Overhead transmission line construction risk assessment visualization method based on BIM model

Abstract

The application discloses a kind of overhead transmission line construction risk assessment visualization methods based on BIM model, including the risk intelligent identification evaluation based on construction scheme text analysis, the risk assessment visualization marking based on BIM model and large language model combination and the risk assessment visualization result directly marked in the construction scene of BIM model as the suggestion basis of current construction scheme optimization, and form the complete closed loop of risk identification, risk visualization, risk feedback and scheme optimization.The present application realizes the intelligent identification evaluation of risk to the process that drives large language model to carry out risk marking pre-rehearsal in BIM model by text analysis model, large language model and BIM model, changes the traditional discrete, static processing mode into dynamic correlation mode, significantly improves risk identification accuracy and visualization efficiency.

Description

Technical Field

[0001] This invention relates to a visualization method for risk assessment of overhead transmission line construction, specifically a visualization method for risk assessment of overhead transmission line construction based on BIM model, belonging to the field of project engineering information technology. Background Technology

[0002] In recent years, with the advancement of power energy transformation and the construction of new power upgrades, the scale of overhead transmission line renovation and construction has continued to expand. However, the construction of overhead transmission lines often takes place in complex and dangerous outdoor environments, such as crossing mountains, rivers, railways, and national highways. These environments are often accompanied by variable geological conditions, uncertain environmental factors, complex and diverse risk factors, and significant challenges in real-time adjustments and rescue. Inaccurate or incomplete identification and assessment of complex risks such as tension loss of control, falling objects from heights, extreme weather, and live-line work in the construction plan can lead to huge economic losses and even major casualties, resulting in severe social impacts. Furthermore, the construction of overhead transmission lines involves fieldwork throughout, requiring adherence to environmental protection, ecological protection, and legal restrictions, and often necessitates avoiding existing power lines, communication lines, highways, and railways. Given these characteristics of overhead transmission line construction, risk identification in the construction plan is extremely important, as it directly affects the safety of workers on the construction site and the quality and progress of the transmission line construction. In construction plans, potential risks such as foundation collapse, uncontrolled tension of power transmission poles, and live-line work can all lead to accidents resulting in significant casualties and economic losses. Therefore, accurate and comprehensive risk identification for overhead power transmission line construction plans is a core and essential step in the power transmission line construction process.

[0003] Currently, traditional risk identification relies heavily on the human experience of safety officers, resulting in strong subjectivity. Furthermore, with the increasing scale of transmission line projects, manual identification is inefficient, prone to omissions, and cannot be dynamically adjusted to keep pace with construction progress. In addition, traditional risk identification results are often presented in static formats such as documents and tables, making it difficult for relevant personnel to intuitively grasp the existing construction risks and leaving them with a vague understanding of the specific locations and process nodes where risks exist. Based on these factors, traditional risk assessment and identification methods have several limitations: Firstly, manual identification results are highly subjective and experience-based, heavily influenced by human experience, and cannot keep pace with the ever-expanding scale of transmission line construction, resulting in low efficiency and accuracy. Moreover, manual identification results are often presented as unstructured text descriptions, making them unusable in subsequent modules and lacking portability in risk information use. Secondly, traditional risk identification results are often based on text and static warning diagrams, lacking precise dynamic visualization effects.

[0004] In summary, current risk identification technologies for power transmission line construction have significant limitations in terms of scale, dynamism, and visualization efficiency.

[0005] To achieve the objectives described above, the main technologies used are project construction risk management technology, construction project information technology, artificial intelligence large-scale modeling technology, and Building Information Modeling (BIM) technology. However, the intelligent identification and visualization of construction risks still faces certain shortcomings and challenges in practical applications.

[0006] The first category concerns the problems in intelligent risk identification during power transmission line construction. Traditional risk identification in power transmission line construction often relies on the safety officer's experience and historical case databases, which has the following limitations: First, risk identification is highly subjective and ineffective for unfamiliar or hidden risks. Second, with the expansion of power transmission line construction and the diversification of construction scenarios (mountainous areas, river crossings, high-speed rail crossings, etc.), manual verification is inefficient and prone to omissions, especially in new construction conditions lacking historical cases, leading to knowledge gaps regarding unknown risks. Third, risk identification is usually a phased process, lacking dynamic correlation with the construction period, and the results are mostly unstructured text descriptions, unable to be directly applied to subsequent risk management and analysis modules. These problems result in a severe lack of comprehensiveness and timeliness in traditional risk identification, failing to match the rapid pace and efficiency of power transmission line construction.

