CN115630492A - Intelligent inversion method, system and storage medium for tunnel lining disease change characteristics - Google Patents
Intelligent inversion method, system and storage medium for tunnel lining disease change characteristics Download PDFInfo
- Publication number
- CN115630492A CN115630492A CN202211246860.4A CN202211246860A CN115630492A CN 115630492 A CN115630492 A CN 115630492A CN 202211246860 A CN202211246860 A CN 202211246860A CN 115630492 A CN115630492 A CN 115630492A
- Authority
- CN
- China
- Prior art keywords
- scan data
- tunnel lining
- phase
- dimensional
- inversion
- 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
- 238000000034 method Methods 0.000 title claims abstract description 44
- 230000008859 change Effects 0.000 title claims abstract description 39
- 201000010099 disease Diseases 0.000 title claims abstract description 36
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 36
- 238000003062 neural network model Methods 0.000 claims abstract description 49
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 21
- 230000007547 defect Effects 0.000 claims abstract description 20
- 238000012545 processing Methods 0.000 claims abstract description 20
- 238000012549 training Methods 0.000 claims description 33
- 238000005070 sampling Methods 0.000 claims description 23
- 238000000819 phase cycle Methods 0.000 claims description 18
- 238000004088 simulation Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 13
- 230000004044 response Effects 0.000 claims description 10
- 238000003384 imaging method Methods 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 7
- 230000001427 coherent effect Effects 0.000 claims description 4
- 238000002592 echocardiography Methods 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract description 6
- 230000000875 corresponding effect Effects 0.000 description 24
- 238000013527 convolutional neural network Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000002596 correlated effect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000009659 non-destructive testing Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses an intelligent inversion method, a system and a storage medium for tunnel lining disease change characteristics, wherein the intelligent inversion method comprises the following steps: performing geological radar data acquisition on the tunnel lining area to be detected in a plurality of different time periods to obtain B-scan data of the tunnel lining area to be detected in a plurality of time periods; respectively carrying out synthetic aperture focusing algorithm processing on the B-scan data of each time period to correspondingly obtain focused B-scan data, and forming two-dimensional B-scan data pairs corresponding to each time period of the tunnel lining area to be detected; respectively inputting the two-dimensional B-scan data pairs corresponding to each time period into the trained neural network model; the neural network model is formed by cascading a time phase slice inversion module and a phase slice inversion module through a four-dimensional inversion module; and outputting the dielectric constant model corresponding to each time period by the neural network model to obtain the change characteristics of the defects of the tunnel lining area to be detected. The invention realizes the dynamic detection of the diseases in the area to be detected and can greatly improve the working efficiency and inversion stability.
Description
Technical Field
The invention belongs to the technical field of civil engineering, geotechnical engineering and nondestructive testing, and particularly relates to an intelligent inversion method, system and storage medium for tunnel lining disease change characteristics.
Background
Geological radar (GPR) is widely applied to the fields of civil engineering, geotechnical engineering, nondestructive testing and the like by virtue of the characteristics of no damage, high efficiency, high anti-interference level, strong penetration capacity and the like. The work flow of the geological radar is to determine test parameters such as survey line layout, grid size, survey point spacing and the like after selecting a region to be detected, and detect the internal structure of an object by transmitting electromagnetic waves and receiving echoes. The geological radar can obtain a piece of B-scan data reflecting underground information in a common offset two-dimensional profile measurement mode. The method only detects and identifies single B-Scan data, accurate positioning and quantitative judgment of the three-dimensional diseases are difficult to carry out due to the fact that information in the width direction of a three-dimensional space is lacked, and more complete acquisition and analysis of underground information can be achieved by processing a plurality of two-dimensional B-Scan data in the space in consideration of continuity of scanning of geological radars acquired by adjacent channels in the width direction.
