Bridge construction safety early warning system and method based on artificial intelligence

Abstract

The invention relates to a bridge construction safety early warning system based on artificial intelligence. The system comprises a data acquisition layer used for acquiring actual data in a bridge construction stage; the risk identification layer is used for calling a pre-trained AI model to analyze the data according to the actual data acquired by the data acquisition layer, and identifying potential safety risk data; the risk decision-making layer is used for importing the security risk data into a pre-trained risk model when the risk identification layer identifies the security risk, and determining a risk priority and a coping strategy; and the early warning notification layer is used for receiving the risk priority and the coping strategy sent by the risk decision-making layer and sending a prompt and early warning notification. According to the invention, early warning notification can be efficiently carried out on security risks in real time, and corresponding decision suggestions can be given.

Description

Technical Field

[0001] The present invention relates to a bridge construction safety early warning system and method based on artificial intelligence, belonging to the technical field of artificial intelligence safety early warning. Background Art

[0002] In the field of bridge construction, traditional safety management mainly relies on means such as manual inspections, fixed monitoring devices, and simple sensor monitoring, and it faces many difficulties.

[0003] Firstly, the construction site is complex and changeable, with a large number of dangerous working environments such as high altitude, deep foundation pits, and over water. There are many and scattered safety risk points. Relying on manual inspections and experience judgments to detect potential hazards is inefficient and prone to omissions.

[0004] Secondly, a huge amount of data is generated at the construction site, including structural data, environmental data, equipment data, power data, image data, as well as physiological data and location information of construction workers.

[0005] Traditional methods are difficult to effectively integrate and analyze these multi-source heterogeneous data, and cannot timely detect potential safety risks. At the same time, the safety risks faced in different stages of bridge construction are different, and traditional monitoring methods cannot dynamically adjust the monitoring focus and early warning strategies according to the changes in the construction stage.

[0006] Therefore, developing a bridge construction safety early warning system based on artificial intelligence has important practical significance and broad application prospects for improving the safety management level of bridge construction, ensuring the lives of construction workers, and the quality of the project. Summary of the Invention

[0007] The purpose of the present invention is to provide a bridge construction safety early warning system based on artificial intelligence, which can have real-time and efficient early warning notifications of safety risks and can give corresponding decision-making suggestions.

[0008] To solve the above problems, the technical solution adopted by the present invention is: a bridge construction safety early warning system based on artificial intelligence, including

[0009] A data acquisition layer for collecting actual data during the bridge construction stage;

[0010] A risk identification layer for analyzing the data collected by the data acquisition layer by calling a pre-trained AI model to identify potential safety risk data;

[0011] A risk decision layer for, when the risk identification layer identifies a safety risk, importing the safety risk data into a pre-trained risk model to determine the risk priority and response strategy;

[0012] The early warning notification layer is used to receive the risk priority and response strategies sent by the risk decision-making layer, and issue prompts and early warning notifications.

[0013] As a further improvement of the present invention, it further includes

[0014] The visualization display layer is used to generate a three-dimensional model of the construction site based on the actual data collected by the data collection layer, and associate the existing construction workers and working equipment in the generated three-dimensional model according to the risk priority and response strategies determined by the risk decision-making layer, and present the construction site situation in the three-dimensional model in real time.

[0015] As a further improvement of the present invention, the data collection layer includes

[0016] A camera, which is used to collect image and video data of the construction site to obtain construction worker operation behavior data and equipment operation status data;

[0017] A sensor network, which includes multiple sensors and is used to collect one or more of the following data: structural data, geological exploration data, foundation settlement data, groundwater level data, pier inclination data, concrete pouring data, beam stress and strain data, bridge deck paving data, equipment operation parameters, lighting system data, drainage system data, and protection facility data in different areas of the bridge construction site;

[0018] Wearable devices, which are used to collect the physiological data and location information of construction workers.

[0019] As a further improvement of the present invention, the risk identification layer includes

[0020] The construction stage division module, which is used to divide the bridge construction project into a foundation construction stage, a lower structure construction stage, an upper structure construction stage, and an accessory facility installation stage;

[0021] The algorithm model storage module, which is used to store the pre-trained AI models corresponding to different construction stages;

[0022] The data screening module, which is used to screen out the actual data corresponding to the current construction stage from the actual data of each construction stage collected by the data collection layer;

[0023] The stage risk identification module, which is used to retrieve the pre-trained AI model corresponding to the current construction stage, process the data screened by the data screening module, and identify potential safety risk data.

