CN114444187A - Bridge damage diagnosis method integrating vibration transmission big data and capsule network - Google Patents
Bridge damage diagnosis method integrating vibration transmission big data and capsule network Download PDFInfo
- Publication number
- CN114444187A CN114444187A CN202210108414.0A CN202210108414A CN114444187A CN 114444187 A CN114444187 A CN 114444187A CN 202210108414 A CN202210108414 A CN 202210108414A CN 114444187 A CN114444187 A CN 114444187A
- Authority
- CN
- China
- Prior art keywords
- damage
- layer
- capsule
- transfer rate
- vibration
- 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
Links
- 230000006378 damage Effects 0.000 title claims abstract description 98
- 239000002775 capsule Substances 0.000 title claims abstract description 56
- 230000005540 biological transmission Effects 0.000 title claims abstract description 24
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000003745 diagnosis Methods 0.000 title claims abstract description 23
- 238000012546 transfer Methods 0.000 claims abstract description 56
- 230000004044 response Effects 0.000 claims abstract description 27
- 238000003062 neural network model Methods 0.000 claims abstract description 20
- 239000011159 matrix material Substances 0.000 claims abstract description 18
- 238000011002 quantification Methods 0.000 claims abstract description 18
- 238000000605 extraction Methods 0.000 claims abstract description 15
- 210000002569 neuron Anatomy 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 40
- 238000010586 diagram Methods 0.000 claims description 12
- 238000011176 pooling Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 230000001133 acceleration Effects 0.000 claims description 7
- 230000004807 localization Effects 0.000 claims description 7
- 238000001228 spectrum Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 claims description 4
- 208000027418 Wounds and injury Diseases 0.000 claims description 3
- 230000003213 activating effect Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000006073 displacement reaction Methods 0.000 claims description 3
- 208000014674 injury Diseases 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 3
- 230000007704 transition Effects 0.000 claims description 3
- 239000010410 layer Substances 0.000 description 52
- 238000012544 monitoring process Methods 0.000 description 6
- 229910000831 Steel Inorganic materials 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 4
- 239000010959 steel Substances 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000003902 lesion Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003137 locomotive effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- 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
- G01M5/0066—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings by exciting or detecting vibration or acceleration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Aviation & Aerospace Engineering (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Architecture (AREA)
- Civil Engineering (AREA)
- Structural Engineering (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Medical Informatics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention provides a bridge damage diagnosis method integrating vibration transmission big data and a capsule network, which comprises the following steps: calculating to obtain a vibration transfer rate function according to the vibration response data, establishing a transfer rate function matrix, and converting the transfer rate function matrix into an image form to obtain a transfer rate image; constructing a neural network model for structural damage positioning and quantification based on the capsule network, wherein a feature extraction layer in the capsule network is changed into a dense layer, and a convolution attention module is added after a second convolution layer; and inputting the transmission rate image into the trained neural network model, and carrying out damage positioning and quantification on the structure to obtain a prediction result of the damage position and degree. According to the method, a dense layer and a convolution attention mode are introduced, so that the damaged image is extracted more efficiently, vector neurons are output through a dynamic route of a capsule network layer, attitude information in vibration transmission data information is stored, and finally the damage position and the damage quantity value are predicted through a full connection layer.
Description
Technical Field
The invention relates to the technical field of structural damage diagnosis, in particular to a bridge damage diagnosis method integrating vibration transmission big data and a capsule network.
Background
The bridge structure is complex in structure, various in load types and severe in environment, and inevitably damaged in the long-term service process, so that the structure damage seriously threatens the safe operation of the structure. Therefore, the damage diagnosis of the structure, the early detection of the structural damage and the grasping of the damage degree are important means for ensuring the safe operation of the bridge structure.
At present, a plurality of large bridges are provided with a structural health monitoring system so as to seek to master the running state of the bridge in real time and ensure the safe operation of the bridge. The structural health monitoring system collects and stores massive monitoring data, the massive monitoring data contain rich information representing structural damage, the massive monitoring data are fully utilized, and the conversion from a data warehouse to data productivity is still a challenge.