In the context of geological radar data interpretation, experts are required to interpret various subsurface objects in geological radar data, including environmental complexity and experimental noise. Furthermore, analyzing large amounts of geological radar data with the naked eye is an inefficient and time consuming task. Therefore, it is important to develop a fast and accurate automatic detection technology for underground targets. In recent years, a Convolutional Neural Network (CNN) based method has been widely used to learn a feature expression directly from input data. Researchers developed CNN-based data interpretation methods for automatically detecting targets from GPR data. The main problem of the existing CNN-based method for geological radar data interpretation is that a single piece of B-scan data is used for interpretation and analysis only for a certain time state of a region to be detected, and the work of three-dimensional dynamic defect detection, target inversion and the like by utilizing spatial characteristic information of a plurality of pieces of B-scan data at different time periods is not considered.
Disclosure of Invention
The invention aims to solve the defects in the background art, and provides an intelligent inversion method, a system and a storage medium for tunnel lining disease change characteristics.
The technical scheme adopted by the invention is as follows: an intelligent inversion method for tunnel lining disease change characteristics comprises the following steps:
performing geological radar data acquisition on a tunnel lining area to be detected in a plurality of different time periods to obtain B-scan data of the tunnel lining area to be detected in the plurality of time periods;
respectively carrying out synthetic aperture focusing algorithm processing on the B-scan data of each time period to correspondingly obtain focused B-scan data, and combining the B-scan data and the focused B-scan data of each time period to form a two-dimensional B-scan data pair corresponding to each time period of the tunnel lining area to be detected;
respectively inputting the two-dimensional B-scan data pairs corresponding to each time period into the trained neural network model; the neural network model is formed by cascading a time phase slice inversion module and a phase slice inversion module through a four-dimensional inversion module of time phase + phase;
the neural network model outputs a dielectric constant model corresponding to each time period;
and acquiring the positions and shapes of the defects of the region to be detected in different time periods through the dielectric constant models corresponding to the time periods, so as to obtain the change characteristics of the defects of the tunnel lining region to be detected.
In the technical scheme, after B-scan data of a region to be detected is obtained, direct wave removal and noise removal pretreatment operations are carried out on the B-scan data; after preprocessing, the B-scan data is processed by a synthetic aperture focusing algorithm. By converging and focusing the echo signals, the influence of interference signals can be weakened, the transverse resolution can be improved, the diseases in the data image are closer to the real size and position, the characteristic response signals in the image are enhanced, and a high-quality two-dimensional B-scan pair required by the neural network model training is convenient to construct, so that the neural network model can better extract and analyze data characteristic information, the accuracy of the neural network model training is improved, and the more accurate identification effect is achieved.
In the technical scheme, a plurality of parallel geological radar survey lines are uniformly distributed in a lining area of a tunnel to be measured, and the geological radar survey lines cover the lining area of the tunnel to be measured; and collecting geological radar survey line data through a geological radar in the same time period to obtain B-scan data of the tunnel lining area to be measured in the time period. By enlarging the transverse coverage area, on the basis that the traditional single measuring line can only position the diseases in the tunnel lining, the three-dimensional space characteristic information of the diseases can be efficiently acquired, and a data basis is provided for the space-time change of the space characteristics of the diseases. The duration of each time period is the time required for the geological radar data acquisition equipment to acquire all geological radar survey line data in the tunnel lining area to be measured. The operator can set the geological radar data acquisition frequency (such as once per week, every three days or every day) of the tunnel lining area to be detected according to actual requirements. In a set period (such as within a plurality of months or a year), selecting a certain consistent time point according to a set frequency, and starting to collect all geological radar survey line data in a tunnel lining area to be measured by adopting geological radar data collection equipment. And taking the B-scan data acquired each time as the corresponding B-scan data of a certain time period.
In the technical scheme, the geological radar acquires data of the geological radar survey line at different time periods; the geological radar carries the positioning system and the ranging system to realize the initial calibration of the geological radar survey line phase, and the survey line phase, the grid size and the grid size of data acquisition are adjusted, so that the phase information repeatedly measured at the same appointed position in a plurality of time periods is kept consistent. By ensuring accurate registration of the B-scan data of a plurality of different time periods at the same position in space, the information difference of the geological radar data of a plurality of time periods is conveniently utilized, and the accuracy of dynamic disease change analysis in the tunnel lining model is improved.
