A method for detecting the bearing capacity of a structure engineering on site

Abstract

The application provides a structural engineering bearing capacity on-site detection method, and belongs to the technical field of structural engineering detection.The application installs multi-axial acceleration sensors and strain gauge arrays on the surface of a structure component to be detected, applies incremental load by using an intelligent loading device, collects vibration response data and strain response data in real time by using a mobile data collection terminal, extracts modal parameters based on a frequency domain analysis method to establish a load-modal parameter relationship curve, analyzes and calculates the bearing capacity of the component, optimizes a load transfer path analysis by using a graph theory shortest path algorithm, and accurately evaluates the bearing performance of the component by using a structure response prediction model based on a Transformer architecture, so that technical integration of multi-dimensional detection data fusion, intelligent loading control, quantitative bearing capacity evaluation and accurate performance prediction is realized, and the technical problem of insufficient detection precision of the bearing capacity of the structure component is solved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of structural engineering detection, and in particular relates to a structural engineering bearing capacity on-site detection method. BACKGROUND

[0002] The structural engineering bearing capacity detection is a key technology to ensure the safety of buildings and engineering structures. The traditional detection method mainly relies on static loading test, material sampling detection and experience evaluation to judge the bearing performance of structural members. These methods are widely used in the fields of bridge detection, building safety evaluation, industrial equipment detection, etc. The bearing capacity of the structure is evaluated by applying a predetermined load and observing the structural response. However, the traditional detection method has significant defects. The static loading test can only obtain response data under specific load conditions and cannot fully reflect the dynamic behavior characteristics of the member under complex load history. Although material sampling detection can obtain basic performance parameters of materials, it is difficult to accurately reflect the overall bearing performance and damage state of the member. The experience evaluation method is subjective and lacks quantitative analysis basis. In the prior art, due to the lack of comprehensive analysis ability of multi-dimensional detection data and accurate quantitative evaluation means, the traditional method cannot accurately reflect the real bearing performance and damage state of the structural member, and cannot meet the technical requirements of modern engineering for accurate evaluation of structural safety. That is to say, the prior art has the technical problem of insufficient detection precision of the bearing capacity of the structural member. SUMMARY

[0003] Therefore, the present application provides a structural engineering bearing capacity on-site detection method, which can solve the technical problem of insufficient detection precision of the bearing capacity of the structural member in the prior art.

[0004] The application is implemented in the following manner: the application provides a structural engineering bearing capacity on-site detection method, which comprises the following steps: installing multi-axial acceleration sensors on the surface of a structural member to be detected, arranging the multi-axial acceleration sensors at a density of 4 per square meter, setting the sampling frequency of the multi-axial acceleration sensors to 1000 Hz, simultaneously installing a strain gauge array at a key load-bearing part of the member, and measuring the cross-sectional dimension of the member by using a digital electronic caliper; applying an incremental load to the structural member to be detected by using an intelligent loading device, wherein the incremental load comprises static loading and dynamic loading, the loading amplitude is gradually increased from 10% of the design load to 120%, and the surface hardness of the member is detected by using a digital electronic rebound hammer; collecting vibration response data and strain response data of the structural member to be detected under different loads in real time by using a mobile data acquisition terminal, and sending the vibration response data and the strain response data to a field data processing system by using a wireless data transmission module; processing the vibration response data based on a frequency domain analysis method, extracting modal parameters of the structural member to be detected, wherein the modal parameters comprise natural frequency, mode shape and damping ratio, and establishing a load-modal parameter relationship curve; analyzing and calculating the bearing capacity of the structural member to be detected by using an elastic-plastic damage constitutive equation of a material, and outputting a residual bearing capacity evaluation result of the member; optimizing a load transfer path analysis by using a shortest path algorithm in graph theory; and predicting the subsequent bearing performance of the structural member to be detected by using a structural response prediction model.

[0005] The structural response prediction model is a time series prediction network based on a Transformer architecture, and comprises an encoder and a decoder. The encoder is used to extract feature representations of historical detection data, and the decoder is used to generate prediction results of future bearing performance. The attention mechanism parameters in the structural response prediction model are dynamically adjusted according to the type of the member, the load characteristics and the detection accuracy requirements.

[0006] The structural response prediction model is based on current detection data and historical performance data, and outputs a bearing capacity degradation trend and a residual service life evaluation result of the structural member to be detected.

[0007] The digital electronic caliper is a high-precision digital caliper of Mitutoyo CD-6"CSX, has a measurement accuracy of ±0.01 mm, a measurement range of 0-150 mm, and is equipped with a Bluetooth data transmission function, and is used for accurately measuring length, width and thickness geometric dimension parameters of the member. The measured geometric parameters of the member are directly transmitted to the mobile data acquisition terminal for recording and subsequent analysis.

[0008] The intelligent loading device is an integrated detection platform of an integrated hydraulic loading system, a load control system and a data acquisition system, the intelligent loading device is equipped with high-precision load sensors and displacement sensors, the load sensor has an accuracy of ±0.5%, the displacement sensor has an accuracy of ±0.005 mm, the intelligent loading device has a self-adaptive load control function, and the loading rate and the loading mode are automatically adjusted according to the real-time response characteristics of the structure member to be detected.

[0009] The digital electronic rebound hammer is a concrete rebound hammer with a model number of HT-225, and has an impact energy of 2.207 J, and is used for detecting the rebound value of the surface material of the member, and the rebound value is converted into a surface hardness value by a built-in data processing chip, and the surface hardness value is used for evaluating the strength characteristics and uniformity of the material of the member.

[0010] The mobile data acquisition terminal has precise positioning function, Bluetooth communication function, information display function and Internet communication function, the precise positioning function obtains the spatial coordinates of the mobile data acquisition terminal through a satellite positioning system, the positioning accuracy is centimeter level, and the spatial coordinates are used for determining the site position of the structure member to be detected.

[0011] The load-mode parameter relationship curve is a mathematical curve for describing the change relationship between the load size and the natural frequency, the mode shape and the damping ratio.

[0012] The material elastic-plastic damage constitutive equation is used for describing the nonlinear behavior and damage evolution process of the structure member to be detected under the action of the load, the input of the material elastic-plastic damage constitutive equation includes the geometric parameters of the member, the material elastic modulus, the load history and the measured strain data, and the output is the residual bearing capacity evaluation result and the safety factor of the member.

[0013] The optimal load transmission path is a path with the minimum stress when the load is transmitted in the member under the given load condition.

[0014] The shortest path algorithm in the graph theory optimizes the steps of analyzing the load transmission path, and the steps are that the structure member to be detected is discretized into nodes and edges, and the optimal load transmission path and the stress distribution of the key nodes under different load conditions are calculated.

[0015] The multi-axis acceleration sensor is a sensor device capable of simultaneously measuring acceleration components in X, Y and Z directions.