The key to the diagnosis of structural damage is the determination of the damage location and the quantification of the damage level. The structural damage diagnosis method based on the structural dynamic response signal and the artificial intelligence algorithm is favored because of high-efficiency data processing capability and excellent feature extraction capability, and can realize real-time diagnosis of structural damage. The deep learning intelligent algorithm represented by the convolutional neural network achieves remarkable effect on structural damage diagnosis, but the convolutional neural network loses a lot of important information in the pooling process, and the used scalar neurons cannot extract posture information such as positions, angles and the like when extracting features. However, in actual engineering, the bridge structure is subjected to the common action of complex multi-source loads, similar characteristics of the same type of damage are reduced due to load change and uncertainty, the diagnosis performance of the convolutional neural network is deteriorated due to the defects, and the method is particularly obvious in structural damage information requiring extraction of mass monitoring data. In addition, the traditional time domain signal of the time series or the frequency domain signal based on the fourier transform is used as the data of the damage identification, and high instability exists under the excitation interference, so that the damage identification is not sensitive enough, and the damage characteristics of the structure are difficult to fully characterize. Therefore, a more effective structural damage diagnosis method combining the structural vibration response signal and the artificial intelligence algorithm is needed to realize accurate positioning of structural damage and quantification of damage degree.
Disclosure of Invention
In order to solve the problems, the invention provides an intelligent bridge damage diagnosis method based on fusion of vibration transfer big data and a capsule network, which combines a transfer rate function with the capsule network and integrates various machine vision learning algorithms.
The invention provides the following technical scheme.
A bridge damage diagnosis method integrating vibration transmission big data and a capsule network comprises the following steps:
calculating to obtain a vibration transfer rate function according to vibration response data of a bridge structure to be predicted, establishing a transfer rate function matrix, and converting the transfer rate function matrix into an image form to obtain a transfer rate image;
constructing a neural network model for structural damage positioning and quantification based on the capsule network, wherein a feature extraction layer in the capsule network is changed into a dense layer, and a convolution attention module is added after a second convolution layer;
and inputting the transmission rate image into the trained neural network model, and carrying out damage positioning and quantification on the structure to obtain a prediction result of the damage position and degree.
Preferably, the training of the neural network model for structural lesion localization and quantification includes the following steps:
constructing a bridge structure finite element model for simulating various damage conditions of a bridge, and extracting vibration response data of the structure in various damage states;
according to the vibration response data, calculating and obtaining vibration transfer rate functions under various damage states, establishing a transfer rate function matrix, and converting the transfer rate function matrix into an image form to obtain a transfer rate image;
classifying the transmission rate images, and attaching corresponding damage labels, wherein the label contents comprise damage positions and damage degrees of the structures; constructing an image sample data set according to the transfer rate image;
and determining a loss function, inputting an image sample data set, and performing neural network model training to obtain a trained neural network model.
Preferably, the vibrational response data comprises acceleration and displacement, noted as:
xi={x(tn)},i=1,2,...k,n=1,2,...,N
wherein k is the number of data acquisition points, and N is the length of response data.
Preferably, the vibration transfer rate function is:
yi={y(fm)},m=1,2,...,M
wherein M is a transfer rate data length;
the transfer rate function matrix is:
Yj={y1,y2,...,yi}
wherein j represents the jth injury condition;
the calculation of the vibration transfer rate function includes:
wherein the content of the first and second substances,in a degree of freedom i1A sum of i2The rate of transfer between the responses at (a),is a cross-spectrum of the two or more,is a self-spectrum, x (ω) and x*(ω) is the vibration response x (t), respectivelyn) Its fourier transform and its conjugate.
Preferably, the image form of the transfer rate function matrix conversion is a contour map.
Preferably, the neural network model for locating and quantifying the structural damage comprises a first convolutional layer, a dense layer, a second convolutional layer, an attention module, a capsule layer and a full connection layer; the dense layer comprises a plurality of groups of dense blocks, and the dense blocks are linked by a transition layer; the capsule layers include Primary-Caps layers and second capsule layers.