In the above technical solution, the process of constructing the neural network model includes:
carrying out multi-position geological radar data acquisition on a known tunnel lining simulation model in different time periods; b-scan data of different positions of a known tunnel lining model at different time periods are obtained;
b-scan data of different positions of a known tunnel lining model in different time periods are processed by a synthetic aperture focusing algorithm to obtain focused B-scan data of different positions of the corresponding known tunnel lining model in different time periods;
defining data of different time periods as time phase sequences and data of different positions as phase sequences to form two-dimensional B-scan data time phase sequence and two-dimensional B-scan data phase sequence of a known tunnel lining simulation model;
respectively constructing training sets of a time phase slice inversion module, a phase slice inversion module and a four-dimensional inversion module:
the single sample information of the training set of the time phase slice inversion module comprises: taking a previous time phase two-dimensional B-scan data pair, a current time phase two-dimensional B-scan data pair and a next time phase two-dimensional B-scan data pair as input; taking a dielectric constant model of the current-phase two-dimensional B-scan data pair as an inverted label;
the single sample information of the training set of the phase slice inversion module includes: taking a former phase two-dimensional B-scan data pair, a current phase two-dimensional B-scan data pair and a next phase two-dimensional B-scan data pair as input, and taking a dielectric constant model of the current phase two-dimensional B-scan data pair as an inversion label;
the single sample information of the training set of the four-dimensional inversion module includes: the output of the time phase slice inversion module and the current output of the phase slice inversion module are used as input, and a dielectric constant model of a current time phase and current phase two-dimensional B-scan data pair is used as an inversion label.
And training the neural network model by adopting a training set.
The neural network model is constructed and trained through a cascade time phase slice inversion module, a phase slice inversion module and a four-dimensional inversion module. The neural network model is classified in the time dimension and the space dimension respectively, then disease feature identification is carried out in different dimensions, the feature information of diseases in the space and the time can be further correlated, and the highly correlated characteristics of the disease development in the time dimension and the space dimension are fully utilized, so that the accuracy and the stability of the inversion of the tunnel lining change features of the neural network model are improved.
In the technical scheme, the time phase slice inversion module and the phase slice inversion module both adopt Unet network structures fused with focused imaging processing branches, SAR image data input is added in the traditional Unet structure, ground penetrating radar data reach high-level resolution, feature information extraction is clearer and more comprehensive, the Unet network uses channel splicing to serve as a fusion mode of a feature map, the input of convolution can be richer, and the inversion results obtained by the two inversion modules in time dimension and space dimension can reflect the original feature information of lining disease data more.
The encoder modules of the time phase slice inversion module and the phase slice inversion module adopt three-channel input of B-scan data, focusing B-scan data and a dielectric constant model, and data obtained by the last down-sampling in the up-sampling stage are subjected to up-sampling after passing through a convolution layer; performing channel splicing on the data obtained by the up-sampling, the B-scan data with the same size and the same channel number and the focused B-scan data in the down-sampling process, changing the channel number through the convolution layer, performing up-sampling again, and repeatedly performing channel splicing, convolution and up-sampling operations; up-sampling is carried out for 3 times in total; extracting and fusing channel characteristic information from the finally up-sampled data through convolution to realize data dimension reduction and obtain an output result;
a two-channel Unet network is used as a four-dimensional inversion module.
In the above technical solution, the process of performing synthetic aperture focusing algorithm processing on the B-scan data to obtain focused B-scan data includes: the method comprises the steps of firstly calculating the two-way time delay from a point to be imaged to each channel in a regional image, then searching for a scattering response amplitude corresponding to a time delay point on each channel received echo, finally carrying out time domain coherent superposition on the scattering response amplitudes of the point in all channel echoes, obtaining a backscattering intensity value of a target according to the amplitude result, carrying out the operation on each point in the imaging region, realizing the focusing processing on the regional image, enabling the target on an imaging section to be focused into a small-range strong energy region, improving the transverse resolution of B-scan data, and improving the accurate identification capability of the target.
In the technical scheme, a small amount of known real tunnel lining region geological radar data are adopted for the trained neural network model to finely adjust the trained neural network model parameters, and the model is optimized through the real tunnel data, so that the accuracy and the stability of the application of the cascade neural network model in the actual engineering are improved.