[0024] As a further improvement of the present invention, the risk identification layer further includes a conventional risk identification module, and this conventional risk identification module includes

[0025] The personnel risk identification unit is used to demarcate dangerous operation areas during each construction stage and region, analyze the image data of the construction site, judge whether the safety equipment worn by construction personnel and their locations are compliant, and control the alarm equipment in the dangerous operation area to send corresponding alarm information when construction personnel enter the dangerous operation area;

[0026] The equipment risk identification unit is used to connect all construction equipment in bridge construction, monitor the equipment operation data in real time, judge whether the operation data of each construction equipment exceeds the threshold, and control the alarm equipment in the area where the construction equipment is located to send corresponding alarm information when there is equipment operation data exceeding the threshold;

[0027] The power risk identification unit is used to connect the distribution box at the construction site, analyze the current, voltage fluctuations and key point temperature data. If there are abnormal fluctuations or overheating, it is judged that there is a fire risk, and the power supply is immediately disconnected and the alarm equipment in the area where the power supply is located is controlled to alarm.

[0028] As a further improvement of the present invention, the conventional risk identification module further includes an environmental risk identification unit, which is used to calculate the environmental safety risk assessment value of each area according to the temperature, humidity, wind speed, precipitation data, geological data, surrounding environment data and construction personnel activity frequency at each area of the construction site. If the environmental safety risk assessment value of a certain area exceeds the threshold, it is considered that there is an environmental safety risk in this area;

[0029] The formula for the environmental risk identification unit to calculate the environmental safety risk assessment value is:

[0030] R i =ω1T i +ω2H i +ω2Wi+ω4R i +ω5G i +ω6E i +ω7P i +σ

[0031] Where R i represents the environmental safety risk assessment value of the i-th area, ω1, ω2, ω3, ω4, ω5, ω6, ω7 respectively represent the weight coefficients of temperature, humidity, wind speed, precipitation, geological data, surrounding environment data and construction personnel activity frequency, T i 、H i 、W i 、R i 、G i 、E i 、P i respectively represent the side effect indexes of the actual temperature data, humidity data, wind speed data, precipitation data, geological condition data, surrounding environment data and construction personnel activity frequency in the i-th area on construction safety, and σ represents the error.

[0032] As a further improvement of the present invention, the risk decision-making layer includes

[0033] a risk assessment module, which quantitatively assesses the safety risk data identified by the risk identification layer to determine the level and impact degree of the risk;

[0034] a priority determination module, which is used to determine the risk priority order according to the level and impact degree of the risk when there are multiple risk data;

[0035] a coping strategy generation module, which is used to generate targeted and operable coping strategies for different levels and priority risks according to the pre-imported decision model, the actual construction conditions, site conditions and relevant specification standards of the bridge construction project;

[0036] a decision result output module, which is used to integrate the risk priority and coping strategies into a decision result and transmit it to the early warning notification layer.

[0037] As a further improvement of the present invention, the early warning notification layer includes

[0038] an early warning information generation module, which is used to generate detailed early warning content according to information such as the type, level, location and time of the risk;

[0039] a multi-channel early warning notification module, which is used to transmit the early warning letter content to the on-site construction personnel and the responsible personnel in the relevant area;

[0040] an early warning feedback and recording module, which is used to receive the feedback from relevant personnel on the early warning information and record and store the release and processing process of the early warning information.

[0041] As a further improvement of the present invention, the visualization display layer includes

[0042] a three-dimensional model construction module, which is used to generate a three-dimensional visualization model of each area of the bridge construction project according to a preset ratio, scale it according to the actual data collected by the data collection layer according to the actual ratio size, and present the spatial positions and states of the construction personnel and working equipment in the visual three-dimensional model;

[0043] an early warning information module, which is used to generate warning signs with gradually deepening rendering colors from light to dark according to the risk level in the corresponding area in the visual three-dimensional model when there are safety risks during the construction process, and display the early warning information notification bar;

[0044] a historical data query module, which is used to record, query and display the historical safety data during the bridge construction process.

[0045] Another object of the present invention is to provide a safety warning method for bridge construction projects based on artificial intelligence, which adopts a safety warning system for bridge construction based on artificial intelligence, and includes the following steps:

[0046] S1. Collect various construction data of different areas at the construction site of the bridge construction project in real time;

[0047] S2. According to the construction stage and the category of construction data, retrieve the corresponding AI model to identify potential safety risks in each area. When there is a safety risk in a certain area, control the on-site alarm device in that area to send out an alarm message;

[0048] S3. Generate a three-dimensional model of the construction site of the bridge construction project at different construction stages, and identify the spatial positions of construction personnel and construction equipment at the construction site according to the construction data. Dynamically present the spatial positions and states of the operators and operating equipment in the visual three-dimensional model. When there is a safety risk in a certain area, convey the risk information and the corresponding recommended risk decision to the relevant personnel and display relevant warning information at the corresponding position in the three-dimensional model, and record and display the accident data in the model list after the accident ends.

[0049] In summary, the beneficial effects of the present invention are as follows: The present invention is equipped with a risk decision-making function, which quantitatively evaluates and prioritizes the identified safety risks, and generates corresponding coping strategies in combination with the actual situation and specification standards. It provides a scientific and reasonable decision-making basis for construction management personnel, helps them quickly and accurately formulate effective risk control measures, improves the scientificity and effectiveness of decision-making, and enhances the ability to control construction safety risks.

[0050] The visual display layer constructed by the present invention can generate a three-dimensional model of bridge construction and dynamically present the actual situation of the construction site in real time according to the construction data, including information such as the spatial positions and states of construction personnel and equipment. When there is a safety risk, the system can also display relevant warning information at the corresponding position in the three-dimensional model and intuitively display the decision results of the risk decision-making layer in the form of a notification bar. This visual display method enables construction management personnel to more intuitively and clearly understand the safety status and risk distribution of the construction site, facilitating their overall control and refined management. At the same time, it also provides a more intuitive safety warning for construction personnel, enhancing their safety awareness.

[0051] The present invention can store models and information related to safety risks for ready access at any time. This not only facilitates risk traceability and accident investigation during the construction process but also accumulates valuable data resources for subsequent risk analysis and experience summary. Through in-depth mining and analysis of these historical data, the risk identification model and early warning strategy can be further optimized, improving the early warning accuracy and reliability of the system. At the same time, it provides rich experience for the safety management of bridge construction projects, contributing to the improvement of the safety management level of the entire industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a framework diagram of the present invention.

[0053] Figure 2 is a flowchart of the early warning method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0054] The following further describes the specific embodiments of the present invention with reference to the accompanying drawings.

[0055] As Figure 1 shown, the bridge construction safety early warning system based on artificial intelligence includes a data acquisition layer, a risk identification layer, a risk decision layer, an early warning notification layer, and a visualization display layer.

[0056] The data acquisition layer is used to collect actual data in different areas during the bridge construction stage. The actual data includes structural data, environmental data, equipment data, power data, image data, as well as physiological data and location information of construction workers during the bridge construction process.

[0057] The risk identification layer is connected to the data acquisition layer, and the connection between the two can be wired or wireless. The risk identification layer is used to analyze the actual data collected by the data acquisition layer in each construction stage by calling a pre-trained AI model, and identify potential safety risk data. In the present invention, the bridge construction project is divided into a foundation construction stage, a lower structure construction stage, an upper structure construction stage, and an auxiliary facility installation stage, and pre-trained AI models corresponding to each stage are stored in the foundation construction stage, the lower structure construction stage, the upper structure construction stage, and the auxiliary facility installation stage.

[0058] The risk decision layer is connected to the data acquisition layer and the risk identification layer by wire or wireless. The risk decision layer is used to import the safety risk data into a pre-trained risk model when the risk identification layer identifies a safety risk, determine the risk priority and response strategy, and transmit the risk priority and corresponding strategy to the early warning notification layer.

[0059] The early warning notification layer is connected to the risk decision-making layer by wire or wirelessly. The early warning notification layer is used to receive the risk priority and response strategies sent by the risk decision-making layer, issue voice alarms and flashing light prompts to the construction site, and send early warning notifications to relevant personnel and departments and track the feedback of the early warning notifications. The early warning notification includes the content, priority, response strategies, etc. of the early warning information.

[0060] The visualization display layer is used to generate a three-dimensional model of the construction site based on the actual data collected by the data collection layer, and associate the existing construction personnel and working equipment in the generated three-dimensional model according to the risk priority and response strategies determined by the risk decision-making layer, and present the construction site situation in real time and dynamically in the three-dimensional model. When there is a safety risk in a certain construction area, relevant warning information is displayed at the corresponding position in the three-dimensional model, and the decision results of the risk decision-making layer are displayed in the form of a notification bar, and the models and information related to the safety risk are stored for ready access.

[0061] The data collection layer includes cameras, sensor networks and wearable devices: Cameras are used to collect image and video data of the construction site to obtain construction personnel operation behavior data and equipment operation status data. The number and location of the cameras are selected according to the actual situation to cover the entire area of bridge construction; The sensor network includes various types of sensors, which are respectively used to collect one or more of the structural data, geological exploration data, foundation settlement data, groundwater level data, pier inclination data, concrete pouring data, beam stress and strain data, bridge deck paving data, equipment operation parameters, lighting system data, drainage system data, protection facility data, etc. in different areas of the bridge construction site; Among them, the bridge foundation settlement data, pier inclination data, beam stress and strain data, etc. are structural data, and the equipment operation parameters include the vibration frequency, oil temperature, oil pressure, current, voltage, etc. data during the operation of the construction equipment; The wearable device is worn by the construction personnel and is used to collect the physiological data and location information of the construction personnel.

[0062] The risk identification layer includes a construction stage division module, an algorithm model storage module, a data screening module and a stage risk identification module:

[0063] The construction stage division module is used to divide the bridge construction project into a foundation construction stage, a lower structure construction stage, an upper structure construction stage and an auxiliary facility installation stage.

[0064] The algorithm model storage module is used to store the pre-trained AI models corresponding to different construction stages. The AI models include a geological analysis model, a structural stress analysis model, a beam erection model, and an installation process risk assessment model. During the foundation construction stage, the geological analysis model is used to judge the geological stability of the pier foundation. During the lower structure construction stage, the structural stress analysis model is used to evaluate the stress risks of piers and abutments. During the upper structure construction stage, the beam erection model can analyze the risks in processes such as hoisting. During the bridge deck paving stage, the bridge deck construction quality analysis model is used to analyze the construction quality of the bridge deck. During the installation of auxiliary facilities stage, the installation process risk assessment model is used to evaluate the installation risks of auxiliary facilities.

[0065] The data screening module is used to screen out the actual data corresponding to the current construction stage from the actual data of each construction stage collected by the data acquisition layer.

[0066] Specifically, during the foundation construction stage, the focus is on screening geological exploration data, foundation settlement data, and groundwater level data. Geological exploration data contains information such as stratum structure and rock and soil properties, which is an important basis for judging the stability of the bridge foundation. Foundation settlement data reflects the settlement changes during the foundation construction process in real time. If the settlement rate is abnormal, it may indicate problems with the foundation. Groundwater level data has a significant impact on foundation construction. High water levels or fluctuations may trigger risks such as foundation pit water inrush. These data are crucial for identifying geological safety risks such as excessive foundation settlement, underground karsts, and foundation pit water inrush.

[0067] During the pier construction stage, pier inclination data, concrete pouring data, and construction equipment operation data are screened. Pier inclination data directly reflects the verticality state of the pier. Even minor inclination changes may affect the overall structural safety of the bridge. Concrete pouring data such as pouring speed, temperature, and slump are related to the pouring quality of concrete and are closely related to the strength and durability of the pier. Construction equipment operation data includes the operation parameters of equipment such as cranes and concrete pump trucks. Equipment failures may lead to construction accidents and affect the construction progress and safety of the pier. Data screening at this stage provides key information for identifying safety risks such as pier inclination, instability of the formwork support system, and collapse of the pier column construction platform.

[0068] During the beam construction stage, mainly beam stress and strain data, hoisting equipment data, and high-altitude operation personnel data are screened. Beam stress and strain data can judge the stress condition of the beam under construction loads, timely detect stress concentration or over-limit areas, and prevent the generation of beam cracks. Hoisting equipment data such as the lifting capacity, lifting height, and wire rope tension of the crane play a key role in the safety of beam hoisting. High-altitude operation personnel data involves personnel operation behaviors and position information, which can effectively identify the risks of illegal operations by high-altitude operation personnel and ensure the safety of beam construction.

[0069] During the bridge deck construction stage, data on bridge deck paving is screened. The data on bridge deck paving includes paving thickness, flatness, material properties, etc., which directly affect the driving comfort and safety of the bridge deck.

[0070] During the installation stage of auxiliary facilities, data on lighting systems, drainage systems, and protective facilities are screened; data on lighting systems such as lamp brightness and lighting range ensure construction and passage safety at night; data on drainage systems involve the slope, diameter, and drainage capacity of drainage pipes to prevent water accumulation on the bridge deck or under the bridge; data on protective facilities include the integrity and firmness of crash barriers, falling prevention nets, etc. Data screening at this stage helps identify lighting system failures, drainage system blockages, and risks of damage to protective facilities, ensuring the normal functioning of the bridge's auxiliary facilities.

[0071] The stage risk identification module is used to retrieve a pre-trained AI model corresponding to the current construction stage, process the data screened by the data screening module, and identify potential safety risk data.

[0072] Specifically, during the foundation construction stage, the stage risk identification module uses a pre-trained first AI model to analyze the screened data. The first AI model contains a geological analysis model that can judge whether there is a problem of excessive foundation settlement, identify underground karst caves, and detect the risk of foundation pit water inrush based on the collected data. The first AI model is trained using supervised learning methods. A large amount of historical data on different geological conditions is collected as the training set, and the data is labeled with categories such as normal, minor risk, and severe risk. Algorithms such as support vector machines and random forests are used for training. The support vector machine separates data of different risk categories by finding an optimal classification hyperplane, while the random forest is based on the ensemble learning of multiple decision trees to improve the generalization ability and stability of the model. At the same time, combined with geomechanics theory, the geological parameters are associated with the risk categories, enabling the model to accurately judge risks based on geological exploration data, foundation settlement data, and groundwater level data, etc.

[0073] During the pier construction stage, the stage risk identification module uses a pre-trained second AI model for analysis. The second AI model has a structural stress analysis function and can determine whether the pier is tilted due to factors such as improper construction technology, uneven foundation settlement, or external forces. For the concrete pouring data, it evaluates the concrete pouring quality. Based on the operation data of construction equipment, it identifies the risk of the pier column construction platform collapsing. The second AI model can be constructed as an integrated architecture containing multiple sub-models. One of the sub-models uses a deep learning model based on the attention mechanism to process images of pier tilting and the concrete pouring process. The attention mechanism can focus on key areas in the images, such as the bottom of the pier and the concrete pouring connection, to more accurately identify the tilting angle of the pier and concrete defects. Another sub-model uses a time series analysis model based on long short-term memory networks (LSTM) to process the time series of the operation data of construction equipment and predict the trend of equipment failures. By weighted fusion of the results of these two sub-models, it comprehensively judges whether there are safety risks such as pier tilting, instability of the formwork support system, and collapse of the pier column construction platform.

[0074] During the beam construction stage, the stage risk identification module uses a pre-trained third AI model for analysis. The third AI model includes a beam erection model, which is used to identify safety risks such as beam cracks, hoisting equipment failures, and violations by high-altitude workers. The third AI model adopts a hybrid architecture based on generative adversarial networks and object detection networks and can be trained by combining computer vision and finite element analysis methods. It uses computer vision technology to process image and video data during the beam construction process and extracts information such as the surface crack characteristics of the beam, the attitude of the hoisting equipment, and the actions of high-altitude workers. Through a large amount of labeled data, it trains deep learning-based object detection and recognition models, such as YOLO, Faster RCNN, etc., to enable them to accurately identify risks such as beam cracks, hoisting equipment failures, and violations by high-altitude workers.

[0075] During the bridge deck construction stage, a pre-trained fourth AI model is used for analysis. The fourth AI model includes a bridge deck construction quality analysis model. The fourth AI model can adopt a hybrid architecture based on a rule engine and deep learning. The rule engine formulates rules such as the thickness, flatness, and railing installation strength of the bridge deck pavement according to the specifications and standards of bridge deck construction, and conducts preliminary screening and judgment on the input data. The deep learning part uses a multi-layer perceptron (MLP) or a convolutional neural network to learn complex data patterns during the bridge deck construction process, such as the change in the compaction degree of paving materials and the image characteristics of railing welding quality. By comprehensively analyzing the results of the rule engine and deep learning, it identifies risks such as non-compliance with the flatness of the bridge deck pavement, insufficient railing installation strength, and incorrect traffic signs.

[0076] During the installation stage of the auxiliary facilities, the pre-trained fifth AI model is used for analysis. The fifth AI model includes an installation process risk assessment model, which can utilize an architecture based on a knowledge graph and a decision tree. First, a knowledge graph of bridge auxiliary facilities is constructed to associate and represent knowledge such as the components, installation processes, and common fault modes of lighting systems, drainage systems, protection facilities, etc. The decision tree model classifies and judges the risks of the auxiliary facilities based on the knowledge in the knowledge graph and the actual installation data.

[0077] The risk identification layer of the present invention further includes a conventional risk identification module, which includes a personnel risk identification unit, an equipment risk identification unit, an electricity risk identification unit, and an environmental risk identification unit.

[0078] The personnel risk identification unit is used to demarcate dangerous operation areas at each construction stage and region, such as the high-altitude operation area of the pier and the dangerous area under the erection of the beam body, etc. Analyze the image data of the construction site, judge whether the safety equipment worn by the construction personnel and their positions are compliant, prevent personnel from invading the dangerous area, and when the construction personnel enter the dangerous operation area, control the alarm equipment in the dangerous operation area to send out corresponding alarm information.

[0079] The equipment risk identification unit is used to connect to all construction equipment in bridge construction, monitor the equipment operation data in real time, judge whether the operation data of each construction equipment exceeds the threshold, and when there is equipment operation data exceeding the threshold, control the alarm equipment in the area where the construction equipment is located to send out corresponding alarm information to avoid safety accidents caused by equipment failures.

[0080] The electricity risk identification unit is used to connect to the distribution box at the construction site, analyze the current, voltage fluctuations, and key point temperature data. If there are abnormal fluctuations or overheating, judge that there is a fire risk, immediately cut off the power supply and control the alarm equipment in the area where the power supply is located to alarm, prevent electrical fires from occurring, and ensure the safe and stable operation of the construction site power system.

[0081] The environmental risk identification unit is used to calculate the environmental safety risk assessment value of each area according to the temperature, humidity, wind speed, precipitation data, geological data, surrounding environment data, and construction personnel activity frequency at each area of the construction site. If the environmental safety risk assessment value of a certain area exceeds the threshold, it is considered that there is an environmental safety risk in that area.

[0082] The formula for the environmental risk identification unit to calculate the environmental safety risk assessment value of a certain area is:

[0083] R i =ω1T i +ω2H i +ω2Wi+ω4R i +ω5G i +ω6E i +ω7Pi +σ

[0084] where R i represents the environmental safety risk assessment value of the i-th area, and ω1, ω2, ω3, ω4, ω5, ω6, ω7 respectively represent the weight coefficients of temperature, humidity, wind speed, precipitation, geological data, surrounding environment data, and the activity frequency of construction workers. T i 、H i 、W i 、R i 、G i 、E i 、P i respectively represent the side effect indexes of the actual temperature data, humidity data, wind speed data, precipitation data, geological conditions data, surrounding environment data, and the activity frequency of construction workers on construction safety within the i-th area, and σ represents the error.

[0085] The risk decision-making layer in the present invention includes a risk assessment module, a priority determination module, a countermeasure generation module, and a decision result output module.

[0086] The risk assessment module uses quantitative analysis to quantitatively evaluate the safety risk data identified by the risk identification layer, comprehensively considers various factors such as the possibility of risk occurrence, the possible impact range and harm degree, and determines the level and impact degree of the risk.

[0087] The priority determination module is used to determine the risk priority order according to the level and impact degree of the risk when there are multiple risk data.

[0088] The countermeasure generation module is used to generate targeted and operable countermeasures for different levels and priority risks according to the decision-making model imported in advance and verified by a large amount of data training and actual cases, combined with the actual construction situation, site conditions, and relevant specification standards of the bridge construction project.

[0089] The decision result output module is used to integrate the risk priority and countermeasures into the decision result and transmit it to the early warning notification layer in an intuitive manner.

[0090] The decision-making model in the coping strategy generation module of the present invention is first trained using the case-based reasoning (CBR) method. A large number of historical cases during the bridge construction process are collected, including different types of safety risk events, corresponding risk data characteristics, and effective coping strategies adopted. These cases are sorted out and analyzed in detail, key information is extracted, and a case library is constructed. During the training process, when new risk data is input, the model calculates the similarity with the historical cases in the case library, finds the most similar case or combination of cases, refers to its corresponding coping strategy, and makes appropriate adjustments and optimizations according to the current actual situation, so as to generate a coping strategy applicable to the current risk.

[0091] The present invention uses the reinforcement learning method to strengthen the training of the decision-making model. The bridge construction safety management is regarded as a dynamic decision-making process. The decision-making model, as an intelligent agent, interacts with the construction environment. The risk state and coping strategy are defined as actions, and the degree of risk reduction or the effect of construction safety guarantee is defined as the reward signal. In the process of continuously trying different coping strategies, through continuous iterative training, the decision-making model can generate the best coping strategies for different risks.

[0092] The present invention uses the method based on deep learning to assist in training the decision-making model, constructs a deep neural network architecture, such as a multi-layer perceptron (MLP), a convolutional neural network (CNN), or a recurrent neural network (RNN), etc. Select a suitable network structure according to the characteristics of the risk data, and use a large number of labeled risk data and corresponding coping strategy data to train the model. Continuously adjust the network parameters through the backpropagation algorithm, so that the decision-making model can learn the complex mapping relationship between the risk characteristics and the coping strategies, and thus accurately generate coping strategies.

[0093] The early warning notification layer in the present invention includes an early warning information generation module, a multi-channel early warning notification module, and an early warning feedback and recording module.

[0094] The early warning information generation module is used to generate detailed early warning content according to information such as the type, level, location, and time of the risk. The early warning content includes risk description, risk consequences, recommended measures to be taken, and emergency contact information.

[0095] The multi-channel early warning notification module is used to ensure that early warning information can be transmitted to on-site construction workers and relevant area responsible persons in a timely and accurate manner through various communication methods. In the present invention, eye-catching sound and light alarms are set at the construction site. When a high-risk event occurs, strong sound and light alarms are issued to quickly attract the attention of on-site personnel. SMS notifications are sent to the mobile phones of construction management personnel and operators using SMS and the mini-program platform to ensure that even if personnel are not at the construction site, they can obtain early warning information in the first time. Electronic display screens are set in key places such as the construction command center and the project department office to scroll and display early warning information, facilitating centralized management and enabling monitoring personnel to timely understand the risk situation.

[0096] The early warning feedback and recording module is used to receive feedback from relevant personnel on early warning information, and record and store the process of releasing and processing early warning information. Through the analysis of early warning feedback data, the present invention evaluates the effectiveness and timeliness of the early warning system to continuously optimize early warning strategies and processes. At the same time, the recorded early warning historical data can be used for subsequent risk analysis and experience summary, providing data support for improving the safety management level of bridge construction projects.

[0097] The visual display layer includes a three-dimensional model construction module, an early warning information module, and a historical data query module.

[0098] The three-dimensional model construction module is used to generate three-dimensional visualization models corresponding to different regions of the bridge construction project at different stages according to a preset ratio, and scale them according to the actual data collected by the data acquisition layer according to the actual ratio size, and dynamically present the spatial positions and states of construction personnel and working equipment in the visual three-dimensional model.

[0099] The early warning information module is used to, when there are safety risks during the construction process, generate warning signs with gradually deepening rendering colors from light to dark according to the risk level in the corresponding area in the visual three-dimensional model in real time, and display the early warning information notification bar.

[0100] The historical data query module is used to record, query, and display historical safety data during the bridge construction process. The historical safety data includes records of past safety accidents, previously appeared early warning information, and corresponding treatment measures and results.

[0101] The present invention sets a mobile mini-program, which includes functions such as real-time data display, risk early warning reception and reminder, three-dimensional model viewing, feedback and communication, and historical data query.

[0102] The safety early warning method for bridge construction of the present invention, as Figure 2 shown, specifically includes the following steps:

[0103] S1. The data acquisition layer collects various construction data of different regions at the construction site of the bridge construction project in real time.

[0104] S2. According to the construction stage and the category of construction data, retrieve the corresponding AI model to identify potential safety risks in each area. When there is a safety risk in a certain area, control the on-site alarm device in that area to send out an alarm message.

[0105] S3. When a safety risk is identified, import the risk data information into a pre-trained decision-making model, and generate corresponding decision-making suggestions in combination with the actual situation of the bridge construction project and relevant specification standards.

[0106] S4. Generate a three-dimensional model of the construction site of the bridge construction project at different construction stages, and identify the spatial positions of construction personnel and construction equipment at the construction site according to the construction data. Dynamically present the spatial positions and states of the operators and operating equipment in the visual three-dimensional model. When there is a safety risk in a certain area, convey the risk information and the corresponding recommended risk decisions to relevant personnel and display relevant warning and decision-making suggestion information in the notification bar of the three-dimensional model. After the accident, record and display the accident data and make it available for retrieval at any time in the model list.

[0107] The parts not specifically described in the above specification are all prior arts or can be achieved through prior arts. Moreover, the specific implementation cases described in the present invention are only the preferred implementation cases of the present invention, and are not used to limit the implementation scope of the present invention. That is, all equivalent changes and modifications made according to the content of the scope of the present invention patent should be regarded as the technical scope of the present invention.

Claims

1. An artificial intelligence-based bridge construction safety early warning system, characterized by: include The data collection layer is used to collect actual data during the bridge construction phase; The risk identification layer is used to analyze the data based on the actual data collected by the data collection layer and call the pre-trained AI model to identify potential security risk data; The risk decision layer is used to import the security risk data into the pre-trained risk model to determine the risk priority and response strategy when the security risk is identified in the risk identification layer; The early warning notification layer is used to receive risk priorities and response strategies sent by the risk decision-making layer and issue prompts and early warning notifications.

2. The artificial intelligence-based bridge construction safety early warning system according to claim 1 is characterized by: Also includes The visualization layer is used to generate a three-dimensional model of the construction site based on the actual data collected by the data acquisition layer, and to associate the existing construction personnel and work equipment in the generated three-dimensional model according to the risk priorities and response strategies determined by the risk decision-making layer, so as to present the construction site situation in real time in the three-dimensional model.

3. The bridge construction safety early warning system based on artificial intelligence according to claim 1 is characterized by: The data collection layer includes cameras, which are used to collect images and video data of the construction site to obtain the operation behavior data of the construction personnel and the operation status data of the equipment; A sensor network, including a plurality of sensors, for collecting one or more of the following data: structural data, geological exploration data, foundation settlement data, groundwater level data, pier inclination data, concrete pouring data, beam stress and strain data, bridge deck pavement data, equipment operation parameters, lighting system data, drainage system data, protective facility data, etc., at different areas of the bridge construction site; Wearable devices are used to collect construction workers’ physiological data and location information.

4. The artificial intelligence-based bridge construction safety early warning system according to claim 1 is characterized by: The risk identification layer includes a construction phase division module, which is used to divide the bridge construction project into the foundation construction phase, the substructure construction phase, the superstructure construction phase and the ancillary facilities installation phase; Algorithm model storage module, used to store pre-trained AI models corresponding to different construction stages; The data screening module is used to screen out the actual data corresponding to the current construction stage from the actual data of each construction stage collected by the data collection layer; The stage risk identification module is used to call the pre-trained AI model corresponding to the current construction stage, process the data screened by the data screening module, and identify potential safety risk data.