[0007] The second category is the visualization problem of risk identification results for power transmission line construction. Traditional risk identification results are generally expressed in large sections of text or static illustrations, which cannot be correlated with the professional BIM model of the project construction. The risk evolution process cannot be dynamically deduced through accurate 3D images or construction animations. On the other hand, the identification of construction risks is usually a phased process, lacking compatibility with dynamic adjustments to the construction schedule. In addition, the professional tools required for modeling have high technical barriers to entry and are complex to operate, making them difficult for grassroots personnel to widely apply.

[0008] In summary, there are certain technical barriers to the intelligent identification and visualization of risks in power transmission line construction schemes. This patent addresses these issues by proposing a BIM-based method for visualizing risk assessments of power transmission line crossing construction schemes. Summary of the Invention

[0009] The purpose of this invention is to provide a visualization method for risk assessment of overhead transmission line construction based on BIM model in order to solve at least one of the above-mentioned technical problems. By constructing a dynamic process from text analysis to prompt word mapping and then driving a large language model to automatically complete risk labeling and risk inference animation generation in the construction BIM model, the visualization of dynamic pre-simulation of risk identification results of transmission line construction schemes can be realized.

[0010] This invention achieves the above objectives through the following technical solution: a visualization method for construction risk assessment of overhead transmission lines based on BIM models, which includes the following steps: S1. Risk Intelligent Identification and Assessment Based on Construction Plan Text Analysis: The text analysis model is used to perform in-depth semantic analysis on the construction plan text of overhead transmission lines and extract structured prompts for risk assessment. S2. Risk assessment visualization annotation based on the combination of BIM model and large language model: Using structured risk assessment prompts to drive the large language model to automatically adjust and process the BIM model, the risk assessment visualization of the overhead transmission line construction plan is realized, and the risk assessment visualization results are obtained. S3. The risk assessment visualization results directly marked in the construction scenario of the BIM model serve as the basis for suggestions on optimizing the current construction plan, forming a complete closed loop of risk identification, risk visualization, risk feedback, and plan optimization.

[0011] As a further aspect of the present invention: In S1, the extraction of structured risk assessment prompts specifically includes: S11. Semantic Analysis of Construction Plan Text: A text analysis model enhanced with domain knowledge is used to conduct in-depth semantic analysis of the construction plan text for overhead transmission lines; S12. Construction plan text processing: The spatial location information of engineering components in the construction plan text is embedded into the text vector of the corresponding workpiece as the input feature vector. The semantically extracted vector is then output at the output end of the text analysis model according to the constraint decoding method and under the constraint of the mapping rules, the final risk assessment structured prompt words are output.

[0012] As a further aspect of the present invention: In S11, the domain-enhanced text analysis model refers to using professional safety guidelines and historical construction cases related to the field of overhead power transmission line construction as a domain supplementary knowledge corpus, and performing data-enhanced pre-training and comparative pre-training on the text analysis model.

[0013] As a further aspect of the present invention: In S12, the text processing of the construction plan specifically includes the following steps: Based on the construction plan text of overhead transmission lines, and combined with a multi-level standard coding system, the entity information contained in the construction text is dynamically geospatially encoded and temporarily stored. Simultaneously, the Encoder encoding method in Transformer is used to perform conventional semantic encoding of the overhead transmission line construction text, capturing the contextual features between different texts. The specific encoding measures are as follows: The set of construction text recognition results is set as follows: Then, the corresponding text recognition results It can be mapped to geographical location as coordinates ( Therefore, using a multi-level encoding system can map each text containing geographic coordinates into a discrete code: (The specific mapping relationship depends on the encoding system.) Secondly, the semantic information of the construction text is processed using the Encoder encoding method in Transformer, which can be simplified as follows: ; The generated spatial location code is concatenated with features to form a direction containing joint spatial-semantic feature information, which represents the location information after the previous encoding process. Encoding with text information Set the aligned entity and text position pairs as ( The corresponding semantic vectors are extracted accordingly. and spatial encoding spliced ​​together as: By embedding the specific geospatial location code of the corresponding workpiece into the text code of the construction text through feature fusion processing, the model is made capable of perceiving information, thereby improving the accuracy and comprehensiveness of subsequent risk labeling visualization. The fused features are enhanced with key risk features through a multi-layer self-attention mechanism. The result is used as the initial hidden state input for the decoder. Finally, the decoder output format is constrained by a preset mapping rule, and the final structured risk assessment prompts are generated.

