, allowing it to be applied to different project scenarios without repeated training, and significantly reducing data annotation costs.
[0092] 4. Intelligent application for efficient retrieval
[0093] The multimodal retrieval system constructed in this invention supports mixed queries using text keywords, image features, and structured parameters. Through an optimized graph traversal algorithm, the system achieves rapid location of related files, reducing retrieval response time by more than 70%. This breakthrough significantly improves the efficiency of engineering file retrieval and enhances the user experience.
[0094] 5. Real-time risk warning mechanism
[0095] Based on the constructed dynamic causal network, the system can monitor engineering risk paths in real time, accurately predicting typical risk chains such as "delayed construction progress → delayed equipment commissioning → postponed production plan," and triggering early warnings 7-10 days in advance. This function significantly improves the foresight and risk control capabilities of project management.
[0096] 6. Deep integration of domain knowledge
[0097] This invention deeply integrates industry standards such as the "Management Measures for the Compliance and Commissioning of Thermal Power Projects," clearly defines 12 types of entities and 8 types of relationships, and embeds a professional terminology dictionary covering standards such as the "Guidelines for Construction Organization Design of Thermal Power Generation Projects." This deep integration of domain knowledge ensures that the model's output fully conforms to industry standards, improving the accuracy of entity relationship modeling by 40%.
[0098] 7. Data security compliance assurance
[0099] Regarding data security, this invention employs the national cryptographic algorithm SM4 to encrypt classified drawings and key parameters, and strictly implements key lifecycle management in accordance with the requirements of the "Power Data Security Management Standard". This design ensures both efficient data utilization and full protection of privacy, fully complying with the regulatory requirements of the power industry.
[0100] 8. Significant economic benefits
[0101] Practical application data shows that this invention can reduce engineering management costs by 3%-5% while improving file organization efficiency by 45%. This innovative achievement directly responds to the policy requirements of the "Power Digital Transformation Action Plan" and provides strong technical support for "cost reduction and efficiency improvement" and digital transformation and upgrading in the thermal power infrastructure field.
[0102] Example 2
[0103] The thermal power infrastructure project archive map construction system of the present invention includes:
[0104] The alignment module is used to extract multimodal data features from thermal power infrastructure project archives, align the multimodal data features, and obtain multimodal aligned feature vectors.
[0105] The first construction module is used to construct the archive graph structure based on the aligned multimodal data features and to perform causal reasoning;
[0106] The second construction module is used to construct an archive map of thermal power infrastructure projects based on the archive map structure and causal reasoning results.
[0107] The module division in this embodiment is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in each embodiment of this application can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.
[0108] Example 3
[0109] System Deployment Architecture
[0110] (1) Hardware configuration:
[0111] Meta-learning training server: 8×NVIDIA A100 GPU + 1TB memory + 100Gbps InfiniBand network, supporting large-scale pre-trained model optimization (peak computing power of 5PFLOPS per node).
[0112] Graph Inference Server: 2×RTX 6000 + 512GB memory + NVMe SSD storage pool, enabling real-time causal inference (latency <200ms) and ensuring real-time inference performance.
[0113] Edge device: Industrial-grade explosion-proof flat panel (IP68 / Ex ib IIC T4 certified) + 5G module, supporting working environment from -40℃ to 85℃, used for on-site data acquisition and preliminary processing.
[0114] (2) Software stack:
[0115] Underlying framework: PyTorch 2.0 + DGL 1.1, supporting GPU acceleration and efficient graph neural network computation.
[0116] Middleware: Kafka message queue (processes real-time data streams), Neo4j graph database (stores knowledge graphs).
[0117] Upper-layer application: The web interface is built on Vue3 and Element Plus, supporting functions such as multimodal search and risk warning.
[0118] Example 4
[0119] (1) Data governance phase (1-2 months):
[0120] Establish classification standards for thermal power plant archives and complete the cleaning and labeling of historical data.
[0121] Build a domain dictionary and device prototype library, and initialize the pre-trained model.
[0122] (2) Model training phase (3-4 months):
[0123] Train a multimodal alignment model and a causal perception GNN on a pilot project.
[0124] Optimize the change propagation algorithm and real-time update mechanism.
