CN117574110A - Method, device and medium for quickly identifying global bridge damage - Google Patents

Method, device and medium for quickly identifying global bridge damage Download PDF

Info

Publication number
CN117574110A
CN117574110A CN202311563700.7A CN202311563700A CN117574110A CN 117574110 A CN117574110 A CN 117574110A CN 202311563700 A CN202311563700 A CN 202311563700A CN 117574110 A CN117574110 A CN 117574110A
Authority
CN
China
Prior art keywords
damage
strain
vehicle
neural network
load
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202311563700.7A
Other languages
Chinese (zh)
Other versions
CN117574110B (en
Inventor
张青青
赵银安
朱占元
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sichuan Agricultural University
Original Assignee
Sichuan Agricultural University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sichuan Agricultural University filed Critical Sichuan Agricultural University
Priority to CN202311563700.7A priority Critical patent/CN117574110B/en
Publication of CN117574110A publication Critical patent/CN117574110A/en
Application granted granted Critical
Publication of CN117574110B publication Critical patent/CN117574110B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B21/00—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant
    • G01B21/32—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant for measuring the deformation in a solid
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M5/00—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings
    • G01M5/0008—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings of bridges
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M5/00—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings
    • G01M5/0033—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings by determining damage, crack or wear
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M5/00—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings
    • G01M5/0041—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings by determining deflection or stress
    • G01M5/005—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings by determining deflection or stress by means of external apparatus, e.g. test benches or portable test systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Software Systems (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biomedical Technology (AREA)
  • Algebra (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Databases & Information Systems (AREA)
  • Computational Mathematics (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

The invention discloses a method, a device and a medium for quickly identifying global bridge damage, wherein the method comprises the following steps: determining a preliminary damage area according to the action position of a load through the structural response before and after damage, wherein the load is a load for a calibration vehicle to drive through each lane on a bridge according to a set speed; and establishing a neural network according to the relation between the structural response and the damage information, wherein the neural network takes the structural strain response peak value as an input variable, the damage position information and the damage degree as an output variable, and carrying out final damage identification on the preliminary damage region based on the neural network according to the structural response in the preliminary damage region to acquire the damage position information and the damage degree. The method can accurately and rapidly identify potential bridge damage on the global bridge and provide basis for bridge structure health monitoring.