Preferably, the inputting the transmission rate image into the trained neural network model for performing the lesion localization and quantification of the structure includes the following steps:
inputting a transmissibility image with the size of c multiplied by m multiplied by n, wherein c is the number of finger channels, and m multiplied by n is the image size;
the transmissibility image enters a dense layer after the primary extraction of the characteristics of the first convolution layer, and deep characteristic extraction is carried out;
the network computation of the dense layer may be expressed as:
xl=Hl([x0,x1,...,xl-1])
wherein the H () function consists of BN + ReLU +3 x 3Conv, [ x ]0,x1,...,xl-1]The output characteristic diagrams of the layers from 0 to l-1 are connected, namely merging of channels is carried out;
after deep feature extraction is carried out through the dense layer, the second convolution layer is entered and capsule input is initialized;
adding a convolution attention module after the second convolution, wherein the convolution attention module comprises a channel attention module and a space attention module; respectively utilizing different information by adopting two modes of global average pooling and maximum pooling, summarizing spatial feature information and outputting a spatial feature map;
inputting the spatial characteristic diagram into a Primary-Caps layer in a capsule layer of the capsule network to output a plurality of neurons of a low-layer capsule network, and connecting the internal spatial relationship between a network deep learning coding part of information between capsules and the whole by using a dynamic routing algorithm to realize the updating of the neurons of the high-layer capsule network;
flattening the high-level features of the second capsule layer and transmitting the high-level features to two full-connection layers, activating the two full-connection layers by adopting a ReLU function, and finally outputting all unit damage degree values;
and obtaining the positioning and quantification of the bridge structure damage according to the damage degree value of each unit.
The invention has the beneficial effects that:
(1) according to the invention, the traditional time domain and frequency domain data are replaced by the transmission rate data, a mass transmission rate data set is constructed, and signals are converted into images, so that excellent machine vision learning algorithm can be fully utilized.
(2) According to the invention, vector neurons in the capsule network are used for replacing scalar neurons in the traditional convolutional neural network, so that the inherent spatial relationship and semantic information in the data can be effectively considered, the attitude information contained in the transmission rate signal can be fully utilized, and the accuracy of the network is improved.
(3) The invention improves the characteristic extraction layer of the standard capsule network, adopts the Dense Block (Dense Block) to extract the characteristics, and compared with the standard capsule network which only uses a single-layer convolution kernel to extract the characteristics, the Dense Block can directly connect all the layers to each other, thereby ensuring the maximum information flow among the layers in the network, removing the original pooling operation in the Dense Block, avoiding the information loss and extracting more key information in the data set;
(4) introducing the convolution attention module, the convolution attention module will follow two independent dimensions: and (3) deducing an attention diagram in turn by channels and spaces, and then multiplying the attention diagram with the input feature diagram to perform adaptive feature optimization, thereby improving the extraction of useful features.
Drawings
FIG. 1 is a flow chart of a method of an embodiment of the present invention;
FIG. 2 is a schematic diagram of a neural network framework for structural lesion localization and quantification in accordance with an embodiment of the present invention;
FIG. 3 is a diagram illustrating a single-track through type steel truss bridge model according to an embodiment of the present invention;
FIG. 4 is a raw metrology structure acceleration response signal in accordance with an embodiment of the present invention;
FIG. 5 is a graph of the transfer rate function of all the measurement points when a certain structural unit is damaged according to an embodiment of the present invention;
FIG. 6 is a contour plot of transfer rate function transformation for an embodiment of the present invention;
fig. 7 shows the result of localization and quantification of structural damage by the network model according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example 1
The invention discloses a bridge damage diagnosis method integrating vibration transmission big data and a capsule network, which comprises the following steps as shown in figures 1-7:
s1: constructing a bridge structure finite element model for simulating various damage conditions of a bridge, and extracting vibration response data of the structure in various damage states; the vibrational response data includes acceleration and displacement, noted as: x is the number ofi={x(tn) 1,2,. k, N ═ 1,2,. N,; wherein k is the number of data acquisition points, and N is the length of response data.
S2: according to vibration soundAnd calculating to obtain vibration transfer rate functions under various damage states according to the data, establishing a transfer rate function matrix, converting the transfer rate function matrix into an image form, and obtaining a transfer rate image, wherein the image form converted by the transfer rate function matrix is a contour map. Wherein the vibration transfer rate function is: y isi={y(fm) 1,2, say, M; wherein M is a transfer rate data length; the transfer rate function matrix is: y isj={y1,y2,...,yi}; wherein j represents the jth injury condition; calculation of a vibration transfer rate function comprising:
wherein the content of the first and second substances,in a degree of freedom i1A sum of i2The rate of transfer between the responses at (a),is a cross-spectrum of the two or more,is a self-spectrum, x (ω) and x*(ω) is the vibration response x (t), respectivelyn) Its fourier transform and its conjugate.
