House structure health detection method based on multi-modal data

Abstract

The invention provides a house structure health detection method based on multi-modal data, and belongs to the technical field of house structure health detection, and the method comprises the steps: comprehensively collecting the multi-dimensional monitoring data of vibration, optical fiber strain, acoustic emission, temperature field, deformation and the like of a house structure through an advanced multi-modal sensor array; according to the method, firstly, collected data are preprocessed, material performance testing is carried out, and a material parameter database is established; according to the core technology, a Bayesian network is used for calculating the correlation degree of multi-modal monitoring data and performance test data, an accurate relation function is constructed, and a house structure health state equation set is established. Through deep fitting of historical monitoring data, a highly intelligent structure health assessment model is developed, and key indexes such as damage positioning, temperature abnormity, deformation degree and accumulated damage can be accurately calculated. The technical problem that it is difficult to comprehensively and accurately evaluate the overall health state of the structure through an existing single monitoring means is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of building structure health detection, and in particular, relates to a building structure health detection method based on multimodal data. Background Art

[0002] At present, the monitoring and evaluation technology of building structure health status has been widely used in the field of construction engineering. These technologies can provide a scientific basis for the safe operation of building structures and are of great significance for preventing and controlling major safety accidents.

[0003] Commonly used methods for monitoring the health of building structures mainly include vibration monitoring, optical fiber strain monitoring, acoustic emission monitoring, temperature field monitoring and deformation monitoring. Among them, vibration monitoring can obtain the dynamic characteristics of the structure such as vibration frequency, vibration mode and damping ratio, reflecting the changes in the overall performance of the structure. Optical fiber strain monitoring can accurately measure the strain distribution of key components, providing a basis for judging the bearing capacity of the structure. Acoustic emission monitoring can capture the acoustic emission signals released during the expansion of microcracks, which can be used to locate the damaged area. Temperature field monitoring can reflect the abnormal temperature distribution on the surface and inside of the structure, providing clues for analyzing the local failure mechanism. Deformation monitoring can comprehensively reflect the overall deformation trend of the structure such as displacement, tilt and settlement, and provide a reference for structural reinforcement. Although the above single monitoring methods have their own characteristics, it is difficult to comprehensively and accurately evaluate the overall health status of the structure. Summary of the invention

[0004] In view of this, the present invention provides a building structure health detection method based on multimodal data, which can solve the technical problem that the existing single monitoring means is difficult to comprehensively and accurately evaluate the overall health status of the structure.

[0005] The present invention is achieved in that:

[0006] The present invention provides a building structure health detection method based on multimodal data, comprising the following steps:

[0007] S10, deploying a multimodal sensor array to collect multimodal monitoring data of the building structure, wherein the multimodal monitoring data includes vibration monitoring data, optical fiber strain monitoring data, acoustic emission signal data, temperature field monitoring data, and deformation monitoring data;

[0008] S20, preprocessing the multimodal monitoring data;

[0009] S30, perform material performance test, establish a housing structure material parameter database, and obtain housing structure performance test data;

[0010] S40, using a Bayesian network to calculate the correlation between the multimodal monitoring data and the performance test data to obtain a multimodal data correlation matrix;

[0011] S50, selecting a corresponding function relationship from a preset relationship function combination to establish a group of equations for a healthy state of a building structure;

[0012] S60, fitting the building structure health state equation group using historical monitoring data to obtain a fitted structural health assessment model;

[0013] S70, using the fitted structural health assessment model to calculate damage location identification data, temperature anomaly index data, deformation degree index data, and cumulative damage index data;

[0014] S80. Generate a building structure health status assessment report based on the damage location identification data, the temperature anomaly index data, the deformation degree index data, and the cumulative damage index data.

[0015] On the basis of the above technical solution, the housing structure health detection method based on multimodal data of the present invention can also be improved as follows:

[0016] The vibration monitoring data includes vibration frequency, vibration mode data, and damping ratio data.

[0017] Furthermore, the optical fiber strain monitoring data includes optical fiber strain data and sensor temperature data.

[0018] Furthermore, the acoustic emission signal data includes acoustic emission amplitude data, acoustic emission frequency data, acoustic emission duration data, and acoustic emission energy data.

[0019] Furthermore, the temperature field monitoring data includes surface temperature data and ambient temperature data.

[0020] Furthermore, the deformation monitoring data includes displacement data, inclination angle data, and settlement data.

[0021] Furthermore, the preprocessing of the multimodal monitoring data specifically includes:

[0022] Processing the vibration monitoring data, including performing noise reduction processing on the vibration frequency, the vibration shape data, and the damping ratio data;

[0023] Processing the optical fiber strain monitoring data, including performing temperature compensation on the optical fiber strain data using the sensor temperature data to obtain corrected optical fiber strain data;

[0024] Processing the acoustic emission signal data, including extracting waveform features from the acoustic emission amplitude data, the acoustic emission frequency data, the acoustic emission duration data, and the acoustic emission energy data;

[0025] Processing the temperature field monitoring data, including performing temperature correction on the surface temperature data using the ambient temperature data;

[0026] The deformation monitoring data is processed, including benchmarking the displacement data, the inclination angle data, and the settlement data.

[0027] Furthermore, the building structure performance test data includes structural stiffness data, material thermal conductivity data, surface emissivity data, stress intensity factor data, stress cycle number data, and damage threshold data.

[0028] Furthermore, the relationship function combination includes at least a linear relationship, an exponential relationship, a logarithmic relationship, and a power function relationship.

[0029] Furthermore, the group of healthy state equations of the building structure includes a vibration response equation, a strain distribution equation, an acoustic emission characteristic equation, a temperature field distribution equation, a crack extension equation, a structural deformation equation, and a damage accumulation equation.

[0030] The vibration response equation input includes the vibration frequency, the mode shape data, and the damping ratio data, and the output is the structural modal parameters;

[0031] The strain distribution equation input includes the corrected optical fiber strain data and the structural stiffness data, and the output is structural strain distribution data;

[0032] The acoustic emission characteristic equation input includes the acoustic emission amplitude data, the acoustic emission frequency data, the acoustic emission duration data, and the acoustic emission energy data, and the output is damage location identification data;

[0033] The temperature field distribution equation input includes the surface temperature data, the ambient temperature data, the material thermal conductivity data, and the surface emissivity data, and the output is temperature anomaly index data;

[0034] The crack extension equation input includes the structural strain distribution data and the stress intensity factor data, and the output is the crack extension rate data;

[0035] The structural deformation equation input includes the displacement data, the inclination angle data, and the settlement data, and the output is deformation degree index data;

[0036] The damage accumulation equation input includes the stress cycle number data, the damage threshold data, and the crack growth rate data, and the output is the cumulative damage index data;

[0037] The step of collecting the multimodal monitoring data includes:

[0038] The vibration monitoring data collection step includes arranging a piezoelectric acceleration sensor array at key nodes of the building structure, collecting the vibration frequency with a sampling frequency of 1000 Hz, calculating the vibration shape data by an experimental modal analysis method, and calculating the damping ratio data by a half-power bandwidth method;

[0039] The optical fiber strain monitoring data acquisition step includes arranging an optical fiber Bragg grating sensor array along the axial direction on the surface of the main force-bearing component of the house structure, demodulating the optical fiber reflection spectrum through wavelength division multiplexing technology to obtain the optical fiber strain data, and arranging a temperature sensor near the optical fiber Bragg grating sensor to collect the sensor temperature data;

