Steel structure engineering real-time monitoring and early warning method based on BIM

Abstract

The invention discloses a BIM-based steel structure engineering real-time monitoring and early warning method, relates to the technical field of steel structure monitoring, and solves the technical problems of lack of scientificity and adaptability in baseline establishment and weak abnormal judgment and root identification capabilities. A theoretical baseline is corrected through finite element model parameter calibration and multi-working-condition simulation, an actually measured baseline is collected in an environment stable period and purified through wavelet transformation, and model parameters are reversely calibrated when the deviation between the theoretical baseline and the actually measured baseline exceeds a threshold value, so that the deviation between a final baseline and the real health state of the structure is reduced, and the problem of poor adaptability of a fixed baseline is solved. Through double-layer anomaly identification logic of time sequence feature research and judgment and coupling correlation verification, false anomaly is distinguished based on fluctuation amplitude and environment correlation, and then a parameter mapping matrix is constructed through a physical coupling relationship, so that the false alarm rate and the missing report rate are reduced, and double targets of accurate identification of anomaly types and clear root positioning are realized.

Description

Technical Field

[0001] This invention relates to the field of steel structure monitoring technology, specifically a BIM-based real-time monitoring and early warning method for steel structure engineering. Background Technology

[0002] Steel structures are widely used in major infrastructure projects such as super high-rise buildings, large-span stadiums, and bridges due to their advantages of high strength, light weight, and convenient construction. However, during construction, installation, and long-term operation, steel structures are susceptible to factors such as load changes, environmental erosion, material aging, and loosening of joints, which can lead to safety hazards such as stress concentration, excessive displacement, and fatigue damage. If these hazards are not monitored and addressed in a timely manner, they may cause major accidents such as component buckling, structural instability, or even collapse.

[0003] According to patent application number 201910749826.0, a real-time monitoring and early warning method for steel structure engineering based on BIM is disclosed. The system includes a BIM model processing subsystem, a structural analysis and service center subsystem, a communication network subsystem, and a data acquisition equipment terminal. Based on the finite element model of the building structure and real-time sensor data provided by the data network, the system accurately simulates and analyzes the structure, identifies any existing or potential structural problems, issues early warning information, and clearly identifies problematic components in the BIM model using different colors. Simultaneously, structural deformation data and external environmental data transmitted from the acquisition equipment terminal are automatically added to the BIM information model, and the model adjusts synchronously according to structural deformation.

[0004] However, theoretical calculation baselines often rely on ideal parameters from design specifications without considering actual material properties and structural construction conditions. Measured baselines neglect environmental stability control, and the lack of an effective fusion mechanism between theoretical and measured data leads to significant deviations between the baseline and the actual structural health status, easily causing misjudgments. Relying solely on fixed threshold comparisons to identify anomalies fails to distinguish between false anomalies caused by environmental disturbances and true anomalies caused by structural damage. Furthermore, the lack of multi-parameter collaborative verification based on physical coupling makes it difficult to accurately pinpoint the root cause of anomalies, resulting in high false alarm rates and a high risk of missed alarms. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a BIM-based real-time monitoring and early warning method for steel structure engineering, which solves the problems of insufficient scientific rigor and adaptability in baseline establishment, and weak ability to determine anomalies and identify root causes.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a BIM-based real-time monitoring and early warning method for steel structure engineering, which specifically includes the following steps: Multi-source data on stress, displacement, vibration and temperature of steel structures are collected. Denoising, redundancy processing and standardization preprocessing operations are performed on the multi-source data in sequence to obtain preprocessed data. Based on the structural design drawings, a structural mechanics model is established using finite element software to simulate the theoretical values ​​of stress and displacement under different working conditions, which serve as the theoretical reference range for the baseline and generate the theoretical calculation baseline. Under the initial healthy state of structural construction completion, no operational load, and stable environment, the average value or statistical confidence interval of the monitoring data is continuously monitored and taken as the measured initial baseline to generate the measured initial baseline. The overlap between the theoretical reference range and the measured initial baseline interval is calculated. If the deviation is less than the threshold, the average of the two is taken as the final initial baseline. Otherwise, the average of the measured initial baseline is used as the target, and the key parameters in the model are adjusted in reverse to obtain the final initial baseline. The real-time data of the steel structure is collected and matched with the final initial baseline. If the match is found, a normal monitoring signal is generated; otherwise, an abnormal analysis signal is generated. At the same time, the abnormal analysis signal is used to determine the nature of the abnormality and verify the coupling correlation. Calculate the steel structure anomaly index and compare it with the classification threshold to determine the corresponding early warning level. At the same time, generate corresponding early warning and response information based on the early warning level.