[0016] The strain gauge array is a sensor combination composed of a plurality of strain gauges in a predetermined layout, and is used for measuring the strain distribution at different positions of the member.

[0017] The bearing capacity degradation trend is specifically a regularity description of the bearing capacity of the structure member to be detected changing with time.

[0018] The step of establishing the training data set of the structure response prediction model includes collecting historical detection data of different types of structure members, the historical detection data including member geometric parameters, material performance parameters, load history, vibration response data and final bearing capacity, pre-processing and standardizing the original data, organizing the data in time sequence, establishing the correspondence between the input feature vector and the target output vector, and dividing the data set into a training set, a validation set and a test set in a ratio of 8:1:1.

[0019] The step of training the structure response prediction model includes initializing the model parameters, setting the learning rate to 0.001, the batch size to 32, the number of training rounds to 1000 rounds, using mean square error as the loss function, using the Adam optimizer to update the parameters, using the early stopping strategy to prevent overfitting during the training process, stopping the training when the validation set loss does not decrease for 10 consecutive rounds, and evaluating the model performance on the test set after the training is completed.

[0020] The present application integrates multiple axial acceleration sensors, strain gauge arrays, intelligent loading devices and other detection equipment, combines frequency domain analysis, material elastic-plastic damage constitutive equation and time series prediction network based on the Transformer architecture, establishes a technical system for multi-dimensional detection data fusion analysis, can simultaneously obtain key information such as vibration response, strain distribution and modal parameters of the member, and solves the technical defects of single detection data and limited analysis dimension of traditional methods. The present application uses real-time data acquisition and wireless transmission technology, combines adaptive load control and environmental compensation system, realizes intelligent control of the detection process and real-time processing and analysis of data, quantitatively calculates the bearing capacity of the member through the material elastic-plastic damage constitutive equation, optimizes the load transfer path analysis by using the shortest path algorithm of graph theory, improves the accuracy and reliability of the bearing capacity evaluation, and overcomes the technical limitations of strong subjectivity and insufficient precision of traditional methods. The present application realizes accurate quantitative evaluation of the bearing capacity of the member by constructing a structure response prediction model based on historical data and current detection data, solves the technical problem of insufficient detection precision of traditional methods, and provides a scientific decision basis for structure safety management. In summary, the present application solves the technical problem of insufficient detection precision of the bearing capacity of the structure member in the background art. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The flowchart of the method of the present application.

[0022] Figure 2 The neural network structure diagram of the structure response prediction model involved in the present application.

[0023] Figure 3 Data acquisition operation interface diagram of software involved in the present application.

[0024] Figure 4 Data analysis operation interface diagram of software involved in the present application. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0026] As Figure 1 shown is a flowchart of a structural engineering bearing capacity on-site detection method provided by the present application, and the method comprises the following steps:

[0027] S01, a multi-axial acceleration sensor is installed on the surface of a structure member to be detected, the arrangement density of the multi-axial acceleration sensor is 4 per square meter, the sampling frequency of the multi-axial acceleration sensor is 1000 Hz, a strain gauge array is installed at a key part of the bearing of the member, and a digital electronic caliper is used to measure the cross-sectional dimension of the member;

[0028] S02, an incremental load is applied to the structure member to be detected by an intelligent loading device, the incremental load comprises static loading and dynamic loading, the loading amplitude is gradually increased from 10% of the design load to 120%, and a digital electronic rebound instrument is used to detect the hardness of the surface of the member;

[0029] S03, vibration response data and strain response data of the structure member to be detected under different loads are collected in real time by a mobile data acquisition terminal, and the mobile data acquisition terminal sends the vibration response data and the strain response data to an on-site data processing system through a wireless data transmission module;

[0030] S04, the vibration response data are processed based on a frequency domain analysis method, modal parameters of the structure member to be detected are extracted, the modal parameters comprise natural frequency, mode shape and damping ratio, and a load-modal parameter relationship curve is established;

[0031] S05, a material elastic-plastic damage constitutive equation is used to analyze and calculate the bearing capacity of the structure member to be detected, the input of the material elastic-plastic damage constitutive equation comprises member geometric parameters, material elastic modulus, load history and measured strain data, and the output is a member residual bearing capacity evaluation result;

[0032] S06, a shortest path algorithm in graph theory is used to optimize load transfer path analysis, the structure member to be detected is discretized into nodes and edges, and the optimal load transfer path and key node stress distribution under different load conditions are calculated.

[0033] S07, predicting the subsequent bearing performance of the structure member to be detected by a structure response prediction model, the structure response prediction model being based on current detection data and historical performance data, outputting the bearing capacity degradation trend and remaining service life evaluation result of the structure member to be detected.

[0034] The digital electronic caliper is a high-precision digital caliper of model Mitutoyo CD-6"CSX, with a measurement accuracy of ±0.01 mm, a measurement range of 0-150 mm, and a Bluetooth data transmission function, for accurately measuring the length, width, thickness and other geometric size parameters of the member, and the measured geometric parameters of the member are directly transmitted to the mobile data acquisition terminal for recording and subsequent analysis. The digital electronic rebound hammer is a concrete rebound hammer of model HT-225, with an impact energy of 2.207 J, for detecting the rebound value of the surface material of the member, and the rebound value is converted into a surface hardness value by the built-in data processing chip, and the surface hardness value is used to evaluate the strength characteristics and uniformity of the member material.

[0035] The intelligent loading device is an integrated detection platform integrating a hydraulic loading system, a load control system and a data acquisition system, and is equipped with high-precision load sensors and displacement sensors, the load sensor accuracy is ±0.5%, and the displacement sensor accuracy is ±0.005 mm. The intelligent loading device has a self-adaptive load control function, which automatically adjusts the loading rate and loading mode according to the real-time response characteristics of the structure member to be detected, and automatically reduces the loading rate or pauses the loading when detecting abnormal response of the structure member to be detected. The intelligent loading device integrates an environmental compensation system for eliminating the influence of temperature, humidity and other environmental factors on the detection results, with an environmental temperature acquisition accuracy of ±0.1℃ and an environmental humidity acquisition accuracy of ±1%RH. In order to improve the reliability of the detection data, the intelligent loading device uses multi-sensor fusion technology to compare and analyze the measurement results of different types of sensors, identify and eliminate abnormal data points, and ensure the accuracy and consistency of the detection results. The loading rate is adaptively adjusted according to the geometric parameters and material elastic modulus of the member, and the loading mode is dynamically selected according to the load history and measured strain data.

[0036] The mobile data acquisition terminal has precise positioning function, Bluetooth communication function, information display function and Internet communication function. The precise positioning function obtains the spatial coordinates of the mobile data acquisition terminal through the satellite positioning system, and the positioning accuracy is centimeter level. The spatial coordinates are used to determine the field position of the structure member to be detected. The Bluetooth communication function is used to receive the measurement data transmitted by the digital electronic caliper and the digital electronic rebound hammer, the communication distance is 10 m, and the data transmission rate is 2 Mbps. The mobile data acquisition terminal establishes communication connection with the field data processing system through the wireless data transmission module. The wireless data transmission module supports two communication modes of 4G and WiFi, and ensures the stability and real-time performance of data transmission.