[0040] The acoustic emission signal data collection step comprises: arranging a piezoelectric acoustic emission sensor array in an area prone to cracks in the house structure, collecting the acoustic emission amplitude data through a preamplifier, extracting the acoustic emission frequency data through a short-time Fourier transform, recording the start and end time of the acoustic emission signal to obtain the acoustic emission duration data, and integrating the acoustic emission signal to obtain the acoustic emission energy data;

[0041] The temperature field monitoring data collection step includes installing an infrared thermal imager on the outer surface of the house structure to collect the surface temperature data, and arranging a temperature sensor array in the monitoring area to collect the ambient temperature data;

[0042] The deformation monitoring data collection step includes installing a total station on the top of the house structure and arranging prism measuring points on key components to obtain the displacement data, arranging a dual-axis tilt sensor to collect the tilt angle data, and arranging a level to measure the settlement data using a leveling method;

[0043] The housing structure material performance testing steps include:

[0044] The structural stiffness data testing step includes arranging displacement sensors at the loading points of the building structural components, using a static loading method, recording loading stress and deformation data, and calculating the structural stiffness data;

[0045] The material thermal conductivity data testing step includes applying a constant temperature difference at both ends of a housing structure material sample, measuring a steady-state heat flow, and calculating the material thermal conductivity data according to Fourier's thermal conductivity law;

[0046] The surface emissivity data testing step includes using a comparison method at a standard temperature to simultaneously measure the radiation intensity of a building structure material sample and a black body through an infrared thermal imager to obtain the surface emissivity data;

[0047] The stress intensity factor data testing step includes preparing a prefabricated crack on a standard sample of a building structural material, performing a three-point bending test, recording the crack extension length and load data during the loading process, and calculating the stress intensity factor data;

[0048] The stress cycle number data testing step includes performing a fatigue loading test on a sample of a building structure material, recording the number of cyclic loadings of the sample at different stress levels, and obtaining the stress cycle number data;

[0049] The damage threshold data testing step includes cyclically loading a sample of a building structure material, monitoring the crack initiation process by an acoustic emission sensor, recording the stress level when the acoustic emission signal is first detected, and obtaining the damage threshold data.

[0050] Further, the equations or calculations involved in the present invention are described in detail as follows:

[0051] 1. Vibration monitoring data preprocessing:

[0052] The vibration signal denoising process uses the wavelet transform method, which is specifically expressed as follows:

[0053]

[0054] Where, f d (t) is the vibration signal after noise reduction; ψ j,k (t) is the wavelet basis function; φ J,k (t) is the scale function; d j,k is the wavelet coefficient; a J,k is the approximation coefficient; J is the number of decomposition levels, ranging from 3 to 5; k is the translation parameter.

[0055] 2. Fiber optic strain temperature compensation:

[0056] The temperature compensation equation is specifically expressed as follows:

[0057] ε c =ε m +K T (TT 0 );

[0058] In the formula, ε c is the strain value after temperature compensation; ε m is the original strain value measured; K T is the temperature compensation coefficient, usually between 1×10 -6 to 5×10 -6 Between; T is the current temperature; T 0 is the reference temperature.

[0059] 3. Acoustic emission waveform feature extraction:

[0060] The calculation equation of acoustic emission characteristic parameters is specifically expressed as follows:

[0061]

[0062] In the formula, E AE is the acoustic emission energy; V(t) is the acoustic emission signal voltage; t 1 ,t 2 is the start and end time of the signal; f c is the center frequency; X(f) is the Fourier transform of the signal; f is the frequency.

[0063] 4. Temperature field correction:

[0064] The temperature field correction equation is specifically expressed as follows:

[0065] T c =T s +α(T e -T ref )+β;

[0066] Where, T c is the corrected temperature; T s is the surface temperature; T e is the ambient temperature; T ref is the reference ambient temperature; α is the temperature correction coefficient, ranging from 0.8 to 1.2; β is the correction bias term.

[0067] 5. Benchmarking of deformation monitoring data:

[0068] The benchmarking equation is specifically expressed as follows:

[0069]

[0070] In the formula, X norm is the benchmarked data; X is the original data; X min is the minimum value of the data; X max is the maximum value of the data.

[0071] 6. Bayesian network association calculation:

[0072] The correlation calculation equation is specifically expressed as follows:

[0073]

[0074] Where P(H|E) is the posterior probability; P(E|H) is the likelihood probability; P(H) is the prior probability; P(E) is the evidence probability; R ij is the correlation between the i-th and j-th parameters; n is the number of samples.

[0075] 7. Vibration response equation:

[0076] Based on the finite element theory, the vibration response equation is specifically expressed as follows:

[0077]

[0078] Where M is the mass matrix; C is the damping matrix; K is the stiffness matrix; u is the displacement vector; and F(t) is the external excitation force vector.

[0079] Modal parameter extraction equation:

[0080]

[0081] In the formula, φ i is the i-th order mode vibration vector; ω i is the i-th order natural frequency; ξ i is the i-th order damping ratio.

[0082] 8. Strain distribution equation:

[0083] Based on Hooke's law and strain coordination equation, the strain distribution equation is specifically expressed as follows:

[0084] ε total =ε elastic +ε plastic +ε thermal ;

[0085]

[0086] ε plastic =Kσ n ;

[0087] ε thermal =αΔT;

[0088]

[0089] In the formula, ε total is the total strain; ε elastic is the elastic strain; ε plastic is the plastic strain; ε thermal is the thermal strain; σ is the stress; E is the elastic modulus; K is the strength coefficient; n is the strain hardening exponent; α is the thermal expansion coefficient; ΔT is the temperature change; [ε] is the strain tensor.

[0090] 9. Acoustic emission characteristic equation:

[0091] Based on the acoustic emission wave propagation theory, the acoustic emission characteristic equation is specifically expressed as follows:

[0092]

[0093] Where L is the positioning matrix; (x s ,y s ,z s ) is the coordinate of the sound source; (x i ,y i ,z i ) is the coordinate of the i-th sensor; v is the speed of sound wave propagation; t i is the arrival time; t 0 is the occurrence time; DI is the damage index; A i is the amplitude; d i is the propagation distance.

[0094] 10. Temperature field distribution equation:

[0095] Based on Fourier's heat conduction law, the temperature field distribution equation is specifically expressed as follows:

[0096]

[0097] In the formula, ρ is density; c is specific heat capacity; λ is thermal conductivity; q v is the internal heat source; q r is the radiation heat flux density; ε is the emissivity; σ is the Stefan-Boltzmann constant; T s is the surface temperature; T e is the ambient temperature; TAI is the temperature anomaly index; T i is the temperature of the measuring point; T ref is the reference temperature; N is the number of measuring points.

[0098] 11. Crack propagation equation:

[0099] Based on the fracture mechanics theory, the crack propagation equation is specifically expressed as follows:

[0100]

[0101] ΔK=K max -K min ;

[0102] Where a is the crack length; N is the number of cycles; C, m, n are material constants; ΔK is the stress intensity factor amplitude; K max is the maximum stress intensity factor; K IC is the fracture toughness; σ is the stress; F(a / W) is the geometric correction factor; W is the specimen width.

[0103] 12. Structural deformation equation:

[0104] Based on the theory of elastic mechanics, the structural deformation equation is specifically expressed as follows:

[0105]

[0106] Where δ is the deformation vector; K is the stiffness matrix; F is the force vector; M is the moment vector; DI d is the deformation index; w i is the weight coefficient; δ i is the current deformation value; δ i0 is the initial deformation value.