[0014] As a further aspect of the present invention: In S2, the risk assessment visualization annotation specifically includes: S21. Process the structured prompts for risk assessment and the existing BIM model according to the multi-level standard coding system; S22. Risk assessment structured prompts drive a large language model and video generation tool. Risks are labeled and dynamic pre-simulation animations are generated in the BIM model after processing by a multi-level standard coding system, thereby obtaining the risk assessment visualization results of the overhead transmission line construction scheme.

[0015] As a further aspect of the present invention: In S21, the multi-level standard coding system transforms the existing BIM model into a standardized coded BIM model, which facilitates subsequent modeling and processing. The multi-level standard coding system is matched with the structured risk assessment prompts, and the structured risk prompts output by decoding are represented as follows: Each of the risk warning words Each of them contains a corresponding semantic feature vector. Spatial position vector The standard library contains Each engineering component is represented as: The same for each engineering component It contains the corresponding semantic feature vector Spatial position vector ; The semantic matching degree is calculated using cosine similarity. ; The matching degree of spatial location is calculated using a weighted Euclidean distance method, which first transforms the spatial location vector into a three-dimensional coordinate axis representation. Based on this transformation, the corresponding weighted Eulerian distance is calculated as follows:

[0016] Based on the above distance calculation formula, the spatial matching score can be obtained, expressed as: The system identifies the data information of the engineering structure where the risk is located and obtains the prompt words for the risk visualization label by matching.

[0017] As a further aspect of the present invention, the multi-level standard coding system includes four levels: professional category coding, spatial location coding, engineering construction type coding, and construction stage coding.

[0018] As a further aspect of the present invention: in S21, the prompt words for risk visualization annotation obtained by matching are directly driven by the large language model to perform risk annotation in the BIM model through the pre-set API interface.

[0019] A BIM-based visualization system for construction risk assessment of overhead transmission lines, comprising: Intelligent risk identification module: Relying on a text analysis model based on spatial location-based embedding encoding and mapping rule constraint decoding, it dynamically identifies and extracts the direct and hidden risks contained in the construction plan text of overhead transmission lines, and generates the final risk assessment structured prompt words according to the corresponding mapping rule constraints for subsequent risk visualization labeling processing. Risk visualization module: On the BIM model processed by the multi-level standard coding system, the risk assessment structured prompts obtained by the intelligent risk identification module drive the large language model to automatically adjust and process the standardized coded BIM model, thereby realizing the visualization of risk assessment in the overhead transmission line construction plan.

[0020] The beneficial effects of this invention are: (1) From a performance perspective, this invention integrates the embedding of actual spatial location information into the semantic encoding of construction text to achieve the perception of the location of risk occurrence. At the same time, it uses the constraint decoding output architecture to generate structured risk prompts. Finally, it uses the risk prompts to drive the large language model to annotate risks in the existing BIM model through the pre-set API interface and generate dynamic risk pre-show animations. This realizes the automated process from risk identification based on text analysis to risk annotation based on BIM model, transforming the traditional discrete and static processing method into a continuous dynamic pipeline mode. Among them, the early embedding mechanism of spatial location enables the model to have spatial reasoning ability in the encoding stage, which can quickly associate risks with construction progress and improve response speed and decision-making efficiency. The constraint decoding method enables the model to output structured prompts, upgrading the single text description into machine-understandable and executable instructions, making the subsequent processing process more convenient and improving the visualization efficiency of risk assessment. The standard encoding system can realize the rapid updating of parameters, and the large language model with the pre-set API interface can perform risk annotation in the BIM model in batches, improving the efficiency of risk annotation for power transmission line crossing construction schemes.