S3: classifying the transmission rate images, and attaching corresponding damage labels, wherein the label contents comprise damage positions and damage degrees of the structures; constructing an image sample data set according to the transfer rate image; the label is set according to an example, and specifically comprises the following steps:
Label=[0,0,0,15,0,0,30,0,0,0,0,0]
the label indicates that the 4 th unit damage is 15%, the 7 th unit damage is 30%, and 12 units are involved in damage diagnosis.
S4: determining a loss function, inputting an image sample data set to carry out neural network model training, and obtaining a trained neural network model; the neural network model for positioning and quantifying the structural damage comprises a first convolution layer, a dense layer, a second convolution layer, an attention module, a capsule layer and a full-connection layer; the dense layer comprises a plurality of groups of dense blocks, and the dense blocks are linked by a transition layer; the capsule layers include Primary-Caps layers and second capsule layers.
S5: and converting the vibration response data of the bridge structure to be predicted into a transmissibility image through S2, inputting the transmissibility image into the trained neural network model, and positioning and quantifying the damage of the structure to obtain the prediction result of the damage position and degree.
Specifically, the method comprises the following steps:
s5.1: inputting a transmissibility image with the size of c multiplied by m multiplied by n, wherein c is the number of finger channels, and m multiplied by n is the image size; the transfer rate image enters the Dense layer after the characteristics are preliminarily extracted by the first convolution layer, deep-level characteristic extraction is carried out, the size of the characteristic diagram can be adjusted in the step, and the growth rate in the Dense Dense block is determined. The network computation of the dense layer may be expressed as:
xl=Hl([x0,x1,...,xl-1])
wherein the H () function consists of BN + ReLU + 3Conv, [ x ]0,x1,...,xl-1]The output characteristic diagrams of the layers from 0 to l-1 are connected, namely merging of channels is carried out;
s5.2: and after deep feature extraction is carried out through the dense layer, the second convolution layer is entered and capsule input is initialized.
S5.3: adding a convolution attention module after the second convolution, wherein the convolution attention module comprises a channel attention module and a space attention module; different information is respectively utilized by adopting two modes of global average pooling and maximum pooling:
Mc(F)=σ(MLP(AvgPool(F))+MLP(MaxPool(F)))
Ms(F)=σ(f7*7([AvgPool(F),MaxPool(F)]))
and summarizing the spatial feature information and outputting a spatial feature map.
S5.4: inputting the spatial characteristic diagram into a Primary-Caps layer in a capsule layer of the capsule network to output a plurality of neurons of a low-layer capsule network, and connecting the internal spatial relationship between a network deep learning coding part of information between capsules and the whole by using a dynamic routing algorithm to realize the updating of the neurons of the high-layer capsule network.
S5.5: and flattening the high-level features of the second capsule layer and transmitting the high-level features to two full-connection layers, activating the two full-connection layers by adopting a ReLU function, and finally outputting the damage degree values of all units.
S5.6: and obtaining the positioning and quantification of the damage of the bridge structure according to the damage degree value of each unit, and making a corresponding maintenance decision.
In this embodiment:
in order to verify the effectiveness of the intelligent bridge damage diagnosis method fusing vibration transmission big data and a capsule network, an embodiment is provided for explanation.
In this embodiment, a finite element model is built according to structural parameters of an actual bridge, acceleration response of the bridge structure is selected and extracted as vibration analysis data, and analysis is performed according to a transmission rate image database based on vibration transmission big data of S2 and S3.
Referring to fig. 1, in this embodiment, a single-track through steel truss bridge is used as an analysis object, and the steel truss bridge is composed of two parallel triangular trusses which are 5.75m apart and 11m high, and the span of the steel truss bridge is 64m, as shown in fig. 3. The method comprises the steps of adopting a moving load consisting of 1 DF4D locomotive and 4C 80 trailers to drive at the speed of 100km/h to simulate the vibration load condition of the actual running of a train, extracting acceleration response data in the vertical direction, wherein the sampling frequency is 200Hz, the extracted acceleration response time course curve is shown in figure 4, calculating a transfer rate function according to the transfer rate calculation formula, and is shown in figure 5 and converted into a corresponding high line graph, and is shown in figure 6.