[0107] 13. Damage accumulation equation:

[0108] Based on fatigue damage accumulation theory, the damage accumulation equation is specifically expressed as follows:

[0109]

[0110] Where D is the cumulative damage value; n i is the actual number of cycles; N i is the number of cycles allowed; a c is the critical crack length; σ i is the stress level; C, m are material constants; A, n are crack extension parameters; K I is the stress intensity factor.

[0111] The principle and significance of the above equations are explained as follows:

[0112] 1. The vibration response equation adopts the form of a second-order ordinary differential equation, taking into account the mass, damping and stiffness characteristics of the structure, and obtaining the structural dynamic characteristics through modal analysis;

[0113] 2. The strain distribution equation considers the coupling of elastic, plastic and thermal strains and uses tensor form to describe the three-dimensional strain field distribution;

[0114] 3. The acoustic emission characteristic equation is based on the principle of time difference positioning of sound wave propagation and combines the attenuation law to calculate the damage location index;

[0115] 4. The temperature field distribution equation considers heat conduction and radiation heat transfer, and uses partial differential equations to describe the non-steady-state temperature field;

[0116] 5. The crack growth equation is based on Paris's law, taking into account the effects of stress ratio and fracture toughness to describe fatigue crack growth behavior;

[0117] 6. The structural deformation equation adopts matrix form to establish the relationship between load and deformation and calculate the deformation degree index;

[0118] 7. The damage accumulation equation combines Miner's criterion with fracture mechanics theory to consider the coupling effect of fatigue damage and crack extension.

[0119] These equations use nonlinear relationships such as power functions and exponential functions, mainly based on the following considerations:

[0120] 1. The constitutive relationship of materials has nonlinear characteristics, such as the strain hardening exponent reflects the nonlinearity of plastic deformation; in the prior art, the constitutive relationship of materials refers to a mathematical model that describes the relationship between internal stress and strain of a material when it is subjected to force or external conditions change. This relationship reflects the connection between the inherent characteristics of the material and its response under external loads.

[0121] 2. The damage evolution process has a cumulative effect, and an exponential function is needed to describe the accelerated degradation characteristics;

[0122] 3. Sound wave attenuation and heat conduction have exponential decay characteristics;

[0123] 4. The crack growth rate and stress intensity factor have a power function relationship.

[0124] For various functional relationships, the mathematical expressions of different functional relationships are as follows:

[0125] 1. The basic form of linear relationship is specifically expressed as follows:

[0126] y=ax+b+ε;

[0127] In the formula, y is the dependent variable; x is the independent variable; a is the slope, indicating the rate of change; b is the intercept; ε is the random error term, which follows the normal distribution N(0,σ 2 ).

[0128] 2. The basic form of exponential relationship is specifically expressed as follows:

[0129] y=ae bx +ε;

[0130] y=ae bx +ce dx +ε;

[0131] In the formula, y is the dependent variable; x is the independent variable; a, b, c, d are unknown coefficients; e is the base of the natural logarithm; ε is the random error term.

[0132] 3. The basic form of logarithmic relationship is specifically expressed as follows:

[0133] y = a1n(x) + b + ε;

[0134] y=aln(x)+bln(z)+c+ε;

[0135] In the formula, y is the dependent variable; x and z are independent variables; a, b, c are unknown coefficients; ln is the natural logarithm; ε is the random error term.

[0136] 4. The basic form of the power function relationship is specifically expressed as follows:

[0137] y=ax b+ω;

[0138] y=ax b +cz d +ε;

[0139] In the formula, y is the dependent variable; x and z are independent variables; a, b, c, d are unknown coefficients; and ε is the random error term.

[0140] The coefficient determination methods, including the following two, are general methods:

[0141] 1. Least Squares Method:

[0142]

[0143] In the formula, y i is the measured value; is the fitted value; n is the number of samples.

[0144] 2. Maximum Likelihood Estimation:

[0145]

[0146] Where L(θ) is the likelihood function; θ is the parameter to be estimated; f(x i |θ) is the probability density function.

[0147] The goodness of fit evaluation indicators include the following two, both of which are general methods:

[0148] 1. Coefficient of determination:

[0149]

[0150] In the formula, is the average of the measured values.

[0151] 2. Root mean square error:

[0152]

[0153] Applicable conditions of each functional relationship:

[0154] 1. Linear relationships apply when there is a proportional change between variables;

[0155] 2. Exponential relationships are suitable for describing accelerated growth or decay processes;

[0156] 3. The logarithmic relationship is suitable for describing the process of diminishing marginal effects;

[0157] 4. The power function relationship is suitable for describing scaling laws or fractal characteristics.

[0158] Compared with the prior art, the beneficial effects of a housing structure health detection method based on multimodal data provided by the present invention are as follows: first, a variety of sensor arrays are arranged to realize the comprehensive collection of multiple monitoring indicators such as vibration response, strain distribution, temperature field distribution, acoustic emission characteristics and structural deformation. On this basis, the influence of environmental interference factors on the monitoring data is effectively eliminated through preprocessing means such as wavelet transform, temperature compensation, waveform feature extraction, etc. Then, based on the vibration response equation, strain distribution equation, temperature field distribution equation, crack propagation equation and structural deformation equation, a comprehensive mathematical model reflecting the coupling of multiple physical fields is constructed. Using this model, combined with the Bayesian network analysis of the correlation between monitoring data and structural performance parameters, an evaluation equation group describing the structural health status is established. Finally, comprehensive evaluation indicators such as damage location index, temperature anomaly index, deformation degree index and cumulative damage index are extracted from the evaluation equation group, which comprehensively reflects the damage situation, local failure risk and overall degradation trend of the structure.

[0159] Compared with the prior art, the present invention has the following significant advantages: 1) Through data preprocessing and environmental factor compensation, the reliability and accuracy of the monitoring data are effectively improved, laying a solid foundation for the subsequent health status assessment; 2) The established multi-physical field coupling model more comprehensively and accurately describes the evolution law of the structural health status, providing a scientific basis for fault diagnosis and status prediction; 3) The proposed comprehensive evaluation index system can more comprehensively reflect the damage and deterioration of the structure, providing an important reference for the management department to formulate maintenance strategies, and solving the technical problem that the existing single monitoring method is difficult to comprehensively and accurately assess the overall health status of the structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0160] Figure 1 A flow chart of the method provided by the present invention. DETAILED DESCRIPTION

[0161] like Figure 1 As shown, it is a flow chart of a housing structure health detection method based on multimodal data provided by the present invention. The specific implementation methods of the steps of the present invention are described in detail below:

[0162] The specific implementation method of step S10 is: first, multiple sensor arrays are arranged at key nodes of the house structure to collect multi-modal monitoring data such as vibration, optical fiber strain, acoustic emission, temperature field and deformation of the house structure. Among them, the collection of vibration monitoring data includes: arranging a piezoelectric acceleration sensor array at the key nodes of the house structure, using a sampling frequency of 1000 Hz to obtain vibration signals, then extracting vibration frequency and vibration mode data through experimental modal analysis, and calculating damping ratio data using the half-power bandwidth method. The collection of optical fiber strain monitoring data includes: arranging an optical fiber Bragg grating sensor array along the axial direction on the surface of the main load-bearing component of the house structure, using wavelength division multiplexing technology to demodulate the optical fiber reflection spectrum to obtain optical fiber strain data, and arranging a temperature sensor near the optical fiber Bragg grating sensor to obtain sensor temperature data. The collection of acoustic emission signal data includes: arranging a piezoelectric acoustic emission sensor array in the area prone to cracks in the house structure, collecting acoustic emission amplitude data through a preamplifier, extracting acoustic emission frequency data using short-time Fourier transform, recording the start and end time of the acoustic emission signal to obtain acoustic emission duration data, and integrating the acoustic emission signal to obtain acoustic emission energy data. The collection of temperature field monitoring data includes: installing an infrared thermal imager on the outer surface of the house structure to obtain surface temperature data, and deploying a temperature sensor array in the monitoring area to collect ambient temperature data. The collection of deformation monitoring data includes: installing a total station on the top of the house structure and deploying prism measuring points on key components to obtain displacement data, deploying a dual-axis tilt sensor to collect tilt angle data, and using a level to measure and obtain settlement data. These multimodal monitoring data fully reflect the dynamic response and environmental effects of the house structure, laying the foundation for subsequent health status assessment.