The invention provides an intelligent inversion system for tunnel lining disease change characteristics, which comprises a B-scan data acquisition module, a two-dimensional B-scan data pair generation module, a dielectric constant model acquisition module and a change characteristic acquisition module, wherein the B-scan data acquisition module is used for acquiring a change characteristic of a tunnel lining disease;
the B-scan data acquisition module is used for acquiring geological radar data of the tunnel lining area to be detected in a plurality of different time periods to acquire B-scan data of the tunnel lining area to be detected in a plurality of time periods;
the two-dimensional B-scan data pair generation module is used for respectively carrying out synthetic aperture focusing algorithm processing on the B-scan data of each time period to correspondingly obtain focused B-scan data, and the B-scan data of each time period and the focused B-scan data are combined to form a two-dimensional B-scan data pair corresponding to each time period of the tunnel lining area to be detected;
the dielectric constant model acquisition module is used for respectively inputting the two-dimensional B-scan data pairs corresponding to each time period into the trained neural network model; the neural network model is formed by cascading a time phase slice inversion module and a phase slice inversion module through a time phase and phase inversion module; the neural network model outputs a dielectric constant model corresponding to each time period;
the change characteristic acquisition module is used for acquiring the positions and shapes of the defects of the area to be detected in different time periods through the dielectric constant models corresponding to the time periods to obtain the change characteristics of the defects of the tunnel lining area to be detected.
The invention provides a computer readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the processor executes the steps of the intelligent inversion method for tunnel lining disease change characteristics according to the technical scheme. .
The beneficial effects of the invention are: the method adopts an intelligent inversion method of tunnel lining disease change characteristics, and obtains a high-resolution image and a 'B-scan and focusing B-scan' data pair for constructing subsequent CNN network learning training by carrying out synthetic aperture focusing (SAR) algorithm processing on tunnel lining multi-time-period and multi-position geological radar data; and the outputs of the time phase slice inversion module and the phase slice inversion module are cascaded through a dual-channel Unet network, so that four-dimensional time phase + phase inversion is realized, the defect shape profiles of different time sequences and different positions can be quickly obtained, the three-dimensional dynamic rapid detection of lining defects in different time periods is further realized, and the detection working efficiency and the inversion stability are greatly improved.
The neural network model of the invention is based on two-dimensional B-scan data pairs to carry out learning training, and by adding SAR branches in the conventional CNN inversion network architecture, the constraint information based on synthetic aperture focusing algorithm data is added, and the identification precision is improved. The neural network model of the invention fuses time phase sequence information and phase sequence information, introduces SAR branches into a neural network structure, also adds SAR data for splicing in the process of channel splicing of up-sampling and down-sampling data, fuses SAR data characteristic information, and combines local information and overall information for channel splicing, thereby improving the accuracy of identification.
Drawings
FIG. 1 is a flow chart of an intelligent inversion method for tunnel lining disease change characteristics according to the invention;
FIG. 2 is a model structure diagram of inversion of a geological radar three-dimensional disease change characteristic dielectric constant model of the cascade CNN of the invention;
FIG. 3 is a "B-scan, focused B-scan" data pair set constructed in accordance with the present invention and its corresponding dielectric constant model;
FIG. 4 is a diagram of a Unet inversion module model structure merged into SAR branches;
FIG. 5 is a result diagram of a cascaded Unet network identification simulation model of the present invention;
fig. 6 is a diagram showing the result of identifying a real tunnel lining by the cascaded Unet network in the present invention.
Detailed Description
The invention will be further described in detail with reference to the following drawings and specific examples, which are not intended to limit the invention, but are for clear understanding.
The invention relates to an intelligent inversion method for tunnel lining disease change characteristics, which comprises the following steps:
performing geological radar data acquisition on the tunnel lining area to be detected in a plurality of different time periods to obtain B-scan data of the tunnel lining area to be detected in a plurality of time periods;
respectively carrying out synthetic aperture focusing algorithm processing on the B-scan data of each time period to correspondingly obtain focused B-scan data, and combining the B-scan data and the focused B-scan data of each time period to form a two-dimensional B-scan data pair corresponding to each time period of the tunnel lining area to be detected;
respectively inputting the two-dimensional B-scan data pairs corresponding to each time period into the trained neural network model; the neural network model is formed by cascading a time phase slice inversion module and a phase slice inversion module through a time phase and phase inversion module;
the neural network model outputs a dielectric constant model corresponding to each time period;
and acquiring the positions and shapes of the defects of the region to be detected in different time periods through the dielectric constant models corresponding to the time periods, so as to obtain the change characteristics of the defects of the tunnel lining region to be detected.