[0021] (2) From a cost perspective, this invention encapsulates component parameters into reusable templates through a multi-level standard coding system. This eliminates the need for reprocessing across projects; only the corresponding parameters need to be updated, significantly reducing repetitive manpower input. Furthermore, by driving API automation through a large language model, it reduces reliance on the skills of professional operators, saving on personnel training costs. In addition, the automated, dynamic, pipeline-style risk visualization process reduces rework and scheme adjustment costs caused by human error, while simultaneously improving the efficiency of construction scheme inspection and updates.

[0022] (3) From the perspective of safety and reliability, the text analysis model used in this invention utilizes industry-established procedures and domain knowledge for enhanced pre-training of the model, ensuring accurate and comprehensive identification of explicit and implicit risks, and avoiding the influence of subjective experience in manual identification. At the same time, the visualization and dynamic pre-simulation of risks can expose the risks existing in the construction process in advance, curb the source of risks before construction, and significantly improve the systematicness and reliability of construction safety management. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the overall process of the present invention; Figure 2 This is a schematic diagram illustrating the intelligent risk identification and assessment based on construction plan text analysis of the present invention; Figure 3 This is a schematic diagram of the mapping rules generated by the decoding constraints of this invention; Figure 4This is a schematic diagram illustrating the visualization and annotation implementation of risk assessment based on the combination of BIM model and large language model in this invention; Figure 5 This is a schematic diagram of the multi-level BIM model standard coding system of this invention; Figure 6 This is a schematic diagram illustrating the dynamic pre-simulation visualization implementation process based on a large language model and a BIM model, as described in this invention. Detailed Implementation

[0024] 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 embodiments of the present invention, and not all embodiments. 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.

[0025] Examples, such as Figure 1 and Figure 6 As shown in the figure, this embodiment provides a visualization method for construction risk assessment of overhead transmission lines based on BIM models. The visualization method for construction risk assessment of overhead transmission lines specifically includes the following steps: First, risk intelligent identification and assessment based on construction plan text analysis.

[0026] For text analysis of overhead transmission line construction plans, the main approach relies on a text analysis model based on spatial location-based embedding encoding and mapping rule-based constraint decoding. Through a pre-trained large-scale text analysis model, the direct and implicit risks contained in the text of the current overhead transmission line (transmission crossing line) construction plan can be dynamically identified and extracted, and the final risk assessment structured prompts can be generated according to the corresponding mapping rules.

[0027] This step specifically includes: 1) Semantic analysis of construction plan text: A text analysis model enhanced with domain knowledge is used to conduct in-depth semantic analysis of the construction plan text of overhead transmission lines.

[0028] like Figure 2 As shown, the domain-enhanced text analysis model uses professional safety guidelines and historical construction cases related to overhead power transmission line construction as a supplementary domain knowledge corpus, and performs data augmentation pre-training and comparative pre-training on the text analysis model. Data augmentation pre-training improves the text analysis model's understanding of the diversity of risk type descriptions; comparative pre-training improves the text analysis model's ability to distinguish the semantic boundaries of similar risks.

[0029] The pre-trained text analysis model has the ability to identify and extract terminology and unknown risks in the field of overhead power transmission line construction.

[0030] 2) Construction plan text processing: The spatial location information of engineering components in the construction plan text is embedded into the text vector of the corresponding workpiece as the input feature vector. The semantically extracted vector is then output at the output end of the text analysis model in a constrained decoding manner, under the constraints of the mapping rules, to output the final risk assessment structured prompt words.