[0125] (3) Full-scale promotion phase (5-6 months):
[0126] The system was deployed to multiple thermal power projects, and feedback data was collected to continuously optimize the model.
[0127] Establish a cross-project knowledge-sharing mechanism to improve the overall management level of the industry.
[0128] Example 5
[0129] Implement an A / B testing mechanism: Conduct a 72-hour comparative test before the new algorithm goes live to ensure that the performance improvement is significant and has no negative impact.
[0130] Model compression and acceleration: Using knowledge distillation technology, the model size is compressed by 60%, the inference speed is increased by 40%, and mobile deployment is supported.
[0131] Example 6
[0132] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a method for constructing a thermal power infrastructure project archive map. For example, the method includes: extracting multimodal data features from thermal power infrastructure project archives; aligning the multimodal data features to obtain multimodally aligned feature vectors; constructing an archive map structure based on the aligned multimodal data features and performing causal inference; and constructing a thermal power infrastructure project archive map based on the archive map structure and the causal inference results. The memory may include main memory, such as high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry standard architecture bus, a peripheral component interconnection standard bus, an extended industry standard architecture bus, etc. The bus may be classified as an address bus, data bus, control bus, etc. The memory stores the program; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.
[0133] Example 7
[0134] A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method for constructing a thermal power infrastructure project archive map. For example, the method includes: extracting multimodal data features from thermal power infrastructure project archives; aligning the multimodal data features to obtain multimodally aligned feature vectors; constructing an archive map structure based on the aligned multimodal data features and performing causal inference; and constructing a thermal power infrastructure project archive map based on the archive map structure and the causal inference results. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory and / or cache memory, etc. The non-volatile memory may include read-only memory, hard disk, flash memory, optical disk, magnetic disk, etc.
[0135] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.
[0136] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0137] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0138] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0139] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and disclosure of the invention. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.
[0140] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.
[0141] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural changes made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.
Claims
1. A method for constructing an archive map of thermal power plant infrastructure projects, characterized in that, include: Multimodal data features are extracted from thermal power infrastructure project archives, and the multimodal data features are aligned to obtain multimodal aligned feature vectors. An archive graph structure is constructed based on the aligned multimodal data features, and causal reasoning is performed. Based on the structure of the archive diagram and the results of causal reasoning, an archive diagram of thermal power infrastructure projects is constructed.
2. The method for constructing a thermal power infrastructure project archive map according to claim 1, characterized in that, The process of aligning the multimodal data features to obtain the multimodal aligned feature vector is as follows: A dynamic meta-learning alignment algorithm is used to align the features of the multimodal data to obtain the multimodal aligned feature vector.
3. The method for constructing a thermal power infrastructure project archive map according to claim 1, characterized in that, The attributes of nodes in the archive graph structure include multimodal aligned feature vectors, and the attributes of edges include relation confidence and timestamps.
4. The method for constructing a thermal power infrastructure project archive map according to claim 1, characterized in that, The process of causal reasoning is as follows: Causal reasoning based on structural causal models.
5. The method for constructing a thermal power infrastructure project archive map according to claim 4, characterized in that, The resulting causal model is constructed based on a three-layer causal perceptual graph neural network.
6. The method for constructing a thermal power infrastructure project archive map according to claim 4, characterized in that, The loss function of the resulting causal model during training is:
7. The method for constructing a thermal power infrastructure project archive map according to claim 4, characterized in that, The node update formula in the resulting causal model is: in, Let N(i) be the feature vector of node i at layer l; N(i) be the set of neighboring nodes. For causal attention coefficient, for: Where a is the learnable attention vector, β ij The causal weight of edge (i,j) is dynamically updated by the causal rule base.
8. A system for constructing an archive map of thermal power infrastructure projects, characterized in that, include: The alignment module is used to extract multimodal data features from thermal power infrastructure project archives, align the multimodal data features, and obtain multimodal aligned feature vectors. The first construction module is used to construct the archive graph structure based on the aligned multimodal data features and to perform causal reasoning; The second construction module is used to construct an archive map of thermal power infrastructure projects based on the archive map structure and causal reasoning results.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the thermal power infrastructure engineering archive map construction method as described in any one of claims 1-7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for constructing a thermal power infrastructure project archive map as described in any one of claims 1-7.