As shown in fig. 1-2, in an embodiment of the present invention, a method for intelligently inverting a tunnel lining defect change characteristic includes the following steps:
the method comprises the steps of firstly, collecting geological radar data for multiple times at different positions of a known tunnel lining simulation model in different time periods under the same environmental condition, and carrying out data preprocessing operations such as direct wave removal and noise removal on the obtained B-scan data.
Secondly, after preprocessing the B-scan data of different positions at different time intervals of the known tunnel lining simulation model, performing synthetic aperture focusing (SAR) algorithm processing on the B-scan data to construct a 'B-scan and focusing B-scan' data pair, namely a two-dimensional B-scan data pair, wherein the effect is shown in figure 3, the two-dimensional B-scan data pair takes a data format as the input of a neural network model, and can be presented in a picture form when the simulation model is visualized; the neural network model adopts a CNN cascade network and adopts a dielectric constant model as a training label. And performing learning training based on data pairs in a subsequent CNN network, and adding synthetic aperture focusing algorithm-based constraint information by adding SAR branches in a conventional CNN inversion network architecture.
The branch of the synthetic aperture focusing algorithm refers to that resolution characteristic data under different scales are obtained through operations such as convolution, pooling and the like on 'focusing B-scan' data, in the up-sampling process of a CNN inversion network structure, the coded characteristics under the same scale are fused into a decoding process, and the identification result index is improved by adding constraint information based on the synthetic aperture focusing data.
The synthetic aperture focusing (SAR) algorithm processing principle is that a point to be imaged in an area image is firstly calculated to obtain the two-way time delay of the point from each channel, then the scattering response amplitude corresponding to the time delay point is searched on the echo received by each channel, finally the scattering response amplitudes of the point in the echoes of all the channels are subjected to time domain coherent superposition, the backscattering intensity value of a target is obtained according to the amplitude result, each point in the imaging area is subjected to the operation, and then the area can be focused to obtain a high-resolution image. The core idea of the SAR algorithm is delay and accumulation.
The specific SAR algorithm calculation processing steps are as follows:
1) Calculating the two-way time delay of the data point A to each channel, wherein the time delay is tau A ,k To show that, then:
where k is the channel index, v is the propagation velocity, which is related to the dielectric constant of the medium, and x k Is the channel abscissa, x A ，z A The abscissa and ordinate of the imaging point a.
2) Calculating the scattering response amplitude x of the data point A in each channel A,k ：
x A,k ＝r k (t)
In the formula, t is tau A,k ，r k (t) represents the echo data of the geological radar in each channel.
3) And (5) carrying out coherent superposition on the scattering response amplitudes generated by the data point A in each channel to finish focusing on the data point A.
Where EA is the scattering response superposition and M is the number of channels.
4) And processing all points covered by the B-scan data at different positions in different periods of time of the known tunnel lining simulation model according to the steps to correspondingly form focused B-scan data.
And thirdly, defining data acquired in different time periods as time phase sequences, defining data acquired in different positions as phase sequences, and dividing the 'B-scan and focusing B-scan' data into two dimensions of the time phase sequences and the phase sequences to form two-dimensional B-scan data time phase sequences and two-dimensional B-scan data phase sequence of the known tunnel lining simulation model.
The two-dimensional B-scan data pairs acquired by the same phase and different time phases are used for training a time phase slice inversion module, the two-dimensional B-scan data pairs acquired by different phases and the same time phases are used for training a phase slice inversion module, and the time phase slice inversion module and the phase slice inversion module are connected through a four-dimensional time phase + phase inversion module to perform neural network model training, as shown in FIG. 2.
Specifically, the training set of the neural network model is divided into a training set of a time phase slice inversion module, a phase slice inversion module and a four-dimensional inversion module.