[0007] As a further aspect of the present invention, stress, displacement, vibration and temperature data of the steel structure are collected from multiple sources based on resistance strain gauges, laser displacement gauges, accelerometers and temperature sensors. Denoising, redundancy processing and standardization preprocessing operations are performed on the multi-source data in sequence. Denoising adopts Kalman filtering algorithm, redundancy processing removes duplicate and erroneous data, and standardization adjusts the data format to be consistent to obtain preprocessed data.

[0008] As a further aspect of the present invention, the method for generating the theoretical calculation baseline is as follows: An initial finite element model was established based on the structural design drawings. The model material parameters were corrected to the actual test values. A fine mesh was used for key components at the mid-span of the main beam and the column base. For complex nodes of the steel column-steel beam rigid joint, the sub-model technology was used to refine the local mesh. The stress and displacement simulation values ​​were simulated under construction, no-load operation, full-load operation and special working conditions. Theoretical correction coefficients were introduced to obtain the theoretical reference range, where the theoretical stress value range = [simulated value × 0.9, simulated value × 1.1] and the theoretical displacement value range = [simulated value × 0.85, simulated value × 1.15].

[0009] As a further aspect of the present invention, the method for generating the measured initial baseline is as follows: After the structural construction is completed and the environment is stable in its initial healthy state, continuous monitoring is conducted for 24 hours. Sensors are deployed one-to-one with the theoretical monitoring points of the finite element model to obtain measured data. After removing outliers from the measured data, long-term stable components and short-term fluctuating components are separated by wavelet transform. The stable components are extracted as statistical samples, and the mean is calculated based on the stable samples. and standard deviation The measured initial baseline interval was determined using a 99% confidence level, resulting in the measured initial baseline interval = [ -2.58 , +2.58 ].

[0010] As a further aspect of the present invention, the method for matching real-time data of the steel structure with the final initial baseline is as follows: Relying on the sensor network of the perception layer, the core mechanical parameters, environmental parameters and working condition parameters of the steel structure are collected in real time. Based on the final initial baseline, the real-time data is compared bidirectionally. If the real-time data falls within the final initial baseline range, the structure is determined to be in an initial healthy state. The system automatically generates a normal monitoring signal, which includes metadata such as data value, acquisition time, and sensor location and archives it. If the real-time data exceeds the final initial baseline range, an anomaly analysis signal is immediately triggered.

[0011] As a further aspect of the present invention, the method for judging the anomaly properties and verifying the coupling correlation of the anomaly analysis signal is as follows: Obtain the abnormal sensor data corresponding to the abnormal analysis signal, and perform time-series analysis on the abnormal sensor data at a set time period t. If the fluctuation amplitude is ≤ 1.5 times the standard deviation of the final initial baseline and the Pearson correlation coefficient with the environmental parameters is greater than 0.7, it is judged as environmental interference. If the change is greater than 20% of the baseline interval or a discontinuous jump occurs, deep verification is initiated. Based on the physical coupling mechanism, an abnormal parameter-correlated parameter mapping matrix is ​​constructed. If the synchronicity of the change trend of the abnormal parameter and the related parameter is greater than 80% and sensor failure is ruled out, it is characterized as structural damage.