Description

Method, device and medium for quickly identifying global bridge damage
Technical Field
The invention belongs to the field of rapid identification of global bridge damage, and particularly relates to a method, a device and a medium for rapid identification of global bridge damage.
Background
Highway bridges are the most critical components of traffic infrastructure systems and are subject to static loads, live loads, and other extreme loads (e.g., earthquakes and explosions). Damage and failure of bridges, particularly sudden collapse, can result in significant life and property losses. Since the 21 st century, safety accidents caused by structural damage are frequent at home and abroad, and small personnel and property losses are caused. Therefore, it is necessary to perform health monitoring and inspection of the state of the bridge to maintain public safety.
In recent years, damage recognition methods are various in variety and are characterized, but in general, the accuracy, stability and efficiency of the method are further improved to be suitable for practical bridge engineering. Meanwhile, the existing damage identification method is mostly limited in the number of sensors, and can only be applied to one-dimensional bridge structures, so that the damage identification of the whole-domain bridge cannot be realized. In addition, the existing methods mostly adopt sensors based on electric signals, and the traditional sensing technology has the defects of structural health monitoring (various sensors, a large number of sensors, complex system structure, difficult management and high cost, and the sensor elements and the system have the defects of insufficient long-term performance (precision, durability and reliability).
Disclosure of Invention
The present invention has been made to solve the above-mentioned problems occurring in the prior art. Therefore, a method, a device and a medium for quickly identifying global bridge damage are needed, which can accurately and quickly identify potential bridge damage on a global bridge and provide a basis for bridge structure health monitoring.
According to a first aspect of the present invention, there is provided a method for rapidly identifying global bridge damage, the method comprising:
determining a preliminary damage area according to the action position of a load through the structural response before and after damage, wherein the load is a load for a calibration vehicle to drive through each lane on a bridge according to a set speed;
and establishing a neural network according to the relation between the structural response and the damage information, wherein the neural network takes the structural strain response peak value as an input variable, the damage position information and the damage degree as an output variable, and carrying out final damage identification on the preliminary damage region based on the neural network according to the structural response in the preliminary damage region to acquire the damage position information and the damage degree.
Further, the determining the preliminary damage area according to the action position of the load through the structural response before and after the damage specifically comprises the following steps:
calculating the strain response of an mth unit when a vehicle with n axles passes through the bridge at a constant speed v;
calculating the strain transverse distribution ratio of the corresponding section according to the strain response of the mth unit;
determining a damaged main beam according to the strain transverse distribution ratio;
and determining a preliminary damage area according to the strain response difference value of each unit in the damage girder and the health state under the same load.
Further, the calculating the strain response of the mth unit when the vehicle with n axles passes through the bridge at a constant speed v specifically includes:
when a single-axis moving vehicle load is used as a concentrated force and a concentrated force passes over a bridge at a constant speed v, the position of the moving concentrated force is expressed as x (t) =vt. For x on bridge a The strain influence line at this location is converted into a function that varies with the location of the moving load, expressed as:
wherein ε (t) is the strain time-course function; EI is the cross-sectional bending stiffness in the healthy state; h section neutral axis height;
the strain response epsilon of the mth cell when a vehicle having n axles is passing over the bridge at a constant velocity v T (t) is expressed as:
wherein ε k (t) represents the strain response induced by the kth axis of the vehicle load, D k Represents the distance between the kth axis and the 1 st axis of the vehicle, P k The axle weight of the kth axle of the vehicle is represented, k represents the axle number of the vehicle, and n represents the total axle number of the vehicle.
Further, calculating the strain transverse distribution ratio of the corresponding section according to the strain response of the mth unit specifically comprises the following steps:
when a moving load acts on a certain lane, the strain transverse distribution ratio of the section is calculated according to the structural response acquired by a sensor arranged on the ith monitoring section:
wherein j represents a j-th beam, i represents a monitoring section, (ε) T ) ij The strain of the section i on the j-piece beam is represented by the number of main beams b;
when the vehicle is traveling on the j-piece beam and does not reach the mth cell, the strain time course of the mth cell is expressed as:
for the m-th cell of the j-th beam, the damage degree is expressed by alpha when the cell is damaged, represents the elastic modulus of the j main beams when the m-th unit is damaged, wherein d represents a damaged state, alpha is between 0 and 1, 0 represents no damage to the unit, and 1 represents complete degradation of the rigidity of the unit;
substituting the formula (1.4) into the formula (1.2), and calibrating the strain response of the mth unit of the jth beam under the action of the vehicle load to be expressed as:
wherein ε T (t) d Indicating strain response in the damaged state;
TSR of the transverse section in the damaged state is:
according to formula (1.6), the strain lateral distribution change amount TSRC ij Expressed as:
wherein TSRC ij To monitor the change in the strain transverse distribution of the main beam j on the section i.
Further, the determining the damage main beam according to the strain transverse distribution ratio specifically includes:
assume that all main beams are not damaged TSRC ij Value 0, TSRC ij Girders having a value other than 0 were determined as damaged girders.
Further, determining the preliminary damage area according to the difference between the strain response of each unit in the damage girder and the strain response of the health state under the same load action specifically includes:
the strain response of each cell as the calibrated vehicle passes over the j-th beam is expressed as the difference in strain response from the state of health under the same load:
P＝P 1 +P 2 +…P n (1.9)
wherein,the strain difference value of the damaged units m on the j damaged main beams is the time t corresponding to the time when the equivalent resultant force P of the vehicle acts on the middle position of each unit of the damaged main beams, the magnitude of the equivalent resultant force P of the vehicle is equal to the sum of axle weights of all the axles, and the acting position of the equivalent resultant force P of the vehicle is the center of each axle of the vehicle;
will beThe set area around the cell with the largest value is determined as the preliminary damage area.
Further, the neural network comprises an input layer, a hidden layer and an output layer, wherein the input parameter of the input layer is x β (β=1, 2,., λ), the output layer output parameter is y e (e=1, 2,..eta.) the hidden layer node is u γ (γ=1, 2,.,. I.) the neural network employs a Sigmoid function as the activation function, so the outputs of the hidden layer and the output layer are as follows:
wherein omega βγ ，ω γe Network weights between the input layer-hidden layer and the hidden layer-output layer; θ γ ，θ e Is a network threshold; u (u) β ，y e The output of the hidden layer and the output layer respectively;
training the neural network by using a plurality of samples to obtain a trained neural network, and finally identifying the preliminary damage area by the trained neural network, wherein the samples comprise structural strain response, damage position information and damage degree of the suspected damage area.
Further, the training the neural network by using a plurality of samples to obtain a trained neural network specifically includes:
for λ samples, the total error is described as follows:
wherein formula E λ For sample error, q e For the network to expect output, λ is the number of samples.
The error E is changed by adjusting the weight, the network is trained by adopting a gradient descent method, and the weight and the threshold value are continuously adjusted until the error meets the requirement or reaches the upper limit of training times:
and training the neural network by using a nonlinear least squares optimization method.
According to a second technical scheme of the present invention, there is provided a global bridge damage rapid identification device, the device comprising:
the preliminary damage area determining module is configured to determine a preliminary damage area according to the action position of a load through the structural response before and after damage, wherein the load is a load for calibrating a vehicle to drive through each lane on a bridge according to a set speed;
the final damage identification module is configured to establish a neural network according to the relation between the structural response and the damage information, wherein the neural network takes the structural strain response peak value as an input variable, damage position information and damage degree as an output variable, and performs final damage identification on the preliminary damage area according to the structural response in the preliminary damage area based on the neural network, so as to acquire the damage position information and the damage degree.
According to a third aspect of the present invention, there is provided a readable storage medium storing one or more programs executable by one or more processors to implement the method as described above.
The invention has at least the following beneficial effects:
1) The invention provides a novel method for identifying damage of a global bridge based on the combination of monitoring data acquired by a limited long-gauge FBG sensor and a BP neural network. Based on the limited strain data of the monitoring section before and after the damage, the damage main beam is preliminarily determined according to the strain transverse distribution of the monitoring section. The structural damage identification mathematical model established by the BP neural network can effectively process complex nonlinear mapping relation, automatically extract characteristics from low-level sensor data through the neural network as neural network input vectors, and specifically locate and identify the damage degree of the global bridge damage. The method has the advantages that the method not only can accurately locate and quantify the damage of the global bridge, but also can effectively shorten the calculation time of damage identification.
2) The method can realize the global bridge damage identification based on the strain data collected by the limited sensors, saves the monitoring cost, is beneficial to the application of actual engineering projects, and promotes the development of global bridge health monitoring.
3) The method provided by the invention is used for primarily identifying the damaged main beam based on the strain transverse distribution change value, and then further reducing the damage range to find out the suspected damage unit. Finally accurately identifying the position and damage degree of the damaged unit by combining BP neural network
Drawings
Fig. 1 shows a schematic diagram of the overall structure of a test platform according to an embodiment of the present invention, where (a) is a live view diagram and (b) is a layout diagram;
FIG. 2 is a flow chart of a method for quickly identifying global bridge damage according to an embodiment of the present invention;
FIG. 3 illustrates a schematic view of a strain lateral distribution of a global bridge monitoring section according to an embodiment of the present invention;
fig. 4 shows a schematic structural diagram of a BP neural network according to an embodiment of the present invention;
FIG. 5 shows a schematic diagram of a neural network structure according to an embodiment of the present invention;
FIG. 6 shows a schematic view of a multi-piece simply supported girder bridge according to an embodiment of the present invention, wherein: (a) represents a created finite element method model, (b) represents the geometric feature size of the main beam, and (c) represents a simple beam unit division diagram and a plan view;
FIG. 7 illustrates a schematic view of a primary identified damaged main beam according to an embodiment of the invention, where (a) - (c) represent a change in the lateral strain profile for operating mode 1; (d) - (f) represent operating condition 2 strain transverse profile changes;
FIG. 8 shows a schematic diagram of dynamic load sheet damage approximate location according to an embodiment of the invention, wherein (a) represents condition 1 approximate location and (b) represents condition 2 approximate location;
FIG. 9 illustrates a single lesion localization neural network training schematic according to an embodiment of the present invention;
FIG. 10 illustrates a single lesion localization identification schematic according to an embodiment of the present invention;
FIG. 11 illustrates a single lesion quantitative identification schematic according to an embodiment of the present invention;
fig. 12 illustrates a primary identification of strain lateral distribution changes in a damaged main beam according to an embodiment of the invention: wherein (a) - (c) represent operating mode 1 strain transverse profile changes; (d) - (f) represent operating condition 2 strain transverse profile changes;
FIG. 13 illustrates a multi-lesion approximation positioning schematic according to an embodiment of the present invention;
FIG. 14 illustrates a multi-lesion approximation positioning result according to an embodiment of the present invention;
FIG. 15 illustrates a schematic diagram of multi-impairment localization neural network training according to an embodiment of the present invention;
FIG. 16 illustrates a multi-lesion localization identification schematic according to an embodiment of the present invention;
FIG. 17 shows a schematic diagram of multi-lesion quantitative identification according to an embodiment of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the drawings and detailed description to enable those skilled in the art to better understand the technical scheme of the present invention. Embodiments of the present invention will be described in further detail below with reference to the drawings and specific examples, but not by way of limitation. The order in which the steps are described herein by way of example should not be construed as limiting if there is no necessity for a relationship between each other, and it should be understood by those skilled in the art that the steps may be sequentially modified without disrupting the logic of each other so that the overall process is not realized.