[0163] The specific implementation method of step S20 is: pre-processing the collected multimodal monitoring data to improve the data quality. First, the vibration monitoring data is subjected to noise reduction processing. The wavelet transform method is used to filter out high-frequency noise by selecting appropriate wavelet basis functions and decomposition layers, and the intrinsic vibration characteristics of the structure are retained. The optical fiber strain monitoring data is temperature compensated, and the accurate strain value after temperature compensation is calculated using the sensor temperature data and empirical coefficients. The waveform feature extraction of the acoustic emission signal data includes calculating the acoustic emission amplitude, center frequency, duration and energy parameters to provide a basis for subsequent damage analysis. The temperature field monitoring data is corrected, and the influence of ambient temperature changes on the surface temperature is eliminated using ambient temperature data and empirical correction coefficients. The deformation monitoring data is benchmarked, and the displacement, tilt and settlement data of different measuring points are normalized to the [0,1] interval to provide standardized input for comprehensive evaluation. Through the above pre-processing, environmental interference can be effectively eliminated, the physical response characteristics of the structure itself can be highlighted, and a good foundation can be laid for the next step of health status assessment.

[0164] In the prior art, optical fiber strain monitoring technology is a high-precision measurement method based on the principle of optical fiber sensing. When an optical fiber is subjected to external stress, the optical signal inside it will change, including the propagation speed and phase of light. By measuring these changes, the strain of the object can be accurately calculated. The method for obtaining optical fiber strain monitoring data includes:

[0165] 1. Fiber optic laying: One or more specially designed sensing optical fibers are laid on the monitored structure. These optical fibers can sense the strain changes of the structure.

[0166] 2. Signal injection: Light signals are injected into the sensing fiber through a light source. These signals propagate in the fiber and interact with the surrounding environment.

[0167] 3. Signal acquisition: When optical signals propagate in optical fibers, they are affected by environmental factors such as strain and temperature, causing scattering or frequency shift. Using a suitable receiver can capture these changed optical signals.

[0168] 4. Data processing: The collected optical signals need to be processed and analyzed to extract accurate information about the strain. This usually involves using algorithms to analyze the signal’s intensity, spectrum, and other characteristics.

[0169] The specific implementation method of step S30 is: obtaining performance test data of the house structure, including indicators such as structural stiffness, material thermal conductivity, surface emissivity, stress intensity factor, stress cycle number and damage threshold. Specifically, the structural stiffness test is to arrange displacement sensors at the loading points of the house structural components, adopt a static loading method, record the loading stress and deformation data, and then calculate the structural stiffness. The material thermal conductivity test is to apply a constant temperature difference at both ends of the house structural material sample, measure the steady-state heat flow, and calculate the material thermal conductivity according to Fourier's thermal conductivity law. The surface emissivity test is to use a comparison method at standard temperature, and use an infrared thermal imager to simultaneously measure the house structural material sample and the blackbody radiation intensity to obtain the surface emissivity. The stress intensity factor test is to prepare a prefabricated crack on the standard sample of the house structural material, perform a three-point bending test, record the crack extension length and load data during the loading process, and calculate the stress intensity factor. The stress cycle number test is to perform a fatigue loading test on the house structural material sample, and record the number of cyclic loading of the sample at different stress levels. The damage threshold test is to perform cyclic loading on the sample of the building structure material, monitor the crack initiation process through the acoustic emission sensor, record the stress level when the acoustic emission signal is first detected, and obtain the damage threshold. These structural performance parameters reflect the mechanical properties of the building material and provide an important basis for health status assessment.

[0170] The specific implementation method of step S40 is: using the Bayesian network method to calculate the correlation between the multimodal monitoring data and the house structure performance data, and obtaining the multimodal data correlation matrix. Specifically, firstly, the nodes and edges of the Bayesian network are established, wherein the nodes include various types of monitoring data and material performance parameters, and the edges represent the causal relationship between the data. Then, based on the historical monitoring sample data, the conditional probability distribution between each node is calculated using the maximum likelihood estimation method to obtain the posterior probability. Finally, the correlation coefficient between any two nodes is calculated based on the posterior probability to construct the multimodal data correlation matrix. This correlation analysis method based on probabilistic reasoning can effectively characterize the coupling effect of various factors in a complex system and lay the foundation for establishing a health status assessment model.

[0171] The specific implementation of step S50 is: select the corresponding functional relationship from the preset relationship function combination to establish a group of healthy state equations for the building structure. The preset relationship function combination includes linear relationship, exponential relationship, logarithmic relationship and power function relationship. Linear relationship is suitable for describing the proportional change between variables; exponential relationship is suitable for describing the accelerated growth or attenuation process; logarithmic relationship is suitable for describing the process of diminishing marginal effect; power function relationship is suitable for describing scaling law or fractal characteristics. These functional relationships can reflect the inherent laws between the response parameters of the building structure, such as vibration response, strain distribution, temperature field distribution, crack extension and structural deformation. By selecting appropriate functional relationships, a group of healthy state equations covering multi-physical field coupling is established, laying the foundation for the accuracy of the evaluation model.

[0172] The specific implementation method of step S60 is: use historical monitoring data to perform parameter fitting on the established group of equations for the health status of the building structure to obtain a fitted structural health assessment model. Specifically, the least square method or maximum likelihood estimation method is used to determine the unknown coefficients in each equation so as to minimize the sum of square errors between the fitted value and the measured value. At the same time, goodness of fit indicators such as the determination coefficient R^2 and the root mean square error RMSE are calculated to evaluate the prediction accuracy of the model. When the goodness of fit meets the requirements, an assessment model describing the health status of the building structure can be obtained. This model fitting method based on historical data can make full use of existing information and improve the accuracy and reliability of health status assessment.

[0173] The specific implementation method of step S70 is: using the fitted structural health assessment model, calculate the damage location identification data, temperature anomaly index data, deformation degree index data and cumulative damage index data. Among them, the damage location identification data is based on the acoustic emission characteristic equation, combined with the wave propagation time difference positioning principle and attenuation law, to calculate the sound source coordinates and damage index. The temperature anomaly index data is based on the temperature field distribution equation, considering the heat conduction and radiation heat transfer effects, and calculating the temperature anomaly relative to the reference temperature. The deformation degree index data is based on the structural deformation equation, establishing a matrix relationship between load and deformation, and calculating the comprehensive index of structural deformation. The cumulative damage index data is based on the damage accumulation equation, combined with fatigue damage and crack extension theory, to calculate the cumulative damage degree of the entire structure. These health status indicators comprehensively reflect the damage degree and deterioration trend of the building structure, providing a key basis for subsequent comprehensive evaluation.

[0174] The specific implementation method of step S80 is: generate a house structure health status assessment report based on damage location identification data, temperature anomaly index data, deformation degree index data and cumulative damage index data. First, use the damage location identification data to determine the key areas where damage exists in the house structure, and provide targeted guidance for subsequent inspection and maintenance. Secondly, identify the hot spots based on the temperature anomaly index data and analyze the possible local failure mechanism. Furthermore, combine the deformation degree index data to assess the deformation risk of the overall structure, and provide a basis for the formulation of reinforcement measures. Finally, comprehensively consider the cumulative damage index data, predict the remaining service life of the house structure, and provide decision support for the formulation of maintenance and management plans. The above health status assessment report comprehensively summarizes the damage, deterioration trend and safety hazards of the house structure, and provides a scientific basis for the owner or management department to ensure the safe operation of the house structure.

[0175] In general, the core of this method is to establish a structural health status assessment model under the coupling of multiple physical fields, and realize comprehensive analysis based on multi-source monitoring data. By preprocessing to eliminate environmental interference, using Bayesian network to analyze correlation, and selecting appropriate functional relationships to fit the health state equation group, a number of comprehensive indicators reflecting the degree of structural damage and deterioration trend are finally obtained. These indicators provide a scientific basis for the safety management of building structures and have strong practicality. Compared with the existing technology, the innovation of this method is reflected in three aspects: 1) A mathematical model of multi-physical field coupling is established to more accurately characterize the structural response; 2) A correlation analysis method based on probabilistic reasoning is proposed to enhance the expression ability of the coupling relationship between parameters; 3) A comprehensive evaluation index that fully reflects the health status is generated, which provides effective support for structural maintenance management. In short, this method has strong theoretical and application value and has broad promotion prospects in the field of building structure health monitoring and assessment.

[0176] Specifically, the principle of the present invention is: a method for evaluating the health status of a building structure based on multimodal monitoring data. The key to this method is to establish a mathematical model that reflects the coupling of multiple physical fields, and to use Bayesian network analysis technology to achieve the correlation analysis between monitoring data and structural performance parameters. Finally, a comprehensive evaluation index system that fully reflects the health status of the structure is generated. The principle and implementation logic of this method are as follows:

[0177] 1. Collection and preprocessing of multimodal monitoring data: First, multiple sensor arrays are deployed at key locations of the building structure, including vibration sensors, fiber optic strain sensors, acoustic emission sensors, temperature sensors, and displacement sensors, to achieve comprehensive collection of various physical response indicators of the structure. In order to ensure the quality of the monitoring data, the collected raw data needs to be preprocessed. For vibration monitoring data, the wavelet transform method is used for noise reduction to effectively remove high-frequency interference components; for fiber optic strain data, the sensor temperature information is used for temperature compensation to eliminate the measurement error caused by temperature changes; for acoustic emission signals, characteristic parameters such as amplitude, frequency, duration, and energy are extracted to lay the foundation for subsequent damage location analysis; for temperature field monitoring data, ambient temperature information is used for correction to eliminate the influence of ambient temperature changes; for deformation monitoring data, normalization is performed to unify the deformation amounts of different dimensions to the [0,1] interval. Through the above preprocessing, the quality and reliability of monitoring data can be effectively improved.

[0178] 2. Testing of house structure performance parameters: In addition to multimodal monitoring data, the present invention also needs to obtain mechanical performance parameters of house structure materials, including structural stiffness, thermal conductivity, surface emissivity, stress intensity factor, stress cycle number and damage threshold, etc. These parameters reflect the mechanical properties of house materials and are an important basis for establishing a health status model. Specific testing methods include: static loading test to measure structural stiffness, thermal conduction test to determine material thermal conductivity, infrared radiation contrast method to measure surface emissivity, three-point bending test to determine stress intensity factor, fatigue loading test to obtain stress cycle life, acoustic emission monitoring method to determine damage threshold, etc. The above structural performance parameters will be used as input to the health status assessment model together with the multimodal monitoring data.

[0179] 3. Correlation analysis based on Bayesian network: In order to establish a mathematical model to describe the health status of the building structure, it is necessary to first analyze the correlation between the multimodal monitoring data and the structural performance parameters. The present invention adopts the Bayesian network method to perform this analysis. Specifically, first, a node network containing various types of monitoring data and material performance parameters is established, and the conditional probability distribution between each node is calculated using the maximum likelihood estimation method based on historical sample data. Then, the posterior probability distribution is calculated using the Bayesian formula to obtain the correlation coefficient between any two parameters. This analysis method based on probabilistic reasoning can effectively characterize the coupling relationship between various factors in a complex system and provide a reliable data basis for the establishment of a health status assessment model.

[0180] 4. Establishment of health status assessment model: Based on the above correlation analysis results, the present invention selects a suitable mathematical model from the preset functional relationship combination to establish a set of equations describing the health status of the house structure. These equations cover multiple physical processes such as vibration response, strain distribution, temperature field distribution, crack extension and structural deformation, and can accurately reflect the intrinsic relationship between multimodal monitoring data and structural performance parameters. Specifically including: 1) vibration response equation, describing the natural frequency, vibration mode and damping characteristics of the structure; 2) strain distribution equation, describing the strain state of the structure during load bearing; 3) temperature field distribution equation, describing the heat conduction and radiation process under the action of ambient temperature; 4) crack extension equation, describing the expansion law of fatigue cracks; 5) structural deformation equation, describing the overall deformation characteristics under external loads. By fitting historical monitoring data, the unknown parameters in each equation can be determined, and a comprehensive assessment model describing the health status of the house structure can be established.

[0181] 5. Comprehensive evaluation of structural health status: Based on the above health status evaluation model, the present invention extracts four key comprehensive evaluation indicators, namely: 1) damage location index, using acoustic emission feature analysis to determine the key areas of damage in the structure; 2) temperature anomaly index, combining temperature field distribution to analyze local failure risk; 3) deformation degree index, evaluating the structural safety status based on the overall deformation characteristics; 4) cumulative damage index, combining fatigue damage theory to predict the remaining service life of the structure. These indicators comprehensively reflect the damage status, local failure risk and overall deterioration trend of the building structure, providing an important basis for the management department to formulate maintenance plans.

[0182] A specific embodiment 1 of the present invention is provided below. The specific implementation of each step in this embodiment 1 is described in detail as follows: The specific implementation of step S10 is: first, multiple sensor arrays are arranged at key nodes of the house structure to collect multi-modal monitoring data such as vibration, optical fiber strain, acoustic emission, temperature field and deformation of the house structure. Among them, the collection of vibration monitoring data includes: arranging piezoelectric acceleration sensor arrays at key nodes of the house structure, using a sampling frequency of 1000 Hz to obtain vibration signals, and then calculating the vibration frequency ω by experimental modal analysis. i and vibration mode data φ i , and the damping ratio ξ is calculated using the half-power bandwidth method i The basic form of the vibration response equation is Where M is the mass matrix, C is the dust blocking matrix, K is the stiffness matrix, u is the displacement vector, and F(t) is the external excitation force vector. The modal parameter φ can be extracted through modal analysis. i ,ω i ,ξ i The acquisition of optical fiber strain monitoring data includes: arranging an optical fiber Bragg grating sensor array along the axial direction on the surface of the main load-bearing components of the building structure, and using wavelength division multiplexing technology to demodulate the optical fiber reflection spectrum to obtain the optical fiber strain data ε m ,At the same time, a temperature sensor is placed near the fiber Bragg grating sensor to obtain the sensor temperature data T. The strain distribution equation can be expressed as ε total =ε elastic +ε plastic +ε thermal , where ε elastic =σ / E,ε plastic =Kσ n ,ε thermal =αΔT. The collection of acoustic emission signal data includes: deploying a piezoelectric acoustic emission sensor array in the area prone to cracks in the house structure, and collecting acoustic emission amplitude data A through a preamplifier. i , using short-time Fourier transform to extract acoustic emission frequency data f c , record the start and end time of the acoustic emission signal to obtain the acoustic emission duration data, and perform integration operation on the acoustic emission signal to obtain the acoustic emission energy data E AE The acoustic emission characteristic equation can be expressed as and The collection of temperature field monitoring data includes: installing an infrared thermal imager on the outer surface of the building structure to obtain surface temperature data T s , and at the same time, a temperature sensor array is deployed in the monitoring area to collect ambient temperature data T e The temperature field distribution equation can be expressed as and in The collection of deformation monitoring data includes: installing a total station on the top of the building structure and placing prism measuring points on key components to obtain displacement data δ x ,δ y ,δ z , deploy dual-axis tilt sensors to collect tilt angle data θ x ,θ y ,θ z , the settlement data is obtained by measuring with a level. The structural deformation equation can be expressed as and