In this embodiment, three data sets, namely a single-rod damage data set, a two-rod damage data set and a three-rod damage data set, are established, the rod damage is simulated in a rigidity reduction mode, 8640 images are obtained in total, corresponding damage labels are set for all the images, the data sets are divided into a training set and a testing set according to the proportion of 0.85:0.15, and the training times are set to be 50 times.
After training is finished, the test set is used for testing the regression effect of the network model, the accuracy of the fitting effect of the network on the test set is high, the average absolute percentage error is within 3%, and the damage degree of the structure can be well predicted. Fig. 7 shows the predicted values of the three damages, and it can be seen from fig. 7 that the proposed intelligent damage diagnosis system for a bridge well predicts the damaged members and the damage degree values of the structure.
Finally, an appropriate maintenance decision is given according to the damage degree of the structure, for example, when the damage degree is larger than a certain limit value, the replacement of the component can be considered.
The present invention is not limited to the above preferred embodiments, and any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (7)
1. A bridge damage diagnosis method integrating vibration transmission big data and a capsule network is characterized by comprising the following steps:
calculating to obtain a vibration transfer rate function according to vibration response data of a bridge structure to be predicted, establishing a transfer rate function matrix, and converting the transfer rate function matrix into an image form to obtain a transfer rate image;
constructing a neural network model for structural damage positioning and quantification based on the capsule network, wherein a feature extraction layer in the capsule network is changed into a dense layer, and a convolution attention module is added after a second convolution layer;
and inputting the transmission rate image into the trained neural network model, and carrying out damage positioning and quantification on the structure to obtain a prediction result of the damage position and degree.
2. The vibration transfer big data and capsule network fused bridge damage diagnosis method according to claim 1, wherein the training of the neural network model for structural damage localization and quantification comprises the following steps:
constructing a bridge structure finite element model for simulating various damage conditions of a bridge, and extracting vibration response data of the structure in various damage states;
according to the vibration response data, calculating and obtaining vibration transfer rate functions under various damage states, establishing a transfer rate function matrix, and converting the transfer rate function matrix into an image form to obtain a transfer rate image;
classifying the transmission rate images, and attaching corresponding damage labels, wherein the label contents comprise damage positions and damage degrees of the structures; constructing an image sample data set according to the transfer rate image;
and determining a loss function, inputting an image sample data set, and performing neural network model training to obtain a trained neural network model.
3. The method for diagnosing the damage of the bridge fused with the vibration transmission big data and the capsule network according to claim 1, wherein the vibration response data comprises acceleration and displacement, and is recorded as:
xi={x(tn)},i=1,2,...k,n=1,2,...,N
wherein k is the number of data acquisition points, and N is the length of response data.
4. The vibration transfer big data and capsule network fused bridge damage diagnosis method according to claim 1, wherein the vibration transfer rate function is:
yi={y(fm)},m=1,2,...,M
wherein M is the length of the transmission rate data;
the transfer rate function matrix is:
Yj={y1,y2,...,yi}
wherein j represents the jth injury condition;
the calculation of the vibration transfer rate function includes:
wherein the content of the first and second substances,in a degree of freedom i1And (ii) is2The rate of transfer between the responses at (a),is a cross-spectrum of the two or more,is a self-spectrum, x (ω) and x*(ω) is the vibration response x (t), respectivelyn) Its fourier transform and its conjugate.
5. The method for diagnosing the damage of the bridge fusing the vibration transfer big data and the capsule network as claimed in claim 1, wherein the image form of the transfer rate function matrix conversion is a contour map.
6. The vibration transfer big data and capsule network fused bridge damage diagnosis method according to claim 1, wherein the neural network model for structure damage localization and quantification comprises a first convolution layer, a dense layer, a second convolution layer, an attention module, a capsule layer and a full connection layer; the dense layer comprises a plurality of groups of dense blocks, and the dense blocks are linked by a transition layer; the capsule layers include Primary-Caps layers and second capsule layers.
7. The vibration transfer big data and capsule network fused bridge damage diagnosis method according to claim 6, wherein the transmission rate image is input into a trained neural network model for structure damage localization and quantification, comprising the following steps:
inputting a transmissibility image with the size of c multiplied by m multiplied by n, wherein c is the number of finger channels, and m multiplied by n is the image size;
the transmissibility image enters a dense layer after the primary extraction of the characteristics of the first convolution layer, and deep characteristic extraction is carried out;
the network computation of the dense layer may be expressed as:
xl=Hl([x0,x1,...,xl-1])
wherein the H () function consists of BN + ReLU +3 x 3Conv, [ x ]0,x1,...,xl-1]The output characteristic diagrams of the layers from 0 to l-1 are connected, namely merging of channels is carried out;
after deep feature extraction is carried out through the dense layer, the second convolution layer is entered and capsule input is initialized;
adding a convolution attention module after the second convolution, wherein the convolution attention module comprises a channel attention module and a space attention module; respectively utilizing different information by adopting two modes of global average pooling and maximum pooling, summarizing spatial feature information and outputting a spatial feature map;
inputting the spatial characteristic diagram into a Primary-Caps layer in a capsule layer of the capsule network to output a plurality of neurons of a low-layer capsule network, and connecting the internal spatial relationship between a network deep learning coding part of information between capsules and the whole by using a dynamic routing algorithm to realize the updating of the neurons of the high-layer capsule network;
flattening the high-level features of the second capsule layer and transmitting the high-level features to two full-connection layers, activating the two full-connection layers by adopting a ReLU function, and finally outputting all unit damage degree values;
and obtaining the positioning and quantification of the bridge structure damage according to the damage degree value of each unit.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210108414.0A CN114444187B (en) | 2022-01-28 | 2022-01-28 | Bridge damage diagnosis method for fusion of vibration transmission big data and capsule network |
GB2218502.9A GB2615402B (en) | 2022-01-28 | 2022-12-08 | Bridge damage diagnosis method based on fusion of vibration transfer big data and capsule network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210108414.0A CN114444187B (en) | 2022-01-28 | 2022-01-28 | Bridge damage diagnosis method for fusion of vibration transmission big data and capsule network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114444187A true CN114444187A (en) | 2022-05-06 |
CN114444187B CN114444187B (en) | 2023-07-18 |
Family
ID=81371107
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210108414.0A Active CN114444187B (en) | 2022-01-28 | 2022-01-28 | Bridge damage diagnosis method for fusion of vibration transmission big data and capsule network |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN114444187B (en) |
GB (1) | GB2615402B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115410004A (en) * | 2022-09-22 | 2022-11-29 | 安吉县科声磁性器材有限公司 | Intelligent iron powder screening system and method |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117454166B (en) * | 2023-10-11 | 2024-05-10 | 国网四川省电力公司电力科学研究院 | Method for identifying arc faults of induced thermal power based on EffNet lightweight model |
CN117953235B (en) * | 2024-03-27 | 2024-06-28 | 中建国际工程有限公司 | Wall damage degree evaluation method and system based on infrared polarization imaging |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108875705A (en) * | 2018-07-12 | 2018-11-23 | 广州麦仑信息科技有限公司 | A kind of vena metacarpea feature extracting method based on Capsule |
CN111461063A (en) * | 2020-04-24 | 2020-07-28 | 武汉大学 | Behavior identification method based on graph convolution and capsule neural network |
CN111595541A (en) * | 2020-04-27 | 2020-08-28 | 河海大学 | Multi-dimensional structure damage identification method based on mass vibration transmissibility data convolutional neural network processing |
CN111612030A (en) * | 2020-03-30 | 2020-09-01 | 华电电力科学研究院有限公司 | Wind turbine generator blade surface fault identification and classification method based on deep learning |