Fig. 1 shows a schematic overall structure of a test platform according to an embodiment of the present invention, as shown in fig. 1, and this embodiment provides a test platform that can be used to obtain relevant data for implementing the method of the present invention, where the entire test platform includes a 5m long acceleration platform, a 2.5m long deceleration platform, and a 2.34m long test section bridge model, as shown in fig. 1. The 5-piece T beam model is made of organic glass, and the elastic modulus of the organic glass is 2795N/mm 2 Density of 1166.14kg/m 3 Poisson's ratio was 0.36. The material properties of the plexiglass are suitable for simulating concrete materials, and the plexiglass also facilitates cutting to help simulate the occurrence of damage.
It should be noted that the test platform as exemplified above is only for better verifying the validity of the global bridge damage rapid identification method based on the limited monitoring data and the BP neural network, and is not a limitation of the present invention, and the present invention may be implemented without depending on the test platform. For example, the manner of acquiring the relevant data may be stress data acquired on the relevant bridge engineering in the field when the corresponding vehicle passes, and the like.
Fig. 2 is a flowchart illustrating a method for quickly identifying global bridge damage according to an embodiment of the present invention. As shown in fig. 2, an embodiment of the present invention provides a method for quickly identifying damage to a global bridge, which includes the following steps S1-S2.
The method starts with step S1, determining a preliminary damage area according to the action position of a load, namely the load which a calibration vehicle drives on each lane of a bridge according to a set speed, through the structural response before and after damage.
In particular, in combination with the limited sensor arrangement in actual monitoring, the damage range is initially defined from a multitude of structural units of the 2D bridge model. A limited sensor distribution is arranged in cross section at different positions along the longitudinal bridge as shown in fig. 2. The basic parameters of the vehicle, such as the number of axles, axle weight and axle distance, are known by a calibration vehicle, and the vehicle is driven through each lane on the bridge according to the appointed speed. And acquiring structural responses before and after damage by using monitoring points distributed and arranged according to the real-time action positions of the load to preliminarily determine the damage range.
The single axis moving vehicle load is reduced to a concentrated axle load. When a concentrated force passes over the bridge at a constant velocity v, the position of the moving concentrated force is denoted as x (t) =vt. For x on bridge a The strain influence line at this location can be converted into a function that varies with the location of the moving load, expressed as:
wherein ε (t) is the strain time-course function; EI is the cross-sectional bending stiffness in the healthy state; h section neutral axis height.
When a vehicle with n axles passes through a bridge at a constant speed v, the strain response epsilon of the mth unit according to the strain influence line theory T (t) can be expressed as:
wherein ε k (t) represents the strain response induced by the kth axis of the vehicle load, D k Represents the distance between the kth axis and the 1 st axis of the vehicle, P k Represents the axle weight of the kth axle of the vehicle load, k represents the axle number of the vehicle, and n is taken as the generationThe total axle number of the vehicle is shown.
For a multi-girder bridge, the girders are cooperatively operated. As shown in fig. 3, the strain response generated by the vehicle load on a lane on the structure will be distributed to each of the beams in proportion to the strain response generated by the beam only under the lane. Here, the strain transverse distribution ratio (Transverse strain ratio, TSR) is introduced here to explain the strain distribution of the monitored cross section on the bridge. The size of the TSR is related to the structural stiffness distribution, which varies with the bridge configuration. If a bridge is damaged at a location, such damage may result in a redistribution of strain across the corresponding cross-section at that location.
For each monitoring section, a sensor is installed under each beam, and when a moving load acts on a certain lane, the structural response acquired by the sensor installed on the ith monitoring section is used for calculating the TSR of the section, which is:
wherein j represents a j-th beam, i represents a monitoring section, (ε) T ) ij The strain at section i on the j-piece beam is the number of main beams b.
For the m-th cell of the j-th beam, the damage degree is expressed by alpha when the cell is damaged, represents the modulus of elasticity of the j main beams when the m-th unit is damaged, wherein d represents the damage state. Alpha is between 0 and 1, 0 means that the cell is not damaged, and 1 means that the cell stiffness is completely degraded. When the vehicle is traveling on the j-piece beam and does not reach the mth cell, the strain time course of the cell can be expressed as follows according to equation (1.1):
as can be seen from equation (1.4), as the damage level increases, the strain of the cell increases.
Substituting the formula (1.4) into the formula (1.2), and calibrating the strain response of the mth unit of the jth beam under the action of the vehicle load to be expressed as:
wherein ε T (t) d Indicating the strain response in the damaged state. From equation (1.5), it is known that when the moving load speed is constant, the cell strain ε is measured under the load of the multiaxial vehicle T (t) d And increases due to the increase in the degree of damage. TSR of the transverse section in the damaged state is:
wherein,the ratio of the strain transverse distribution at the section i on the j-piece beam after damage. According to the formula (1.6), the strain transverse distribution changes with the increase of the damage degree. Strain lateral distribution change (TSRC) ij ) The method comprises the following steps:
wherein TSRC ij To monitor the change in the strain transverse distribution of the main beam j on the section i. TSRC assuming no damage to all main beams ij A value of 0, when there is damage on a girder, TSRC of the girder ij The value will then mutate instead of 0. Thus, the TSRC is used herein ij The value is used for determining the damage main beam, and the damage recognition range is primarily narrowed.
When there is a flaw on a certain girder, the structural rigidity distribution of the girder is different. For the m unit of the j-th main beam where damage occurs, the damaged area is defined herein by the monitoring unit on the main beam. When the calibration vehicle passes through the j-th beam, the strain response difference between the strain response of each unit and the strain response of the health state under the same load can be expressed as:
P＝P 1 +P 2 +…P n (1.9)
wherein,is the strain difference of the damage unit m on the j damaged main beams. The time t is the time corresponding to the action of the equivalent resultant force P of the vehicle on the middle position of each unit of the damaged girder. As shown in the formula (1.9), the magnitude of the vehicle equivalent resultant force P is equal to the sum of the axle weights of the axles, and the action position of the vehicle equivalent resultant force P is the center of each axle of the vehicle. As can be seen from the formula (1.8), the same main beam is provided with +.>The value will increase correspondingly, in this context in terms of +.>The vicinity of the cell with the large value is a suspected damage cell.
And finally, in step S2, a neural network is established according to the relation between the structural response and the damage information, the neural network takes the structural strain response peak value as an input variable, damage position information and damage degree as an output variable, and final damage identification is carried out on the preliminary damage area according to the structural response in the preliminary damage area based on the neural network, so that the damage position information and the damage degree are obtained.