[0031] The specific processing of construction plan documents includes: 21) Based on the construction plan text of overhead transmission lines, and combined with a multi-level standard coding system, the entity information (such as actual construction procedures, components, etc.) contained in the construction plan text is dynamically geospatially encoded and temporarily stored; at the same time, the Encoder encoding method in Transformer is used to perform conventional text information semantic encoding processing on the construction text of overhead transmission lines to capture the contextual features between different texts. The specific encoding measures are as follows: The set of construction text recognition results is set as follows: Then, the corresponding text recognition results It can be mapped to geographical location as coordinates ( Therefore, using a multi-level encoding system can map each text containing geographic coordinates into a discrete code: (The specific mapping relationship depends on the encoding system); secondly, the semantic information of the construction text is processed using the Encoder encoding method in Transformer, which can be simplified as follows: ; 22) The generated spatial location code is concatenated with features to form a vector containing joint spatial-semantic feature information, which represents the location information after the previous encoding process. Encoding with text information Set the aligned entity and text position pairs as ( The corresponding semantic vectors are extracted accordingly. and spatial encoding spliced ​​together as: By embedding the specific geospatial location code of the corresponding workpiece into the text code of the construction text through feature fusion processing, the text analysis model is able to perceive information (such as "which workpiece will be at risk", "what is the specific content of the risk" and "the actual spatial location of the risk"), thereby improving the accuracy and comprehensiveness of subsequent risk labeling visualization. 3) The fused features are enhanced with key risk features through a multi-layer self-attention mechanism. The result is used as the initial hidden state input of the decoder. Finally, the decoder output format is constrained by a preset mapping rule, and the final structured prompt words for risk assessment are generated.

[0032] This embodiment provides an example of a decoding constraint generation mapping rule, such as... Figure 3 As shown: 1. "The installation of the tensioning machine at the xx tower location for the crossing line is not feasible due to the actual conditions of the foundation, posing a risk of uncontrolled tension and potentially leading to casualties and significant economic losses." 2. "During the construction of the crossing line between towers xx and xx, no professional tensioning machine operator was assigned. This may have resulted in insufficient tension due to improper operation and settings, preventing the transmission line from being pulled up and causing economic losses."

[0033] The corresponding decoding output example is as follows: {{[Risk-ID:"01"];[Risk-TIME:"Pre-construction basic equipment setup stage"];[Risk-Process:"Tension machine installation"];[Risk-Type:"Management risk"];[Risk-Information&Description:"Incorrect site selection for tension machine installation, actual site conditions do not meet the requirements for tension machine erection, posing a risk of tension loss of control"];[Additional:“xxx”].

[0034] {[Risk-ID: "02"];[Risk-TIME:"Line stringing stage"];[Risk-Process:"Tension release of crossing line"];[Risk-Type:"Human risk + Management risk"];[Risk-Information&Description:"No tension machine inspector was set up, which may result in insufficient tension and failure to pull up the transmission line"];[Additional:“xxx”].

[0035] As described above, the core of risk intelligent identification and assessment based on construction plan text analysis lies in the construction of spatial and semantic information association encoding and constraint decoding based on mapping rules. Through pre-training with domain knowledge, the text analysis model has the ability to identify and extract professional risks in the field of power transmission line crossing construction.

[0036] To address the problems of traditional risk identification, such as strong subjectivity, low efficiency, and susceptibility to omissions and knowledge blind spots, this step proposes a text analysis model based on the fusion of actual spatial location information encoding and semantic encoding, and constraint decoding based on predetermined mapping rules. This model enables the identification and extraction of direct and implicit risks in overhead power transmission line construction schemes. After pre-training with enhanced domain knowledge, the text analysis model can directly perform text analysis on power transmission line crossing construction schemes. By embedding spatial location encoding of engineering components, it can achieve location perception of construction risks. Through constraint generation, risk information is extracted according to predetermined mapping rules and transformed into easily processed structured prompt words. The prompt words specifically include information such as risk type, risk level, and risk location, facilitating subsequent risk visualization processing by the model.

[0037] This step involves text-based risk identification of existing power transmission line construction plans and extracting structured risk assessment prompts. This transforms unstructured text into structured prompts that can be executed by machines, providing a foundation for subsequent automatic risk labeling and dynamic simulation of BIM models.

[0038] Second, risk assessment visualization annotation based on the combination of BIM model and large language model.

[0039] like Figure 4 As shown, the visualization of risk identification in the overhead transmission line construction scheme is mainly based on the existing BIM model of the overhead transmission line construction scheme (hereinafter referred to as "existing BIM model"). It uses the risk assessment structured prompt words extracted by the text analysis model to drive the large language model to automatically adjust and process the BIM model, thereby realizing the visualization of risk assessment of the overhead transmission line construction scheme and obtaining the risk assessment visualization results.