KR20210047832A (en) * | 2020-05-08 | 2021-04-30 | 베이징 바이두 넷컴 사이언스 앤 테크놀로지 코., 엘티디. | Processing method and apparatus of neural network model |
CN113255882A (en) * | 2021-04-30 | 2021-08-13 | 南通大学 | Bearing fault diagnosis method based on improved convolution capsule network |
-
2022
- 2022-01-28 CN CN202210108414.0A patent/CN114444187B/en active Active
- 2022-12-08 GB GB2218502.9A patent/GB2615402B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108875705A (en) * | 2018-07-12 | 2018-11-23 | 广州麦仑信息科技有限公司 | A kind of vena metacarpea feature extracting method based on Capsule |
CN111612030A (en) * | 2020-03-30 | 2020-09-01 | 华电电力科学研究院有限公司 | Wind turbine generator blade surface fault identification and classification method based on deep learning |
CN111461063A (en) * | 2020-04-24 | 2020-07-28 | 武汉大学 | Behavior identification method based on graph convolution and capsule neural network |
CN111595541A (en) * | 2020-04-27 | 2020-08-28 | 河海大学 | Multi-dimensional structure damage identification method based on mass vibration transmissibility data convolutional neural network processing |
KR20210047832A (en) * | 2020-05-08 | 2021-04-30 | 베이징 바이두 넷컴 사이언스 앤 테크놀로지 코., 엘티디. | Processing method and apparatus of neural network model |
CN113255882A (en) * | 2021-04-30 | 2021-08-13 | 南通大学 | Bearing fault diagnosis method based on improved convolution capsule network |
Non-Patent Citations (1)
Title |
---|
周航等: "基于改进胶囊神经网络的示功图诊断模型", 《电子科技大学学报》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115410004A (en) * | 2022-09-22 | 2022-11-29 | 安吉县科声磁性器材有限公司 | Intelligent iron powder screening system and method |
Also Published As
Publication number | Publication date |
---|---|
GB2615402B (en) | 2024-07-03 |
GB202218502D0 (en) | 2023-01-25 |
GB2615402A (en) | 2023-08-09 |
CN114444187B (en) | 2023-07-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114444187A (en) | Bridge damage diagnosis method integrating vibration transmission big data and capsule network | |
CN110704801B (en) | Bridge cluster structure operation safety intelligent monitoring and rapid detection complete method | |
CN111274737A (en) | Method and system for predicting remaining service life of mechanical equipment | |
CN111832814A (en) | Air pollutant concentration prediction method based on graph attention machine mechanism | |
CN111709448A (en) | Mechanical fault diagnosis method based on migration relation network | |
CN112487584A (en) | Dynamics-based rolling bearing digital twin modeling method | |
CN113971660B (en) | Computer vision method for bridge health diagnosis and intelligent camera system | |
CN114548375B (en) | Cable-stayed bridge girder dynamic deflection monitoring method based on two-way long-short-term memory neural network | |
CN116011125B (en) | Response prediction method for uncertain axle coupling system | |
CN112631261A (en) | High-power tractor remote fault diagnosis system and method | |
CN116596151B (en) | Traffic flow prediction method and computing device based on time-space diagram attention | |
Moeinifard et al. | Lost vibration test data recovery using convolutional neural network: a case study | |
Rosso et al. | Review on deep learning in structural health monitoring | |
CN111223087A (en) | Automatic bridge crack detection method based on generation countermeasure network | |
CN111383273B (en) | High-speed rail contact net part positioning method based on improved structure reasoning network | |
CN112329520A (en) | Truck bearing fault identification method based on generation countermeasure learning | |
Giglioni et al. | Deep autoencoders for unsupervised damage detection with application to the Z24 benchmark bridge | |
CN117312948A (en) | Rotary machine fault diagnosis method based on multi-sensor fusion and ECA-CNN | |
Ghazvineh et al. | Application of 2-D convolutional neural networks for damage detection in steel frame structures | |
CN116703819A (en) | Rail wagon steel floor damage detection method based on knowledge distillation | |
CN113570111B (en) | Bridge health state on-chip monitoring method based on lightweight network | |
CN115640531A (en) | Fault diagnosis method based on residual error learning and attention mechanism fusion | |
CN114118140B (en) | Multi-view intelligent fault diagnosis method and system for explosion-proof motor bearing | |
CN113033382B (en) | Method, system and device for identifying large-area damage fault of wagon floor | |
CN117493943A (en) | Cable-stayed bridge cable damage identification method, equipment and storage medium |
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 |