[0183] The specific implementation of step S20 is: pre-processing the collected multi-modal monitoring data to improve the data quality. First, the vibration monitoring data is subjected to noise reduction processing, and the wavelet transform method is used to select a suitable wavelet basis function ψ j,k (t) and the number of decomposition levels J, can be expressed as where d j,k is the wavelet coefficient, α J,k is the approximate coefficient, which filters out high-frequency noise and retains the intrinsic vibration characteristics of the structure. Temperature compensation is performed on the optical fiber strain monitoring data, using the sensor temperature data T and the empirical coefficient K T , which can be expressed as ε c =ε m +K T (TT 0 ), and obtain the corrected optical fiber strain data ε c . Extract waveform features of acoustic emission signal data and calculate acoustic emission energy Center frequency Correct the temperature field monitoring data and use the ambient temperature data T e and empirical correction coefficients α, β, can be expressed as T c =T s +α(T e -T ref )+β, and get the corrected surface temperature data T c The deformation monitoring data is benchmarked and can be expressed as The displacement δ of different measuring points x ,δ y ,δ z , tiltθ x ,θ y ,θ z The sedimentation data are normalized to the interval [0,1].

[0184] The specific implementation method of step S30 is: obtaining the performance test data of the house structure. The test method of the structural stiffness data is: arranging displacement sensors at the loading points of the house structure components, using the static loading method, recording the loading stress σ and the deformation data δ, and calculating the structural stiffness K=σ / δ. The test method of the material thermal conductivity data is: applying a constant temperature difference ΔT at both ends of the house structure material sample, measuring the steady-state heat flow q, and calculating the material thermal conductivity according to Fourier's thermal conductivity law. Where L is the thickness of the sample. The test method for surface emissivity data is: using the comparison method at standard temperature, the infrared thermal imager simultaneously measures the radiation intensity I of the building structure material sample and the black body. s and I b , the surface emissivity ε=I can be calculated s / I b The test method of stress intensity factor data is: prepare prefabricated cracks on the standard specimen of building structure material, conduct three-point bending test, record the crack extension length a and load data P during loading, and calculate the stress intensity factor Where σ is stress, W is the specimen width, and F(a / W) is the geometric correction factor. The test method for stress cycle number data is: perform fatigue loading test on the specimen of building structure material, and record the stress of the specimen at different stress levels σ. i The number of cyclic loading times N i , Paris's law can be used The test method for damage threshold data is: cyclically load the sample of building structure material, monitor the crack initiation process through acoustic emission sensor, and record the stress level when the acoustic emission signal is first detected, which is the damage threshold σ th .

[0185] The specific implementation of step S40 is: using the Bayesian network method to calculate the correlation between the multimodal monitoring data and the house structure performance data, and obtaining the multimodal data correlation matrix. The basic idea of ​​the Bayesian network is to use the maximum likelihood estimation method to calculate the conditional probability distribution P(E|H) and the prior probability distribution P(H) between each node based on historical sample data, and then use the Bayesian formula Get the posterior probability distribution. Finally, calculate the correlation coefficient between any two nodes based on the posterior probability Construct the multimodal data association matrix R. This association analysis method based on probabilistic reasoning can effectively characterize the coupling effect of various factors in complex systems.

[0186] The specific implementation of step S50 is: select the corresponding function relationship from the preset relationship function combination to establish the building structure health state equation group. The preset relationship function combination includes the linear relationship y=ax+b+ε, the exponential relationship y=ae bx +ε and y=aebx +ae dx +ε, logarithmic relations y = aln(x) + b + ε and y = aln(x) + bln(z) + c + ε, and power function relations y = ax b +ε and y=ax b +cz d +ε. These functional relationships can reflect the inherent laws between the response parameters of the building structure, such as vibration response, strain distribution, temperature field distribution, crack extension and structural deformation. By selecting appropriate functional relationships, a set of health state equations covering multi-physics field coupling can be established.

[0187] The specific implementation of step S60 is: use historical monitoring data to perform parameter fitting on the established building structure health state equation group to obtain a fitted structural health assessment model. The least squares method can be used for parameter estimation, and the objective function is where y i is the measured value, is the fitting value, and n is the number of samples. At the same time, goodness of fit indicators such as the coefficient of determination are calculated. and RMS error To evaluate the prediction accuracy of the model. When the goodness of fit meets the requirements, an evaluation model describing the health status of the building structure can be obtained.

[0188] The specific implementation of step S70 is: using the fitted structural health assessment model, calculate the damage location identification data, temperature anomaly index data, deformation degree index data and cumulative damage index data. The damage location identification data DI is calculated based on the acoustic emission characteristic equation, combined with the wave propagation time difference positioning principle and attenuation law. The temperature anomaly index data TAI is calculated based on the temperature field distribution equation, taking into account the heat conduction and radiation heat transfer effects. The deformation degree index data DI d It is calculated based on the structural deformation equation and the matrix relationship between load and deformation. The cumulative damage index data D is calculated based on the damage accumulation equation, combined with fatigue damage and crack growth theory. These health status indicators fully reflect the damage degree and deterioration trend of the building structure.

[0189] The specific implementation of step S80 is: based on the various health status indicators calculated above, generate a building structure health status assessment report. First, use the damage location identification data DI to determine the key areas where damage exists in the building structure. Secondly, identify the hot spots based on the temperature anomaly index data TAI and analyze the possible local failure mechanism. Thirdly, combine the deformation degree index data DI dAssess the deformation risk of the overall structure. Finally, consider the cumulative damage index data D comprehensively to predict the remaining service life of the house structure. The above health status assessment report comprehensively summarizes the damage, deterioration trend and safety hazards of the house structure, providing a scientific basis for the owner or management department to ensure the safe operation of the house structure.

[0190] In order to better understand and implement the present invention, the following provides an embodiment 2 of a specific application scenario of the present invention: A high-rise residential building in a certain city was built in 2015, with a total construction area of ​​50,000 square meters, 25 floors above ground and 2 floors underground. The building is located in the city center, with a relatively complex surrounding environment and is affected by many environmental factors such as traffic vibration and temperature changes. In order to ensure the safe operation of the building, the owner decided to use the house structure health detection method based on multimodal data proposed in the present invention to comprehensively evaluate the health status of the building.

[0191] First, a variety of sensor arrays were deployed at key locations in the high-rise residential building, including:

[0192] 1. Vibration monitoring sensor array: A total of 15 piezoelectric acceleration sensors are deployed at the key nodes of the main load-bearing walls and floor slabs of the building, with a sampling frequency of 1000 Hz to obtain dynamic parameters such as vibration frequency, vibration mode and damping ratio.

[0193] 2. Fiber optic strain monitoring sensor array: 20 fiber grating sensors are arranged along the axial direction on the surface of the main load-bearing components of the building, and strain data is obtained through wavelength division multiplexing technology. At the same time, 10 temperature sensors are arranged near the fiber optic sensors to collect the temperature of the sensor body.

[0194] 3. Acoustic emission monitoring sensor array: A total of 12 piezoelectric acoustic emission sensors are deployed in areas of the building where cracks are prone to occur, such as the junction of floor slabs and walls, and the connection nodes of beams and columns, to collect parameters such as the amplitude, frequency, duration and energy of the acoustic emission signals.

[0195] 4. Temperature field monitoring sensor array: 20 infrared thermal imagers are deployed on the exterior wall of the building to obtain surface temperature distribution data. At the same time, 15 temperature sensors are deployed in the monitoring area to collect ambient temperature data.

[0196] 5. Deformation monitoring sensor array: A total station is deployed on the top of the building, and 30 prism measuring points are deployed on the surface of key components to obtain displacement data. In addition, 10 dual-axis inclination sensors and 5 levels are deployed on different floors of the building to collect inclination angle and settlement data.

[0197] The layout of the above multi-modal monitoring sensor array is shown in Table 1:

[0198] Table 1. Layout of multimodal monitoring sensor array

[0199]

[0200] After one year of continuous monitoring, a wealth of multimodal monitoring data has been accumulated. In order to improve the quality and reliability of these original monitoring data, they were preprocessed, including:

[0201] 1. Vibration monitoring data preprocessing:

[0202] The wavelet transform method is used to reduce the noise of the vibration signal. The specific formula is as follows:

[0203]

[0204] Where, f d (t) is the vibration signal after noise reduction; ψ j,k (t) is the wavelet basis function; φ J,k (t) is the scale function; d j,k is the wavelet coefficient; a J,k is the approximate coefficient; J is the number of decomposition layers, ranging from 3 to 5; k is the translation parameter. Through appropriate wavelet basis functions and decomposition layers, high-frequency interference can be effectively filtered out and the intrinsic vibration characteristics of the structure can be retained.

[0205] 2. Temperature compensation of optical fiber strain data:

[0206] The original optical fiber strain data is temperature compensated using the synchronously collected sensor temperature data. The formula is as follows:

[0207] ε c =ε m +K T (TT 0 );

[0208] In the formula, ε c is the strain value after temperature compensation; ε m is the original strain value measured; K T is the temperature compensation coefficient, and its value range is 1×10 -6 to 5×10 -6 ; T is the current temperature; T 0 This can eliminate the influence of temperature changes on strain measurement.

[0209] 3. Acoustic emission signal waveform feature extraction:

[0210] The waveform characteristic parameters of the collected acoustic emission signal are calculated, including the energy E AE , center frequency f c Etc., the formula is as follows:

[0211]

[0212] Where V(t) is the acoustic emission signal voltage; t 1 ,t 2 is the start and end time of the signal; X(f) is the Fourier transform of the signal. These parameters provide a basis for subsequent damage location analysis.

[0213] 4. Temperature field monitoring data correction:

[0214] The surface temperature measurement results are corrected using the ambient temperature data, and the formula is as follows:

[0215] T c =T s +α(T e -T ref )+β;

[0216] Where, T c is the corrected temperature; T s is the surface temperature; T e is the ambient temperature; T ref is the reference ambient temperature; α is the temperature correction coefficient, ranging from 0.8 to 1.2; β is the correction bias term. This can eliminate the influence of ambient temperature changes on surface temperature measurement.

[0217] 5. Benchmarking of deformation monitoring data:

[0218] The deformation monitoring data of different dimensions, such as displacement, tilt and settlement, are uniformly benchmarked, and the formula is as follows:

[0219]

[0220] In the formula, X norm is the benchmarked data; X is the original data; X min is the minimum value of the data; X max is the maximum value of the data. In this way, the deformation of different dimensions can be normalized to the interval [0,1], providing standardized input for comprehensive evaluation.

[0221] After the above preprocessing, high-quality multimodal monitoring data were obtained, laying the foundation for subsequent health status assessment.

[0222] At the same time, the structural material performance parameters of the high-rise residential building were also tested, including:

[0223] 1. Structural stiffness test:

[0224] A total of 10 displacement sensors were placed at the loading points of the main load-bearing components of the building. The static loading method was used to record the loading stress σ and deformation δ data, and the structural stiffness K = σ / δ was calculated. The test results showed that the average structural stiffness of the building was 3.5GPa.

[0225] 2. Material thermal conductivity test:

[0226] A constant temperature difference of 20°C was applied at both ends of the building exterior wall material sample, and the steady-state heat flux density q was measured to be 85W / ㎡. The thermal conductivity of the material was calculated according to Fourier's thermal conductivity law. It is 0.8W / (m·K).

[0227] 3. Surface emissivity test:

[0228] The radiation intensity of building exterior wall material samples and standard black bodies at 25°C was measured using an infrared thermal imager, and the surface emissivity ε was calculated to be 0.85.

[0229] 4. Stress intensity factor test:

[0230] A 30 mm prefabricated crack was prepared on the standard specimen of the main load-bearing component material of the building, and a three-point bending test was performed. The crack extension length a and load data P during the loading process were recorded, and the stress intensity factor was calculated. for

[0231] 5. Stress cycle test:

[0232] Fatigue loading tests were carried out on the material samples of the main load-bearing components of the building, and the number of cyclic loading N of the samples at the stress level of 80MPa was recorded as 5×10^5 times.

[0233] 6. Damage threshold test:

[0234] Cyclic loading tests were carried out on the material samples of the main load-bearing components of the building. The crack initiation process was monitored by acoustic emission sensors. The stress level when the acoustic emission signal was first detected was 60 MPa, which was the damage threshold σ th .

[0235] The test results of the performance parameters of the above structural materials are shown in Table 2:

[0236] Table 2 Test results of structural material performance parameters

[0237]

[0238] With abundant multimodal monitoring data and structural material performance parameters, we need to analyze the correlation between them in order to establish a mathematical model to describe the health status of the building structure. The Bayesian network method is used for analysis here, and the specific steps are as follows:

[0239] 1. First, establish a node network containing various monitoring data and material performance parameters, such as vibration response, strain distribution, temperature field distribution, acoustic emission characteristics, deformation characteristics, structural stiffness, thermal conductivity, emissivity, etc.

[0240] 2. Based on the historical monitoring sample data, the maximum likelihood estimation method is used to calculate the conditional probability distribution P(E|H) and prior probability distribution P(H) between each node.

[0241] 3. Through the Bayesian formula Get the posterior probability distribution.

[0242] 4. Finally, calculate the correlation coefficient between any two nodes based on the posterior probability Constructing a multimodal data association matrix.

[0243] Through the above analysis, the correlation between each monitoring data and the structural performance parameters was obtained, laying the foundation for the subsequent establishment of a health status assessment model.

[0244] Based on the obtained multimodal monitoring data and structural performance parameters, a mathematical model describing the health status of the high-rise residential building was established. Specifically, it includes:

[0245] 1. Vibration response equation:

[0246]

[0247] In the formula, M is the mass matrix; C is the damping matrix; K is the stiffness matrix; u is the displacement vector; F(t) is the external excitation force vector. Through experimental modal analysis, the first five natural frequencies of the building are extracted, which are 0.85Hz,

[0248] 1.35Hz, 2.05Hz, 3.15Hz and 4.25Hz, the corresponding damping ratios are 2.5%, 3.0%, 3.5%, 4.0% and 4.5% respectively.

[0249] 2. Strain distribution equation:

[0250] ε total =ε elastic +ε plastic +ε thermal ;

[0251]

[0252] ε plastic =Kσ 0.3 ;

[0253] ε thermal =1.2×10-5 ΔT;

[0254] In the formula, ε total is the total strain; ε elastic is the elastic strain; ε plastic is the plastic strain; ε thermal is thermal strain; σ is stress; E is elastic modulus of 3.5 GPa; K is strength coefficient of 550 MPa; ΔT is temperature change. According to the optical fiber strain monitoring data, the maximum strain value of the main load-bearing components of the building is calculated to be 650 microstrain.

[0255] 3. Temperature field distribution equation:

[0256]

[0257] Where ρ is the density of 2300 kg / ㎡; c is the specific heat of 840 J / (kg·K); λ is the thermal conductivity of 0.8 W / (m·K); q v is the internal heat source; q r is the radiation heat flux density; ε is the emissivity of 0.85; σ is the Stefan-Boltzmann constant; T s is the surface temperature of the exterior wall; T e is the ambient temperature; TAI is the temperature anomaly index, and the maximum value calculated based on the monitoring data is 0.15.

[0258] 4. Acoustic emission characteristic equation:

[0259]

[0260] Where L is the positioning matrix; (x s ,y s ,z s ) is the coordinate of the sound source; (x i ,y i ,z i ) is the coordinate of the ith sensor; v is the sound wave propagation speed of 4500 m / s; t i is the arrival time; t 0 is the occurrence time; DI is the damage index. According to the acoustic emission monitoring data, the acoustic emission damage index DI of the building in the crack prone area is calculated to be 18.

[0261] 5. Structural deformation equation:

[0262]

[0263] Where δ is the displacement and tilt angle vector; K is the 6×6 stiffness matrix, calculated based on the monitoring data; F and M are the force and torque vectors, respectively; DI dis the deformation degree index, and its maximum value calculated based on the monitoring data is 0.12.

[0264] 6. Crack propagation equation:

[0265]

[0266] ΔK=K max -K min ;

[0267] Where a is the crack length; N is the number of cycles; ΔK is the stress intensity factor amplitude; K max is the maximum stress intensity factor; K IC for is the fracture toughness; σ is stress; F(a / W) is the geometric correction factor. According to the monitoring data and material performance parameters, it is predicted that the crack growth rate of the main load-bearing components of the building is 0.01mm / 10,000 times under the stress level of 80MPa.

[0268] 7. Damage accumulation equation:

[0269]

[0270] Where D is the cumulative damage value; n i is the actual number of cycles; N i is the number of cycles allowed; a c is the critical crack length of 30 mm; i is the stress level; K I is the stress intensity factor. Based on the above parameters, it is predicted that the cumulative damage value of the building will reach the critical value of 0.8 within 5 years under the stress level of 80MPa.

[0271] Combining the above analytical equations, a comprehensive evaluation model describing the health status of the high-rise residential building is formed. Using this model, the following comprehensive evaluation indicators are calculated:

[0272] 1. The damage location index DI = 18, indicating that the building has a certain degree of damage in the crack-prone area.

[0273] 2. Temperature anomaly index TAI = 0.15, indicating that there are serious local temperature anomalies in some areas of the building.

[0274] 3. Deformation index DI d =0.12, indicating that the overall deformation risk of the building is low and in a safe state.

[0275] 4. The cumulative damage index D = 0.6, indicating that the building will reach a critical damage value of 0.8 in the next five years and will require repair and reinforcement.

[0276] It should be noted that the variables involved in the description of the present invention are explained as shown in Table 3 below:

[0277] Table 3 Variable explanation table

[0278]

[0279]

[0280] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A building structure health detection method based on multimodal data, characterized in that: The following steps are involved: S10, deploying a multimodal sensor array to collect multimodal monitoring data of the building structure, wherein the multimodal monitoring data includes vibration monitoring data, optical fiber strain monitoring data, acoustic emission signal data, temperature field monitoring data, and deformation monitoring data; S20, preprocessing the multimodal monitoring data; S30, perform material performance test, establish a housing structure material parameter database, and obtain housing structure performance test data; S40, using a Bayesian network to calculate the correlation between the multimodal monitoring data and the performance test data to obtain a multimodal data correlation matrix; S50, selecting a corresponding function relationship from a preset relationship function combination to establish a group of equations for a healthy state of a building structure; S60, fitting the building structure health state equation group using historical monitoring data to obtain a fitted structural health assessment model; S70, using the fitted structural health assessment model to calculate damage location identification data, temperature anomaly index data, deformation degree index data, and cumulative damage index data; S80. Generate a building structure health status assessment report based on the damage location identification data, the temperature anomaly index data, the deformation degree index data, and the cumulative damage index data.

2. A building structure health detection method based on multimodal data according to claim 1, characterized in that: The vibration monitoring data includes vibration frequency, vibration mode data, and damping ratio data.

3. The building structure health detection method based on multimodal data according to claim 2 is characterized in that: The optical fiber strain monitoring data includes optical fiber strain data and sensor temperature data.

4. The building structure health detection method based on multimodal data according to claim 3 is characterized in that: The acoustic emission signal data includes acoustic emission amplitude data, acoustic emission frequency data, acoustic emission duration data, and acoustic emission energy data.

5. The building structure health detection method based on multimodal data according to claim 4 is characterized in that: The temperature field monitoring data includes surface temperature data and ambient temperature data.

6. A building structure health detection method based on multimodal data according to claim 5, characterized in that: The deformation monitoring data includes displacement data, inclination angle data, and settlement data.

7. A building structure health detection method based on multimodal data according to claim 6, characterized in that: The step of preprocessing the multimodal monitoring data specifically includes: Processing the vibration monitoring data, including performing noise reduction processing on the vibration frequency, the vibration shape data, and the damping ratio data; Processing the optical fiber strain monitoring data, including performing temperature compensation on the optical fiber strain data using the sensor temperature data to obtain corrected optical fiber strain data; Processing the acoustic emission signal data, including extracting waveform features from the acoustic emission amplitude data, the acoustic emission frequency data, the acoustic emission duration data, and the acoustic emission energy data; Processing the temperature field monitoring data, including performing temperature correction on the surface temperature data using the ambient temperature data; The deformation monitoring data is processed, including benchmarking the displacement data, the inclination angle data, and the settlement data.

8. The building structure health detection method based on multimodal data according to claim 7 is characterized in that: The building structure performance test data includes structural stiffness data, material thermal conductivity data, surface emissivity data, stress intensity factor data, stress cycle number data, and damage threshold data.

9. The building structure health detection method based on multimodal data according to claim 8, characterized in that: The relationship function combination includes at least a linear relationship, an exponential relationship, a logarithmic relationship, and a power function relationship.

10. A building structure health detection method based on multimodal data according to claim 9, characterized in that: The building structure health state equation group includes a vibration response equation, a strain distribution equation, an acoustic emission characteristic equation, a temperature field distribution equation, a crack extension equation, a structural deformation equation, and a damage accumulation equation.

Images