For the inverse problems of damage identification of large structures (such as bridges), the number of units to be monitored is huge, in-situ measurement data is greatly smaller than the degree of freedom of the structure, and parameters involved in damage identification are more, other uncertain factors are added, so that the position and degree of damage are difficult to distinguish by using a neural network once. More output will require a lot of impairment information to optimize the network structure, which will lead to a slow training process and unreliable results. In order to simplify the network structure, it is necessary to reduce the number of inputs and outputs. Therefore, it is proposed to initially identify the damaged girder and determine the suspected damaged unit, and then establish a neural network to accurately identify the damaged position and quantify the damage degree.
The present embodiment uses a BP neural network as the final recognition network.
The BP neural network, also called error back propagation multi-layer feedforward network, has a simple and easy-to-implement structure, can effectively solve the learning problem of the multi-layer network, and keeps connection among all layers. The back-propagation (BP) learning algorithm adjusts the weights by back-propagating the error and forward-passing the data to minimize the error. The performance index of the back propagation algorithm is the mean square error between the calculation target and the network output. The process relies on gradient search techniques to reduce the difference between the actual output of the network and the desired output.
Taking a three-layer BP network as an example, the structure is shown in fig. 4. The input parameter of the network is x β (β=1, 2,., λ), the output parameter is y e (e=1, 2,..eta.) the hidden layer node is u γ (y=1, 2, once again, l). The network uses the Sigmoid function as the activation function, so the outputs of the hidden layer and the output layer are as follows:
wherein omega βγ ，ω γe Network weights between the input layer-hidden layer and the hidden layer-output layer; θ γ ，θ e Is a network threshold; u (u) β ，y e The outputs of the hidden layer and the output layer, respectively.
For λ samples, the total error is described as follows:
wherein formula E λ For sample error, q e For the network to expect output, λ is the number of samples.
The output error of the network is related to the weights of the layers, and the error E is changed by adjusting the weights. And training the network by adopting a gradient descent method, and continuously adjusting the weight and the threshold value until the error meets the requirement or reaches the upper limit of training times.
The present study uses the existing Levenberg-Marquardt nonlinear least squares optimization algorithm in MATLAB software to train BP neural networks. The optimization algorithm has the characteristics of high calculation efficiency and fast convergence, the network structure is shown in fig. 5, the output layer adopts a pure linear transformation function, and corresponding learning rate, training targets and the like are set. Once the neural network is sufficiently trained, it can accurately locate the injury and predict the severity of the injury.
The following examples of the invention will further demonstrate the feasibility and advancement of the invention by further experimentation and verification of the method as described above with reference to the specific embodiments.
In this embodiment, a plurality of simply supported T beams are selected to evaluate the performance and feasibility of the proposed damage detection method. The selected bridge is a multi-piece simply supported T-beam bridge with the span of 45m, the bridge deck width of 13.2m and the single-piece beam width of 264m, beam height 2.75m, 5T-beams, C50 concrete. Elastic modulus e=3.45.10 of concrete 4 MPa, gravity density γ=26 KN/m 3 Poisson's ratio μ=0.2. The numerical simulation analysis establishes a finite element model through a finite element method, the whole structure adopts a beam lattice method to establish the model, a virtual cross beam is arranged every 2.5m along the span direction of the bridge, the whole bridge structure establishes 90 units in total, and the length of each unit is 2.5m. The finite element model and cell division are shown in fig. 6.
Dividing the bridge into 5 equal division points, selecting the equal division point of each bridge piece as a collection point of strain response data, wherein the collection point is arranged at the bottom of the T-beam, and 25 response collection points are arranged in total in the full bridge. In the simulation, 25 fiber grating sensors 2.5 meters long were virtually installed under the main beam where the response acquisition points were located, as shown in fig. 6 (b). And extracting corresponding strain response peak values from each simulation result and normalizing the corresponding strain response peak values to be used as the input of the neural network. The sampling rate selected in the numerical simulation is 200Hz.
Single damage identification working condition:
the travelling load was a two-axle vehicle, with a total weight of 400KN, and was driven on a 3# main beam at 36 km/h. The strain at the moment of moving the load through the middle position of each unit is extracted by using 25 FBG sensors on the monitoring sections, and the transverse distribution of the strain on each monitoring section is analyzed. Assuming 25% and 30% stiffness reduction for units 5 near the 3# main beam quarter span and units 11 near the 3# main beam half span of the multi-piece simply supported T beam, respectively, the damage conditions are shown in Table 1.1.
TABLE 1.1 Unit damage Condition
Working condition number Injury unit Degree of injury (%)
1 3# Main girder E5 25
2 3# Main girder E11 30
Preliminary judgment of damaged areas based on strain transverse distribution:
and analyzing the change condition of the transverse distribution of the strain at the moment when the moving load moves to the middle of the No. 3 main beam units 5, 10 and 14, thereby determining the damaged main beam.
Approximate positioning of single damage working conditions:
as can be seen from fig. 8, the suspected damage units in the working condition 1 are units 5, 6 and 7 of the 3# main beam; the suspected damage units of the working condition 2 are units 10, 11 and 12 of the No. 3 main beam.
Accurate positioning of single damage working conditions:
and establishing a sample set of the BP neural network based on suspected damage units of the working conditions 1 and 2, wherein the sample set established by the working condition 1 comprises the health state and the damage degree of 10%, 15%, 20%, 25%, 30%, 35% and 40% of damage of the units of the 3# main beams 5, 6 and 7 respectively. The sample set established in the working condition 2 comprises the health state and the damage degree of 10%, 15%, 20%, 25%, 30%, 35% and 40% of the damage of the units 10, 11 and 12 of the 3# girder respectively. Sample sets for both conditions are shown in tables 1.2 and 1.3.
TABLE 1.2 working condition 1 damage condition sample set for suspected damage units
TABLE 1.3 working condition 2 damage condition sample set for suspected damage units
Sample numbering Injury condition Sample numbering Injury condition
1 3# cell 10 injury 10% 12 3# cell 11 injury 30%
2 3# cell 10 injury 15% 13 3# cell 11 injury 35%
3 3# Unit 10 injury 20% 14 Damage to 3# Unit 11 40%
4 25% of the damage to 3# cell 10 15 3# Unit 12 injury 10%
5 3# cell 10 injury 30％ 16 3# Unit 12 injury 15%
6 3# cell 10 injury 35% 17 3# Unit 12 injury 20%
7 Damage to # 3 cell 10 40% 18 25% of the damage to 3# Unit 12
8 3# cell 11 injury 10% 19 3# Unit 12 injury 30%
9 3# cell 11 injury 15% 20 3# Unit 12 injury 35%
10 3# cell 11 injury 20% 21 Damage to # 3 cell 12 40%
11 25% of the damage to 3# Unit 11 22 No damage state