[0040] This step specifically includes: 1) Based on a multi-level standard coding system, the structured prompts for risk assessment and the existing BIM model are processed respectively.

[0041] On the one hand, the existing BIM model is transformed into a standardized coded BIM model based on a multi-level standard coding system to facilitate subsequent modeling and processing; on the other hand, the structured prompts for risk assessment are matched with the multi-level standard coding system to determine the data information (spatial location parameters and other detailed information) of the engineering construction where the risk is located, and prompts for risk visualization annotation are obtained through matching.

[0042] Generally, conventional BIM models (existing BIM models) are obtained through manual modeling. This modeling process is discrete and relies heavily on subjective human experience, resulting in low modeling efficiency and a lack of repeatability. In contrast, the module designed in this embodiment employs a multi-level standard coding system to standardize the BIM model corresponding to the current construction plan. This achieves a precise and comprehensive mapping between the BIM model and information such as construction procedures and components in the actual construction process. Standardization can be viewed as encapsulating the parameters of the corresponding engineering components, as well as the modeling methods and techniques, into a reusable function. This enables subsequent automated processing of the BIM model.

[0043] like Figure 5 As shown, the multi-level standard coding system includes four levels: professional category coding, spatial location coding, engineering construction type coding, and construction stage coding.

[0044] Multi-level design can transform simple text into structured parameters. A standardized coding system enables the parametric reuse of the same component in different engineering models, requiring only updates to location and parameter information. This improves the efficiency of cross-project modeling of the same workpiece and facilitates BIM modeling and risk labeling using large language models, replacing traditional manual modeling methods and enhancing the automation level of BIM modeling. Multi-level standardized coding essentially upgrades static tags to parametric templates, providing a foundation for automated cross-project modeling and convenient operation driven by large language models.

[0045] Next, the risk assessment structured prompts output by the text analysis model are intelligently matched with a multi-level standard coding system to accurately locate the engineering components corresponding to the risks and their codes and numbers, and to extract the spatial location parameters and process information of the construction corresponding to the risks.

[0046] Set the structured risk warning words in the decoded output as follows Each of the risk warning words Each of them contains a corresponding semantic feature vector. Spatial position vector The standard library contains An engineering component can be represented as: The same for each engineering component It contains the corresponding semantic feature vector Spatial position vector .

[0047] The semantic matching degree is calculated using cosine similarity. .

[0048] The matching degree of spatial location is calculated using a weighted Euclidean distance method, which transforms the spatial location vector into a three-dimensional coordinate axis representation. Based on this transformation, the corresponding weighted Eulerian distance is calculated as follows:

[0049] Based on the above formula, the spatial matching score can be obtained as follows: .

[0050] Once the risk warning words are matched, they can directly drive the large language model to annotate the risks in the BIM model via a pre-built API interface. Bright colors can be used to directly mark the risk locations on the workpiece. For complex risk scenarios, video animation generation models can be used to dynamically simulate the risk evolution process. The entire process can achieve automated annotation with zero human intervention, providing high efficiency for risk visualization.

[0051] 2) Risk assessment structured prompts drive a large language model and video generation tools. Risks are labeled and dynamic pre-simulation animations are generated in the BIM model after processing by a multi-level standard coding system, thereby obtaining the risk assessment visualization results of the overhead transmission line construction scheme.

[0052] Traditional risk identification results are typically expressed in static text and diagrams, failing to establish a corresponding connection with the professional BIM model of the project construction. The location and timing of risk occurrence cannot be presented intuitively in a BIM model or through dynamic simulation animations. This step proposes a risk visualization simulation method based on a BIM model, establishing a multi-level standard coding system that transforms processes, components, and task types into standardized codes, facilitating subsequent risk matching and large-scale model annotation. This system is used to standardize the coding of existing power transmission line crossing construction BIM models. Furthermore, structural risk warning words obtained from text analysis models are intelligently matched with this system to determine the corresponding codes and other detailed information of the risky engineering components in the BIM model, facilitating subsequent annotation. Finally, by pre-setting an API interface between the BIM model and the large language model, the large language model is driven to automatically annotate risks within the BIM model. For complex risk scenarios, video generation tools can be invoked to generate dynamic simulation animations of risk occurrence, thereby visualizing risks in power transmission line crossing construction schemes.

