High-precision steel structure deformation detection method

Abstract

The application discloses a high-precision steel structure deformation detection method, and belongs to the field of steel structure deformation detection. The method comprises the following steps: S1, constructing a control network; S2, arranging monitoring points and sensors; S3, updating a dynamic reference coordinate system in real time; S4, synchronously collecting three-dimensional coordinates of a reflecting target point, strain data and a steel structure surface image under the reference coordinate system; S5, fusing three-dimensional point cloud, strain and visual feature data through a Kalman filtering model to obtain a fusion data set; S6, reconstructing a three-dimensional deformation field; S7, setting a multi-level early warning threshold, determining an early warning level by comparing the multi-level early warning threshold with the three-dimensional deformation field in real time, and sending an early warning value, prompt information and rectification suggestions. The high-precision steel structure deformation detection method can realize high-precision monitoring of the order of 0.08 mm, support full-cycle dynamic early warning, and is suitable for complex steel structure engineering such as large-span and super high-rise buildings.

Description

Technical Field

[0001] This invention relates to the field of steel structure deformation detection technology, and in particular to a high-precision steel structure deformation detection method. Background Technology

[0002] Steel structures, with their core advantages of high strength, large span, and high construction efficiency, are widely used in major engineering projects such as long-span bridges, super high-rise buildings, and stadiums. However, throughout the construction and long-term operation process, steel structures are susceptible to the continuous influence of multiple factors, including dynamic changes in load, fluctuations in ambient temperature, wind vibration, and natural aging of materials, gradually leading to cumulative deformation. If such cumulative deformation is not detected in time through high-precision monitoring, or if the early warning response is delayed, it will directly induce serious safety accidents, threatening the structural safety and reliability of the project.

[0003] Existing methods for detecting deformation in steel structures have significant technical shortcomings and are insufficient to meet the monitoring needs of complex engineering projects. 1. Contact measurement (such as strain gauges and dial gauges) is not only cumbersome to operate, but also has weak resistance to electromagnetic interference, temperature interference and vibration interference, making it difficult to adapt to complex and ever-changing construction scenarios and long-term operating environments. 2. The monitoring accuracy of single non-contact measurement (such as total station, GPS) is limited, and it can only acquire data on the external part of the structure. It cannot simultaneously collect information on the internal strain state of the steel structure and defects such as surface cracks and corrosion, resulting in a one-sided monitoring dimension. 3. Traditional data processing methods lack a systematic multi-source data fusion mechanism and have insufficient ability to suppress environmental noise, which can easily lead to distortion of monitoring data. Ultimately, this greatly reduces the reliability and accuracy of monitoring results, making it difficult to meet the core requirements of high-precision and full-dimensional coverage for monitoring of complex steel structures such as large-span and super high-rise buildings. Summary of the Invention

[0004] The purpose of this invention is to provide a high-precision method for detecting deformation in steel structures, thereby solving the aforementioned technical problems.

[0005] To achieve the above objectives, the present invention provides a high-precision method for detecting deformation of steel structures, comprising the following steps: S1. Construction of the benchmark system: Select more than 3 points outside the steel structure monitoring area, pour reinforced concrete benchmarks with a burial depth of ≥3m, and install anti-interference forced centering structures on the top of the benchmarks to form a triangular or quadrilateral control network composed of 3 or more benchmarks. The positional error of the control network is ≤0.02mm. S2. Monitoring Points and Sensor Layout: Multiple lidar reflective targets are attached to key deformation areas of the steel structure; multiple fiber optic grating sensors are attached along the stress transmission path; and visual cameras are installed in key areas of cracks and corrosion. S3. Dynamic benchmark calibration: A combination of a total station and a dual-frequency GPS receiver is used to measure the coordinates of the benchmark points, and the measurement data is processed by a weighted iterative algorithm to update the dynamic benchmark coordinate system in real time. S4. Multi-source data synchronous acquisition: Under the reference coordinate system, the three-dimensional coordinates of the reflective target point, strain data and steel structure surface image are synchronously acquired using lidar, fiber optic grating sensor and vision camera, and spatiotemporal alignment is performed to obtain multi-source raw dataset. S5. Data Fusion Processing: First, the 3σ criterion is used to remove outliers from the multi-source original dataset; then, the improved wavelet threshold denoising algorithm is used to process the lidar point cloud data obtained based on the three-dimensional coordinates of the reflective target point. At the same time, the temperature drift of the strain data is eliminated by the moving average filter. Finally, the three-dimensional point cloud, strain and visual feature data are fused by the Kalman filter model to obtain the fused dataset. S6. Deformation Reconstruction: By improving the ICP algorithm, register the three-dimensional point cloud to calculate the three-dimensional displacement, combine the material constitutive equation to derive the rotation angle and strain distribution, and then input the finite element model to reconstruct the three-dimensional deformation field. S7. Early Warning: Set multi-level early warning thresholds, determine the early warning level by comparing the multi-level early warning thresholds with the three-dimensional deformation field in real time, and send early warning values, prompts and rectification suggestions.

[0006] Preferably, the interference-proof forced centering structure described in step S1 includes a chassis installed at the top of the reference point and a stainless steel centering shaft that is interference-fitted with the top of the chassis. The top of the stainless steel centering shaft is provided with a standardized interface for connecting the total station and the dual-frequency GPS receiver. The total station and the dual-frequency GPS receiver are provided with anti-magnetic protective covers. The anti-magnetic protective covers are made of permalloy material to shield electromagnetic interference. The antimagnetic protective cover is sealed with a sealing gasket between itself and the stainless steel center axis, achieving an IP68 level of waterproof and dustproof rating. The clearance between the stainless steel central shaft and the chassis is ≤0.005mm; The repeatability of the anti-interference forced centering structure is ≤0.01mm.

[0007] Preferably, in step S2, the key deformation locations include beam end nodes, mid-span, and supports; the diameter of the lidar reflector target is ≥60mm; and the spacing between two adjacent lidar reflector targets... ; The spacing between two adjacent fiber Bragg grating sensors is 1.2m-1.8m. The fiber Bragg grating sensors are in close contact with the steel structure surface and are waterproofed. An external protective shell covers the sensors. Furthermore, the strain measurement accuracy of the fiber Bragg grating sensors is [not specified]. Temperature compensation range: -45℃ to 90℃; Visual camera pixel count: ≥25 million; The stress transfer path is determined as follows: a mechanical model of the steel structure is established using Midas / GTS NX finite element software, and the principal stress traces are extracted as the stress transfer path after the design load is applied.

[0008] Preferably, in step S3, the total station's angle measurement accuracy is ≤0.3″ ​​and its distance measurement accuracy is ≤0.3″. , Indicates the distance being measured; Planar positioning accuracy of dual-frequency GPS receiver And the elevation positioning accuracy ; The combination of a total station and a dual-frequency GPS receiver performs coordinate measurements on the benchmark point every 1.5 hours. The plane stability error of the reference point is ≤0.02mm / day.

[0009] Preferably, the weighted iterative algorithm expression described in step S3 is as follows: ; ; In the formula, and They represent The next iteration and The fused reference point coordinates after the next iteration; and These represent the measurement weights of the dual-frequency GPS receiver and the total station, respectively. express The coordinates of the reference point measured by the dual-frequency GPS receiver after the next iteration; express The coordinates of the reference point measured by the total station after the next iteration; Represents the coordinate residual function The first derivative; The iterative convergence threshold is set to 0.001 mm.

[0010] Preferably, the lidar mentioned in step S4 is a 1550nm fiber lidar with an automatic heating and defogging mechanism installed at the front end of the lens. The automatic heating and defogging mechanism includes a ring heating element installed at the front end of the fiber lidar lens and a temperature and humidity sensor for collecting the temperature and humidity of the lens area. The temperature and humidity sensor is electrically connected to the ring heating element via a controller to dynamically adjust the target temperature of the ring heating element based on the humidity of the lens area. : ; In the formula, This represents the reference temperature value, specifically the industry standard reference temperature for steel structure deformation testing. ; Indicates the temperature-humidity correction factor; This indicates the measured humidity in the lens area; This represents the humidity baseline value, specifically set at 50%RH, which is the industry standard reference humidity for steel structure deformation testing. When the temperature and humidity sensor detects that the humidity in the lens area is ≥85% or the temperature is ≤5℃, the controller activates the ring heating element, which operates at a heating power of 5W, and stops heating when the humidity in the lens area is ≤30% or the temperature is ≥10℃. ranging accuracy of fiber optic lidar angular resolution It collects the three-dimensional coordinates of the reflective target at a frequency of 120Hz.

[0011] Preferably, the improved wavelet thresholding denoising algorithm described in step S5 employs an adaptive threshold function and has a decomposition layer of 4; its expression is as follows: ; In the formula, This indicates the th wavelet thresholding process after improvement. Layer Wavelet coefficients; ; The wavelet decomposition of the th Layer Wavelet coefficients; Indicates the first The adaptive threshold of the layer, and , Indicates the standard deviation of noise; Indicates the data length. ; The moving average filter has a window size of 60 sampling points, and its expression is as follows: ; In the formula, Represents the raw strain data; This represents the strain data after filtering; This indicates the center index of the filtered data point to be calculated. This represents the point index of the original strain data within the sliding window.

[0012] Preferably, the improved ICP algorithm expression described in step S6 is as follows: ; in, ; In the formula, This represents the error objective function of the improved ICP algorithm; This represents the objective function for the position error of the ICP algorithm; Represents the weight of the normal vector; Represents the normal vector of the target point cloud; Represents the rotation matrix; Represents the normal vector of the source point cloud; This indicates the number of points in the point cloud data that participated in ICP registration; Represents the target point cloud; Represents the source point cloud; Represents the translation vector; The iteration termination condition is that the point cloud registration error is ≤0.005mm; The expressions for rotation angle and strain distribution are as follows: ; ; In the formula, Indicates a corner; and These respectively represent the steel structure in , Directional displacement; Indicates strain; Indicates stress, and , Represents Poisson's ratio. and These respectively represent the steel structure in , Directional strain; It represents the elastic modulus.

[0013] Preferably, in step S7, three warning thresholds are set: 55%, 75%, and 90%. The warning values, prompts, and rectification suggestions are encrypted and stored in the system, and then uploaded to the cloud for backup. The storage period is no less than the design service life of the steel structure.

[0014] Therefore, the present invention employs the above-mentioned high-precision steel structure deformation detection method, which has the following beneficial effects: 1. Higher precision: Through dynamic benchmark calibration, multi-source data fusion and targeted noise reduction algorithms, it achieves high-precision monitoring at the ±0.08mm level, which can accurately capture minute deformations; 2. Strong anti-interference capability: The deep-buried design of the reference point, the sensor protection structure, and the multi-level noise reduction algorithm effectively counteract environmental interference such as temperature, vibration, settlement, and electromagnetic interference; 3. Comprehensive monitoring: Simultaneously acquires three-dimensional coordinates, internal strain, and surface defect information of the steel structure, realizing full-dimensional monitoring of external displacement, internal stress, and surface condition; 4. Full-cycle management and control: Supports dynamic early warning and data traceability, with storage period covering the structural design service life, providing complete data support for construction operation, maintenance and rectification, and accident analysis.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] Figure 1 This is a flowchart of a high-precision steel structure deformation detection method according to the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of the present invention and are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout.

[0018] It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, such as a process, method, system, product, or server that includes a series of steps or units, not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or device.

[0019] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0020] Comparison of detection results between the present invention and traditional detection methods: A large-span steel structure bridge in the same scenario was selected, and a comparative monitoring was conducted for 30 days using the method of this invention, the traditional total station detection method, and the single fiber optic sensor detection method. The results are as follows: Monitoring accuracy: The deformation monitoring error of the method of the present invention is ±0.07mm, the error of the traditional total station detection method is ±1.2mm, and the error of the single fiber optic sensor detection method is ±0.5mm. The accuracy of the present invention is improved by about 17 times (compared to the total station) and 7 times (compared to the single fiber optic sensor). Anti-interference capability: Under strong electromagnetic interference (1000V / m) and temperature fluctuation (-10℃~40℃) environments, the data fluctuation rate of this invention is ≤2%, while the data fluctuation rate of traditional total station reaches 15% and the data fluctuation rate of single fiber optic sensor reaches 8%; Monitoring dimensions: This invention can simultaneously output three types of data: three-dimensional deformation field, internal strain, and surface defects, while traditional methods can only obtain data in a single dimension; Early warning response: The response time of this invention from deformation exceeding the threshold to sending early warning information is ≤10s, while the response time of traditional manual monitoring is ≥30min, which improves the response efficiency by 180 times, thus proving the effectiveness of this invention.

[0021] like Figure 1 As shown, a high-precision steel structure deformation detection method includes the following steps: S1. Construction of the benchmark system: Select more than 3 points outside the steel structure monitoring area, pour reinforced concrete benchmarks with a burial depth of ≥3m, and install anti-interference forced centering structures on the top of the benchmarks to form a triangular or quadrilateral control network composed of 3 or more benchmarks. The positional error of the control network is ≤0.02mm. The interference-proof forced centering structure described in step S1 includes a chassis installed at the top of the reference point and a stainless steel centering shaft that is interference-fitted with the top of the chassis. The top of the stainless steel centering shaft is provided with a standardized interface for connecting the total station and the dual-frequency GPS receiver. The total station and the dual-frequency GPS receiver are covered with anti-magnetic protective covers, which are made of permalloy to shield electromagnetic interference. The antimagnetic protective cover is sealed with a sealing gasket between itself and the stainless steel center axis, achieving an IP68 level of waterproof and dustproof rating. The clearance between the stainless steel central shaft and the chassis is ≤0.005mm; The repeatability of the anti-interference forced centering structure is ≤0.01mm.

[0022] S2. Monitoring Points and Sensor Layout: Multiple lidar reflective targets are attached to key deformation areas of the steel structure; multiple fiber optic grating sensors are attached along the stress transmission path; and visual cameras are installed in key areas of cracks and corrosion. In step S2, the key deformation locations include beam end nodes, mid-span, and supports. The diameter of the lidar reflector target is ≥60mm, and the spacing between two adjacent lidar reflector targets is... ; The spacing between two adjacent fiber Bragg grating sensors is 1.2m-1.8m. The fiber Bragg grating sensors are in close contact with the steel structure surface and are waterproofed. An external protective shell covers the sensors. Furthermore, the strain measurement accuracy of the fiber Bragg grating sensors is [not specified]. Temperature compensation range: -45℃ to 90℃; Visual camera pixel count: ≥25 million; The stress transfer path is determined as follows: a mechanical model of the steel structure is established using Midas / GTS NX finite element software. After applying the design load, the principal stress traces are extracted as the stress transfer path. Sensors are densely arranged in sections of the path where the stress gradient is >5MPa / m.

[0023] S3. Dynamic benchmark calibration: A combination of a total station and a dual-frequency GPS receiver is used to measure the coordinates of the benchmark points, and the measurement data is processed by a weighted iterative algorithm to update the dynamic benchmark coordinate system in real time. In step S3, the total station's angle measurement accuracy is ≤0.3″ ​​and distance measurement accuracy is ≤0.3″. , Indicates the distance being measured; Planar positioning accuracy of dual-frequency GPS receiver And the elevation positioning accuracy ; The combination of a total station and a dual-frequency GPS receiver performs coordinate measurements on the benchmark point every 1.5 hours. The plane stability error of the reference point is ≤0.02mm / day.

[0024] The weighted iterative algorithm described in step S3 is expressed as follows: ; ; In the formula, and They represent The next iteration and The fused reference point coordinates after the next iteration; and These represent the measurement weights of the dual-frequency GPS receiver and the total station, respectively. , ; express The coordinates of the reference point measured by the dual-frequency GPS receiver after the next iteration; express The coordinates of the reference point measured by the total station after the next iteration; Represents the coordinate residual function The first derivative; The iterative convergence threshold is set to 0.001 mm.

[0025] S4. Multi-source data synchronous acquisition: Under the reference coordinate system, the three-dimensional coordinates of the reflective target point, strain data and steel structure surface image are synchronously acquired using lidar, fiber optic grating sensor and vision camera, and spatiotemporal alignment is performed to obtain multi-source raw dataset. The lidar mentioned in step S4 is a 1550nm fiber lidar with an automatic heating and defogging mechanism installed at the front of the lens. The automatic heating and defogging mechanism includes a ring heating element installed at the front of the fiber lidar lens and a temperature and humidity sensor for collecting the temperature and humidity of the lens area. The temperature and humidity sensor is electrically connected to the ring heating element via a controller to dynamically adjust the target temperature of the ring heating element based on the humidity of the lens area. : ; In the formula, This represents the reference temperature value, specifically the industry standard reference temperature for steel structure deformation testing. ; Indicates the temperature-humidity correction factor; This indicates the measured humidity in the lens area; This represents the humidity baseline value, specifically set at 50%RH, which is the industry standard reference humidity for steel structure deformation testing. When the temperature and humidity sensor detects that the humidity in the lens area is ≥85% or the temperature is ≤5℃, the controller activates the ring heating element, which operates at a heating power of 5W, and stops heating when the humidity in the lens area is ≤30% or the temperature is ≥10℃. ranging accuracy of fiber optic lidar angular resolution It collects the three-dimensional coordinates of the reflective target at a frequency of 120Hz.

[0026] S5. Data Fusion Processing: First, the 3σ criterion is used to remove outliers from the multi-source original dataset; then, the improved wavelet threshold denoising algorithm is used to process the lidar point cloud data obtained based on the three-dimensional coordinates of the reflective target point. At the same time, the temperature drift of the strain data is eliminated by the moving average filter. Finally, the three-dimensional point cloud, strain and visual feature data are fused by the Kalman filter model to obtain the fused dataset. The improved wavelet thresholding algorithm described in step S5 employs an adaptive threshold function and has a decomposition layer of 4 (i.e., by replacing the fixed threshold with an adaptive threshold that varies with the number of decomposition layers, the noise removal accuracy is improved); its expression is as follows: ; In the formula, This indicates the th wavelet thresholding process after improvement. Layer Wavelet coefficients; ; The wavelet decomposition of the th Layer Wavelet coefficients; Indicates the first The adaptive threshold of the layer, and , Indicates the standard deviation of noise; Indicates the data length. ; The moving average filter has a window size of 60 sampling points, and its expression is as follows: ; In the formula, Represents the raw strain data; This represents the strain data after filtering; This indicates the center index of the filtered data point to be calculated. This represents the point index of the original strain data within the sliding window.

[0027] S6. Deformation Reconstruction: By improving the ICP algorithm, register the three-dimensional point cloud to calculate the three-dimensional displacement, combine the material constitutive equation to derive the rotation angle and strain distribution, and then input the finite element model to reconstruct the three-dimensional deformation field. The improved ICP algorithm expression described in step S6 is as follows: ; in, ; In the formula, This represents the error objective function of the improved ICP algorithm; This represents the objective function for the position error of the ICP algorithm; Represents the weight of the normal vector, and ; Represents the normal vector of the target point cloud; Represents the rotation matrix; Represents the normal vector of the source point cloud; This indicates the number of points in the point cloud data that participated in ICP registration; Represents the target point cloud; Represents the source point cloud; Represents the translation vector; The iteration termination condition is that the point cloud registration error is ≤0.005mm; The expressions for rotation angle and strain distribution are as follows: ; ; In the formula, Indicates a corner; and These respectively represent the steel structure in , Directional displacement; Indicates strain; Indicates stress, and , Represents Poisson's ratio. , and These respectively represent the steel structure in , Directional strain; Indicates the elastic modulus. .

[0028] S7. Early Warning: Set multi-level early warning thresholds, determine the early warning level by comparing the multi-level early warning thresholds with the three-dimensional deformation field in real time, and send early warning values, prompts and rectification suggestions.

[0029] In step S7, three warning thresholds are set: 55%, 75%, and 90%. The warning values, prompts, and rectification suggestions are encrypted and stored in the system, and then uploaded to the cloud for backup. The storage period is no less than the design service life of the steel structure.

[0030] Specifically, the three-level early warning system is divided as follows: Level 1 early warning: Deformation ≥ 55% of the design allowable value, with a warning value of Level II warning: Deformation ≥ 75% of the design allowable value, warning value is... Level 3 warning: Deformation ≥ 90% of the design allowable value, warning value is... , The maximum allowable deformation for steel structure design; The corresponding warning messages are as follows: Level 1 warning: "The deformation of the steel structure is within a safe range. It is recommended to increase the monitoring frequency to once every 0.5 hours"; Level 2 warning: "The deformation of the steel structure is approaching the warning value. It is recommended to suspend the application of unnecessary loads and check the sensor status"; Level 3 warning: "The deformation of the steel structure has reached the danger threshold. It is recommended to evacuate personnel immediately and start emergency reinforcement measures." The corresponding rectification suggestions are as follows: Level 1 warning rectification suggestion: review the monitoring data collection frequency and check whether there is settlement at the benchmark point; Level 2 warning rectification suggestion: provide temporary support for the deformed parts and recalibrate the sensors; Level 3 warning rectification suggestion: entrust a third-party testing agency to conduct a structural safety assessment and formulate a reinforcement or repair plan.

[0031] Example Taking a long-span steel structure bridge (main span 120m) as an example, the monitoring method of this invention is used, and the steps are as follows: S1: Benchmark system construction: Three benchmark points are set up around the bridge, buried at a depth of 3.5m, to form a triangular control network; S2: Monitoring point and sensor layout: 72 lidar reflective target points (3.8m spacing), 56 fiber optic grating sensors (1.5m spacing) and 6 high-definition vision cameras are deployed to cover easily deformable parts and areas with high incidence of defects. S3: Dynamic benchmark calibration: The benchmark coordinate system is updated every 1.5 hours through a combination of dual-frequency GPS and high-precision total station measurement; S4: Multi-source data synchronous acquisition: LiDAR, fiber optic sensor and vision camera acquire data synchronously to achieve spatiotemporal alignment; S5: Data fusion processing: Outliers are removed using the 3σ criterion. After improved wavelet thresholding and moving average filtering, the data is fused using Kalman filtering to calculate displacement and strain distribution. S6: Deformation and Reconstruction; S7: Warning: The allowable deformation of the bridge is set to 5mm, corresponding to a warning value of 2.75mm, an alarm value of 3.75mm, and an emergency response value of 4.5mm. The monitoring accuracy reaches ±0.07mm, successfully capturing minute deformations caused by vehicle load and temperature changes, and the warning response is timely and accurate.

[0032] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. 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 still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A high-precision method for detecting deformation of steel structures, characterized in that: Includes the following steps: S1. Construction of the benchmark system: Select more than 3 points outside the steel structure monitoring area, pour reinforced concrete benchmarks with a burial depth of ≥3m, and install anti-interference forced centering structures on the top of the benchmarks to form a triangular or quadrilateral control network composed of 3 or more benchmarks. The positional error of the control network is ≤0.02mm. S2. Monitoring Points and Sensor Layout: Multiple lidar reflective targets are attached to key deformation areas of the steel structure; multiple fiber optic grating sensors are attached along the stress transmission path; and visual cameras are installed in key areas of cracks and corrosion. S3. Dynamic benchmark calibration: A combination of a total station and a dual-frequency GPS receiver is used to measure the coordinates of the benchmark points, and the measurement data is processed by a weighted iterative algorithm to update the dynamic benchmark coordinate system in real time. S4. Multi-source data synchronous acquisition: Under the reference coordinate system, the three-dimensional coordinates of the reflective target point, strain data and steel structure surface image are synchronously acquired using lidar, fiber optic grating sensor and vision camera, and spatiotemporal alignment is performed to obtain multi-source raw dataset. S5. Data fusion processing: First, the 3σ criterion is used to remove outliers from the multi-source original dataset; Then, the improved wavelet threshold denoising algorithm is used to process the lidar point cloud data obtained based on the three-dimensional coordinates of the reflective target. At the same time, the temperature drift of the strain data is eliminated by moving average filtering. Then, the three-dimensional point cloud, strain and visual feature data are fused by Kalman filtering model to obtain the fused dataset. S6. Deformation Reconstruction: By improving the ICP algorithm, register the three-dimensional point cloud to calculate the three-dimensional displacement, combine the material constitutive equation to derive the rotation angle and strain distribution, and then input the finite element model to reconstruct the three-dimensional deformation field. S7. Early Warning: Set multi-level early warning thresholds, determine the early warning level by comparing the multi-level early warning thresholds with the three-dimensional deformation field in real time, and send early warning values, prompts and rectification suggestions.

2. The high-precision steel structure deformation detection method according to claim 1, characterized in that: The interference-proof forced centering structure described in step S1 includes a chassis installed at the top of the reference point and a stainless steel centering shaft that is interference-fitted with the top of the chassis. The top of the stainless steel centering shaft is provided with a standardized interface for connecting the total station and the dual-frequency GPS receiver. The total station and the dual-frequency GPS receiver are covered with anti-magnetic protective covers, which are made of permalloy to shield electromagnetic interference. The antimagnetic protective cover is sealed with a sealing gasket between itself and the stainless steel center axis, achieving an IP68 level of waterproof and dustproof rating. The clearance between the stainless steel central shaft and the chassis is ≤0.005mm; The repeatability of the anti-interference forced centering structure is ≤0.01mm.

3. The high-precision steel structure deformation detection method according to claim 1, characterized in that: In step S2, the key deformation locations include beam end nodes, mid-span, and supports. The diameter of the lidar reflector target is ≥60mm, and the spacing between two adjacent lidar reflector targets is... ; The spacing between two adjacent fiber Bragg grating sensors is 1.2m-1.8m. The fiber Bragg grating sensors are in close contact with the steel structure surface and are waterproofed. An external protective shell covers the sensors. Furthermore, the strain measurement accuracy of the fiber Bragg grating sensors is [not specified]. Temperature compensation range: -45℃ to 90℃; Visual camera pixel count: ≥25 million; The stress transfer path is determined as follows: a mechanical model of the steel structure is established using Midas / GTS NX finite element software, and the principal stress traces are extracted as the stress transfer path after the design load is applied.

4. The high-precision steel structure deformation detection method according to claim 1, characterized in that: In step S3, the total station's angle measurement accuracy is ≤0.3″ ​​and distance measurement accuracy is ≤0.3″. , Indicates the distance being measured; Planar positioning accuracy of dual-frequency GPS receiver And the elevation positioning accuracy ; The combination of a total station and a dual-frequency GPS receiver measures the coordinates of the benchmark point every 1.5 hours. The plane stability error of the reference point is ≤0.02mm / day.

5. The high-precision steel structure deformation detection method according to claim 1, characterized in that: The weighted iterative algorithm described in step S3 is expressed as follows: ； ； In the formula, and They represent The next iteration and The fused reference point coordinates after the next iteration; and These represent the measurement weights of the dual-frequency GPS receiver and the total station, respectively. express The coordinates of the reference point measured by the dual-frequency GPS receiver after the next iteration; express The coordinates of the reference point measured by the total station after the next iteration; Represents the coordinate residual function The first derivative; The iterative convergence threshold is set to 0.001 mm.

6. The high-precision steel structure deformation detection method according to claim 1, characterized in that: The lidar mentioned in step S4 is a 1550nm fiber lidar with an automatic heating and defogging mechanism installed at the front of the lens. The automatic heating and defogging mechanism includes a ring heating element installed at the front of the fiber lidar lens and a temperature and humidity sensor for collecting the temperature and humidity of the lens area. The temperature and humidity sensor is electrically connected to the ring heating element via a controller to dynamically adjust the target temperature of the ring heating element based on the humidity of the lens area. : ； In the formula, Indicates the reference temperature value; Indicates the temperature-humidity correction factor; This indicates the measured humidity in the lens area; Indicates the humidity reference value; When the temperature and humidity sensor detects that the humidity in the lens area is ≥85% or the temperature is ≤5℃, the controller activates the ring heating element, which operates at a heating power of 5W, and stops heating when the humidity in the lens area is ≤30% or the temperature is ≥10℃. ranging accuracy of fiber optic lidar angular resolution It collects the three-dimensional coordinates of the reflective target at a frequency of 120Hz.

7. The high-precision steel structure deformation detection method according to claim 1, characterized in that: The improved wavelet thresholding algorithm described in step S5 uses an adaptive threshold function and has a decomposition layer of 4 layers. Its expression is as follows: ； In the formula, This indicates the th wavelet thresholding process after improvement. Layer Wavelet coefficients; ; The wavelet decomposition of the th Layer Wavelet coefficients; Indicates the first The adaptive threshold of the layer, and , Indicates the standard deviation of noise; Indicates the data length. ; The moving average filter has a window size of 60 sampling points, and its expression is as follows: ； In the formula, This represents the raw strain data; This represents the strain data after filtering; This indicates the center index of the filtered data point to be calculated. This represents the point index of the original strain data within the sliding window.

8. The high-precision steel structure deformation detection method according to claim 1, characterized in that: The improved ICP algorithm expression described in step S6 is as follows: ； in, ； In the formula, This represents the error objective function of the improved ICP algorithm; This represents the objective function for the position error of the ICP algorithm; Represents the weight of the normal vector; Represents the normal vector of the target point cloud; Represents the rotation matrix; Represents the normal vector of the source point cloud; This indicates the number of points in the point cloud data that participated in ICP registration. Represents the target point cloud; Represents the source point cloud; Represents the translation vector; The iteration termination condition is that the point cloud registration error is ≤0.005mm; The expressions for rotation angle and strain distribution are as follows: ； ； In the formula, Indicates a corner; and These respectively represent the steel structure in , Directional displacement; Indicates strain; Indicates stress, and , Represents Poisson's ratio. and These respectively represent the steel structure in , Directional strain; It represents the elastic modulus.

9. The high-precision steel structure deformation detection method according to claim 1, characterized in that: In step S7, three warning thresholds are set: 55%, 75%, and 90%. The warning values, prompts, and rectification suggestions are encrypted and stored in the system, and then uploaded to the cloud for backup. The storage period is no less than the design service life of the steel structure.

Images