CN115310482A - Radar intelligent identification method for bridge reinforcing steel bar - Google Patents
Radar intelligent identification method for bridge reinforcing steel bar Download PDFInfo
- Publication number
- CN115310482A CN115310482A CN202210919790.8A CN202210919790A CN115310482A CN 115310482 A CN115310482 A CN 115310482A CN 202210919790 A CN202210919790 A CN 202210919790A CN 115310482 A CN115310482 A CN 115310482A
- Authority
- CN
- China
- Prior art keywords
- bridge
- radar
- network
- data
- steel bar
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 229910001294 Reinforcing steel Inorganic materials 0.000 title claims abstract description 19
- 230000000149 penetrating effect Effects 0.000 claims abstract description 43
- 229910000831 Steel Inorganic materials 0.000 claims abstract description 38
- 239000010959 steel Substances 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 25
- 230000004044 response Effects 0.000 claims abstract description 16
- 230000002787 reinforcement Effects 0.000 claims abstract description 15
- 238000013528 artificial neural network Methods 0.000 claims abstract description 9
- 230000000694 effects Effects 0.000 claims abstract description 6
- 239000010410 layer Substances 0.000 claims description 34
- 230000006870 function Effects 0.000 claims description 10
- 238000012360 testing method Methods 0.000 claims description 10
- 238000011176 pooling Methods 0.000 claims description 9
- 230000008034 disappearance Effects 0.000 claims description 7
- 238000009933 burial Methods 0.000 claims description 6
- 239000004567 concrete Substances 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 230000007786 learning performance Effects 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 claims description 3
- 239000011241 protective layer Substances 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000012935 Averaging Methods 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000004364 calculation method Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 12
- 238000001514 detection method Methods 0.000 description 11
- 238000005516 engineering process Methods 0.000 description 9
- 238000010276 construction Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000005672 electromagnetic field Effects 0.000 description 2
- 239000011150 reinforced concrete Substances 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000002547 anomalous effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
Images
Landscapes
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention provides a radar intelligent identification method of bridge reinforcing steel bars, which changes the dielectric constant and the conductivity of a bridge structure and obtains the response characteristics of the bridge structure through forward modeling; collecting a response typical map of the existing bridge steel bar ground penetrating radar, combing response characteristics by a system, and constructing a sample data set; forward calculation is carried out through a time domain finite difference method to obtain a data set of sample data pairs; designing a radar intelligent recognition network of bridge reinforcements for a backbone network based on an M-Net framework of a full convolution neural network, and training and optimizing the radar intelligent recognition network; and (4) inverting the actually measured ground penetrating radar data of the physical model, and verifying whether the position, the dielectric constant and the diameter of the steel bar are consistent with the real condition. The invention at least has the following beneficial effects: the multi-scale input is integrated into the coding module, so that the parameter is prevented from being greatly increased; increasing the network width of the path of the coding module; the generalization of the BIG-inv net network is increased; and multi-scale information can be extracted, and the identification effect of the ground penetrating radar bridge reinforcing steel bar is improved.
Description
Technical Field
The invention relates to a radar intelligent identification method for bridge reinforcing steel bars, in particular to the fields of bridge detection, ground penetrating radar and deep learning.
Background
China is a big bridge country, and bridges are indispensable in modern people going out as important components of traffic. Bridge engineering is generally a reinforced concrete structure, and steel bar displacement and steel bar quality are common problems during concrete structure construction, so that a plurality of potential safety hazards are caused, and even safety accidents occur. Therefore, the nondestructive detection (the position, the diameter and the like of the steel bars) of the bridge reinforced concrete structure has great significance for the normal operation of the bridge building.
The ground penetrating radar technology is a nondestructive detection technology for detecting a medium (such as the ground and concrete) and an object (such as a steel bar, a pipeline and a hole) inside the medium. The ground penetrating radar technology has the advantages of high detection speed, high resolution, flexibility in operation, low detection cost, wide detection range and the like, so that the technology is widely applied to various fields such as municipal engineering, environmental engineering, geological engineering and the like. Over a decade, the rapid development of electronic components and radio technology has enabled the ground penetrating radar hardware system to have the characteristics of high acquisition precision, high speed, large data volume and the like, while the ground penetrating radar data processing technology matched with the hardware system is relatively weak, which hinders the further popularization of the ground penetrating radar technology. The inversion accuracy of the ground penetrating radar data depends on the experience of professionals. In addition, the data volume collected by the ground penetrating radar in the actual engineering is large, and great effort and time are required to process and explain the radar data. Therefore, it is a necessary trend of the steel bar detection technology to improve the identification efficiency and accuracy of the diameter, position and the like of the steel bar.
With the arrival of the big data era, the demands of various industries on data analysis are continuously increased, so that machine learning is rapidly developed. At present, a plurality of machine learning methods are widely applied to solving the geophysical inverse problem and achieve better effect. Deep learning is an important branch of machine learning, and nowadays becomes a core algorithm of artificial intelligence. Which is essentially a complex function with generic universally applicable approximation capabilities. For any non-linear complex function, the network function can approximate it arbitrarily, as long as there are enough neurons. The results obtained in the fields of target detection, natural language processing, medical image segmentation and the like far exceed the prior related technologies.
Based on the actual requirement of intelligent identification of the bridge steel bar, the invention provides the radar intelligent identification method of the bridge steel bar by utilizing the unique advantage of the full convolution neural network in processing the nonlinear problem.
Disclosure of Invention
In view of the problems in the prior art, the application provides a radar intelligent identification method for bridge reinforcements, which comprises the following steps:
s1, obtaining response characteristics of a bridge structure through forward modeling by changing the dielectric constant and the conductivity of the bridge structure;
s2, collecting a response typical map of the existing bridge steel bar ground penetrating radar, combing response characteristics by a system, and constructing a sample data set;
s3, dividing grids for a two-dimensional dielectric model, wherein the two-dimensional dielectric model comprises reinforcing steel bars with different diameters and different burial depths; designing steel bar dielectric models with different diameters, different burial depths and different positions in concrete, and forward calculating by a time domain finite difference method to obtain a data set of sample data pairs; each sample data pair consists of a dielectric model and corresponding ground penetrating radar data;
s4, designing a radar intelligent identification network of the bridge steel bars for the backbone network based on the M-Net architecture of the full convolution neural network, constructing a radar intelligent identification network model frame by using Keras, and adopting tensorflow as a rear end;
s5, training and optimizing the radar intelligent recognition network, randomly dividing a sample database into training data, testing data and verification data, standardizing all input and output values to be in a range of [0,1] in the training process, and verifying after each training;
and S6, inverting the actually measured ground penetrating radar data of the physical model, and verifying whether the position, the dielectric constant and the diameter of the steel bar are consistent with the real condition.
In one embodiment, the dielectric constant and the conductivity of the bridge structure are changed by adjusting the diameter of the steel bars, the thickness of the protective layer and the distance between the steel bars.
In one embodiment, the size of the georadar data in step S3 is the same as the model data by bilinear interpolation.
In one embodiment, the radar intelligent recognition network comprises a ground penetrating radar bridge reinforcing steel bar multi-scale input module, an encoding module, a pyramid convolution module, a decoding module and a multi-scale output module.
In one embodiment, the ground penetrating radar bridge steel bar multi-scale input module is used for constructing data pyramid input and realizing different levels of receptive field fusion; it naturally down-samples the radar data using an averaging pooling layer and constructs a multi-scale input in the encoder path.
In one embodiment, in step S4, on the basis of the M-net input layer, a pyramid convolution layer is added, and multi-scale input is integrated into the coding module, so as to avoid a large increase in parameters, increase the network width of the coding module path, and increase the generalization of the radar intelligent identification network.
In one embodiment, the encoding module includes 8 convolutional layers, 3 pooling layers; the convolution layer is responsible for acquiring local characteristics of ground penetrating radar bridge steel bar data, the pooling layer down-samples the data and transmits the characteristics with unchanged size to the next layer; in order to improve the learning speed of the network under the condition of ensuring stable network learning performance, a batch normalization algorithm is added after a convolution layer, and the problems of gradient disappearance and convergence speed during network training are solved through an activation function.
In one embodiment, after the encoding stage is finished, the pyramid convolution module performs pyramid convolution operation on the feature map, and the pyramid convolution module comprises convolution kernels of different scales, so that multi-scale information can be extracted, and the identification effect of the ground penetrating radar bridge reinforcing steel bars is improved.
In one embodiment, the decoding module performs up-sampling on the reduced feature image, and then performs convolution processing, so as to perfect the geometry of the anomalous body and make up for the loss of detail caused by the previous operation.
In one embodiment, the multi-scale output module generates a corresponding local output map for the earlier layer; taking the average value of the plurality of local output mappings as a final output; therefore, deep monitoring is realized, the problem of gradient disappearance can be relieved, and training of an early layer is facilitated.
The technical features mentioned above can be combined in various suitable ways or replaced by equivalent technical features as long as the purpose of the invention can be achieved.
Compared with the prior art, the radar intelligent identification method for the bridge reinforcing steel bar provided by the invention at least has the following beneficial effects:
1) The multi-scale input is integrated into the coding module, so that the parameter is prevented from being greatly increased; 2) Increasing the network width of the path of the coding module; 3) The generalization of the BIG-inv net network is increased; 4) Multi-scale information can be extracted, and the identification effect of the ground penetrating radar bridge reinforcing steel bar is improved; 5) The problem of gradient disappearance can be relieved; 6) The intelligent recognition result is highly consistent with the position, the diameter and the dielectric constant of the model.
Drawings
The invention will be described in more detail hereinafter on the basis of embodiments and with reference to the accompanying drawings. Wherein:
FIG. 1 shows a technical general route diagram of the method for radar intelligent identification of bridge reinforcing steel bars according to the present invention;
FIG. 2 is a schematic diagram illustrating the operation of the ground penetrating radar of the present invention;
FIG. 3 shows a BIG-inv net architecture diagram of the present invention;
FIG. 4 shows a multi-scale input module diagram of the ground penetrating radar bridge steel bar data of the present invention;
FIG. 5 shows a schematic diagram of an encoding module of the present invention;
FIG. 6 is a schematic diagram of a pyramid convolution module according to the present invention;
FIG. 7 shows a schematic diagram of a decoding module of the present invention;
FIG. 8 shows a schematic diagram of a multi-scale output module of the present invention;
FIGS. 9 (a) -9 (c) are schematic diagrams illustrating intelligent identification of two rebar models according to the present invention;
FIGS. 10 (a) -10 (c) are schematic diagrams illustrating intelligent recognition of three rebar models according to the present invention;
FIG. 11 shows a physical model diagram of the present invention;
fig. 12 shows a schematic diagram of the intelligent identification of measured data according to the present invention.
Detailed Description
The invention will be further explained with reference to the drawings.
The technical general route of the invention is shown in figure 1.
The ground penetrating radar adopts high-frequency electromagnetic waves for detection. As shown in fig. 2, the ground penetrating radar transmits high-frequency pulse electromagnetic waves to the underground through a transmitting antenna T, when the electromagnetic waves are transmitted in an underground medium, part of the radar waves are reflected at a place where the electromagnetic property is changed, the reflected electromagnetic waves are received by a receiving antenna R, and then a two-dimensional radar data image is formed after the radar data image is processed by radar special software, so that the inference can be carried out on the medium distribution rule (such as medium thickness, interface, buried depth, size, shape, trend and the like of internal buried objects or defects). The steel bar is a cylindrical target, and is represented in a hyperbolic shape in a ground penetrating radar sectional view.
The dielectric constant abnormal body existing in the underground half space can cause electromagnetic field response on an observation surface, and the electromagnetic field response caused by different abnormal bodies also shows different characteristics. I.e. having a certain spatial correlation and local presence. The full convolution neural network method is a special deep learning method, and a convolution operator is used for replacing matrix multiplication, so that the locality and the spatiality of an input image and an output label are mainly learned. Therefore, it is feasible to perform inversion on the ground penetrating radar data by using a full convolution neural network method.
The inversion target of the invention is to obtain a mapping function from the ground penetrating radar echo data to the dielectric constant of the model body by a full convolution neural network method. The physical parameters of the model body medium are shown in Table 1.
ε r =f(E Z ),
In the formula: epsilon r Is the model dielectric constant; f is a full convolution network mapping function; e Z Is echo data.
TABLE 1 model physical parameters of bulk media
The radar intelligent identification of the bridge steel bar mainly comprises the steps of response characteristic analysis of the bridge steel bar ground penetrating radar, sample data set construction, building of a BIG-inv net network model, network training and optimization, model test, instance detection and the like.
(1) Response characteristic analysis of bridge steel bar ground penetrating radar
Because the bridge structure is composed of multiple media, the dielectricity of the bridge structure is different, the response characteristics of the bridge structure can be obtained through forward evolution by changing the dielectric constant and the conductivity of concrete and adjusting the diameter of the reinforcing steel bars, the thickness of the protective layer, the distance between the reinforcing steel bars and the like, and thus the propagation rules of electromagnetic waves under different conditions can be mastered. In addition, response typical maps of existing bridge steel bar ground penetrating radars are widely collected, response characteristics are combed by a system, and a basis is provided for sample data set construction.
(2) Sample data set construction
The two-dimensional dielectric model is divided into 256 × 128 grids, each grid having a size of 0.25cm × 0.25cm. The dielectric model comprises reinforcing steel bars with different diameters and different burial depths. Steel bar dielectric models with different diameters, different burial depths and different positions in homogeneous concrete are designed, forward calculation is performed through a time domain finite difference method, and a data set of 1152 sample data pairs is obtained. Each sample data pair consists of a dielectric model (256 x 128 in size) and corresponding georadar data (which is the same size as the model data by bilinear interpolation).
(3) BIG-inv net network model building
The invention designs a radar intelligent identification network (BIG-inv Net) of bridge reinforcements for a backbone network based on an M-Net framework of a full convolution neural network, as shown in figure 3. BIG-inv net comprises 5 modules.
The ground penetrating radar bridge steel bar data multi-scale input module is shown in figure 4, and the module is used for constructing data pyramid input and realizing different levels of receptive field fusion. Particular implementations naturally down-sample radar data using an average pooling layer and construct a multi-scale input in the encoder path. The invention adds the pyramid convolution layer on the basis of M-net measuring input layer, and the advantages of the pyramid convolution layer are as follows: 1) The multi-scale input is integrated into the coding module, so that the parameter is prevented from being greatly increased; 2) Increasing the network width of the path of the coding module; 3) The generalization of the BIG-inv net network is increased.
The coding module is shown in fig. 5, and the module consists of 8 convolutional layers and 3 pooling layers. The convolution layer is responsible for obtaining local characteristics of ground penetrating radar bridge steel bar data, and the pooling layer down-samples the data and transmits the characteristics with unchanged size to the next layer. In order to improve the learning speed of the network under the condition of ensuring the stable network learning performance, a batch normalization algorithm is added after the convolution layer; the activation function can solve the problems of gradient disappearance and convergence speed in network training, and the mathematical expression of the linear rectification function is as follows:
f ReLU (x)=max(x,0),
when the input is greater than 0, the output is equal to the input; when the input is less than 0, the output is 0.
And (4) after the coding stage is finished, performing pyramid convolution operation on the characteristic graph by a pyramid convolution module which contains convolution kernels with different scales, so that multi-scale information can be extracted, and the identification effect of the ground penetrating radar bridge reinforcing steel bar is improved.
The decoding module, see fig. 7, has already obtained all the anomaly information and the approximate location information, and then needs to correspond these anomalies to specific pixel-level locations. The decoding module performs up-sampling on the reduced characteristic image and then performs convolution processing, so that the geometric shape of the abnormal body is perfected, and the detail loss caused by the previous operation is made up.
The multi-scale output module, see fig. 8, generates a corresponding local output map for the earlier layer. And taking the average value of the plurality of local output maps as a final output. Therefore, deep monitoring is realized, the problem of gradient disappearance can be relieved, and training of an early layer is facilitated.
(4) Network training and optimization
The network training and optimization comprises two parts of training data and testing the network. Randomly dividing a sample database into training data, testing data and verification data, wherein the corresponding proportion is 10:1:1 (training set: 920 pairs; validation set: 116 pairs; test set 116 pairs). During the training process, all input and output values are normalized to a range of [0,1 ]. The BIG-inv net network model framework is constructed by using Keras and adopting tensiorflow as a back end. During training, an Adam optimizer with batch size of 10, learning rate of 0.001, momentum of 0.9 and weight decay of 1e-8 is set. A total of 60 rounds of training were performed. To prevent overfitting, validation was performed after each training. Setting the mean square error of the theoretical model and the inversion result as an inversion model evaluation index, wherein a mean square error function expression is as follows:
in the formula: n is the number of inversion parameters, m i In order to test the theoretical value of the dielectric constant parameter of the model after normalization,to normalize the inverse dielectric constant parameters.
(5) Model test
Fig. 9 (a) and 10 (a) are echo data generated by forward evolution of a dielectric constant model FDTD; FIGS. 9 (b) and 10 (b) are dielectric constant models; fig. 9 (c) and 10 (c) show the recognition results of the trained neural network. The comparison table of the model identification value and the theoretical value is as follows:
TABLE 2 comparison table of intelligent identification value and theoretical value of two steel bar models
TABLE 3 comparison table of intelligent identification values and theoretical values of three steel bar models
Model tests show that the intelligent recognition result is highly consistent with the position, the diameter and the dielectric constant of the model.
(6) Example detection
The method is adopted to carry out inversion on the actually measured ground penetrating radar data of the physical model. The physical model is shown in fig. 11, the direction of the blue arrow in the figure is the direction of the measuring line, 7 steel bars with the radius of 0.006m and 8 steel bars with the radius of 0.005m are vertically arranged in the measured body, and the position parameters of the steel bars are shown in table 4.
TABLE 4 measured body Steel Bar parameters
The acquisition equipment used by the invention is an LTD-2600 type intelligent ground penetrating radar. The length of the measuring wire is 2.56m, the time window is 10ns, and the frequency of the antenna is 1500MHz. The actually measured ground penetrating radar data and the identification result are shown in figure 12. It can be seen that the position, dielectric constant and diameter of the steel bar are basically consistent with the real situation.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to limit the present invention, and those skilled in the art can make variations and modifications of the present invention without departing from the spirit and scope of the present invention by using the methods and technical contents disclosed above.
Claims (10)
1. The radar intelligent identification method of the bridge reinforcing steel bar is characterized by comprising the following steps:
s1, acquiring response characteristics of a bridge structure through forward modeling by changing the dielectric constant and the conductivity of the bridge structure;
s2, collecting a response typical map of the existing bridge steel bar ground penetrating radar, combing response characteristics by a system, and constructing a sample data set;
s3, dividing grids for a two-dimensional dielectric model, wherein the two-dimensional dielectric model comprises reinforcing steel bars with different diameters and different burial depths; designing steel bar dielectric models with different diameters, different burial depths and different positions in concrete, and forward calculating by a time domain finite difference method to obtain a data set of sample data pairs; each sample data pair consists of a dielectric model and corresponding ground penetrating radar data;
s4, designing a radar intelligent identification network of bridge reinforcements for a backbone network based on an M-Net framework of a full convolution neural network, constructing a radar intelligent identification network model framework by using Keras, and adopting tensorflow as a rear end;
s5, training and optimizing the radar intelligent recognition network, randomly dividing a sample database into training data, testing data and verification data, standardizing all input and output values to be in a range of [0,1] in the training process, and verifying after each training;
and S6, inverting the actually measured ground penetrating radar data of the physical model, and verifying whether the position, the dielectric constant and the diameter of the steel bar are consistent with the real condition.
2. The method for radar intelligent identification of bridge reinforcements according to claim 1, wherein the dielectric constant and the conductivity of the bridge structure are changed by adjusting the diameter of the reinforcements, the thickness of the protective layer and the distance between the reinforcements.
3. The method for radar intelligent recognition of bridge reinforcements according to claim 1, wherein the ground penetrating radar data in step S3 is the same size as the model data by bilinear interpolation.
4. The method for radar intelligent recognition of the bridge reinforcement according to claim 1, wherein the radar intelligent recognition network comprises a ground penetrating radar bridge reinforcement multi-scale input module, an encoding module, a pyramid convolution module, a decoding module and a multi-scale output module.
5. The method for radar intelligent identification of the bridge reinforcing steel bars according to claim 4, wherein a ground penetrating radar bridge reinforcing steel bar multi-scale input module is used for constructing data pyramid input and realizing fusion of different levels of receptive fields; it naturally down-samples the radar data using an averaging pooling layer and constructs a multi-scale input in the encoder path.
6. The method for radar-based intelligent identification of bridge reinforcements according to any one of claims 1 to 5, wherein in step S4, on the basis of the M-net test input layer, a pyramid type convolution layer is added, and a multi-scale input is integrated into the coding module, so that the parameters are prevented from being greatly increased, the network width of the coding module path is increased, and the generalization of a radar-based intelligent identification network is increased.
7. The method for radar intelligent recognition of bridge reinforcements according to claim 4, wherein the coding module comprises 8 convolutional layers and 3 pooling layers; the convolution layer is responsible for acquiring local characteristics of ground penetrating radar bridge steel bar data, the pooling layer down-samples the data and transmits the characteristics with unchanged size to the next layer; in order to improve the learning speed of the network under the condition of ensuring stable network learning performance, a batch normalization algorithm is added after a convolution layer, and the problems of gradient disappearance and convergence speed during network training are solved through an activation function.
8. The radar intelligent identification method for the bridge reinforcements according to claim 4, wherein after the encoding stage is finished, the pyramid convolution module performs pyramid convolution operation on the feature map, and the pyramid convolution module comprises convolution kernels with different scales, so that multi-scale information can be extracted, and the identification effect of the ground penetrating radar bridge reinforcements is improved.
9. The method for radar intelligent identification of bridge reinforcements according to claim 4, wherein the decoding module performs up-sampling on the reduced characteristic image and then performs convolution processing, so as to perfect the geometric shape of an abnormal body and make up for detail loss caused by previous operations.
10. The method of claim 4, wherein the multi-scale output module generates a corresponding local output map for an early layer; taking the average value of the plurality of local output mappings as a final output; therefore, deep monitoring is realized, the problem of gradient disappearance can be relieved, and training of an early layer is facilitated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210919790.8A CN115310482A (en) | 2022-07-31 | 2022-07-31 | Radar intelligent identification method for bridge reinforcing steel bar |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210919790.8A CN115310482A (en) | 2022-07-31 | 2022-07-31 | Radar intelligent identification method for bridge reinforcing steel bar |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115310482A true CN115310482A (en) | 2022-11-08 |
Family
ID=83858430
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210919790.8A Pending CN115310482A (en) | 2022-07-31 | 2022-07-31 | Radar intelligent identification method for bridge reinforcing steel bar |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115310482A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115598714A (en) * | 2022-12-14 | 2023-01-13 | 西南交通大学(Cn) | Time-space coupling neural network-based ground penetrating radar electromagnetic wave impedance inversion method |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107239751A (en) * | 2017-05-22 | 2017-10-10 | 西安电子科技大学 | High Resolution SAR image classification method based on the full convolutional network of non-down sampling contourlet |
CN112381172A (en) * | 2020-11-28 | 2021-02-19 | 桂林电子科技大学 | InSAR interference image phase unwrapping method based on U-net |
CN112462346A (en) * | 2020-11-26 | 2021-03-09 | 西安交通大学 | Ground penetrating radar roadbed defect target detection method based on convolutional neural network |
CN114548278A (en) * | 2022-02-22 | 2022-05-27 | 西安建筑科技大学 | In-service tunnel lining structure defect identification method and system based on deep learning |
US11397242B1 (en) * | 2021-02-23 | 2022-07-26 | Tsinghua University | 3D object detection method based on multi-view feature fusion of 4D RaDAR and LiDAR point clouds |
-
2022
- 2022-07-31 CN CN202210919790.8A patent/CN115310482A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107239751A (en) * | 2017-05-22 | 2017-10-10 | 西安电子科技大学 | High Resolution SAR image classification method based on the full convolutional network of non-down sampling contourlet |
CN112462346A (en) * | 2020-11-26 | 2021-03-09 | 西安交通大学 | Ground penetrating radar roadbed defect target detection method based on convolutional neural network |
CN112381172A (en) * | 2020-11-28 | 2021-02-19 | 桂林电子科技大学 | InSAR interference image phase unwrapping method based on U-net |
US11397242B1 (en) * | 2021-02-23 | 2022-07-26 | Tsinghua University | 3D object detection method based on multi-view feature fusion of 4D RaDAR and LiDAR point clouds |
CN114548278A (en) * | 2022-02-22 | 2022-05-27 | 西安建筑科技大学 | In-service tunnel lining structure defect identification method and system based on deep learning |
Non-Patent Citations (3)
Title |
---|
汤浩;何楚;: "全卷积网络结合改进的条件随机场-循环神经网络用于SAR图像场景分类", 计算机应用, no. 12, 10 December 2016 (2016-12-10) * |
肖雨彤;张继贤;黄国满;顾海燕;卢丽君;: "FCN与CRF结合的PolSAR影像建筑区域提取", 遥感信息, no. 03, 20 June 2020 (2020-06-20) * |
赵泉华;谢凯浪;王光辉;李玉;: "全卷积网络和条件随机场相结合的全极化SAR土地覆盖分类", 测绘学报, no. 01, 15 January 2020 (2020-01-15) * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115598714A (en) * | 2022-12-14 | 2023-01-13 | 西南交通大学(Cn) | Time-space coupling neural network-based ground penetrating radar electromagnetic wave impedance inversion method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108182450B (en) | Airborne ground penetrating radar target identification method based on deep convolutional network | |
CN111812732B (en) | Geoelectromagnetic nonlinear inversion method based on convolutional neural network | |
CN110968826B (en) | Magnetotelluric deep neural network inversion method based on spatial mapping technology | |
WO2022227206A1 (en) | Fully convolutional neural network-based magnetotelluric inversion method | |
CN108447057B (en) | SAR image change detection method based on significance and depth convolution network | |
CN111126591A (en) | Magnetotelluric deep neural network inversion method based on space constraint technology | |
Li et al. | Deep learning–based nondestructive evaluation of reinforcement bars using ground‐penetrating radar and electromagnetic induction data | |
CN113848252B (en) | Corrosion imaging method based on ultrasonic guided wave and convolutional neural network | |
CN111638249B (en) | Water content measuring method based on deep learning and application of water content measuring method in oil well exploitation | |
CN115310482A (en) | Radar intelligent identification method for bridge reinforcing steel bar | |
CN112163669A (en) | Pavement subsidence prediction method based on BP neural network | |
CN113468804B (en) | Underground pipeline identification method based on matrix bundles and deep neural network | |
Vy et al. | Damage localization using acoustic emission sensors via convolutional neural network and continuous wavelet transform | |
Xiong et al. | Automatic detection and location of pavement internal distresses from ground penetrating radar images based on deep learning | |
Imai et al. | Reflectivity-consistent sparse blind deconvolution for denoising and calibration of multichannel GPR volume images | |
CN115598714B (en) | Time-space coupling neural network-based ground penetrating radar electromagnetic wave impedance inversion method | |
CN116879960A (en) | Advanced mining detection abnormal body positioning and identifying method based on deep learning | |
CN116340770A (en) | Intelligent identification method for tunnel lining geological radar data | |
CN115826059A (en) | Transient electromagnetic rapid advanced prediction method | |
CN112948916B (en) | Rock-soil body mechanical parameter acquisition method | |
CN110045362B (en) | RNN-based radar target track identification method and device | |
CN112882022A (en) | Ground penetrating radar data time-frequency combination full waveform inversion method | |
Tang et al. | Fast Advanced Prediction of water-filled Structure using Transient Electromagnetic Method based on Deep Learning | |
CN115542315B (en) | ADI-FDTD-based ground penetrating radar forward modeling method and system | |
Tang et al. | A Deep learning estimation for probing Depth of Transient Electromagnetic Observation |
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 |