[0053] Based on the above description, this step constructs a multi-level BIM standard coding system, standardizes the coding of existing overhead power transmission line BIM models, and uses structured risk prompts to drive a large language model to visualize risks in the standardized coded power transmission line crossing construction BIM model. Furthermore, for complex risk scenarios, video generation tools can be used to generate risk simulation animations. This automates the process from static text-based risk identification to visual risk annotation, significantly improving the intuitiveness and efficiency of construction plan risk assessment.

[0054] Third, the risk assessment visualization results directly marked in the construction scenario of the BIM model can serve as a basis for suggestions on optimizing the current construction plan, forming a complete closed loop of risk identification, risk visualization, risk feedback and plan optimization.

[0055] This technical solution, based on a BIM model, transforms the traditional risk identification and assessment method into a closed-loop automated process. Through text analysis models, large language models, and BIM models, this method achieves intelligent risk identification and assessment, driving the large language model to perform risk labeling and pre-simulation within the BIM model. This transforms the traditional discrete, static processing method into a dynamic, relational one, significantly improving the accuracy of risk identification and visualization efficiency.

[0056] Example 2: This example provides a BIM model-based visualization system for risk assessment of overhead transmission line construction. This system is used to implement the method in Example 1, and specifically includes: Intelligent risk identification module: Relying on a text analysis model based on spatial location-based embedding encoding and mapping rule-constrained decoding, the pre-trained large text analysis model can dynamically identify and extract the direct and hidden risks contained in the text of the current overhead power transmission line construction plan, and generate the final risk assessment structured prompt words according to the corresponding mapping rule constraints for subsequent risk visualization and annotation processing. Risk visualization module: On the BIM model processed by the multi-level standard coding system, the risk assessment structured prompts obtained by the intelligent risk identification module drive the large language model to automatically adjust and process the standardized coded BIM model, thereby realizing the visualization of risk assessment in the overhead transmission line construction plan.

[0057] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0058] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A visualization method for construction risk assessment of overhead transmission lines based on BIM models, characterized in that, The visualization method for risk assessment of overhead transmission line construction includes the following steps: S1. Risk Intelligent Identification and Assessment Based on Construction Plan Text Analysis: The text analysis model is used to perform in-depth semantic analysis on the construction plan text of overhead transmission lines and extract structured prompts for risk assessment. S2. Risk assessment visualization annotation based on the combination of BIM model and large language model: The risk assessment structured prompt words are used to drive the large language model to automatically adjust and process the BIM model, realize the risk assessment visualization of the overhead transmission line construction plan, and obtain the risk assessment visualization result; S3. The risk assessment visualization results directly annotated in the construction scenario of the BIM model serve as the basis for suggestions on optimizing the current construction plan, forming a complete closed loop of risk identification, risk visualization, risk feedback, and plan optimization.

2. The visualization method for construction risk assessment of overhead transmission lines according to claim 1, characterized in that, In S1, the extraction of structured risk assessment prompts specifically includes: S11. Semantic Analysis of Construction Plan Text: A text analysis model enhanced with domain knowledge is used to conduct in-depth semantic analysis of the construction plan text for overhead transmission lines; S12. Construction plan text processing: The spatial location information of engineering components in the construction plan text is embedded into the text vector of the corresponding workpiece as the input feature vector. The semantically extracted vector is then output at the output end of the text analysis model according to the constraint decoding method and under the constraint of the mapping rules, the final risk assessment structured prompt words are output.

3. The visualization method for construction risk assessment of overhead transmission lines according to claim 2, characterized in that: In S11, the domain-enhanced text analysis model refers to using professional safety guidelines and historical construction cases related to the field of overhead power transmission line construction as a supplementary domain knowledge corpus, and performing data-enhanced pre-training and comparative pre-training on the text analysis model.