The single sample information of the training set of the phase slice inversion module comprises: taking a previous time phase two-dimensional B-Scan data pair, a current time phase two-dimensional B-Scan data pair and a next time phase two-dimensional B-Scan data pair as input; and taking the dielectric constant model of the current phase slice as a label of inversion.
The single sample information of the training set of the phase slice inversion module comprises: taking a previous phase two-dimensional B-Scan data pair, a current phase two-dimensional B-Scan data pair and a next phase two-dimensional B-Scan data pair as input; and taking the dielectric constant model of the current phase slice as an inversion label.
The single sample information of the training set of the four-dimensional inversion module comprises: and taking the output of the time phase slice inversion module and the output of the phase slice inversion module as the input of a four-dimensional time phase + phase inversion module, and taking a dielectric constant model of the current phase and the current phase slice as an inversion label.
And training the neural network model by adopting the training set, and applying the trained neural network model to output a corresponding dielectric constant model based on the output B-Scan data pair.
Loss in FIG. 2 is a loss function; and (4) continuously optimizing parameters of the neural network model by using the recognition result and the training label in the training process of the neural network model to perform loss.
The time phase slice inversion module and the phase slice inversion module are both formed by improving a Unet network structure, the Unet network structure is a network improved based on a full convolution neural network, and an encoder and a decoder of the Unet network are approximately bilaterally symmetrical like a capital letter U, so that the Unet network is named as the Unet network.
The neural network model improvement is divided into two aspects, one aspect is that a coder module of the Unet network structure is changed from single-channel input into three-channel input, and time phase sequence information and phase sequence information are convenient to fuse. The improvement of the other aspect is that SAR branches are introduced into a Unet network structure, SAR data are added for splicing in the process of channel splicing of data obtained by up-sampling and data with the same size in the down-sampling process, SAR data characteristic information is fused, and channel splicing is carried out by combining local information and overall information, so that the accuracy of identification is improved. Specifically, in the up-sampling stage, the data obtained by the last down-sampling is up-sampled after passing through the convolutional layer in the first step. And secondly, performing channel splicing on the data, the simulation data with the same size and the same channel number and SAR imaging data in the downsampling process, changing the channel number through the convolutional layer, performing upsampling again, repeatedly performing channel splicing, convolution and upsampling operations, and performing upsampling for 3 times in the second step. And thirdly, extracting and fusing channel characteristic information from the finally up-sampled data through convolution to realize data dimension reduction and obtain an output identification result. The outputs of the time phase slice inversion module and the phase slice inversion module are cascaded through a dual-channel Unet network to realize four-dimensional time phase + phase inversion, and the flow is shown in FIG. 4.
Geological radar data (a GPR simulation image shown in figure 5 and representing B-Scan data after preprocessing) generated by the simulation model, a recognition result (shown in figure 5 and merged into an SAR branch inversion result) of a certain sample of the trained neural network model, and a training label (shown in figure 5) corresponding to the training sample. As shown in fig. 5, the tunnel lining simulation model recognition results including cracks (upper defects) and voids (lower defects) are shown. And verifying the effectiveness of the trained neural network model by comparing the training label with the recognition result of the trained neural network model.
And fourthly, performing multi-time-period and multi-position data acquisition work on a target detection area (namely the tunnel lining area to be detected) through a positioning system and a ranging system which are constructed by a laser ranging module, a real-time quality control module, a data acquisition and storage module, a target coarse positioning module and the like, obtaining B-scan data of the tunnel lining area to be detected in multiple time periods, and performing preprocessing operations such as denoising, estimation and removal of direct wave interference.
The geological radar carries a positioning system and a ranging system to realize the initial calibration of geological radar survey line phase so as to adjust the survey line phase, grid size and other acquisition parameters of tunnel lining disease data acquisition, and the consistency of phase information of a plurality of time phase repeated measurements is ensured. And analyzing and monitoring the dynamic change of the diseases under the tunnel lining model by using the information difference of geological radar data acquired at certain time intervals.
Based on the characteristics of the direct wave of the geological radar signal, the two-dimensional physical wavelet is used as a basic wavelet to perform wavelet transformation on the geological radar signal, and a proper wavelet scale is selected to estimate the interference of the direct wave so as to remove the direct wave. Based on the characteristic that effective components of adjacent reflected waves have strong correlation in waveform and energy, the KL transformation can be used for reserving a correlation signal in a certain direction so as to suppress uncorrelated noise and other waves and achieve the purpose of removing noise in geological radar signals. In the aspect of noise removal, the KL transform is used for carrying out global decomposition on echo signals based on the difference of signal correlation degrees in the depth direction, namely the KL transform can be used for retaining the correlated signals in a certain direction so as to suppress uncorrelated noise and other waves and obtain a good denoising effect easily.
The transmitting antenna directly reaches the receiving antenna through the space and the earth surface without reflecting part of energy by the underground target, and the received signal is called direct wave. From the two-dimensional physical wavelet characteristics, when the spatial scale factor becomes large, the two-dimensional physical wavelet function becomes flat, which is consistent with the characteristics of the direct wave. Two-dimensional physical wavelet transformation is carried out on the two-dimensional geological radar signal, and only a large space scale is selected for signal reconstruction, so that estimation and removal of direct wave interference can be realized.
And fifthly, respectively carrying out synthetic aperture focusing algorithm processing on the B-scan data of each time period by adopting the method same as the step two to correspondingly obtain focused B-scan data, and combining the B-scan data of each time period and the focused B-scan data to form a two-dimensional B-scan data pair corresponding to each time period of the tunnel lining area to be detected.
Sixthly, respectively inputting the two-dimensional B-scan data pairs corresponding to each time period into the trained neural network model; the neural network model outputs a dielectric constant model corresponding to each time period, as shown in fig. 6, which shows the dielectric constant model of the same region to be measured and corresponding GPR data obtained at two different time periods.
And seventhly, acquiring the positions and shapes of the defects of the to-be-detected region in different time periods through the dielectric constant models corresponding to the time periods, and obtaining the change characteristics of the defects of the to-be-detected tunnel lining region.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Those not described in detail in this specification are well within the skill of the art.
Claims (10)
1. An intelligent inversion method for tunnel lining disease change characteristics is characterized by comprising the following steps: the method comprises the following steps:
performing geological radar data acquisition on a tunnel lining area to be detected in a plurality of different time periods to obtain B-scan data of the tunnel lining area to be detected in the plurality of time periods;
respectively carrying out synthetic aperture focusing algorithm processing on the B-scan data of each time period to correspondingly obtain focused B-scan data, and combining the B-scan data and the focused B-scan data of each time period to form a two-dimensional B-scan data pair corresponding to each time period of the tunnel lining area to be detected;
respectively inputting the two-dimensional B-scan data pairs corresponding to each time period into the trained neural network model; the neural network model is formed by cascading a time phase slice inversion module and a phase slice inversion module through a four-dimensional inversion module;
the neural network model outputs a dielectric constant model corresponding to each time period;
and acquiring the positions and shapes of the defects of the area to be detected in different time periods through the dielectric constant models corresponding to the time periods to obtain the change characteristics of the defects of the tunnel lining area to be detected.
2. The intelligent inversion method for tunnel lining disease change characteristics according to claim 1, characterized in that: after B-scan data of a region to be detected is obtained, direct wave removal and noise removal preprocessing operations are carried out on the B-scan data; after preprocessing, the B-scan data is processed by a synthetic aperture focusing algorithm.
3. The intelligent inversion method for tunnel lining disease change characteristics according to claim 1, characterized in that: uniformly distributing a plurality of parallel geological radar survey lines in a tunnel lining area to be measured, wherein the geological radar survey lines cover the tunnel lining area to be measured; and collecting geological radar survey line data through a geological radar in the same time period to obtain B-scan data of the tunnel lining area to be measured in the time period.
4. The intelligent inversion method for tunnel lining disease change characteristics according to claim 3, characterized in that: the geological radar carries out data acquisition on the geological radar measuring line at different time periods; the geological radar carries the positioning system and the ranging system to realize the initial calibration of the geological radar survey line phase, and the survey line phase, the grid size and the grid size of data acquisition are adjusted, so that the phase information repeatedly measured at the same appointed position in a plurality of time periods is kept consistent.