The test samples with sample numbers of 5 and 13 are randomly selected in the working condition 1, the test samples with sample numbers of 13 and 18 are randomly selected in the working condition 2, the rest are training samples, the data of the training samples are used for training the BP neural network, the test samples do not participate in the training of the BP neural network, and after the network is trained, the identification capability of the BP neural network is checked by the test samples.
By constructing and training the neural network, the embodiment successfully realizes the machine learning understanding of structural mechanical behaviors, and establishes a complex nonlinear mapping relationship between strain and damage. In this section, test conditions are specifically designed to verify the performance of the neural network. Since a part of data is already contained in the training sample, the test working condition of the section does not use the strain data contained in the previous training data any more, but adopts the new strain data to conduct damage prediction. After the new strain data is processed by the neural network trained herein, the damage index of each test sample is predicted, as shown in fig. 10. The predicted results of these test samples are shown, showing predicted values of dimensionless lesion localization indicators. According to the prediction result, the dimensionless damage positioning index in the graph can accurately identify the real damage position in each group of test working conditions. Although there is a prediction error at some location of the test sample, it is negligible.
And (5) identifying the damage degree of the single damage working condition:
the above damage units are then identified for damage extent:
multiple damage identification working conditions:
according to bridge vulnerability analysis, units near the four points and the two points of the multiple simply supported T beams along the span direction are easy to damage, and the dynamic load multi-damage working condition design considers that two damages occur on different main beam positions and two positions of damages occur on the same main beam. Therefore, the dynamic load multi-damage working condition design is shown in table 1.4.
TABLE 1.4 Multi-injury Unit operating conditions
Working condition number Injury unit Degree of injury
1 3# girder E11 and 4# girder E15 30% and 25%
2 3# main beam E11 and 3# main beam E15 30% and 25%
Preliminary judgment of damaged areas based on strain transverse distribution:
the damage designed in the working condition 1 is on the 3# girder and the damage designed in the working condition 2 is on the 4# girder, the damage is only on the 3# girder, the dynamic load is applied in a mode that the bridge head of the 3# girder runs to the 3# bridge tail at a speed of 36Km/h, the moving load is a two-axle vehicle, and the total weight is 400KN. The strain transverse distribution change of each girder monitoring section when the moving load is loaded in the middle of the 3# unit 5, the 3# unit 10 and the 3# unit 14 is calculated.
As can be seen from fig. 12, in the working condition 1, the strain transverse distribution change amounts of the 3# girder and the 4# girder are the largest, and then the 3# girder and the 4# girder are primarily determined to be damaged. And in the working condition 2, only the middle 3# girder strain transverse distribution change amount is the largest, and the 3# girder preliminary judgment position damages the girder.
As can be seen from fig. 13, the condition 1: when the calibration vehicle runs to the middle of the 11 th unit, the strain of the 3 rd monitoring point of the 3# main beam is greatly suddenly changed, so that the damage can be judged to occur near the 11 th unit of the middle one-piece beam, therefore, the unit 10, the unit 11 and the unit 12 with the damage unit of 3# can be approximately positioned, when the 4# main beam is approximately positioned, when the calibration vehicle runs to the middle of the 4# main beam unit 15, the strain suddenly changed of the 4 th monitoring point of the 4# main beam is maximum, so that the units 14, 15 and 16 of the 4# main beam can be judged to be suspected damage units, but a complete structural body is formed through the action of transverse connection among various main beams of a plurality of simple T beams, the damage results of the adjacent main beams can cause mutual influence, and due to the influence of the damage of the 3# main beam, the 4 th monitoring unit of the 4# main beam has a great positive value suddenly changed, and the position of the suddenly changed point is consistent with the suspected damage position of the 3# main beam. In the working condition 2, the two potential injuries of the middle 3# girder can be judged to be respectively generated near the 11 th unit and near the 14 th unit, the first suspected injury unit is the 10 th, 11 th and 12 th units of the 3# girder, and the second suspected injury unit is the 13 th, 14 th and 15 th units of the 3# girder.
Accurate positioning of multiple damage working conditions:
according to the change amount of strain generated by the damage girder monitoring point under the action of movement, a suspected damage unit is successfully found, and the specific position is shown in fig. 14.
The suspected lesion unit numbers for each condition are shown in fig. 14, where 2 conditions in fig. 14 are taken as an example, and a total of 3×3=9 random combinations with lesion severity of 15%, 25%, 30% are assumed. In the working condition 1, the 46, 47 and 48 units of the 3# main beam and the 68, 69 and 70 units of the 4# main beam are combined in pairs, and a total of 3×3=9 combinations of multiple damage positions are added, so that the working condition 1 has 9×9+1=82 groups of sample data, 2 groups of sample data are randomly extracted from the sample data to serve as test sample data, and the rest 80 groups of sample data are used as training sample data of a neural network. Meanwhile, the 46, 47 and 48 units of the 3# main beam and the 49, 50 and 51 units of the 3# main beam are combined in pairs to form 9 kinds of multi-damage position combinations, a group of 82 groups of sample data are shared by the working condition 2 after the healthy working condition is added as the working condition 1,2 groups of sample data are randomly extracted from the sample data to serve as test sample data, and the rest sample data are training sample data. Test samples for conditions 1 and 2 are shown in table 1.5.
Table 1.5 mobile load multiple injury test specimens
And training the BP neural network through training sample data to obtain a neural network training regression diagram shown in fig. 15.
The test sample is input into the trained neural network, and the multi-damage working condition damage position recognition result is shown in fig. 16.
And (3) identifying the damage degree of the multi-damage working condition:
the damage amount is predicted based on the BP neural network for the identified damage position, and the prediction result is shown in fig. 17.
The above description is intended to be illustrative and not restrictive. For example, the above-described examples (or one or more aspects thereof) may be used in combination with each other. For example, other embodiments may be used by those of ordinary skill in the art upon reading the above description. In addition, in the above detailed description, various features may be grouped together to streamline the invention. This is not to be interpreted as an intention that the features of the claimed invention are essential to any of the claims. Rather, inventive subject matter may lie in less than all features of a particular inventive embodiment. Thus, the following claims are hereby incorporated into the detailed description as examples or embodiments, with each claim standing on its own as a separate embodiment, and it is contemplated that these embodiments may be combined with one another in various combinations or permutations. The scope of the invention should be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims (10)

