Station building construction worker safety knowledge extraction and knowledge graph construction method

Abstract

The invention discloses a station building construction worker safety knowledge extraction and knowledge graph construction method, and relates to the technical field of station building construction, and the method comprises the steps: deploying a sensor at a construction site to collect construction environment data, collecting traditional text information in real time, and constructing a multi-modal data set; in combination with a natural language processing (NLP) technology, security-related knowledge is automatically extracted from an unstructured text, and a preliminary knowledge graph is generated. Through combination with the BIM model, the risk coefficient of each construction sub-region is calculated, visual display is carried out, and the knowledge graph is updated in time. Furthermore, by establishing a dynamic updating model, continuous optimization of the knowledge graph is realized, and a sharing platform is established, so that safety experience exchange between different construction sites is promoted, and optimization of accident prevention and safety measures is promoted. According to the method, the intelligent level of construction safety management is improved, and personalized safety support is provided for construction workers.

Description

Technical Field

[0001] The present invention relates to the technical field of station building construction, and specifically to a method for extracting safety knowledge of construction workers in station buildings and constructing a knowledge graph. Background Art

[0002] During the construction process of a station building, construction workers face various safety risks. These risks not only come from the construction operations themselves but are also significantly affected by the environmental conditions at the construction site. For example, environmental factors such as temperature, humidity, noise, and vibration, as well as the operating conditions of construction equipment, can all have an important impact on construction safety. At the same time, the risk management and control at the construction site involve multiple aspects of information and knowledge, including safety guidance content such as building structure design, equipment configuration, and electrical pipeline layout.

[0003] A large amount of construction safety knowledge is contained in various engineering document materials, such as construction safety standards and specifications, accident investigation reports, operation manuals, etc. These documents have accumulated rich safety experience and lessons. However, the management and application of this safety knowledge are often not systematic and efficient enough, resulting in their inability to fully play their roles. Especially in the safety training of construction workers, there is a lack of personalized and real-time safety knowledge support.

[0004] In addition, existing rule-based text knowledge extraction methods usually rely on manually defined rules or fixed syntactic patterns. This method is inefficient and less applicable when dealing with large-scale and complex unstructured safety texts, and it is difficult to cope with the complex and changeable risks at the construction site. Even when using natural language processing (NLP) and deep learning technologies for entity extraction, the existing technologies still have difficulty in automatically synchronously extracting the relationships between various knowledge entities, resulting in the construction and update of the knowledge graph being not timely and accurate enough. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention provides a method for extracting safety knowledge of construction workers in station buildings and constructing a knowledge graph to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for extracting safety knowledge of construction workers in station buildings and constructing a knowledge graph, including the following steps: Step 1: Deploy sensors at the station building construction site to collect construction environment data, collect various types of traditional text data in real time, and establish a multi-modal data set; Step 2: Extract data from the multi-modal data set, combine with natural language processing NLP technology, automatically extract safety-related knowledge elements from unstructured texts, identify important safety information in the texts, and convert it into structured data to initially generate a knowledge graph; Step 3: Combine the BIM model with the preliminarily generated knowledge graph, divide the station building construction area into N sub-areas, calculate and obtain the construction risk coefficient Qyxs of the sub-areas by combining the data in the multi-modal data set, update the knowledge graph through threshold comparison and analysis, and then convert the safety information in the updated knowledge graph into a visual format through the 3D visualization function of BIM to further generate a visualizable knowledge graph; Step 4: Calculate and obtain the environmental impact coefficient HJx by combining the data in the multi-modal data set through the further generated visual knowledge graph, update the visual knowledge graph through threshold comparison and analysis, and further obtain the visual knowledge graph KSG1; Step 5: Establish a dynamic update model of the knowledge graph, train and test it, and use the trained dynamic update model of the knowledge graph as data for operation to continuously and dynamically update the knowledge graph; Step 6: Based on the dynamically updated knowledge graph, establish a sharing platform for the safety knowledge graph, connect different station building construction sites, conduct online communication, enable different sites, managers and construction workers to share their safety experiences, form collective wisdom, and promote the prevention of accidents and the optimization of safety measures.

[0007] Preferably, Step 1 includes: S11: Deploy sensors at the station building construction site, and collect data through the sensors including the temperature value, humidity value, noise value and vibration value at the construction site; collect traditional texts including construction specifications, accident reports and operation manuals; and combine the collected sensor data and traditional text safety knowledge to form a multi-modal data set.

[0008] Preferably, Step 2 includes: S21: Preprocess all the collected unstructured texts, including removing noise data, standardizing terms, word segmentation and part-of-speech tagging, and convert the texts into a format that can be processed by a computer.