[0037] The material elastic-plastic damage constitutive equation is used to describe the nonlinear behavior and damage evolution process of the structure member to be detected under the action of load. The input of the material elastic-plastic damage constitutive equation includes the material elastic modulus, yield strength, ultimate strength and damage parameter, and the output is the stress-strain relationship and damage state of the structure member to be detected under different load levels. The material elastic modulus is derived from the detection results of the digital electronic rebound hammer and material test data. The yield strength and ultimate strength are derived from the component design data and field sampling test results. The damage parameter is calculated according to the strain response data and loading history.

[0038] The bearing capacity evaluation function is used to comprehensively analyze the bearing performance and safety state of the structure member to be detected. The input of the bearing capacity evaluation function includes the measured modal parameters, the strain response data, the load history and the material performance parameters. The output is the residual bearing capacity evaluation result and the safety factor of the component. The measured modal parameters are derived from the frequency domain analysis result of step S04. The material performance parameters include the material elastic modulus, yield strength and ultimate strength. The safety factor is used to evaluate the safety margin and use state of the structure member to be detected.

[0039] As Figure 2As shown, the structure response prediction model is a time series prediction network based on a Transformer architecture, which includes an encoder for extracting feature representations of historical detection data and a decoder for generating prediction results of future bearing capacity, and the attention mechanism parameters in the structure response prediction model are dynamically adjusted according to the type of structural member, load characteristics and detection accuracy requirements; the steps of establishing the training data set of the structure response prediction model include collecting historical detection data of different types of structural members, the historical detection data including the geometric parameters of the member, material performance parameters, the load history, the vibration response data and the final bearing capacity, preprocessing and standardizing the original data, organizing the data in time series, establishing the corresponding relationship between the input feature vector and the target output vector, dividing the data set into a training set, a validation set and a test set in a ratio of 8:1:1; the steps of training the structure response prediction model include initializing the model parameters, setting the learning rate to 0.001, the batch size to 32, the number of training rounds to 1000, using mean square error as the loss function, using the Adam optimizer to update the parameters, using the early stopping strategy to prevent overfitting during the training process, stopping the training when the validation set loss does not decrease for 10 consecutive rounds, and evaluating the model performance on the test set after training to ensure that the prediction accuracy meets the requirements of engineering applications.

[0040] The attention weight adjustment function is used to adjust the attention mechanism parameters of the structure response prediction model, and the attention weight adjustment function is calculated based on the type of structural member, load characteristics, detection accuracy and historical data quality to obtain a weight adjustment coefficient, when the weight adjustment coefficient belongs to the range of 0 to 0.3, a linear weight adjustment function is used to adjust the attention mechanism parameters of the structure response prediction model, when the weight adjustment coefficient belongs to the range of 0.3 to 0.7, an exponential weight adjustment function is used to adjust the attention mechanism parameters of the structure response prediction model, and when the weight adjustment coefficient belongs to the range of 0.7 to 1.0, a logarithmic weight adjustment function is used to adjust the attention mechanism parameters of the structure response prediction model, the type of structural member is derived from engineering design data, the load characteristics are derived from the analysis results of the load history, the detection accuracy is derived from sensor calibration data, and the historical data quality is determined by data integrity and consistency evaluation.

[0041] The multi-axis acceleration sensor is a sensor device capable of simultaneously measuring acceleration components in X, Y and Z directions.

[0042] The strain gauge array is a sensor combination composed of a plurality of strain gauges arranged in a predetermined layout for measuring strain distribution at different positions of the member.

[0043] The load-mode parameter relationship curve is a mathematical curve describing the change relationship between the load size and the natural frequency, mode shape and damping ratio.

[0044] The optimal load transfer path is a path in which stress is minimum when a load is transferred inside a component under a given load condition.

[0045] The load-carrying capacity degradation trend is a regular description of the load-carrying capacity of the structure component to be detected changing over time.

[0046] Further, a host computer is also included, and a detection data analysis software is arranged on the host computer to support summarizing and analyzing the collected data, as shown in Figures 3-4

[0047] The specific implementation of the above steps is described in detail below.

[0048] The specific implementation of step S01 is to first determine the sensor arrangement scheme, calculate the arrangement point coordinate according to the geometric shape and size of the structure component to be detected, divide the component surface into a plurality of measurement regions by using the grid arrangement principle, control the area of each region within the range of 0.25 square meters, and ensure that the sensor coverage rate reaches 100%. Then, the calibration work of the multi-axis acceleration sensor is performed, the sensor is checked for three-axis accuracy by using the gravity acceleration reference method, the calibration error threshold is controlled within the range of ±0.01g, and the frequency response characteristics of the sensor are checked to ensure that the linearity meets the requirements within the frequency band of 1-500Hz. Then, the sensor installation and fixation operation is performed, the sensor is firmly bonded to the component surface by using epoxy resin adhesive, the curing time is not less than 24 hours, and the coupling effect of the sensor and the component is verified by tapping test after installation. The installation of the strain gauge array adopts temperature compensation technology, the component surface is first cleaned to remove oil stains and oxide layers, the strain gauge is pasted after coating the primer, the strain gauge is arranged according to the principle of main stress direction distribution, and the longitudinal and transverse strain gauge groups are arranged at the maximum bending moment section and the maximum shear section of the component, respectively. The measurement work of the digital electronic caliper is performed according to the multi-point average principle, the length, width and thickness of the same section are measured 5 times to take the average value, the measurement accuracy is controlled within the range of ±0.01mm, and the measurement data is automatically transmitted to the mobile data acquisition terminal for recording through the Bluetooth module. The maximum bending moment section of the component refers to the position of the cross section with the maximum bending moment value along the length direction of the component under the action of external load, which is usually located at the concentrated load action point or the uniformly distributed load midspan position. The maximum shear section refers to the position of the cross section with the maximum absolute value of shear along the length direction of the component under the action of external load, which is usually located near the support or the load mutation point. The determination of these parameters needs to be obtained by structural mechanics analysis and calculation to provide a theoretical basis for the reasonable arrangement of the sensor.

[0049] ​The specific implementation of step S02 is to establish a control system of the intelligent loading device, to realize accurate application of the load by adopting a closed-loop feedback control principle. The control system automatically adjusts the output pressure of the hydraulic system according to a preset loading program, and the load control precision reaches ±0.5%. The formulation of the loading scheme is based on the design load value of the component. The static force loading adopts a step loading mode, and the load increment of each step is 10% of the design load. The holding time of each step is not less than 5 minutes. The dynamic force loading adopts a sinusoidal wave excitation signal, and the excitation frequency range is 1-100 Hz. The excitation amplitude is gradually increased. The detection work of the digital electronic rebound hammer follows the statistical principle. Not less than 16 measuring points are uniformly selected on the surface of the component. Each measuring point is measured continuously for 3 times to take the average value. The variation coefficient of the rebound value is controlled within 15%. The environmental compensation system monitors the temperature and humidity changes of the detection site in real time. When the temperature change exceeds ±2℃ or the humidity change exceeds ±5%, the system automatically performs environmental correction calculation to eliminate the influence of environmental factors on the detection result. The multi-sensor fusion technology adopts a weighted average algorithm to comprehensively process the measurement results of different sensors. The weight coefficient is determined according to the sensor accuracy level and historical reliability data. The abnormal data identification adopts a three times standard deviation criterion. The data points exceeding the normal range are automatically excluded.

[0050] The specific implementation of step S03 is to establish a communication network of the data acquisition system. The mobile data acquisition terminal serves as a data aggregation node, receives various sensor signals through wired and wireless modes, adopts a time synchronization technology to ensure the synchronicity of multi-channel data, and controls the time synchronization precision within 1 ms. The data preprocessing module performs filtering and amplification processing on the original signal. A Butterworth low-pass filter is adopted to remove high-frequency noise, and the cutoff frequency is set to 500 Hz. The signal amplification multiple is adaptively adjusted according to the sensor output characteristics. The data storage adopts a circular buffer mechanism. When the storage capacity reaches 80%, the data uploading program is automatically started to ensure that the data is not lost. The wireless data transmission module adopts a dual-mode communication mode. WiFi is preferentially used for data transmission to obtain a higher transmission rate. When the WiFi signal is unstable, it is automatically switched to the 4G network. The transmission protocol adopts TCP to ensure the reliability of data transmission. The field data processing system immediately performs format verification and integrity check after receiving the data. The abnormal data triggers a retransmission mechanism. The data reception confirmation signal is automatically returned to the mobile terminal.

[0051] The specific implementation of step S04 is to convert the vibration response data in the frequency domain by using the fast Fourier transform algorithm, convert the time domain signal into a frequency domain signal to extract modal information, set the transform window length to 2048 sampling points, and set the overlap rate to 50% to improve the spectral resolution. The modal parameter identification adopts the frequency domain decomposition method, and the natural frequency of the structure is identified through the power spectral density function, and the frequency identification accuracy is controlled within ±0.1 Hz. The mode shape extraction adopts the singular value decomposition algorithm, and the frequency response function matrix is decomposed to obtain the mode shape vector of each order, and the mode shape normalization processing adopts the mass normalization method. The damping ratio calculation adopts the half-power point method, and two frequency points with a power drop of 3dB near the resonance peak are selected, and the damping ratio value is calculated through the bandwidth. The relationship curve between the load and the modal parameters is established by using the least square fitting method, and the modal parameters under different load levels are polynomial fitted, the fitting order is determined according to the data distribution characteristics, the fitting accuracy is evaluated by the correlation coefficient, and the correlation coefficient is required to be not less than 0.95.

[0052] The specific implementation of step S05 is to establish a material elastic-plastic damage constitutive model, which is based on the continuous damage mechanics theory and uses the strain equivalent principle to describe the nonlinear behavior of the material. The input parameters include the geometric parameters of the component, the basic mechanical parameters such as the material elastic modulus, the Poisson's ratio, the yield strength, the ultimate strength, and the measured load history and strain response data. The solution of the constitutive equation adopts the incremental iteration method, which discretizes the load history into several incremental steps, and uses the Newton-Raphson iteration algorithm to solve the nonlinear equation set in each incremental step, and the iteration convergence criterion is that the relative error is less than . The calculation of the damage variable is based on the strain energy release rate theory, and the damage degree is determined by comparing the current strain state with the historical maximum strain state, and the damage threshold is set to 80% of the yield strain of the material. The calculation results of the bearing capacity include the current residual bearing capacity of the component, the safety factor and the damage degree, and the safety factor is calculated by using the limit state design method, and when the safety factor is less than 1.5, it is determined that the bearing capacity is insufficient.

[0053] The specific implementation of step S06 is to establish a load transfer path optimization model using Dijkstra's algorithm in graph theory. The structure to be detected is discretized into a network structure of nodes and edges, with nodes representing key cross-section positions of the components and edges representing load transfer paths between adjacent cross-sections. The node weight is determined according to the cross-sectional geometric characteristics and material properties, and the edge weight is calculated according to the distance and stiffness distribution between the two nodes. The algorithm starts from the load application point and gradually searches for the shortest path to each node. The path length is represented by a stress level weighted distance, with higher stress resulting in higher weight. The key node identification uses centrality analysis methods to calculate the betweenness centrality and closeness centrality of each node. Nodes with high centrality values are key control points for load transfer. Stress distribution calculation is based on the finite element discretization principle, which discretizes the continuous structure into a finite number of elements. The stress distribution within each element is calculated using shape function interpolation. The element type is selected according to the geometric shape of the component, with beam elements for slender components and plate elements for planar components.

[0054] The specific implementation of step S07 is to establish a time series prediction neural network model based on the Transformer architecture. This model uses an encoder-decoder structure to process historical detection data and predict future bearing performance trends. The encoder part uses a multi-head self-attention mechanism to extract time series features from historical data, with 8 attention heads and a hidden layer dimension of 512. The decoder part generates prediction results for future time steps, with a prediction time span of 10% of the remaining design service life of the component. Model training uses supervised learning, with a mean squared error loss function and an Adam optimizer. The learning rate initial value is set to 0.001, and the cosine annealing scheduling strategy is used to dynamically adjust the learning rate. Data preprocessing includes standardization, missing value filling, and outlier processing. Standardization uses the Z-score method, missing values are filled using linear interpolation, and outliers are identified and processed using the interquartile range method. Model validation uses time series cross-validation, dividing the data into multiple training and test sets according to time order to verify the model's generalization ability. The prediction results include bearing capacity degradation curves, residual service life probability distribution, and risk assessment levels. Risk levels are divided into low, medium, and high risk, corresponding to different maintenance strategy recommendations.

[0055] The implementation of the attention weight adjustment function is based on a multi-factor comprehensive evaluation method. The component type factor is determined according to the structural form and stress characteristics. The weight coefficient of beam components is 0.8, the weight coefficient of column components is 0.9, and the weight coefficient of plate components is 0.7. The load characteristic factor is calculated according to the load type and variation law. The weight coefficient of static load is 0.6, the weight coefficient of dynamic load is 0.8, and the weight coefficient of alternating load is 1.0. The detection accuracy factor is determined according to the sensor calibration results. When the accuracy level is A level, the weight coefficient is 1.0, when the level is B, the weight coefficient is 0.8, and when the level is C, the weight coefficient is 0.6. The historical data quality factor is determined by evaluating the data integrity and consistency. When the data integrity is higher than 95% and the consistency index is greater than 0.9, the weight coefficient is 1.0. The weight adjustment coefficient is calculated by using the weighted average method, and the weights of each factor are 0.3, 0.3, 0.2, and 0.2 respectively. According to the numerical range of the weight adjustment coefficient, different adjustment functions are used. Linear function is suitable for small adjustment, exponential function is suitable for medium adjustment, and logarithmic function is suitable for large adjustment, ensuring that the adjustment of the attention mechanism parameters is accurate and stable.

[0056] It should be noted that the structural response prediction model adopts a time series prediction network based on the Transformer architecture. The overall structure consists of two core modules, encoder and decoder. Through multi-head self-attention mechanism and position encoding technology, the deep learning and prediction analysis of the bearing performance time series data of structural components are realized. The encoder module is responsible for receiving and processing historical detection data, including component geometric parameters, material performance parameters, load history, vibration response data and strain response data, etc. Multi-dimensional feature information is extracted through multi-layer neural network and attention mechanism to extract the potential law and feature representation in the data, and the original high-dimensional time series data is mapped to a low-dimensional feature vector space, providing an effective feature basis for subsequent prediction analysis. The decoder module receives the feature representation output by the encoder, combines the current detection data and the preset prediction time window, and generates a prediction sequence of future bearing performance through a recurrent neural network structure and an attention mechanism, outputting the bearing capacity degradation trend and the remaining service life evaluation results.

[0057] The attention mechanism of the structural response prediction model adopts a dynamic weight adjustment strategy. The weight adjustment coefficient is calculated according to the component type, load characteristics, detection accuracy, and historical data quality through an attention weight adjustment function. When the weight adjustment coefficient is in the range of 0 to 0.3, a linear weight adjustment function is used; when the weight adjustment coefficient is in the range of 0.3 to 0.7, an exponential weight adjustment function is used; and when the weight adjustment coefficient is in the range of 0.7 to 1.0, a logarithmic weight adjustment function is used. This segmented weight adjustment mechanism enables the model to adaptively adjust the attention degree to different characteristics, improving the accuracy and robustness of the prediction results. The model also integrates a position encoding layer and a multi-head attention layer. The position encoding layer is used to maintain the time sequence information of the time series data, and the multi-head attention layer captures different levels and angles of feature relationships in the data through parallel calculation of multiple attention heads, enhancing the model's learning ability for complex time series patterns.

[0058] The establishment process of the structural response prediction model training data set first collects historical detection data of different types of structural components from multiple engineering projects and laboratory tests. The data collection range covers steel structures, concrete structures, composite material structures, and other component types, ensuring the diversity and representativeness of the data set. The collected raw data includes component geometric parameters such as length, width, thickness, cross-sectional area, and other size information, material performance parameters such as elastic modulus, yield strength, ultimate strength, Poisson's ratio, and other mechanical properties, load history data including loading time series, load amplitude variation, loading frequency, and other dynamic information, vibration response data including acceleration time history, frequency domain response, modal parameters, and other dynamic characteristics, strain response data including strain time history, strain distribution, damage indicators, and other static mechanical characteristics, as well as final bearing capacity test results and failure mode information.

[0059] In the data preprocessing stage, the collected raw data is subjected to quality control and standardization processing. First, data cleaning is performed to remove outliers and noise data, missing data is identified and processed through statistical analysis methods, and incomplete data records are processed using interpolation algorithms or deletion strategies. Then, features with different dimensions and value ranges are normalized to ensure that each feature is learned and compared on the same numerical scale. The data standardization process uses the Z-score standardization method to convert the value of each feature to a distribution with a mean of 0 and a standard deviation of 1, eliminating the influence of dimension differences between different features on model training.

[0060] The time series data organization stage rearranges and organizes the preprocessed data in chronological order, establishes the correspondence between the time window and the prediction window, the time window is used for inputting historical data for feature learning, and the prediction window is used for outputting future prediction results for supervised learning. By using the sliding window technique, continuous time series data is divided into multiple training samples, each sample contains fixed-length historical data as an input feature vector and corresponding future data as a target output vector, establishing the mapping relationship between input and output. The data set is finally divided into training set, validation set and test set according to the ratio of 8:1:1, the training set is used for the optimization learning of model parameters, the validation set is used for hyperparameter adjustment and model selection, and the test set is used for objective evaluation of the final model performance, ensuring the generalization ability and actual application effect of the model.

[0061] The model training process adopts a standard deep learning training process, first initializes the model parameters including weight matrix and bias vector, sets the key hyperparameters including learning rate 0.001, batch size 32, training rounds 1000, selects mean square error as the loss function to measure the difference between the predicted value and the true value, and uses Adam optimizer for gradient descent and parameter update. During the training process, the early stopping strategy is implemented to prevent model overfitting, when the loss function on the validation set does not decrease for 10 consecutive rounds, the training is automatically stopped, ensuring that the model learns sufficiently on the training set while having good generalization performance. After training, the final performance is evaluated on an independent test set, the prediction accuracy and stability of the model are evaluated by multiple evaluation indicators such as root mean square error, mean absolute error, and determination coefficient, to ensure that the model meets the accuracy requirements and reliability standards of actual engineering applications.

[0062] It should be noted that the sensor arrangement density formula is based on the spatial sampling theory, by ensuring the density of 4 sensors per square meter, it can meet the spatial resolution requirements of structural vibration signals, compared with the traditional empirical arrangement method, this formula method ensures the scientificity of sensor arrangement and the integrity of measurement data. The incremental load formula adopts a linear incremental strategy, from 10% of the design load to 120%, which can fully reflect the response characteristics of the component under different load levels, compared with single load level detection, this method can obtain more rich structural performance information.

[0063] The multi-sensor fusion weighted average formula is based on the information fusion theory. By assigning appropriate weights to sensors of different accuracy levels, it can effectively improve the reliability and accuracy of measurement results. Compared with single sensor measurement, this fusion method has the advantages of strong anti-interference ability and small measurement error. The three standard deviation criterion for abnormal data identification is based on the normal distribution theory. It can effectively identify and eliminate abnormal data points in the measurement process, ensuring the accuracy of subsequent analysis and calculation. The identification accuracy of this method reaches 99.7%. The Z-score method of data standardization can eliminate the differences in dimensions and numerical ranges of different sensor outputs, providing a unified data basis for multi-source data fusion and machine learning model training.

[0064] The fast Fourier transform formula converts time-domain vibration signals into frequency-domain signals, which can effectively extract the modal characteristics of the structure. Its calculation efficiency is much higher than that of traditional frequency-domain analysis methods, providing technical support for real-time online detection. The damping ratio calculation formula is based on the half-power point method. By identifying the frequency points on both sides of the resonance peak, it can accurately obtain the damping characteristics of the structure. Compared with other damping identification methods, this method has the advantages of simple calculation and high precision.

[0065] The elastoplastic damage constitutive equation introduces damage variables into the stress-strain relationship, which can describe the nonlinear behavior and damage evolution process of materials under load. Compared with traditional linear elastic analysis methods, this equation can more accurately predict the load capacity degradation process of the structure. The damage variable evolution equation is based on the theory of continuum damage mechanics. It describes the development law of damage through an exponential function. This function form can reflect the nonlinear characteristics and cumulative effects of damage development. The safety factor calculation formula is based on the limit state design theory. By comparing the residual carrying capacity of the structure with the actual load effect, it can quantitatively evaluate the safety state and performance of the structure.

[0066] The load transfer path weight function considers two key factors: physical distance and stress level. It quantifies the transfer efficiency of the path through weighted combination. Compared with the traditional method that only considers geometric distance, this function can more accurately identify the weak links of the structure. The distance update formula of Dijkstra's algorithm uses dynamic programming ideas. By gradually optimizing the path length, it can efficiently find the global optimal solution. The time complexity of this algorithm is , which is significantly more efficient than the exhaustive search method.

[0067] The attention weight adjustment function comprehensively evaluates the model adjustment demand through a multi-factor weighted average manner, can adaptively adjust the model parameters according to different detection conditions, and compared with the prediction model with fixed parameters, the adjustment mechanism significantly improves the adaptability and prediction accuracy of the model. The design of the segmented adjustment function is based on the characteristics of different adjustment amplitudes, the linear function is suitable for small amplitude adjustment to maintain stability, the exponential function is suitable for medium amplitude adjustment to provide sufficient adjustment range, and the logarithmic function is suitable for large amplitude adjustment to avoid parameter mutation, and this segmented processing strategy ensures the smoothness and effectiveness of parameter adjustment.

[0068] The key technical ideas of the present application mainly embody in the following aspects.

[0069] The first key technical idea is an intelligent detection system of multi-sensor fusion. The traditional structure detection method usually relies on a single type of sensor for measurement, which is easily affected by environmental interference and sensor errors, resulting in insufficient reliability of the detection result. The present application adopts multi-axis acceleration sensor, strain gauge array, digital electronic caliper and digital electronic rebound hammer and other sensors to work cooperatively, and through multi-sensor fusion technology, the different types of measurement data are comprehensively processed, which can effectively improve the detection accuracy and reliability. This technical idea is based on the information fusion theory, through the combination of redundant measurement and complementary measurement, not only can verify the correctness of the measurement result, but also can obtain comprehensive information that single sensor cannot provide, so as to realize the comprehensive and accurate evaluation of the structure bearing capacity.

[0070] The second key technical idea is the load transfer path optimization analysis based on graph theory algorithm. The traditional structure analysis method mainly relies on finite element numerical calculation, the calculation efficiency is relatively low and it is difficult to intuitively reflect the load transfer rule in the structure. The present application innovatively introduces the shortest path algorithm in graph theory, which discretizes the complex three-dimensional structure into a network model of nodes and edges, and identifies the weak links and key control points of the structure by finding the optimal load transfer path. This method not only has high calculation efficiency, but also can clearly reveal the load transfer mechanism, providing an important basis for structure design optimization and safety evaluation. Compared with the traditional method, this technical idea has the advantages of fast calculation speed, clear physical meaning and wide application range.

[0071] The third key technical idea is an intelligent prediction model based on the Transformer architecture. Traditional structural performance prediction mainly relies on empirical formulas or simple statistical models, which have limited prediction accuracy and are difficult to handle complex nonlinear relationships. The present application uses advanced deep learning techniques to establish a time series prediction neural network based on the Transformer architecture, which can automatically learn the internal laws of structural performance changes from historical test data, and accurately predict the future degradation trend of carrying capacity. This model has strong feature extraction and long sequence modeling capabilities, and can handle complex dependency relationships in multi-dimensional time series data, with significantly better prediction accuracy than traditional methods.

[0072] The synergistic effect of these three key technical ideas produces significant technical effects and advantages. The multi-sensor fusion system provides high-quality input data for the graph theory algorithm, ensuring the accuracy of load transfer path analysis. The analysis results of the graph theory algorithm provide important structural feature information for the Transformer prediction model, improving the prediction accuracy of the model. The prediction results of the Transformer model can guide the optimization of sensor placement and adjustment of detection schemes, forming a complete closed-loop optimization system. This synergistic effect makes the entire detection method have comprehensive advantages such as high precision, fast efficiency, and strong adaptability, which can meet the needs of on-site detection of complex engineering structures and provide scientific and reliable technical support for structural safety assessment and maintenance decision-making.

[0073] Specifically, the principle of the present application is that the core principle of the present application that can solve the problems of the prior art lies in establishing a multi-dimensional detection data fusion analysis technical framework, which simultaneously collects dynamic response and static strain information of the member through multi-axis acceleration sensors and strain gauge arrays, obtaining more comprehensive and accurate structural behavior data than traditional single detection methods, providing a reliable data foundation for accurate assessment of carrying capacity. The self-adaptive control function of the intelligent loading device can dynamically adjust the loading parameters according to the real-time response characteristics of the member, ensuring the safety of the detection process and the effectiveness of the data. The application of the environmental compensation system eliminates the interference of external factors on the detection results, improving the stability and consistency of the detection accuracy.

[0074] The combination of the frequency domain analysis method and the material elastic-plastic damage constitutive equation enables the present application to analyze the carrying behavior of the member from the mechanical mechanism level, the extraction of modal parameters reveals the change law of the dynamic characteristics of the member, and the damage constitutive equation describes the nonlinear behavior and damage evolution process of the material under load. This physical mechanism-based analysis method is more scientific and accurate than traditional empirical evaluation. The introduction of the graph shortest path algorithm optimizes the load transfer path analysis, which can identify the key stress areas inside the member and provide more detailed analysis results for carrying capacity assessment.

[0075] A time-series prediction network based on the Transformer architecture establishes a mathematical model of how a structural member's load-bearing capacity changes over time by learning regular features from historical inspection data. The application of an attention mechanism enables the model to automatically identify key factors affecting load-bearing capacity and dynamically adjust prediction weights, improving the accuracy and reliability of the prediction results. This deep learning-based prediction method combines the statistical regularities of a large amount of historical data with real-time information from current inspection data, enabling quantitative prediction of the future load-bearing performance of structural members and providing forward-looking technical support for structural safety management.

[0076] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0077] The specific implementation of step S01 involves determining the sensor layout coordinates based on the component's geometric characteristics and load distribution. The sensor layout density calculation formula is as follows:

[0078] ;

[0079] In the formula, Sensor density, expressed in units per square meter; This represents the total number of sensors arranged on the surface of the component. This represents the total surface area of ​​the component, expressed in square meters.

[0080] Sensor coordinate calculation uses a gridded layout method, the first The coordinates of each sensor are represented as follows:

[0081] ;

[0082] In the formula, For the first The spatial coordinates of each sensor; These are the three-dimensional coordinate components of the sensor, in meters; This is the row number where the sensor is located, and its value is a positive integer. The value is the column number where the sensor is located, and it is a positive integer. The grid is in direction and The directional spacing, in meters, is typically 0.5 meters.

[0083] The formula for calculating the calibration error of a multi-axis accelerometer is:

[0084] ;

[0085] In the formula, The relative error of sensor calibration is expressed as a percentage. The measured acceleration value is given by the sensor, in m / s². The reference value for gravitational acceleration is 9.8 m / s². Sensor calibration requirements. .

[0086] The specific implementation of step S02 is to establish an incremental load loading sequence, and the load increment formula is expressed as:

[0087] ;

[0088] In the formula, For the first Level load value, in Newtons; The reference value for the design load of the component is given in Newtons. This represents the loading level, with a value ranging from 1 to 12.

[0089] Multi-sensor fusion employs a weighted average algorithm, and the fusion result is calculated using the following formula:

[0090] ;

[0091] In the formula, The measurement results are obtained after multi-sensor fusion. For the first The measured values ​​of each sensor; For the first The weighting coefficients for each sensor range from 0 to 1, and ; The number of sensors participating in the fusion.

[0092] Outlier identification uses the three-standard-deviation criterion, and the judgment formula is as follows:

[0093] ;

[0094] In the formula, For the first One data point to be detected; The mean of the data sequence to be detected; denoted as the standard deviation of the data sequence to be tested.

[0095] The dynamic loading excitation signal is in the form of a sine wave, and the excitation force is expressed as:

[0096] ;

[0097] In the formula, for The dynamic stimulus of a moment, measured in Newtons; The excitation amplitude is expressed in Newtons. The excitation frequency is measured in Hz and ranges from 1 to 100 Hz. The initial phase angle is expressed in radians. This is a time variable, with the unit being seconds.

[0098] The statistical processing of rebound values ​​uses the arithmetic mean method, and the calculation formula is as follows:

[0099] ;

[0100] In the formula, This represents the average rebound value. The number of measurement points is required. ; For the first The rebound value at each measuring point.

[0101] The specific implementation of step S03 is to establish a data preprocessing system, and the data standardization adopts the Z-score method. The standardization formula is:

[0102] ;

[0103] In the formula, For the first Standardized values ​​of each data point, dimensionless; For the first One original data point; The mean of the original data sequence; denoted as the standard deviation of the original data sequence. The standardized data follows a standard normal distribution with a mean of 0 and a standard deviation of 1.

[0104] The specific implementation of step S04 is to perform frequency domain analysis on the vibration signal using Fast Fourier Transform, and the transformation formula is expressed as follows:

[0105] ;

[0106] In the formula, For the frequency domain signal at the 1st... Complex values ​​at each frequency point; For the first The time-domain signal value of each sampling point; This represents the total number of sampling points, typically 2048. The imaginary unit satisfies ; The frequency point number, with a value range of 0 to... .

[0107] Natural frequency identification is achieved through power spectral density peak detection. The power spectral density is calculated using the following formula:

[0108] ;

[0109] In the formula, is the power spectral density at frequency , in (m / s²)² / Hz; is the Fourier transform result at frequency ; is the signal duration, in seconds; is the sampling frequency, taking value 1000 Hz.

[0110] The damping ratio is calculated by the half-power point method, and the calculation formula is:

[0111] ;

[0112] wherein, is the damping ratio, dimensionless, typically taking value in the range of 0.01 to 0.05; is the resonance frequency, in Hz; is the frequency value corresponding to a 3dB power drop on both sides of the resonance peak, in Hz.

[0113] The relationship between the load and the modal parameter is fitted by a polynomial, and the fitting function is expressed as:

[0114] ;

[0115] wherein, is the natural frequency of the th order under the load , in Hz; is the applied load, in Newton; is the fitting coefficient, wherein is in Hz, is in Hz / N, is in Hz / N², is in Hz / N³; is the fitting error term, in Hz, and the requirement is Hz; is the modal order, typically taking value in the range of 1 to 5.

[0116] The specific implementation of step S05 is to establish a material elastic-plastic damage constitutive model, and the stress-strain relationship is expressed as:

[0117] ;

[0118] wherein, is the effective stress, in MPa; is the damage variable, taking value in the range of 0 to 1, wherein 0 represents no damage and 1 represents complete damage; is the material elastic modulus, in GPa; is the material strain, dimensionless.

[0119] The evolution equation of damage variable is expressed as:

[0120] ;

[0121] In the formula, is the damage initial strain, dimensionless, and the typical value is 80% of the yield strain; is the historical maximum strain, dimensionless; is the ultimate strain, dimensionless.

[0122] The carrying capacity evaluation function is expressed as:

[0123] ;

[0124] In the formula, is the residual carrying capacity, unit: Newton; is the initial carrying capacity, unit: Newton; is the material performance reduction coefficient, the value range is 0.8 to 1.0; is the environmental influence coefficient, the value range is 0.9 to 1.0.

[0125] The safety factor calculation adopts the limit state design method, and the calculation formula is:

[0126] ;

[0127] In the formula, is the safety factor, dimensionless, and the requirement is ; is the actual action load effect, unit: Newton.

[0128] The specific implementation of step S06 is to establish a load transfer path model by using a graph theory algorithm, and the distance weight calculation formula between nodes is:

[0129] ;

[0130] In the formula, is the path weight of node to node , dimensionless; is the physical distance between two nodes, unit: meter; is the average stress on the path, unit: MPa; is the allowable stress of the material, unit: MPa; is the weight coefficient, dimensionless, and the typical values are 0.6 and 0.4 respectively.

[0131] The shortest path search adopts the Dijkstra algorithm, and the distance update formula is:

[0132] ;

[0133] wherein, is the shortest distance from the origin to node , dimensionless; is the shortest distance from the origin to node , dimensionless; is the edge weight from node to node , dimensionless.

[0134] The key node identification adopts the betweenness centrality calculation, and the formula is expressed as:

[0135] ;

[0136] wherein, is the betweenness centrality of node , dimensionless, the value range is 0 to 1; is the total number of shortest paths from node to node ; is the number of shortest paths passing through node .

[0137] The specific implementation of step S07 is to establish a prediction model based on deep learning, and the attention weight adjustment function is expressed as:

[0138] ;

[0139] wherein, is a weight adjustment coefficient, dimensionless, the value range is 0 to 1; is a component type factor, dimensionless, 0.8 for beam type components, 0.9 for column type components, and 0.7 for plate type components; is a load characteristic factor, dimensionless, 0.6 for static load, 0.8 for dynamic load, and 1.0 for alternating load; is a detection accuracy factor, dimensionless, 1.0 for A-level accuracy, 0.8 for B-level accuracy, and 0.6 for C-level accuracy; is a historical data quality factor, dimensionless, the value range is 0.6 to 1.0; is the weight coefficient of each factor, dimensionless, respectively 0.3, 0.3, 0.2, and 0.2.

[0140] According to the weight adjustment coefficient, different adjustment functions are adopted, and the linear adjustment function is expressed as:

[0141] when ;

[0142] The exponential adjustment function is expressed as:

[0143] When ;

[0144] The logarithmic adjustment function is expressed as:

[0145] When ;

[0146] In the formula, is the adjusted attention parameter, dimensionless; is the attention parameter before adjustment, dimensionless; is the adjustment coefficient, dimensionless, and the typical values are 0.1, 0.5, and 0.2, respectively.

[0147] It should be noted that the variables involved in the present application are explained in detail as shown in Tables 1 and 2.

[0148] Table 1 Variable explanation table (first part)

[0149]

[0150] Table 2 Variable explanation table (second part)

[0151]

[0152] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for detecting the load-carrying capacity of a structural engineering in situ, characterized by, The method comprises the following steps: A multi-axial acceleration sensor is installed on the surface of the structure member to be detected, the arrangement density of the multi-axial acceleration sensor is 4 per square meter, the sampling frequency of the multi-axial acceleration sensor is 1000 Hz, a strain gauge array is installed at a key part of the member to be detected, and a digital electronic caliper is used to measure the cross-sectional size of the member; an incremental load is applied to the structure member to be detected by an intelligent loading device, the incremental load comprises static loading and dynamic loading, the amplitude of the load is gradually increased from 10% of the design load to 120%, and a digital electronic rebound instrument is used to detect the surface hardness of the member; vibration response data and strain response data of the structure member to be detected under different loads are collected in real time by a mobile data acquisition terminal, and the mobile data acquisition terminal sends the vibration response data and the strain response data to a field data processing system through a wireless data transmission module; The vibration response data is processed based on a frequency domain analysis method, modal parameters of the structure member to be detected are extracted, the modal parameters comprise natural frequency, mode shape and damping ratio, and a load-modal parameter relationship curve is established; the residual carrying capacity of the structure member to be detected is analyzed and calculated by using an elastic-plastic damage constitutive equation of a material, and a residual carrying capacity evaluation result of the member is output; after the residual carrying capacity evaluation result of the member is output, a step of optimizing a load transfer path analysis by using a shortest path algorithm in graph theory and predicting subsequent carrying performance of the structure member to be detected through a structure response prediction model is further included; the structure of the structure response prediction model is a time series prediction network based on a Transformer architecture, the structure response prediction model comprises two main parts of an encoder and a decoder, the encoder is used to extract feature representation of historical detection data, and the decoder is used to generate a prediction result of future carrying performance, and attention mechanism parameters in the structure response prediction model are dynamically adjusted according to a member type, a load feature and a detection accuracy requirement, and specifically, an attention weight adjustment function is calculated, and a specific formula is as follows: ； In the formula, is a weight adjustment coefficient, dimensionless, with a value range of 0 to 1; is a component type factor, dimensionless, with a value of 0.8 for a beam type component, 0.9 for a column type component, and 0.7 for a plate type component; is a load characteristic factor, dimensionless, with a value of 0.6 for static load, 0.8 for dynamic load, and 1.0 for alternating load; is a detection accuracy factor, dimensionless, with a value of 1.0 for A-level accuracy, 0.8 for B-level accuracy, and 0.6 for C-level accuracy; is a historical data quality factor, dimensionless, with a value range of 0.6 to 1.0; is a weight coefficient of each factor, dimensionless, with a value of 0.3, 0.3, 0.2, and 0.2, respectively; Different adjustment functions are used according to the weight adjustment coefficient, a linear adjustment function is represented as follows: when ; An exponential adjustment function is represented as follows: when ; A logarithmic adjustment function is represented as follows: when ; In the formula, is the adjusted attention parameter, is the attention parameter before adjustment, is the adjustment coefficient, and the values are 0.1, 0.5, and 0.2, respectively. The load transfer path analysis optimized by using the shortest path algorithm in graph theory specifically comprises the following steps: a load transfer path optimization model is established by using a Dijkstra algorithm in graph theory, the structure member to be detected is discretized into a network structure of nodes and edges, the nodes represent key cross-sectional positions of the member, and the edges represent load transfer paths between adjacent cross sections; node weights are determined according to cross-sectional geometric characteristics and material performance, and edge weights are calculated according to distances between two nodes and stiffness distribution; the shortest paths reaching each node are searched step by step from a load application point, and a path length is represented by using a stress level weighted distance, and the greater the stress, the higher the weight; Key node identification is performed by using a centrality analysis method, betweenness centrality and closeness centrality of each node are calculated, and a node with a high centrality value is a key control point of load transfer.

2. The method of field testing the load bearing capacity of structural engineering according to claim 1, wherein, The measurement accuracy of the digital electronic caliper is ±0.01 mm, the measurement range is 0-150 mm, and the digital electronic caliper is provided with a Bluetooth data transmission function and is used for accurately measuring length, width and thickness geometric size parameters of the member.

3. The method of field testing the load bearing capacity of structural engineering according to claim 2, wherein, The intelligent loading device is an integrated detection platform of an integrated hydraulic loading system, a load control system and a data acquisition system, the intelligent loading device is equipped with high-precision load sensors and displacement sensors, the load sensor has an accuracy of ±0.5%, and the displacement sensor has an accuracy of ±0.005 mm.

4. The method of field testing the load bearing capacity of structural engineering according to claim 3, wherein, The impact energy of the digital electronic rebound hammer is 2.207 J, which is used for detecting the rebound value of the surface material of the component, the rebound value is converted into a surface hardness value through the built-in data processing chip, and the surface hardness value is used for evaluating the strength characteristics and uniformity of the component material.

5. The method of field testing the load bearing capacity of a structural engineering according to claim 4, wherein, The mobile data acquisition terminal has precise positioning function, Bluetooth communication function, information display function and Internet communication function, the precise positioning function obtains the spatial coordinates of the mobile data acquisition terminal through a satellite positioning system, the positioning accuracy is centimeter level, and the spatial coordinates are used for determining the on-site position of the structure component to be detected.

6. The method of field testing the load bearing capacity of a structural engineering according to claim 5, wherein, The load-modal parameter relationship curve is a mathematical curve for describing the change relationship between the load size and the natural frequency, the mode shape and the damping ratio.

7. The method of field testing the load bearing capacity of a structural engineering according to claim 6, wherein, The material elastic-plastic damage constitutive equation is used for describing the nonlinear behavior and damage evolution process of the structure component under the action of the load, the input of the material elastic-plastic damage constitutive equation includes the component geometric parameters, the material elastic modulus, the load history and the measured strain data, and the output is the component residual bearing capacity evaluation result and the safety factor.

Images