1. A method for quickly identifying global bridge damage, the method comprising:
determining a preliminary damage area according to the action position of a load through the structural response before and after damage, wherein the load is a load for a calibration vehicle to drive through each lane on a bridge according to a set speed;
and establishing a neural network according to the relation between the structural response and the damage information, wherein the neural network takes the structural strain response peak value as an input variable, the damage position information and the damage degree as an output variable, and carrying out final damage identification on the preliminary damage region based on the neural network according to the structural response in the preliminary damage region to acquire the damage position information and the damage degree.
2. The method according to claim 1, wherein the determining the preliminary damage area by the structural response before and after the damage according to the action position of the load specifically comprises:
calculating the strain response of an mth unit when a vehicle with n axles passes through the bridge at a constant speed v;
calculating the strain transverse distribution ratio of the corresponding section according to the strain response of the mth unit;
determining a damaged main beam according to the strain transverse distribution ratio;
and determining a preliminary damage area according to the strain response difference value of each unit in the damage girder and the health state under the same load.
3. The method according to claim 2, wherein said calculating the strain response of the mth cell of a vehicle having n axles passing from the bridge at a constant velocity v comprises:
when a single-axis moving vehicle load is used as a concentrated force and a concentrated force passes over a bridge at a constant speed v, the position of the moving concentrated force is expressed as x (t) =vt. For x on bridge a The strain influence line at this location is converted into a function that varies with the location of the moving load, expressed as:
wherein ε (t) is the strain time-course function; EI is the cross-sectional bending stiffness in the healthy state; h section neutral axis height;
when there are n axlesThe strain response epsilon of the mth cell as the vehicle passes over the bridge at a constant velocity v T (t) is expressed as:
wherein ε k (t) represents the strain response induced by the kth axis of the vehicle load, D k Represents the distance between the kth axis and the 1 st axis of the vehicle, P k The axle weight of the kth axle of the vehicle is represented, k represents the axle number of the vehicle, and n represents the total axle number of the vehicle.
4. A method according to claim 3, characterized in that the calculation of the strain transverse distribution ratio of the corresponding section from the strain response of the mth cell comprises in particular:
when a moving load acts on a certain lane, the strain transverse distribution ratio of the section is calculated according to the structural response acquired by a sensor arranged on the ith monitoring section:
wherein j represents a j-th beam, i represents a monitoring section, (ε) T ) ij The strain of the section i on the j-piece beam is represented by the number of main beams b;
when the vehicle is traveling on the j-piece beam and does not reach the mth cell, the strain time course of the mth cell is expressed as:
for the m-th cell of the j-th beam, the damage degree is expressed by alpha when the cell is damaged, represents the elastic modulus of the j main beams when the m-th unit is damaged, wherein d represents a damaged state, alpha is between 0 and 1, 0 represents no damage to the unit, and 1 represents complete degradation of the rigidity of the unit;
substituting the formula (1.4) into the formula (1.2), and calibrating the strain response of the mth unit of the jth beam under the action of the vehicle load to be expressed as:
wherein ε T (t) d Indicating strain response in the damaged state;
TSR of the transverse section in the damaged state is:
according to formula (1.6), the strain lateral distribution change amount TSRC ij Expressed as:
wherein TSRC ij To monitor the change in the strain transverse distribution of the main beam j on the section i.
5. The method according to claim 4, wherein determining the damaged main beam according to the strain transverse distribution ratio comprises:
assume that all main beams are not damaged TSRC ij Value 0, TSRC ij Girders having a value other than 0 were determined as damaged girders.
6. The method according to claim 4, wherein the determining the preliminary damage area according to the difference between the strain response of each unit in the damaged main beam and the strain response of the health status under the same load comprises:
the strain response of each cell as the calibrated vehicle passes over the j-th beam is expressed as the difference in strain response from the state of health under the same load:
P＝P 1 +P 2 +…P n (1.9)
wherein,the strain difference value of the damaged units m on the j damaged main beams is the time t corresponding to the time when the equivalent resultant force P of the vehicle acts on the middle position of each unit of the damaged main beams, the magnitude of the equivalent resultant force P of the vehicle is equal to the sum of axle weights of all the axles, and the acting position of the equivalent resultant force P of the vehicle is the center of each axle of the vehicle;
will beThe set area around the cell with the largest value is determined as the preliminary damage area.
7. The method of claim 1, wherein the neural network comprises an input layer, a hidden layer, and an output layer, the input layer having an input parameter x β (β=1, 2,., λ), the output layer output parameter is y e (e=1, 2,..eta.) the hidden layer node is u γ (γ=1, 2,.,. I.) the neural network employs a Sigmoid function as the activation function, so the outputs of the hidden layer and the output layer are as follows:
wherein omega βγ ，ω γe Network weights between the input layer-hidden layer and the hidden layer-output layer; θ γ ，θ e Is a network threshold; u (u) β ，y e The output of the hidden layer and the output layer respectively;
training the neural network by using a plurality of samples to obtain a trained neural network, and finally identifying the preliminary damage area by the trained neural network, wherein the samples comprise structural strain response, damage position information and damage degree of the suspected damage area.
8. The method of claim 7, wherein training the neural network using the plurality of samples results in a trained neural network, comprising:
for λ samples, the total error is described as follows:
wherein formula E λ For sample error, q e For the network to expect output, λ is the number of samples.
The error E is changed by adjusting the weight, the network is trained by adopting a gradient descent method, and the weight and the threshold value are continuously adjusted until the error meets the requirement or reaches the upper limit of training times:
and training the neural network by using a nonlinear least squares optimization method.
9. A method and apparatus for quickly identifying global bridge damage, the apparatus comprising:
the preliminary damage area determining module is configured to determine a preliminary damage area according to the action position of a load through the structural response before and after damage, wherein the load is a load for calibrating a vehicle to drive through each lane on a bridge according to a set speed;
the final damage identification module is configured to establish a neural network according to the relation between the structural response and the damage information, the neural network takes the structural strain response peak value as an input variable, damage position information and damage degree as output variables, and performs final damage identification on the preliminary damage area according to the structural response in the preliminary damage area based on the neural network, so as to acquire the damage position information and the damage degree.
10. A readable storage medium storing one or more programs executable by one or more processors to implement the method of any of claims 1-8.
CN202311563700.7A 2023-11-22 2023-11-22 Method, device and medium for quickly identifying global bridge damage Active CN117574110B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311563700.7A CN117574110B (en) 2023-11-22 2023-11-22 Method, device and medium for quickly identifying global bridge damage