[0012] As a further aspect of the present invention, the method for determining the warning level is as follows: The anomaly index AI is calculated using the formula AI = |real-time data - median dynamic threshold| / (upper limit of dynamic threshold - median dynamic threshold), where the median dynamic threshold = (upper limit of dynamic threshold + lower limit of dynamic threshold) / 2. When AI ≤ 1.0, it is a blue warning; when 1.0 < AI ≤ 2.0, it is a yellow warning; when 2.0 < AI ≤ 3.0, it is an orange warning; and when AI > 3.0, it is a red warning. Warning level information is generated.

[0013] As a further embodiment of the present invention, the handling method for blue alerts is as follows: triggering environmental-data correlation verification and sensor status check; updating the environmental correction factor library and adjusting the dynamic threshold median when there is environmental interference; remotely calibrating and increasing the monitoring frequency for 48 hours when there is sensor disturbance; and generating blue prediction and handling information. The procedure for handling a yellow alert is as follows: highlight the abnormal component flashing yellow in the BIM model, automatically link the component's design parameters, construction records, and historical monitoring data, and push the information to the relevant person in charge. Then, conduct a visual inspection and load check. If anomalies are found during the visual inspection, reinforce the corresponding abnormal component at the nodes. If anomalies are found during the load check, suspend temporary loading of the abnormal area and generate yellow alert handling information.

[0014] As a further aspect of the present invention, the handling method for an orange alert is as follows: an emergency response is initiated within 1 hour, temporary monitoring points such as fiber optic strain gauges and laser displacement gauges are added, the sampling frequency is increased to 1 time / second, the data is transmitted to the emergency command platform in real time, and parameter mutation and secondary alert triggering conditions are set. The procedure for handling a red alert is to immediately initiate a full evacuation and multi-departmental coordination, freeze all encrypted data 10 minutes before the anomaly, implement liquid nitrogen cooling and guy rope fixation instability prevention measures, and formulate a component replacement plan after the risk has stabilized.

[0015] As a further aspect of the present invention, the triggering condition for the secondary early warning of parameter mutation is that the stress increase exceeds 20MPa within 10 minutes. The control parameters for the local cooling of liquid nitrogen under the red warning are that the cooling rate is ≤10℃ / minute and the minimum temperature is ≥-20℃. The guy rope adopts a specification of 16mm or more in diameter and the tensile force is ≥1.5 times the overturning force of the component.

[0016] This invention provides a BIM-based real-time monitoring and early warning method for steel structure engineering. Compared with existing technologies, it has the following advantages: This invention employs a dual-baseline fusion strategy, combining theoretically calculated baselines and measured initial baselines. The theoretical baseline is corrected through finite element model parameter calibration and multi-condition simulation, while the measured baseline is refined through environmental stabilization period acquisition and wavelet transform. When the deviation between the two exceeds a threshold, the model parameters are reverse-calibrated to reduce the deviation between the final baseline and the actual health state of the structure, thus solving the problem of poor adaptability of fixed baselines. Through a two-layer anomaly identification logic of time-series feature analysis and coupled correlation verification, false anomalies are first distinguished based on the correlation between fluctuation amplitude and environment, and then a parameter mapping matrix is ​​constructed through physical coupling relationships to reduce false alarm rate and false negative rate, achieving the dual goals of accurate anomaly type identification and clear root cause location. Attached Figure Description

[0017] Figure 1 This is a diagram illustrating the steps and methods of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 This application provides a BIM-based real-time monitoring and early warning method for steel structure engineering, which specifically includes the following steps: Step 1: Collect multi-source data of the steel structure based on multi-source sensors. The multi-source data includes stress, displacement, vibration, and temperature. Specifically, the data is collected using resistance strain gauges, laser displacement gauges, accelerometers, and temperature sensors, respectively. The obtained multi-source data is then preprocessed, including denoising, redundancy removal, and standardization. Denoising is performed using a Kalman filter algorithm, redundancy removal removes duplicate and erroneous data, and standardization modifies the data format to ensure consistency and compliance with subsequent usage requirements, resulting in preprocessed data. Step 2: Determine the final initial baseline based on the obtained preprocessed data. The final initial baseline is determined by comprehensively analyzing the theoretically calculated baseline and the measured initial baseline. For the theoretically calculated baseline, based on the structural design drawings, a structural mechanics model is established using finite element software to simulate the theoretical values ​​of stress and displacement under different working conditions, which serve as the theoretical reference range for the baseline. The specific processing method is as follows: An initial model is established based on the design drawings. During modeling, material parameters need to be corrected to actual measured values. A fine mesh is used for key components, such as the mid-span of the main beam and column bases, while a sparse mesh is used for non-critical areas. Sub-model technology is used for complex nodes, such as the steel column-steel beam rigid node, where the local mesh is refined separately to capture stress concentration characteristics. Then, simulated stress and displacement values ​​are obtained for construction stage conditions, no-load operation conditions, full-load operation conditions, and special conditions. Theoretical correction coefficients are introduced into the simulation results. Specifically, the corrected reference ranges are: theoretical stress value range = [simulated value × 0.9, simulated value × 1.1]; theoretical displacement value range = [simulated value × 0.85, simulated value × 1.15]; For the measured initial baseline, under the initial healthy state of structural completion, no operational load, and stable environment, the average value or statistical confidence interval of continuous monitoring data is taken as the measured initial baseline, and the specific processing method is as follows: Historical data was acquired, with a data collection window of 7-15 days after structural construction completion. A corresponding continuous monitoring period was also determined, e.g., 24-72 hours, ensuring stable temperature, load, and environment during this period. Specifically, stable temperature meant ambient temperature fluctuations ≤5℃ within 24 hours; stable load meant no temporary loads or construction machinery parking; and a quiet environment meant wind speed ≤3m / s. Sensors were placed at the theoretical monitoring points of the finite element model, and corresponding measured data were collected. Outliers were removed from the measured data. Then, the preprocessed data was separated into long-term stable components and short-term fluctuating components using wavelet transform. The former reflects the true state of the structure, while the latter represents environmental disturbances. The stable components were extracted as statistical samples, and the mean was calculated based on these stable samples. and standard deviation The measured initial baseline interval was determined using a 99% confidence level, resulting in the measured initial baseline interval = [ -2.58 , +2.58 ]; Calculate the overlap between the theoretical interval and the measured interval. If the deviation is ≤5%, it is determined that the model and the measured values ​​are consistent. The average of the two is directly taken as the final initial baseline. If the deviation exceeds the threshold, the error tracing process is initiated. If the deviation originates from the theoretical model, the average of the measured initial baseline is used as the target. Key parameters in the model, such as elastic modulus and nodal stiffness, are adjusted in reverse to ensure that the deviation between the simulation results and the measured values ​​is ≤5%, and the final initial baseline is determined.

[0020] Step 3: Obtain real-time data of the steel structure and compare it with the final initial baseline. If the real-time data matches the final initial baseline, it means that the steel structure is within the range of the final initial baseline, which indicates that the current steel structure is in normal condition and a normal monitoring signal is generated. Conversely, if the real-time data does not match the final initial baseline, it means that the steel structure is not within the range of the final initial baseline, which indicates that the current steel structure is in abnormal condition and an abnormal analysis signal is generated. Relying on the sensor network of the perception layer, the core mechanical parameters, environmental parameters, and operating parameters of the steel structure are collected in real time. After the data is preprocessed by the edge gateway, it is synchronized to the monitoring platform through transmission technologies such as 5G / LoRa. The real-time data is then compared bidirectionally with the final initial baseline as a reference. If the real-time data falls within the final initial baseline range, the structure is determined to be in an initial healthy state. The system automatically generates a normal monitoring signal, which includes metadata such as data value, acquisition time, and sensor location, and archives it. If the real-time data exceeds the final initial baseline range, an anomaly analysis signal is immediately triggered. The generated anomaly analysis signals are processed to obtain the corresponding anomaly sensor data. The data changes are analyzed with time t as the period. If the data changes are stable fluctuations, it indicates that the problem may be caused by environmental interference. Conversely, if the data changes abruptly, it indicates that the steel structure may be damaged and further verification and analysis are required. The anomaly sensor data is obtained and the corresponding correlation data is determined through the correlation algorithm. Specifically, this is determined by coupling correlation, which indicates that multiple parameters are affected by the same core variable or that there is physical coupling between parameters, such as force-displacement or voltage-current. At the same time, the obtained correlation parameters are judged for anomalies. If the correlation parameters are co-anomalies, the steel structure is qualitatively damaged. Time-series analysis is performed on abnormal sensor data with a time period t, the value of which is set by the operator. A sliding window is used to extract the data trend. If the data fluctuation within period t is ≤ 1.5 times the standard deviation of the final initial baseline, and the trend is strongly correlated with environmental parameters (specifically, Pearson coefficient > 0.7), it is determined to be a false anomaly caused by environmental interference. The system marks this state for observation and continuously tracks it. If the data change within period t is > 20% of the final initial baseline interval, or if discontinuous jumps occur, it is determined to be a suspected structural damage signal. A deep verification process is initiated to locate parameters intrinsically related to the abnormal data based on physical coupling mechanisms. Focusing on parameter combinations driven by the same core variable or bound to physical formulas, such as displacement associated with stress anomalies, bolt preload associated with nodal strain, and stiffness associated with vibration frequency, an anomaly parameter-associated parameter mapping matrix is ​​formed. Through cross-correlation analysis, if the synchronicity of the changing trends of the anomaly parameter and the associated parameter is >80%, and sensor failure is ruled out, it is characterized as a true anomaly caused by structural damage; if only a single parameter is abnormal, and all associated parameters are within the normal range, system problems such as loose sensor installation and signal interference need to be investigated. The anomaly index AI is calculated using the formula: AI = |Real-time Data - Median Dynamic Threshold| / (Upper Dynamic Threshold - Median Dynamic Threshold), where the median dynamic threshold = (Upper Dynamic Threshold + Lower Dynamic Threshold) / 2. The anomaly index AI is then compared with the corresponding tiered thresholds to determine the anomaly level. Anomaly levels include blue, yellow, orange, and red alerts. Specifically, if AI ≤ 1.0, it is classified as a blue alert; if 1.0 < AI ≤ 2.0, it is classified as a yellow alert; if 2.0 < AI ≤ 3.0, it is classified as an orange alert; and if AI > 3.0, it is classified as a red alert. Corresponding alert level information is then generated.

[0021] Step 4: Take corresponding measures based on the obtained warning level information. For blue warnings, trigger environmental-data correlation verification and sensor status check, and combine manual review for confirmation. If it is environmental interference, the system automatically updates the environmental correction factor library and adjusts the dynamic threshold median. If it is sensor disturbance, remotely calibrate the sensor and continuously track it to generate blue warning handling information. In response to a yellow alert, abnormal components that flash yellow are highlighted in the BIM model. The design parameters, construction records, and historical monitoring data of the component are automatically linked and pushed to the relevant person in charge. Visual inspection and load investigation are then carried out. If anomalies are found during the visual inspection, the corresponding abnormal components are reinforced at the nodes. If anomalies are found during the load investigation, the temporary loading of the abnormal area is suspended, and yellow alert handling information is generated. In response to the orange alert, an emergency response was initiated. Temporary monitoring points were added to the abnormal components and several adjacent related components. Fiber optic strain gauges were used for stress monitoring, and laser displacement gauges were used for displacement monitoring. The sampling frequency was uniformly increased to 1 time / second, and the data was transmitted to the emergency command platform in real time. Parameter mutation and secondary warning trigger conditions were set, and the parameter mutation secondary warning trigger condition was that the stress increase exceeded 20MPa within 10 minutes.

[0022] In response to the red alert, an immediate evacuation of all personnel and multi-departmental coordination were initiated. Ten minutes before the anomaly, all encrypted data was frozen, liquid nitrogen cooling was implemented, and measures to prevent instability of guy ropes were taken. After the risk stabilized, a component replacement plan was formulated. The control parameters for localized liquid nitrogen cooling during the red alert were: cooling rate ≤ 10℃ / minute, minimum temperature ≥ -20℃; guy ropes with a diameter of 16mm or more and a tensile strength ≥ 1.5 times the overturning force of the component.

[0023] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0024] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A BIM-based real-time monitoring and early warning method for steel structure engineering, characterized in that, The method specifically includes the following steps: Multi-source data on stress, displacement, vibration and temperature of steel structures are collected. Denoising, redundancy processing and standardization preprocessing operations are performed on the multi-source data in sequence to obtain preprocessed data. Based on the structural design drawings, a structural mechanics model is established using finite element software to simulate the theoretical values ​​of stress and displacement under different working conditions, which serve as the theoretical reference range for the baseline and generate the theoretical calculation baseline. Under the initial healthy state of structural construction completion, no operational load, and stable environment, the average value or statistical confidence interval of the monitoring data is continuously monitored and taken as the measured initial baseline to generate the measured initial baseline. The overlap between the theoretical reference range and the measured initial baseline interval is calculated. If the deviation is less than the threshold, the average of the two is taken as the final initial baseline. Otherwise, the average of the measured initial baseline is used as the target, and the key parameters in the model are adjusted in reverse to obtain the final initial baseline. The real-time data of the steel structure is collected and matched with the final initial baseline. If the match is found, a normal monitoring signal is generated; otherwise, an abnormal analysis signal is generated. At the same time, the abnormal analysis signal is used to determine the nature of the abnormality and verify the coupling correlation. Calculate the steel structure anomaly index and compare it with the classification threshold to determine the corresponding early warning level. At the same time, generate corresponding early warning and response information based on the early warning level.

2. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 1, characterized in that, Based on resistance strain gauges, laser displacement gauges, accelerometers, and temperature sensors, multi-source data on stress, displacement, vibration, and temperature of the steel structure are collected. The multi-source data are then subjected to denoising, redundancy processing, and standardization preprocessing operations. Denoising uses the Kalman filter algorithm, redundancy processing removes duplicate and erroneous data, and standardization adjusts the data format to be consistent to obtain preprocessed data.

3. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 1, characterized in that, The method for generating the theoretical calculation baseline is as follows: An initial finite element model was established based on the structural design drawings. The model material parameters were corrected to the actual test values. A fine mesh was used for key components at the mid-span of the main beam and the column base. For complex nodes of the steel column-steel beam rigid joint, the sub-model technology was used to refine the local mesh. The stress and displacement simulation values ​​were simulated under construction, no-load operation, full-load operation and special working conditions. Theoretical correction coefficients were introduced to obtain the theoretical reference range, where the theoretical stress value range = [simulated value × 0.9, simulated value × 1.1] and the theoretical displacement value range = [simulated value × 0.85, simulated value × 1.15].

4. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 1, characterized in that, The method for generating the measured initial baseline is as follows: After the structural construction is completed and the environment is stable in its initial healthy state, continuous monitoring is conducted for 24 hours. Sensors are deployed one-to-one with the theoretical monitoring points of the finite element model to obtain measured data. After removing outliers from the measured data, long-term stable components and short-term fluctuating components are separated by wavelet transform. The stable components are extracted as statistical samples, and the mean is calculated based on the stable samples. and standard deviation The measured initial baseline interval was determined using a 99% confidence level, resulting in the measured initial baseline interval = [ -2.58 , +2.58 ].

5. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 1, characterized in that, The method for matching real-time data of the steel structure with the final initial baseline is as follows: Relying on the sensor network of the perception layer, the core mechanical parameters, environmental parameters and working condition parameters of the steel structure are collected in real time. Based on the final initial baseline, the real-time data is compared bidirectionally. If the real-time data falls within the final initial baseline range, the structure is determined to be in an initial healthy state. The system automatically generates a normal monitoring signal, which includes metadata such as data value, acquisition time, and sensor location and archives it. If the real-time data exceeds the final initial baseline range, an anomaly analysis signal is immediately triggered.

6. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 1, characterized in that, The method for anomaly analysis signals to determine their anomaly properties and verify their coupling correlation is as follows: Obtain the abnormal sensor data corresponding to the abnormal analysis signal, and perform time-series analysis on the abnormal sensor data at a set time period t. If the fluctuation amplitude is ≤ 1.5 times the standard deviation of the final initial baseline and the Pearson correlation coefficient with the environmental parameters is greater than 0.7, it is judged as environmental interference. If the change is greater than 20% of the baseline interval or a discontinuous jump occurs, deep verification is initiated. Based on the physical coupling mechanism, an abnormal parameter-correlated parameter mapping matrix is ​​constructed. If the synchronicity of the change trend of the abnormal parameter and the related parameter is greater than 80% and sensor failure is ruled out, it is characterized as structural damage.

7. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 1, characterized in that, The warning level is determined as follows: The anomaly index AI is calculated using the formula AI = |real-time data - median dynamic threshold| / (upper limit of dynamic threshold - median dynamic threshold), where the median dynamic threshold = (upper limit of dynamic threshold + lower limit of dynamic threshold) / 2. When AI ≤ 1.0, it is a blue warning; when 1.0 < AI ≤ 2.0, it is a yellow warning; when 2.0 < AI ≤ 3.0, it is an orange warning; and when AI > 3.0, it is a red warning. Warning level information is generated.

8. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 7, characterized in that, The handling method for a blue alert is to trigger environmental-data correlation verification and sensor status check, update the environmental correction factor library and adjust the dynamic threshold median when there is environmental interference, remotely calibrate and increase the monitoring frequency for 48 hours when there is sensor disturbance, and generate blue prediction and handling information. The procedure for handling a yellow alert is as follows: highlight the abnormal component flashing yellow in the BIM model, automatically link the component's design parameters, construction records, and historical monitoring data, and push the information to the relevant person in charge. Then, conduct a visual inspection and load check. If anomalies are found during the visual inspection, reinforce the corresponding abnormal component at the nodes. If anomalies are found during the load check, suspend temporary loading of the abnormal area and generate yellow alert handling information.

9. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 7, characterized in that, The procedure for handling an orange alert is as follows: activate the emergency response within 1 hour, add temporary monitoring points such as fiber optic strain gauges and laser displacement gauges, increase the sampling frequency to 1 time / second, transmit the data to the emergency command platform in real time, and set conditions for triggering secondary alerts by parameter mutations. The procedure for handling a red alert is to immediately initiate a full evacuation and multi-departmental coordination, freeze all encrypted data 10 minutes before the anomaly, implement liquid nitrogen cooling and guy rope fixation instability prevention measures, and formulate a component replacement plan after the risk has stabilized.

10. The BIM-based real-time monitoring and early warning method for steel structure engineering according to claim 9, characterized in that, The trigger condition for the secondary warning of parameter mutation is that the stress increase exceeds 20MPa within 10 minutes. The control parameters for the local cooling of liquid nitrogen under the red warning are: cooling rate ≤10℃ / minute, minimum temperature ≥-20℃; the guy ropes should be made of 16mm or more in diameter and have a tensile force ≥1.5 times the overturning force of the component.

Images