[0009] Preferably, Step 2 further includes: S22: Automatically extract safety-related knowledge elements from the unstructured traditional texts by using NLP technology, identify important safety information in the texts, including hazard sources, emergency measures and precautions, and convert them into structured data; S23: Use named entity recognition technology to identify key entities from the texts, identify the semantic relationships between the entities through the relationship extraction algorithm, convert the extracted entities and relationships into a graph structure, create knowledge nodes and the edges between the nodes, converge the node relationships, and preliminarily generate the knowledge graph G. The formula is as follows:

[0010]

[0011] ;

[0012] In the formula, V represents all entities extracted from the text, and E represents the relationships between the entities in the edge set entity. represents a node. represents the a-th node. represents the z-th node. represents the s-th node relationship, and R represents the set of all relationships extracted from the text.

[0013] Preferably, step three includes: S31. By combining the BIM model with the safety knowledge graph, using the three-dimensional spatial data of BIM including the structural layout, equipment configuration, and pipeline and electrical layout of the station building, after dimensionless processing, calculate and obtain the building structure risk coefficient Wstu, the equipment configuration risk coefficient Sequ, and the electrical pipeline layout risk coefficient Wqy respectively. The formulas are as follows:

[0014] In the formula, h represents the height of the building, lc represents the number of floors of the building, Sx represents the support coefficient of the building, and the common values are: temporary support is 1.5, permanent support is 1, and support missing is 2. , and represent the weight coefficients; ; In the formula, n represents the total number of devices. represents the weight of the i-th device. represents the volume of the i-th device. represents the health coefficient of the i-th device. The common values are: devices with no maintenance history within one year are 1, devices with no maintenance history from 1 to 3 years are 1.5, devices with a maintenance history from 1 to 3 years are 2, devices over 3 years are 2.5, and devices over 3 years with a maintenance history are 3. , and represent the weight coefficients;

[0015] In the formula, m represents the types of electrical pipelines. represents the length of the j-th electrical pipeline. represents the crossing coefficient of the j-th electrical pipeline. The common value is: no crossing point is 0. , Denote the safety measure coefficient of the j-th electrical pipeline. The common values are: with insulation protection and pipeline protection is 1, with insulation protection but without pipeline protection is 1.5, with pipeline protection but without insulation protection is 1.5, without insulation protection and without pipeline protection is 2. 、 and are denoted as weight coefficients.

[0016] Preferably, step three further includes: S32. Combine the BIM model with the safety knowledge graph, and divide the construction area of the substation building into N sub-areas. Combine the calculated building structure risk coefficient Wstu, equipment configuration risk coefficient Sequ, and electrical pipeline layout risk coefficient Wqy. After dimensionless processing, calculate the construction risk coefficient Qyxs of the sub-area. The formula is as follows: ; In the formula, N represents the total number of sub-areas, represents the building structure risk coefficient of the b-th sub-area, represents the equipment configuration risk coefficient of the b-th sub-area, represents the electrical pipeline layout risk coefficient of the b-th sub-area, 、 and are denoted as weight coefficients; S33. By presetting the first standard threshold Q1 in advance, compare and analyze the construction risk coefficient Qyxs of the sub-area with the first standard threshold Q1. The first evaluation results obtained include: When the construction risk coefficient Qyxs of the sub-area < the first standard threshold Q1, it means that there is no construction risk when construction personnel enter the current sub-area, no adjustment is made, and continuous monitoring is carried out; When the construction risk coefficient Qyxs of the sub-area ≥ the first standard threshold Q1, it means that there is construction risk when construction personnel enter the current sub-area, trigger the first warning instruction, and generate the first strategy including: converting the risk information into a graph node, updating the knowledge graph, and conducting a safety hazard investigation on the construction area site.

[0017] Preferably, step three further includes: S34. Convert the safety information in the updated knowledge graph into a visual format through the 3D visualization function of BIM. After dimensionless processing, further generate a visual knowledge graph KSG. The formula is as follows:

[0018]

[0019] In the formula, It is represented as a visualized BIM 3D model, and G is represented as a knowledge graph. It represents the data synthesized from the safety knowledge extracted from the knowledge graph G and the BIM model.

[0020] Preferably, step four includes: S41. Through the further generated visualized knowledge graph, combined with the data of the multimodal dataset, after dimensionless processing, calculate and obtain the environmental impact coefficient HJx. The formula is as follows: ;

[0021] In the formula, represents the temperature value of the construction environment. represents the influence degree of the environmental temperature factor on construction safety. represents the humidity value of the construction environment. The lowest safe temperature value. represents the highest safe temperature value. represents the influence degree of the environmental humidity factor on construction safety. represents the ideal humidity value. represents the safety threshold of the humidity value. represents the noise value of the construction environment. represents the influence degree of the environmental noise factor on construction safety. represents the maximum noise value. represents the safety threshold of the noise. represents the vibration value of the construction equipment. represents the influence degree of the construction equipment vibration factor on construction safety. represents the safety threshold of the construction equipment vibration. represents the maximum vibration value of the construction equipment. , , and represent the weight coefficients; S42. By presetting the second standard threshold Q2 in advance and comparing and analyzing the environmental impact coefficient HJx with the second standard threshold Q2, the second evaluation result is obtained, including: When the environmental impact coefficient HJx < the second standard threshold Q2, it means there is no risk of environmental impact during construction, no adjustment is made, and continuous monitoring is carried out; When the environmental impact coefficient HJx ≥ the second standard threshold Q2, it means there is a risk of environmental impact during construction, trigger the second warning instruction, and generate the second strategy, including: converting the environmental impact risk information into graph nodes, embedding them into the knowledge graph, updating the visualized knowledge graph KSG, and further obtaining the visualized knowledge graph KSG1; taking preventive measures and safety inspections at the construction site.

[0022] Preferably, step five includes: S51. Using a convolutional neural network, construct an initial convolutional neural network model, and train and test the initial convolutional neural network model with the visual knowledge graph KSG1 data. Use the trained initial convolutional neural network model as the knowledge graph dynamic update model. At the same time, use the intermediate layer outputs of the construction risk coefficient Qyxs and the environmental impact coefficient HJx of the sub-region as feature vectors to identify feature information, and train and test the knowledge graph dynamic update model with the obtained feature information. Use the trained knowledge graph dynamic update model for data operation to continuously and dynamically update the knowledge graph.

[0023] Preferably, step six includes: S61. Based on the dynamically updated knowledge graph, establish a sharing platform for the safety knowledge graph, connect different station house construction sites, conduct online communication, and enable workers, project managers, and technicians to share safety experiences and feedback through functions such as comments, ratings, and experience exchanges. Integrate safety knowledge and historical safety events on the platform, and through a traceability mechanism, identify the root causes of accidents and promote accident prevention and safety measure optimization.

[0024] The present invention provides a method for extracting safety knowledge of station house construction workers and constructing a knowledge graph, which has the following beneficial effects: (1) For the method for extracting safety knowledge of station house construction workers and constructing a knowledge graph, by combining multi-modal data sets including sensor data and traditional texts, this method can collect and analyze various data on the construction site in real time, and realize the extraction and update of dynamic and safety knowledge. Construction workers can obtain more targeted and safer knowledge support. Especially in a complex construction environment, they can timely respond to potential risks and reduce safety hazards.

[0025] (2) For the method for extracting safety knowledge of station house construction workers and constructing a knowledge graph, by comprehensively evaluating factors such as the building structure, equipment configuration, and electrical pipeline layout on the construction site with environmental impacts including temperature values, humidity values, noise values, and vibration values, and presenting them visually through a BIM model. This multi-dimensional evaluation method can help construction personnel and managers comprehensively understand the risks and environmental factors on the construction site, so as to provide a more accurate basis for risk control and safety management.

[0026] (3)The method for extracting safety knowledge of construction workers in this station building and constructing a knowledge graph uses deep learning technologies such as convolutional neural networks to establish a dynamic update model of the knowledge graph. The system can automatically update the knowledge graph in real time according to the changes in on-site data, the dynamic adjustment of construction risks, and the changes in environmental impacts. This automated update method greatly reduces manual intervention, improves the timeliness and accuracy of the knowledge graph, and provides safety guidance that better meets the actual needs of construction workers.

[0027] (4)The method for extracting safety knowledge of construction workers in this station building and constructing a knowledge graph is based on the dynamically updated knowledge graph. This method also provides a safety knowledge sharing platform that allows online communication and experience sharing between different station building construction sites. Through the feedback and interaction of workers, project managers, and technical personnel, integrating the safety experience and historical events of different construction sites can not only optimize safety measures but also improve the safety management level of the entire industry and promote the formation and improvement of construction safety culture. Description of the Drawings

[0028] Figure 1 Schematic diagram of the steps of the method for extracting safety knowledge of construction workers in the station building of the present invention and constructing a knowledge graph. Detailed Embodiments

[0029] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying 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 of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0030] Embodiment 1 Please refer to Figure 1 , the present invention provides a method for extracting safety knowledge of construction workers in a station building and constructing a knowledge graph, including the following steps: Step 1: Deploy sensors at the construction site of the station building to collect construction environment data, collect various types of traditional text data in real time, and establish a multi-modal data set; Step 2: Extract data from the multi-modal data set, combine natural language processing (NLP) technology, automatically extract safety-related knowledge elements from unstructured text, identify important safety information in the text, and convert it into structured data to initially generate a knowledge graph; Step 3: Combine the BIM model with the preliminarily generated knowledge graph, divide the station building construction area into N sub-areas, calculate and obtain the construction risk coefficient Qyxs of the sub-areas by combining the data of the multi-modal data set, update the knowledge graph through threshold comparison and analysis, and then convert the safety information in the updated knowledge graph into a visual format through the 3D visualization function of BIM to further generate a visualizable knowledge graph; Step 4: Calculate and obtain the environmental impact coefficient HJx by combining the data of the further generated visual knowledge graph with the data of the multi-modal data set, update the visual knowledge graph through threshold comparison and analysis, and further obtain the visual knowledge graph KSG1; Step 5: Establish a dynamic update model of the knowledge graph, conduct training and testing, and use the trained dynamic update model of the knowledge graph as data to run and continuously update the knowledge graph dynamically; Step 6: Based on the dynamically updated knowledge graph, establish a sharing platform for the safety knowledge graph, connect different station building construction sites, conduct online communication, enable different construction sites, managers and construction workers to share their safety experiences, form collective wisdom, and promote the prevention of accidents and the optimization of safety measures.

[0031] In this embodiment, through multi-modal data collection and processing, combined with the visualization functions of the BIM model and the knowledge graph, the present invention can calculate and evaluate the risk coefficient and environmental impact coefficient of the construction site in real time. When the risk coefficient or environmental impact coefficient exceeds the preset threshold, the system will automatically update the knowledge graph and generate a safety warning. This real-time risk monitoring and warning mechanism effectively improves the safety management ability of the construction site, ensures that construction personnel can obtain targeted safety guidance in a timely manner, and thus greatly reduces the probability of safety accidents.

[0032] Embodiment 2 This embodiment is an explanatory description based on Embodiment 1. Specifically, Step 1 includes: S11: Deploy sensors at the station building construction site, and collect data through the sensors including the temperature value, humidity value, noise value and vibration value of the construction site; collect traditional texts including construction specifications, accident reports and operation manuals; and combine the collected sensor data and traditional text safety knowledge to form a multi-modal data set.

[0033] In this embodiment, by deploying sensors at the construction site of the station building and combining the collected environmental data of temperature, humidity, noise, and vibration, the system can comprehensively and real-time monitor various safety indicators at the construction site. At the same time, the integration of traditional text data such as construction specifications, accident reports, and operation manuals forms a multi-modal data set. Such data fusion not only provides more accurate and multi-dimensional information support for safety risk assessment, but also improves the decision-making efficiency of on-site management, ensuring the early detection and timely response to potential risks during the construction process.

[0034] Embodiment 3 This embodiment is an explanatory description based on Embodiment 2. Specifically, Step 2 includes: S21. Preprocess all the collected unstructured text, including removing noise data, standardizing terms, word segmentation, and part-of-speech tagging, and convert the text into a format that can be processed by a computer.

[0035] In this embodiment, by preprocessing the collected unstructured text, including removing noise data, standardizing terms, word segmentation, and part-of-speech tagging, the quality and usability of the text data can be significantly improved. Preprocess the image data to make the data consistent. This process enables the originally messy and information-dense safety documents to be converted into a standardized and structured data format, facilitating subsequent computer processing and analysis.

[0036] Embodiment 4 This embodiment is an explanatory description based on Embodiment 3. Specifically, Step 2 further includes: S22. Automatically extract safety-related knowledge elements from the unstructured traditional text by using NLP technology, identify important safety information in the text, including hazard sources, emergency measures, and precautions, and convert them into structured data; S23. Use named entity recognition technology to identify key entities from the text, identify the semantic relationships between entities through a relationship extraction algorithm, convert the extracted entities and relationships into a graph structure, create knowledge nodes and the edges between the nodes, and converge the node relationships to initially generate a knowledge graph G. The formula is as follows:

[0037]

[0038] ;

[0039] In the formula, V represents all the entities extracted from the text, E represents the relationships between the edge set entities, represents a node, represents the a-th node, represents the z-th node, represents the s-th node relationship, and R represents the set of all relationships extracted from the text.

[0040] In this embodiment, by using NLP technology to automatically extract safety-related knowledge elements from unstructured text, including hazard sources, emergency measures, and precautions, and converting them into structured data, it helps to quickly identify and organize safety information, avoiding the cumbersome and inefficient manual sorting. At the same time, by using named entity recognition technology and relationship extraction algorithms, a safety knowledge graph can be automatically identified and constructed to accurately depict the relationships and connections between entities.

[0041] Embodiment 5 This embodiment is an explanatory description based on Embodiment 4. Specifically, Step 3 includes: S31. By combining the BIM model with the safety knowledge graph, using the three-dimensional spatial data of BIM, including the structural layout, equipment configuration, and pipeline and electrical layout of the station building, after dimensionless processing, the building structure risk coefficient Wstu, equipment configuration risk coefficient Sequ, and electrical pipeline layout risk coefficient Wqy are calculated respectively. The formulas are as follows:

[0042] In the formula, h represents the height of the building, lc represents the number of floors of the building, Sx represents the support coefficient of the building, and the common values are: 1.5 for temporary support, 1 for permanent support, and 2 for lack of support. , and represent weight coefficients, which are adjusted and set by the user. , , and ; ; In the formula, n represents the total number of devices. represents the weight of the i-th device. represents the volume of the i-th device. represents the health coefficient of the i-th device. The common values are: 1 for devices with no maintenance history within one year, 1.5 for devices with no maintenance history from 1 to 3 years, 2 for devices with a maintenance history from 1 to 3 years, 2.5 for devices over 3 years, and 3 for devices over 3 years with a maintenance history. , and represent weight coefficients, which are adjusted and set by the user. , , and ;

[0043] In the formula, m represents the type of electrical pipeline, represents the length of the j-th type of electrical pipeline, represents the crossing coefficient of the j-th type of electrical pipeline, and the common value is: 0 for no crossing point, , represents the safety measure coefficient of the j-th type of electrical pipeline, and the common value is: 1 for having insulation protection and pipeline protection, 1.5 for having insulation protection but no pipeline protection, 1.5 for having pipeline protection but no insulation protection, and 2 for having no insulation protection and no pipeline protection, , and represent the weight coefficients, which are adjusted and set by the user, , , , and .

[0044] In this embodiment, by combining the BIM model with the safety knowledge graph, detailed risk assessment can be carried out using three-dimensional spatial data including: the structural layout of the building, equipment configuration, and electrical pipeline layout. Based on the building structure risk coefficient, equipment configuration risk coefficient, and electrical pipeline layout risk coefficient calculated from the BIM model, the potential safety risks at the construction site can be accurately reflected. This assessment method takes into account multiple factors such as the size, weight, and health status of each component, which helps to comprehensively and accurately identify and predict risks.

[0045] Embodiment 6 This embodiment is an explanatory description carried out in Embodiment 5. Specifically, Step 3 further includes: S32. By combining the BIM model with the safety knowledge graph and dividing the station house construction area into N sub-areas, and combining the calculated building structure risk coefficient Wstu, equipment configuration risk coefficient Sequ, and electrical pipeline layout risk coefficient Wqy, after dimensionless processing, calculate the construction risk coefficient Qyxs of the sub-area, and the formula is as follows: ; In the formula, N represents the total number of sub-areas, represents the building structure risk coefficient of the b-th sub-area, represents the equipment configuration risk coefficient of the b-th sub-area, represents the electrical pipeline layout risk coefficient of the b-th sub-area, , and represent the weight coefficients, which are adjusted and set by the user, , , , and ; S33. By presetting a first standard threshold Q1 in advance and comparing and analyzing the construction risk coefficient Qyxs of the sub-region with the first standard threshold Q1, the first evaluation result obtained includes: When the construction risk coefficient Qyxs of the sub-region < the first standard threshold Q1, it indicates that there is no construction risk when construction personnel enter the current sub-region, no adjustment is made, and continuous monitoring is carried out; When the construction risk coefficient Qyxs of the sub-region ≥ the first standard threshold Q1, it indicates that there is construction risk when construction personnel enter the current sub-region, triggering a first warning instruction, and generating a first strategy including: converting the risk information into a graph node, updating the knowledge graph, and conducting a safety hazard investigation on the construction area site.

[0046] In this embodiment, by dividing the station building construction area into multiple sub-regions and calculating the risk coefficients in combination with the building structure, equipment configuration, and electrical pipeline layout, the construction risks of each sub-region can be dynamically evaluated. By setting a standard threshold and comparing it with the calculated construction risk coefficient, the safety status of each region can be judged in real time. If the construction risk exceeds the threshold, the system will automatically trigger an alarm to remind construction personnel and managers to take corresponding safety measures for risk investigation and rectification. This dynamic risk management method can not only timely discover potential safety hazards but also achieve precise safety control during the construction process.

[0047] Embodiment 7 This embodiment is an explanatory description based on Embodiment 6. Specifically, step three further includes: S34. Through the three-dimensional visualization function of BIM, the safety information in the updated knowledge graph is converted into a visual format. After dimensionless processing, a visual knowledge graph KSG is further generated, and the formula is as follows:

[0048]

[0049] In the formula, represents the BIM three-dimensional model after visualization processing, G represents the knowledge graph, represents the data synthesized from the safety knowledge extracted from the knowledge graph G and the BIM model.

[0050] In this embodiment, by combining the updated safety knowledge graph with the BIM three-dimensional model and utilizing the three-dimensional visualization function of BIM, complex safety information can be converted into an intuitive visual format. This visual knowledge graph not only enables construction personnel to understand the safety status of the construction site in real time but also clearly shows potential risk areas and safety hazards. Through the visual presentation of the three-dimensional model, construction personnel can more effectively identify and prevent possible safety risks.

[0051] Embodiment 8 This embodiment is an explanatory description based on Embodiment 7. Specifically, Step 4 includes: S41. Through the further generated visual knowledge graph, combined with the data of the multimodal dataset, after dimensionless processing, the environmental impact coefficient HJx is calculated as follows: ;

[0052] In the formula, represents the temperature value of the construction environment, represents the degree of influence of the environmental temperature factor on construction safety, represents the humidity value of the construction environment, the minimum safety temperature value, represents the maximum safety temperature value, represents the degree of influence of the environmental humidity factor on construction safety, represents the ideal humidity value, represents the safety threshold of the humidity value, represents the noise value of the construction environment, represents the degree of influence of the environmental noise factor on construction safety, represents the maximum noise value, represents the safety threshold of the noise, represents the vibration value of the construction equipment, represents the degree of influence of the construction equipment vibration factor on construction safety, represents the safety threshold of the construction equipment vibration, represents the maximum vibration value of the construction equipment, . . and represent weight coefficients, which are adjusted and set by the user, , , , and ; S42. By presetting the second standard threshold Q2 in advance and comparing and analyzing the environmental impact coefficient HJx with the second standard threshold Q2, the second evaluation result is obtained, including: When the environmental impact coefficient HJx < the second standard threshold Q2, it indicates that there is no risk of environmental impact during construction, and no adjustment is made, and continuous monitoring is carried out; When the environmental impact coefficient HJx ≥ the second standard threshold Q2, it indicates that there is a risk of environmental impact during construction, triggering the second warning instruction, and generating the second strategy including: converting the environmental impact risk information into a graph node, embedding it into the knowledge graph, updating the visual knowledge graph KSG, and further obtaining the visual knowledge graph KSG1; taking preventive measures and safety inspections at the construction site.

[0053] In this embodiment, by calculating the environmental impact coefficient and comparing it with the preset threshold, potential safety risks in the construction environment can be identified in a timely manner, such as factors like excessive temperature, humidity, noise, or equipment vibration. This method can accurately evaluate the impact of environmental factors on construction safety and give early warnings of possible safety hazards. When the environmental impact coefficient exceeds the safety threshold, the system can automatically trigger a warning and update the visual knowledge graph, ensuring that construction personnel can obtain the latest safety information in a timely manner and take necessary preventive measures. This process effectively enhances the construction site's ability to respond to environmental risks, improves the overall safety management level, and reduces safety accidents caused by environmental factors.

[0054] Embodiment 9 This embodiment is an explanatory description based on Embodiment 8. Specifically, Step 5 includes: S51. Using a convolutional neural network, construct an initial convolutional neural network model, train and test the initial convolutional neural network model with the data of the visual knowledge graph KSG1, and use the trained initial convolutional neural network model as the knowledge graph dynamic update model. At the same time, use the intermediate layer output of the construction risk coefficient Qyxs and the environmental impact coefficient HJx of the sub-region as the feature vector to identify the feature information, and train and test the knowledge graph dynamic update model with the obtained feature information, and use the trained knowledge graph dynamic update model as the data operation to continuously and dynamically update the knowledge graph.

[0055] In this embodiment, by training the data of the visual knowledge graph KSG1 through a convolutional neural network (CNN) to construct a dynamic update model, the characteristics of complex data such as construction risks and environmental impacts can be automatically learned. This method continuously trains and tests the model to optimize the knowledge graph in real time to ensure that it reflects the latest risk situation at the construction site. Using the intermediate layer output of the construction risk coefficient Qyxs and the environmental impact coefficient HJx of the sub-region as the feature vector, the model can accurately identify potential safety hazards and risk changes, providing timely and accurate safety decision-making support for construction management personnel. This technology greatly improves the adaptive ability and response speed of the knowledge graph, ensuring that the safety management at the construction site always maintains the best state in dynamic changes.

[0056] Example 10 This embodiment is explained in Embodiment 9. Specifically, step 6 includes: S61. Based on the dynamically updated knowledge graph, a shared platform for the safety knowledge graph is established to connect different station construction sites for online communication. Through the comment, rating and experience exchange functions, workers, project managers and technicians can share safety experiences and feedback. Safety knowledge and historical safety events are integrated on the platform. Through the traceability mechanism, the root causes of accidents can be identified to promote accident prevention and optimization of safety measures.

[0057] In this embodiment, by establishing a shared platform based on a dynamically updated knowledge graph, information can be exchanged between construction sites of different stations, and workers, project managers and technicians can share safety experiences and feedback online. This platform integrates historical safety events and safety knowledge, and promotes the formation of collective wisdom with the help of comment, scoring and experience exchange functions. Through the accident tracing mechanism, the platform can deeply analyze and identify the root causes of accidents, thereby promoting the continuous optimization of accident prevention measures and improving the overall safety management level of the construction site.

[0058] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technicians in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0059] The above formulas are obtained by collecting a large amount of data for software simulation and selecting a formula that is close to the actual value. The coefficients in the formula are set by technical personnel in this field according to actual conditions. The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited to this. Any technical personnel familiar with the technical field within the technical scope disclosed by the present invention, according to the technical solution and the inventive concept of the present invention, make equivalent replacement or change, which should be covered within the protection scope of the present invention.

Claims

1. A method for extracting safety knowledge of construction workers in the station building and constructing a knowledge graph, characterized in that, It includes the following steps: Step 1: Deploy sensors at the construction site of the station building to collect construction environment data, collect various types of traditional text data in real time, and establish a multi-modal data set; Step 2: By extracting the data of the multi-modal data set and combining with natural language processing NLP technology, automatically extract safety-related knowledge elements from unstructured text, identify important safety information in the text, and transform it into structured data, and initially generate a knowledge graph; Step 3: Combine the BIM model with the initially generated knowledge graph, divide the construction area of the station building into N sub-areas, combine the data of the multi-modal data set, calculate and obtain the construction risk coefficient Qyxs of the sub-areas, update the knowledge graph through threshold comparison analysis, and then convert the safety information in the updated knowledge graph into a visual format through the three-dimensional visualization function of BIM, and further generate a visualizable knowledge graph; Step 4: Through the further generated visual knowledge graph, combine the data of the multi-modal data set, calculate and obtain the environmental impact coefficient HJx, update the visual knowledge graph through threshold comparison analysis, and further obtain the visual knowledge graph KSG1; Step 5: Establish a dynamic update model of the knowledge graph, and conduct training and testing. Use the trained dynamic update model of the knowledge graph as data to run and continuously update the knowledge graph dynamically; Step 6: Based on the dynamically updated knowledge graph, establish a sharing platform for the safety knowledge graph, connect different station building construction sites, conduct online communication, and let different construction sites, managers and construction workers share their safety experiences, form collective wisdom, and promote accident prevention and safety measure optimization.

2. The method for extracting safety knowledge of construction workers in the station building and constructing the knowledge graph according to claim 1, characterized in that Step 1 includes: S11: Deploy sensors at the construction site of the station building. The data collected by the sensors include the temperature value, humidity value, noise value and vibration value at the construction site; collect traditional texts including construction specifications, accident reports and operation manuals; and combine the collected sensor data and traditional text safety knowledge to form a multi-modal data set.

3. The method for extracting safety knowledge of construction workers in a station building and constructing a knowledge graph according to claim 2, wherein Step 2 includes: S21: Preprocess all the collected unstructured text, including removing noise data, standardizing terms, word segmentation and part-of-speech tagging, and convert the text into a format that can be processed by a computer.

4. The method for extracting safety knowledge of construction workers in the station building and constructing the knowledge graph according to claim 3, wherein, Step 2 also includes: S22: Automatically extract safety-related knowledge elements from unstructured traditional text by using NLP technology, identify important safety information in the text, including hazard sources, emergency measures and precautions, and transform them into structured data; S23: Use named entity recognition technology to identify key entities from the text, identify the semantic relationships between entities through relationship extraction algorithms, transform the extracted entities and relationships into a graph structure, create knowledge nodes and edges between nodes, and converge the node relationships to initially generate a knowledge graph G. The formula is as follows: ； Wherein, V represents all entities extracted from the text, and E represents the relationships between the edge set entities. represents a node. represents the a-th node. represents the z-th node. represents the s-th node relationship, and R represents the set of all relationships extracted from the text.

5. The method for extracting safety knowledge of construction workers in a station building and constructing a knowledge graph according to claim 4, characterized in that Step 3 includes: S31. By combining the BIM model with the safety knowledge graph, using the three-dimensional spatial data of BIM, including the structural layout, equipment configuration, and pipeline and electrical layout of the station building, after dimensionless processing, calculate and obtain the building structure risk coefficient Wstu, the equipment configuration risk coefficient Sequ, and the electrical pipeline layout risk coefficient Wqy respectively. The formulas are as follows: Where h represents the height of the building, lc represents the number of floors of the building, and Sx represents the support coefficient of the building, 、 and are represented as weight coefficients; ； Where n represents the total number of devices, represents the weight of the i-th device, represents the volume of the i-th device, represents the health factor of the i-th device, 、 and represent weight coefficients; Wherein, m represents the type of electrical conduit, represents the length of the j-th type of electrical conduit, represents the intersection coefficient of the j-th type of electrical conduit, represents the safety measure coefficient of the j-th type of electrical conduit, 、 and represent weight coefficients.