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311563700.7A CN117574110B (en) 2023-11-22 2023-11-22 Method, device and medium for quickly identifying global bridge damage

Publications (2)

Publication Number Publication Date
CN117574110A true CN117574110A (en) 2024-02-20
CN117574110B CN117574110B (en) 2024-07-05

Family

ID=89884010

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311563700.7A Active CN117574110B (en) 2023-11-22 2023-11-22 Method, device and medium for quickly identifying global bridge damage

Country Status (1)

Country Link
CN (1) CN117574110B (en)

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103048102A (en) * 2012-11-30 2013-04-17 江苏省交通科学研究院股份有限公司 Beam bridge state evaluation method
CN103853932A (en) * 2014-03-26 2014-06-11 上海同豪土木工程咨询有限公司 Method for evaluating bending rigidity damage degree of plate girder bridge plate
CN104200005A (en) * 2014-07-28 2014-12-10 东北大学 Bridge damage identification method based on neural network
CN105716534A (en) * 2016-02-22 2016-06-29 东南大学 Tied-arch bridge deflection identification method based on long gauge length fiber grating sensors
CN106092623A (en) * 2016-05-26 2016-11-09 东南大学 A kind of bridge structural damage identification appraisal procedure based on long gauge length stiffness coefficient
WO2017202139A1 (en) * 2016-05-26 2017-11-30 东南大学 Bridge damage identification method based on long-gauge-length strain influence envelope
CN112665695A (en) * 2020-12-16 2021-04-16 四川农业大学 Axle information recognition early warning method and system based on bridge transverse strain
CN114723042A (en) * 2022-03-24 2022-07-08 郑州大学 Bridge damage position and damage degree identification method based on optimized neural network
CN116502496A (en) * 2023-04-27 2023-07-28 湖北工业大学 Simple bridge damage detection method based on mobile load and convolutional neural network
CN116541920A (en) * 2023-03-20 2023-08-04 安徽建筑大学 Bridge finite element model correction method based on BP neural network
CN117029711A (en) * 2023-08-10 2023-11-10 四川农业大学 Full-bridge strain response reconstruction method based on machine vision and optical fiber sensing technology

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103048102A (en) * 2012-11-30 2013-04-17 江苏省交通科学研究院股份有限公司 Beam bridge state evaluation method
CN103853932A (en) * 2014-03-26 2014-06-11 上海同豪土木工程咨询有限公司 Method for evaluating bending rigidity damage degree of plate girder bridge plate
CN104200005A (en) * 2014-07-28 2014-12-10 东北大学 Bridge damage identification method based on neural network
CN105716534A (en) * 2016-02-22 2016-06-29 东南大学 Tied-arch bridge deflection identification method based on long gauge length fiber grating sensors
CN106092623A (en) * 2016-05-26 2016-11-09 东南大学 A kind of bridge structural damage identification appraisal procedure based on long gauge length stiffness coefficient
WO2017202139A1 (en) * 2016-05-26 2017-11-30 东南大学 Bridge damage identification method based on long-gauge-length strain influence envelope
CN112665695A (en) * 2020-12-16 2021-04-16 四川农业大学 Axle information recognition early warning method and system based on bridge transverse strain
CN114723042A (en) * 2022-03-24 2022-07-08 郑州大学 Bridge damage position and damage degree identification method based on optimized neural network
CN116541920A (en) * 2023-03-20 2023-08-04 安徽建筑大学 Bridge finite element model correction method based on BP neural network
CN116502496A (en) * 2023-04-27 2023-07-28 湖北工业大学 Simple bridge damage detection method based on mobile load and convolutional neural network
CN117029711A (en) * 2023-08-10 2023-11-10 四川农业大学 Full-bridge strain response reconstruction method based on machine vision and optical fiber sensing technology

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
雷建平, 陈优敏, 李之达: "大跨度箱形拱桥的横向联系试验与分析", 武汉理工大学学报(交通科学与工程版), no. 03, 30 June 2005 (2005-06-30) *

Also Published As

Publication number Publication date
CN117574110B (en) 2024-07-05

Similar Documents

Publication Publication Date Title
CN108775993B (en) A kind of damage detection for bridges method
Alamdari et al. Damage diagnosis in bridge structures using rotation influence line: Validation on a cable-stayed bridge
Rageh et al. Steel railway bridge fatigue damage detection using numerical models and machine learning: Mitigating influence of modeling uncertainty
Nguyen et al. Damage detection in girder bridges using modal curvatures gapped smoothing method and Convolutional Neural Network: Application to Bo Nghi bridge
CN106092623B (en) A kind of bridge structural damage identification appraisal procedure based on long gauge length stiffness coefficient
CN105973619A (en) Bridge local damage identification method based on influence line under structure health monitoring system
Deng et al. Identification of dynamic vehicular axle loads: Demonstration by a field study
Xu Making good use of structural health monitoring systems of long-span cable-supported bridges
CN112949131B (en) Probability damage positioning vector method for continuous bridge cluster damage diagnosis
CN110728089B (en) Large-span bridge stay cable structure damage diagnosis method based on BOTDA technology
Entezami et al. Improving feature extraction via time series modeling for structural health monitoring based on unsupervised learning methods
Al-Khateeb et al. Structural health monitoring of a cable-stayed bridge using regularly conducted diagnostic load tests
Martinez et al. Bridge flexural rigidity calculation using measured drive-by deflections
CN112326787A (en) Beam bridge identification method based on multipoint rapid static acquisition of exclusive test car
Amanollah et al. Damage detection of structures based on wavelet analysis using improved AlexNet
CN112945490A (en) Method for testing bearing capacity of bridge based on deflection influence line
Sigurdardottir et al. The neutral axis location for structural health monitoring: an overview
CN112945489A (en) Method for testing bridge deflection based on millimeter wave radar
Karimi et al. Damage identification in bridge structures: review of available methods and case studies
Barrile et al. Road infrastructure monitoring: an experimental geomatic integrated system
CN110147595B (en) Method for evaluating transverse cooperative working performance of multi-piece type beam bridge
Ataei et al. Sensor fusion of a railway bridge load test using neural networks
Pamuncak et al. Estimation of structural response using convolutional neural network: Application to the Suramadu bridge
Ardani et al. Evaluating pod-based unsupervised damage identification using controlled damage propagation of out-of-service bridges
CN117574110B (en) Method, device and medium for quickly identifying global bridge damage

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant