Long-term monitoring and early warning method suitable for large-span steel frame beam

Abstract

The invention relates to a long-term monitoring and early warning method suitable for a large-span steel frame beam, and the method specifically comprises the steps: laying a distributed fiber grating sensor network at a key stress part of a target steel frame beam, and synchronously collecting an original deformation signal in real time; a temperature compensation algorithm and a wavelet noise reduction algorithm are used for preprocessing the collected various original deformation signals; identifying the deformation state of the key stress part through an improved conjugate beam method on the basis of the multiple original deformation signals after preprocessing and the structural model information of the target steel frame beam; constructing a steel frame beam deformation prediction model, and predicting the future deformation trend of the target steel frame beam by using the preprocessed historical and real-time signal time sequence data; and a steel frame beam abnormal deformation diagnosis model based on the graph neural network is constructed, abnormal recognition is performed on the deformation state of the key stress part based on a preset threshold value, and early warning is performed on the abnormal part in combination with the future deformation trend of the target steel frame beam.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of steel structure monitoring, in particular to a long-term monitoring and early warning method suitable for large-span steel frame beams. BACKGROUND

[0002] With the rapid development of large public buildings, the long-term service safety of large-span steel frame beams as the core load-bearing components faces severe challenges. Under the combined action of complex environmental loads and material aging, such structures are prone to hidden damage such as crack propagation, bolt loosening, and support settlement, and traditional manual detection methods are difficult to find hidden dangers in time.

[0003] The existing monitoring methods have the following technical bottlenecks: 1. Insufficient environmental interference suppression; 2. Weak damage positioning ability: point sensor network based on threshold discrimination cannot perceive the spatial conduction effect of damage; 3. Distortion of long-term prediction model: traditional time series model lacks physical mechanism constraint. 4. Lack of multi-source data fusion: the heterogeneity of strain, vibration, and temperature signals leads to information islands.

[0004] The patent number CN118536385A discloses a bridge support structure state intelligent monitoring and evaluation method based on a neural network model, which uses a Transformer model to process multi-source data and combines three-dimensional laser scanning and GNN to process spatial relationships.

[0005] However, in the prior art, the anti-interference ability of conventional sensors is weak and the long-term stability is poor, and the use of Transformer model to process time series data does not combine actual structural mechanics, which cannot adapt to the complex stress characteristics of steel structures, and the deduced data may have errors. In the existing GNN model, multi-source data is not fused and utilized, and the positioning result of deformation anomalies is not accurate enough. SUMMARY

[0006] To solve the problems existing in the prior art, the present application proposes a long-term monitoring and early warning method suitable for large-span steel frame beams.

[0007] The technical solution of the present application is as follows:

[0008] On the one hand, the present application proposes a long-term monitoring and early warning method suitable for large-span steel frame beams, the specific steps of which include:

[0009] Distributing a distributed fiber Bragg grating sensor network at the key stress parts of the target steel frame beam and collecting original deformation signals in real time and synchronously;

[0010] Using temperature compensation algorithm and wavelet denoising algorithm to preprocess the collected various original deformation signals;

[0011] Based on the pre-processed multiple original deformation signals and the structural model information of the target steel frame beam, the deformation state of the key stress position is identified by improving the conjugate beam method;

[0012] A steel frame beam deformation prediction model is constructed, and historical and real-time signal time series data after preprocessing are used to predict the future deformation trend of the target steel frame beam;

[0013] An abnormal deformation diagnosis model of the steel frame beam based on a graph neural network is constructed, and the deformation state of the key stress position is abnormally identified based on a preset threshold, and an abnormal position is warned in combination with the future deformation trend of the target steel frame beam.

[0014] As a preferred implementation, the key stress position includes a beam-column joint area, a potential or existing crack area, a mid-span area, and a high stress concentration area.

[0015] As a preferred implementation, the structural model information of the target steel frame beam includes set topology information, material constitutive information, boundary condition parameters, and load distribution benchmarks.

[0016] As a preferred implementation, the multiple original deformation signals at least include temperature signals, strain signals, and vibration response signals.

[0017] As a preferred implementation, the temperature compensation algorithm and the wavelet denoising algorithm perform a preprocessing step on the collected multiple original deformation signals, specifically:

[0018] The temperature component in the original deformation signal is stripped using the temperature compensation algorithm;

[0019] The strain signal and the vibration signal processed by the temperature compensation algorithm are filtered and denoised using the wavelet denoising algorithm.

[0020] As a preferred implementation, the step of identifying the deformation state of the key stress position by improving the conjugate beam method specifically includes:

[0021] A conjugate beam model with the same geometry and boundary conditions as the target steel frame beam is established;

[0022] The bending moment distribution of the conjugate beam is solved by a numerical iteration algorithm, and the real-time deflection curve and key section rotation angle of the target steel frame beam are obtained by inversion;

[0023] An adaptive weight factor is used to interpolate and compensate the solution results of the sensor distribution sparse area;

[0024] The deformation state of the target steel frame beam is identified according to the real-time deflection curve and key section rotation angle of the final target steel frame beam.

[0025] As a preferred implementation mode, the specific method for identifying the deformation state of the target steel frame beam according to the real-time deflection curve and the key section rotation angle of the final target steel frame beam comprises at least one of the following:

[0026] The real-time deflection curve and the key section rotation angle identified by the improved conjugate beam method are compared with historical baseline values or design allowable values, and if the values exceed a preset threshold, the values are determined to be abnormal;

[0027] Statistical analysis is performed on the continuously identified deformation results to detect abnormal points or mutations that exceed the normal fluctuation range, and the abnormal points or mutations are determined to be abnormal;

[0028] The predicted future deformation or change trend is compared with a limit bearing state or a critical failure threshold, and the abnormality is determined according to the comparison result.

[0029] As a preferred implementation mode, the specific method for identifying the deformation state of the target steel frame beam according to the real-time deflection curve and the key section rotation angle of the final target steel frame beam comprises at least one of the following:

[0030] The preprocessed multiple original deformation signals are respectively input as independent modalities, the spatial topological relationship of the sensor network is modeled by using the graph neural network, the input of each original signal is dynamically weighted by using the cross-modal attention mechanism, and whether the deformation state of the target steel frame beam exceeds the expected state is diagnosed in real time according to the preset threshold.

[0031] On the other hand, the present application proposes an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to realize the long-term monitoring and early warning method for a large-span steel frame beam according to any one of the embodiments of the present application.

[0032] On the other hand, the present application proposes a computer readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to realize the long-term monitoring and early warning method for a large-span steel frame beam according to any one of the embodiments of the present application.

[0033] The present application has the following advantages:

[0034] 1、The distributed fiber grating sensor network is laid in key mechanical sensitive parts such as beam-column joints and crack areas, temperature compensation algorithm is combined to eliminate the interference of thermal expansion and cold contraction on strain measurement, and wavelet denoising algorithm is used to filter environmental noise, so that high-precision, anti-interference real-time deformation monitoring capability is realized;

[0035] 2. Based on the structural model information of the steel frame beam, this application uses an improved conjugate beam method to invert the real-time deflection curves and cross-sectional rotation angles of key parts, and introduces an adaptive weight factor to interpolate and compensate for sensor sparse areas, thereby solving the problem of local measurement blind spots and improving the overall deformation recognition error rate;

[0036] 3. This application constructs a graph neural network spatial topology model, dynamically associates the spatial relationships of the sensor network, and uses a cross-modal attention mechanism to dynamically weighted fuse temperature, strain, and vibration signals to achieve accurate positioning of abnormal parts and multi-level early warning, avoiding misjudgment of a single data source;

[0037] 4. This application integrates historical and real-time data to predict the long-term deformation trends of steel beams, such as creep and fatigue, through a combined prediction model, supporting preventive maintenance decisions and extending the service life of the structure;

[0038] 5. This application reduces the density of sensor deployment through sparse area interpolation compensation technology, while realizing automated real-time diagnosis, reducing the frequency of manual inspections, and providing early warning of potential failure risks to avoid major safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0041] It should be understood that the step numbers used herein are only for convenience of description and are not intended to limit the order in which the steps are to be executed.

[0042] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0043] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0044] The term "and / or" means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0045] Embodiment one:

[0046] Referring to Figure 1 A long-span steel frame beam long-term monitoring and early warning method, the specific steps include:

[0047] Distributed fiber Bragg grating sensor network is arranged at the key stress position of the target steel frame beam, and the original deformation signal is collected in real time and synchronously;

[0048] The collected various original deformation signals are preprocessed by using temperature compensation algorithm and wavelet denoising algorithm;

[0049] Based on the preprocessed various original deformation signals and the structure model information of the target steel frame beam, the deformation state of the key stress position is identified by improving the conjugate beam method;

[0050] A steel frame beam deformation prediction model is constructed, and the historical and real-time signal time series data after preprocessing are used to predict the future deformation trend of the target steel frame beam;

[0051] A steel frame beam abnormal deformation diagnosis model based on a graph neural network is constructed, the deformation state of the key stress position is abnormally identified based on a preset threshold, and an early warning is issued for the abnormal position in combination with the future deformation trend of the target steel frame beam.

[0052] As a preferred embodiment of the present embodiment, the key stress position includes beam-column joint area, potential or existing crack area, mid-span area and high stress concentration area.

[0053] As a preferred embodiment of the present embodiment, the structure model information of the target steel frame beam includes set topology information, material constitutive information, boundary condition parameters and load distribution benchmark.

[0054] In the present embodiment, the structure model information of the target steel frame beam specifically includes:

[0055] (1) Geometric topology information:

[0056] Geometric size of beam: span length, cross-section height / width, variable cross-section position;

[0057] Sensor spatial coordinates: accurate installation position (x, y, z coordinates) of each FBG sensor on the beam body;

[0058] Structural connection relationship: connection node distribution of beam-column, beam-support;

[0059] (2) Material constitutive information:

[0060] Elastic modulus (E): Static / dynamic elastic modulus value of steel material;

[0061] Density (p): Used for associating vibration signal with mass distribution;

[0062] Thermal expansion coefficient (a): Auxiliary correction for temperature compensation algorithm;

[0063] (3) Boundary condition parameters:

[0064] Node constraint type:

[0065] Fixed connection (infinite bending stiffness);

[0066] Hinged connection (bending moment = 0);

[0067] Semi-rigid connection (measured or designed value of rotational stiffness k r ) needs to be provided;

[0068] Support settlement data: Historical record of foundation displacement in long-term monitoring;

[0069] (4) Load distribution reference

[0070] Design load spectrum: Standard distribution mode of dead load and live load;

[0071] Initial stress state: Residual stress distribution after construction is completed.

[0072] As a preferred embodiment of the present embodiment, the plurality of original deformation signals at least include temperature signals, strain signals, and vibration response signals.

[0073] In the present embodiment, in addition to the above three kinds of original deformation signals, acoustic emission signals, inclination angle signals, ultrasonic time domain reflection signals, distributed optical fiber acoustic sensing signals, digital image correlation signals, etc. can also be included.

[0074] Acoustic emission signals are used to monitor the initiation and propagation of damage such as internal micro-cracks in materials, weld cracking, and bolt loosening;

[0075] Inclination angle signals are used to monitor changes in spatial pose such as support settlement, overall inclination, and local buckling;

[0076] Ultrasonic time domain reflection signals are used to monitor changes in interface state such as loss of bolt pretension and slip of concealed connectors;

[0077] Distributed optical fiber acoustic sensing signals are used to monitor dynamic distributed responses such as wind-induced vortex vibration and vehicle load;

[0078] Digital image correlation signals are used to monitor full-field non-contact deformation such as local buckling and out-of-plane deformation.

[0079] As a preferred embodiment of the present embodiment, the temperature compensation algorithm and the wavelet denoising algorithm are used to preprocess the collected original deformation signals, and the preprocessing steps are specifically as follows:

[0080] The temperature component in the original deformation signal is stripped by using the temperature compensation algorithm.

[0081] The strain signal and the vibration signal processed by the temperature compensation algorithm are filtered and denoised by using the wavelet denoising algorithm.

[0082] In the present embodiment, the temperature component in the original deformation signal is stripped by using the temperature compensation algorithm, and the specific steps are as follows:

[0083] A temperature compensation reference grating is arranged at a non-stress part, and the grating only senses temperature changes and is not affected by mechanical strain. The wavelength shift of the reference grating is subtracted from the original strain signal through mathematical operation to separate a pure strain signal.

[0084] A temperature-wavelength mapping model is established by using the wavelength shift (pure temperature response) of the reference grating to dynamically strip the temperature component from the mixed signal; the stripped signal is restored to a physical quantity reflecting only mechanical strain to eliminate the measurement error caused by thermal expansion and contraction of the steel.

[0085] As a preferred embodiment of the present embodiment, the deformation state of the key stress part is identified by improving the conjugate beam method, and the specific steps are as follows:

[0086] A conjugate beam model with the same geometry and boundary conditions as the target steel frame beam is established.

[0087] The bending moment distribution of the conjugate beam is solved by a numerical iteration algorithm to inversely obtain the real-time deflection curve and the key cross-section angle of the target steel frame beam.

[0088] An adaptive weight factor is used to interpolate and compensate the solving results of the sparse sensor distribution area.

[0089] The deformation state of the target steel frame beam is identified according to the real-time deflection curve and the key cross-section angle of the final target steel frame beam.

[0090] In the present embodiment, the Newton iteration method is used to solve the deflection differential equation in the step of solving the bending moment distribution of the conjugate beam by a numerical iteration algorithm.

[0091]

[0092] In the formula, w is the deflection of the beam; M(x) is the cross-section bending moment distribution; E is the elastic modulus of the steel; I is the cross-section moment of inertia; and EI is the cross-section bending stiffness.

[0093] As a preferred embodiment of the present embodiment, the specific method for identifying the deformation state of the target steel frame beam according to the real-time deflection curve and the key section rotation angle of the final target steel frame beam comprises at least one of the following:

[0094] The real-time deflection curve and the key section rotation angle identified by the improved conjugate beam method are compared with the historical baseline value or the design allowable value, and if it exceeds the preset threshold, it is determined to be abnormal;

[0095] Statistical analysis is performed on the continuously identified deformation results to detect abnormal points or mutations that exceed the normal fluctuation range, and it is determined to be abnormal;

[0096] The predicted future deformation or change trend is compared with the limit bearing state or critical failure threshold, and the abnormality is determined according to the comparison result.

[0097] As a preferred embodiment of the present embodiment, the specific method for identifying the deformation state of the target steel frame beam according to the real-time deflection curve and the key section rotation angle of the final target steel frame beam comprises at least one of the following:

[0098] The preprocessed multiple original deformation signals are respectively input as independent modalities, the spatial topological relationship of the sensor network is modeled using the graph neural network, the input of each original signal is dynamically weighted using the cross-modality attention mechanism, and whether the deformation state of the target steel frame beam exceeds the expected state is diagnosed in real time according to the preset threshold.

[0099] In the present embodiment, the specific architecture of the steel frame beam deformation diagnosis model based on the graph neural network is as follows:

[0100] 1. Input layer:

[0101] Modality 1: preprocessed strain signal;

[0102] Modality 2: vibration spectrum;

[0103] Modality 3: temperature field distribution map.

[0104] 2. Graph neural network layer:

[0105] Construct a sensor spatial topology graph: node = sensor position, edge = Euclidean distance (adjacency matrix A);

[0106] Aggregate neighbor information through graph convolution layer:

[0107]

[0108] In the formula, is the sensor spatial topological relationship matrix with self-loop; is the diagonal matrix of ; and is the standardized result; H(l) is the original signal information of the input layer; W (l) is a weight matrix.

[0109] 3. Cross-modal attention mechanism connection layer:

[0110] The specific formula of dynamically calculating the modal weight is as follows:

[0111]

[0112] In the formula, alpha k is the maximum weight of the kth modal; h k is the high-level feature expression of the kth modal; W is a trainable projection matrix; tanh is a hyperbolic tangent function; u k is a trainable attention parameter.

[0113] The above only describes the embodiments of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation using the content of the specification and drawings, or directly or indirectly applied to other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A long-term monitoring and early warning method for large-span steel frame beams, characterized in that: The specific steps include: A distributed fiber Bragg grating sensor network is deployed at the key stress-bearing locations of the target steel frame beams to synchronously collect the original deformation signals in real time; The temperature compensation algorithm and wavelet noise reduction algorithm are used to pre-process the collected various original deformation signals; Based on the pre-processed original deformation signals and the structural model information of the target steel frame beam, the deformation state of the key stress-bearing parts is identified by using the improved conjugate beam method. Construct a steel frame beam deformation prediction model and use the pre-processed historical and real-time signal time series data to predict the future deformation trend of the target steel frame beam; A graph neural network-based steel frame beam abnormal deformation diagnosis model is constructed to identify abnormal deformation states of key stress-bearing parts based on preset thresholds, and to issue early warnings for abnormal parts based on the future deformation trend of the target steel frame beam.

2. The long-term monitoring and early warning method for large-span steel frame beams according to claim 1 is characterized in that: The key stress-bearing locations include beam-column joint areas, potential or existing crack areas, mid-span areas, and high stress concentration areas.

3. The long-term monitoring and early warning method for large-span steel frame beams according to claim 1 is characterized in that: The structural model information of the target steel frame beam includes set topology information, material constitutive information, boundary condition parameters and load distribution benchmark.

4. The long-term monitoring and early warning method for large-span steel frame beams according to claim 1 is characterized in that: The multiple original deformation signals include at least a temperature signal, a strain signal and a vibration response signal.

5. The long-term monitoring and early warning method for large-span steel frame beams according to claim 4 is characterized in that: The temperature compensation algorithm and wavelet noise reduction algorithm pre-process the collected multiple original deformation signals in the following steps: Use temperature compensation algorithm to remove the temperature component in the original deformation signal; The wavelet denoising algorithm is used to filter and denoise the strain signal and vibration signal processed by the temperature compensation algorithm.

6. The long-term monitoring and early warning method for large-span steel frame beams according to claim 4 is characterized in that: The steps of identifying the deformation state of the key stress-bearing parts by using the improved conjugate beam method are specifically as follows: Establish a conjugate beam model with the same geometry and boundary conditions as the target steel frame beam; The bending moment distribution of the conjugate beam is solved by numerical iterative algorithm, and the real-time deflection curve and key section rotation angle of the target steel frame beam are obtained by inversion. Adaptive weight factors are used to interpolate and compensate the solution results in areas where sensors are sparsely distributed; Then, the deformation state of the target steel frame beam is identified based on the real-time deflection curve and key section rotation angle of the final target steel frame beam.

7. The long-term monitoring and early warning method for large-span steel frame beams according to claim 6 is characterized in that: The specific method for identifying the deformation state of the target steel frame beam according to the real-time deflection curve and key section angle of the final target steel frame beam includes at least one of the following: The real-time deflection curve and key section rotation angle obtained by the improved conjugate beam method are compared with their historical baseline values ​​or design allowable values. If they exceed the preset threshold, they are judged as abnormal. Perform statistical analysis on the deformation results of continuous recognition to detect abnormal points or mutations that exceed the normal fluctuation range and judge them as abnormal; Compare the predicted future deformation or change trend with the ultimate load state or critical failure threshold, and judge the anomaly based on the comparison results.

8. The long-term monitoring and early warning method for large-span steel frame beams according to claim 1 is characterized in that: The steps of constructing a steel frame beam abnormal deformation diagnosis model based on a graph neural network and identifying abnormal deformation states of key stress-bearing parts based on preset thresholds are as follows: The preprocessed multiple original deformation signals are used as independent modal inputs respectively, and the spatial topological relationship of the sensor network is modeled using a graph neural network. The cross-modal attention mechanism is used to dynamically weight the input of each original signal, and the preset threshold is used to diagnose in real time whether the deformation state of the target steel frame beam exceeds the expected state.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, it implements a long-term monitoring and early warning method for large-span steel frame beams as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, a long-term monitoring and early warning method for large-span steel frame beams as described in any one of claims 1 to 8 is implemented.

Images