5. The artificial intelligence-based bridge construction safety early warning system according to claim 4 is characterized by: The risk identification layer also includes a conventional risk identification module, which includes The personnel risk identification unit is used to identify dangerous work areas in each construction stage and area, analyze the construction site image data, determine whether the construction personnel's safety equipment and location are in compliance with regulations, and control the alarm equipment in the dangerous work area to send out corresponding alarm information when the construction personnel enter the dangerous work area; The equipment risk identification unit is used to connect all construction equipment in bridge construction, monitor equipment operation data in real time, and determine whether the operation data of each construction equipment exceeds the threshold. When the equipment operation data exceeds the threshold, the alarm device in the area where the construction equipment is located is controlled to send out the corresponding alarm information; The power risk identification unit is used to connect to the distribution box at the construction site to analyze the current, voltage fluctuations and temperature data of key points. If abnormal fluctuations or overtemperature occur, it is judged that there is a fire risk, and the power supply is immediately disconnected and the alarm equipment in the area where the power supply is located is controlled to sound an alarm.

6. The artificial intelligence-based bridge construction safety early warning system according to claim 5 is characterized by: The conventional risk identification module also includes an environmental risk identification unit, which is used to calculate the environmental safety risk assessment value of each area based on the temperature, humidity, wind speed, precipitation data, geological data, surrounding environment data and construction personnel activity frequency of each area of ​​the construction site. If the environmental safety risk assessment value of a certain area exceeds the threshold, it is considered that there is an environmental safety risk in the area; The formula for calculating the environmental safety risk assessment value of the environmental risk identification unit is: R i =ω1T i +ω2H i +ω2Wi+ω4R i +ω5G i +ω6E i +ω7P i +s Where Ri represents the environmental safety risk assessment value of the ith area, ω1, ω2, ω3, ω4, ω5, ω6, and ω7 represent the weight coefficients of temperature, humidity, wind speed, precipitation, geological data, surrounding environmental data, and construction personnel activity frequency, respectively, and T i , H i , W i , R i , G i 、E i , P i They respectively represent the side effect index of the actual temperature data, humidity data, wind speed data, precipitation data, geological data, surrounding environment data and construction personnel activity frequency in the i-th area on construction safety, and σ represents the error.

7. The artificial intelligence-based bridge construction safety early warning system according to claim 1 is characterized by: The risk decision layer includes a risk assessment module, which conducts quantitative assessment of the security risk data identified by the risk identification layer to determine the level and impact of the risk; The priority determination module is used to determine the risk priority order according to the risk level and impact when there are multiple risk data; The response strategy generation module is used to generate targeted and operational response strategies for risks of different levels and priorities based on the previously imported decision-making model and the actual construction situation, site conditions and relevant specifications and standards of the bridge construction project; The decision result output module is used to integrate the risk priority and response strategy into the decision result and pass it to the early warning notification layer.

8. The artificial intelligence-based bridge construction safety early warning system according to claim 1 is characterized by: The warning notification layer includes a warning information generation module, which is used to generate detailed warning content based on risk type, level, location and time; Multi-channel early warning notification module, used to deliver the content of the early warning letter to on-site construction personnel and responsible personnel in related areas; The early warning feedback and recording module is used to receive feedback from relevant personnel on early warning information, and to record and store the release and processing process of early warning information.

9. The artificial intelligence-based bridge construction safety early warning system according to claim 2 is characterized in that: The visualization layer includes The 3D model building module is used to generate a 3D visualization model of each area of ​​the bridge construction project according to a preset scale, and scale it according to the actual scale size based on the actual data collected by the data acquisition layer, so as to present the spatial position and status of the construction personnel and working equipment in the visualization 3D model; The early warning information module is used to generate warning signs with rendering colors from light to dark in the corresponding areas of the visual 3D model according to the risk level from low to high when there are safety risks during the construction process, and display the early warning information notification bar; The historical data query module is used to record, query and display the historical safety data during the bridge construction process.

10. A safety early warning method for bridge construction projects based on artificial intelligence, characterized in that: The bridge construction safety early warning system based on artificial intelligence according to any one of claims 1 to 9 comprises the following steps: S1. Real-time collection of various construction data from different areas of the bridge construction site; S2. According to the construction stage and construction data category, the corresponding AI model is retrieved to identify the potential safety risks in each area. When there is a safety risk in a certain area, the on-site alarm equipment in the area is controlled to send out an alarm message; S3. Generate a three-dimensional model of the construction site of the bridge construction project at different construction stages, and identify the spatial positions of construction personnel and construction equipment on the construction site based on the construction data, and dynamically present the spatial positions and status of the workers and equipment in the visual three-dimensional model. When there is a safety risk in a certain area, the risk information and the corresponding recommended risk decisions will be conveyed to the relevant personnel and the relevant warning information will be displayed at the corresponding position of the three-dimensional model. After the accident, the accident data will be recorded and displayed in the model list.

Images