4. The visualization method for construction risk assessment of overhead transmission lines according to claim 2, characterized in that: In step S12, the text processing of the construction plan specifically includes the following steps: Based on the construction plan text of overhead transmission lines, and combined with a multi-level standard coding system, the entity information contained in the construction plan text is dynamically geospatial location encoded and temporarily stored; at the same time, the Encoder encoding method in Transformer is used to perform conventional text information semantic encoding processing on the construction text of overhead transmission lines to capture the contextual features between different texts. The specific coding measures are as follows: The set of construction text recognition results is set as follows: The corresponding text recognition results Corresponding to geographical location are coordinates ( Therefore, using a multi-level encoding system maps each text containing geographic coordinates to a discrete code: ; Secondly, the semantic information of the construction text is processed using the Encoder encoding method in Transformer, simplifying its representation as follows: ; The generated spatial location code is concatenated with features to form a vector containing joint spatial-semantic feature information, which represents the location information after the previous encoding process. Encoding with text information Set the aligned entity and text position pairs as ( The corresponding semantic vectors are extracted accordingly. and spatial encoding spliced ​​together as: By embedding the specific geospatial location code of the corresponding workpiece into the text code of the construction text through feature fusion processing, the text analysis model is made capable of perceiving information, thereby improving the accuracy and comprehensiveness of subsequent risk labeling visualization. The fused features are enhanced with key risk features through a multi-layer self-attention mechanism. The result is used as the initial hidden state input for the decoder. Finally, the decoder output format is constrained by a preset mapping rule, and the final structured risk assessment prompts are generated.

5. The visualization method for construction risk assessment of overhead transmission lines according to claim 4, characterized in that: In S2, the specific annotations for risk assessment visualization include: S21. Process the structured prompts for risk assessment and the existing BIM model according to the multi-level standard coding system; S22. The risk assessment structured prompt word-driven large language model and video generation tool are used to perform risk annotation processing and generate dynamic risk pre-show animation in the BIM model after processing by a multi-level standard coding system, thereby obtaining the risk assessment visualization results of the overhead transmission line construction scheme.

6. The visualization method for construction risk assessment of overhead transmission lines according to claim 5, characterized in that: In S21, the multi-level standard coding system transforms the existing BIM model into a standardized coded BIM model, which facilitates subsequent modeling and processing. The multi-level standard coding system is matched with the structured risk assessment prompts, and the structured risk prompts output by decoding are represented as follows: Each of the risk warning words Each of them contains a corresponding semantic feature vector. Spatial position vector The standard library contains Each engineering component is represented as: The same for each engineering component It contains the corresponding semantic feature vector Spatial position vector ; The semantic matching degree is calculated using cosine similarity. ; The matching degree of spatial location is calculated using a weighted Euclidean distance method, which transforms the spatial location vector into a three-dimensional coordinate axis representation. Based on this transformation, the corresponding weighted Eulerian distance is calculated as follows: Based on the above distance calculation formula, the spatial matching score can be obtained, expressed as: The system identifies the data information of the engineering structure where the risk is located and obtains the prompt words for the risk visualization label by matching.

7. The visualization method for construction risk assessment of overhead transmission lines according to claim 5, characterized in that: The multi-level standard coding system includes four levels: professional category coding, spatial location coding, engineering construction type coding, and construction stage coding.

8. The visualization method for construction risk assessment of overhead transmission lines according to claim 6, characterized in that: In step S21, the prompt words obtained by matching the risk visualization annotation directly drive the large language model to perform risk annotation in the BIM model through the pre-set API interface.

9. A BIM-based visualization system for overhead transmission line construction risk assessment, used to implement the visualization method for overhead transmission line construction risk assessment as described in any one of claims 1 to 8, characterized in that, The overhead transmission line construction risk assessment visualization system includes: Intelligent risk identification module: Relying on a text analysis model based on spatial location-based embedding encoding and mapping rule constraint decoding, it dynamically identifies and extracts the direct and hidden risks contained in the construction plan text of overhead transmission lines, and generates the final risk assessment structured prompt words according to the corresponding mapping rule constraints for subsequent risk visualization labeling processing. Risk visualization module: On the BIM model processed by the multi-level standard coding system, the risk assessment structured prompts obtained by the intelligent risk identification module drive the large language model to automatically adjust and process the standardized coded BIM model, thereby realizing the visualization of risk assessment in the overhead transmission line construction plan.

Images