6. The method for extracting safety knowledge of construction workers in the station building and constructing the knowledge graph according to claim 5, characterized in that, Step three also includes: S32. By combining the BIM model with the safety knowledge graph, and dividing the station construction area into N sub-areas, combining the calculated building structure risk coefficient Wstu, equipment configuration risk coefficient Sequ, and electrical pipeline layout risk coefficient Wqy, after dimensionless processing, calculate and obtain the construction risk coefficient Qyxs of the sub-areas. The formula is as follows: ； Where N represents the total number of sub-regions, represents the building structure risk coefficient of the b-th sub-region, represents the equipment configuration risk coefficient of the b-th sub-region, represents the electrical pipeline layout risk coefficient of the b-th sub-region, , and represent the weight coefficients; S33. By presetting the first standard threshold Q1 in advance, and comparing and analyzing the construction risk coefficient Qyxs of the sub-areas with the first standard threshold Q1, obtain the first evaluation result including: When the construction risk coefficient Qyxs of the sub-area < the first standard threshold Q1, it means that there is no construction risk when construction personnel enter the current sub-area, no adjustment is made, and continuous monitoring is carried out; When the construction risk coefficient Qyxs of the sub-area ≥ the first standard threshold Q1, it means that there is construction risk when construction personnel enter the current sub-area, trigger the first warning instruction, and generate the first strategy including: converting the risk information into a graph node, updating the knowledge graph, and conducting a safety hazard inspection on the construction area site.

7. The method for extracting safety knowledge of construction workers in the station building and constructing the knowledge graph according to claim 6, wherein, Step three also includes: S34. By converting the safety information in the updated knowledge graph into a visual format through the three-dimensional visualization function of BIM, after dimensionless processing, further generate the visual knowledge graph KSG. The formula is as follows: In the formula, represents the visualized BIM three-dimensional model, and G represents the knowledge graph. represents the data synthesized from the safety knowledge extracted from the knowledge graph G and the BIM model.

8. The method for extracting safety knowledge of construction workers in the station building and constructing the knowledge graph according to claim 7, characterized in that, Step four includes: S41. By combining the further generated visual knowledge graph with the data of the multi-modal data set, after dimensionless processing, calculate and obtain the environmental impact coefficient HJx. The formula is as follows: ； In the formula, represents the temperature value of the construction environment, represents the influence degree of the environmental temperature factor on construction safety, represents the humidity value of the construction environment, the minimum safety temperature value, represents the maximum safety temperature value, represents the influence degree of the environmental humidity factor on construction safety, represents the ideal humidity value, represents the safety threshold of the humidity value, represents the noise value of the construction environment, represents the influence degree of the environmental noise factor on construction safety, represents the maximum noise value, represents the safety threshold of the noise, represents the vibration value of the construction equipment, represents the influence degree of the construction equipment vibration factor on construction safety, represents the safety threshold of the construction equipment vibration, represents the maximum vibration value of the construction equipment, , , and are represented as weight coefficients; S42. By presetting the second standard threshold Q2 in advance, and comparing and analyzing the environmental impact coefficient HJx with the second standard threshold Q2, obtain the second evaluation result including: When the environmental impact coefficient HJx < the second standard threshold Q2, it means that there is no risk of environmental impact during construction, no adjustment is made, and continuous monitoring is carried out; When the environmental impact coefficient HJx ≥ the second standard threshold Q2, it means that there is a risk of environmental impact during construction, trigger the second warning instruction, and generate the second strategy including: converting the environmental impact risk information into a graph node, embedding it into the knowledge graph, updating the visual knowledge graph KSG, and further obtaining the visual knowledge graph KSG1; taking preventive measures and safety inspections on the construction site.

9. The method for extracting safety knowledge of construction workers in the station building and constructing the knowledge graph according to claim 8, wherein, Step five includes: S51. Use a convolutional neural network to construct an initial convolutional neural network model, train and test the initial convolutional neural network model with the visualized knowledge graph KSG1 data, and use the trained initial convolutional neural network model as a knowledge graph dynamic update model. At the same time, use the intermediate layer outputs of the construction risk coefficient Qyxs and the environmental impact coefficient HJx of the sub-region as feature vectors to identify feature information, and train and test the knowledge graph dynamic update model with the obtained feature information. Use the trained knowledge graph dynamic update model for data operation to continuously and dynamically update the knowledge graph.

10. The method for extracting safety knowledge of construction workers in a station building and constructing a knowledge graph according to claim 9, characterized in that, Step six includes: S61. Based on the dynamically updated knowledge graph, establish a sharing platform for the safety knowledge graph, connect different station building construction sites for online communication, and allow workers, project managers, and technicians to share safety experiences and feedback through functions such as comments, ratings, and experience exchanges. Integrate safety knowledge and historical safety events on the platform, and through the traceability mechanism, identify the root causes of accidents to promote accident prevention and optimization of safety measures.

Images