5. The intelligent inversion method for tunnel lining disease change characteristics according to claim 1, characterized in that: the construction process of the neural network model comprises the following steps:
carrying out multi-position geological radar data acquisition on a known tunnel lining simulation model in different time periods; b-scan data of different positions of a known tunnel lining model at different time periods are obtained;
b-scan data of different positions of a known tunnel lining model at different time periods are subjected to synthetic aperture focusing algorithm processing to obtain corresponding focused B-scan data of different positions of the known tunnel lining model at different time periods;
defining data of different time periods as time phase sequences and data of different positions as phase sequences to form two-dimensional B-scan data to time phase sequences and two-dimensional B-scan data to phase sequences of a known tunnel lining simulation model;
respectively constructing training sets of a time phase slice inversion module, a phase slice inversion module and a four-dimensional inversion module:
the single sample information of the training set of the time phase slice inversion module comprises: taking a previous time phase two-dimensional B-scan data pair, a current time phase two-dimensional B-scan data pair and a next time phase two-dimensional B-scan data pair as input; taking the dielectric constant model of the current-phase two-dimensional B-scan data pair as an inversion label;
the single sample information of the training set of the phase slice inversion module includes: taking a former phase two-dimensional B-scan data pair, a current phase two-dimensional B-scan data pair and a next phase two-dimensional B-scan data pair as input, and taking a dielectric constant model of the current phase two-dimensional B-scan data pair as an inversion label;
the single sample information of the training set of the four-dimensional inversion module includes: the output of the time phase slice inversion module and the current output of the phase slice inversion module are used as input, and a dielectric constant model of a current time phase and current phase two-dimensional B-scan data pair is used as an inversion label.
And training the neural network model by adopting a training set.
6. The intelligent inversion method for tunnel lining disease change characteristics according to claim 1, characterized in that: the time phase slice inversion module and the phase slice inversion module both adopt a Unet network structure which is fused into a focusing imaging processing branch;
the encoder modules of the time phase slice inversion module and the phase slice inversion module adopt three-channel input of B-scan data, focusing B-scan data and a dielectric constant model, and data obtained by the last down-sampling in the up-sampling stage are subjected to up-sampling after passing through a convolution layer; performing channel splicing on the data obtained by the up-sampling, B-scan data with the same size and the same channel number and focused B-scan data in the down-sampling process, changing the channel number through a convolution layer, performing up-sampling again, and repeating the operations of channel splicing, convolution and up-sampling; up-sampling is carried out for 3 times in total; extracting and fusing channel characteristic information from the finally up-sampled data through convolution to realize data dimension reduction and obtain an output result;
a two-channel Unet network is used as a four-dimensional inversion module.
7. The intelligent inversion method for tunnel lining disease change characteristics according to claim 1, characterized in that: the process of carrying out synthetic aperture focusing algorithm processing on the B-scan data to obtain focused B-scan data comprises the following steps: the method comprises the steps of firstly calculating the two-way time delay from a point to be imaged to each channel in a regional image, then searching for a scattering response amplitude corresponding to a time delay point on each channel echo, finally carrying out time domain coherent superposition on the scattering response amplitudes of the point in all channel echoes, obtaining a backscattering intensity value of a target according to the amplitude result, and carrying out the operation on each point in the imaging region to realize the focusing processing on the regional image.
8. The intelligent inversion method for tunnel lining disease change characteristics according to claim 1, characterized in that: and (3) fine-tuning the trained neural network model parameters by adopting a small amount of known real geological radar data of the tunnel lining region for the trained neural network model.
9. An intelligent inversion system for tunnel lining disease change characteristics is characterized by comprising a B-scan data acquisition module, a two-dimensional B-scan data pair generation module, a dielectric constant model acquisition module and a change characteristic acquisition module;
the B-scan data acquisition module is used for acquiring geological radar data of the tunnel lining area to be detected in a plurality of different time periods to acquire B-scan data of the tunnel lining area to be detected in a plurality of time periods;
the two-dimensional B-scan data generation module is used for respectively carrying out synthetic aperture focusing algorithm processing on the B-scan data of each time period to correspondingly obtain focused B-scan data, and the B-scan data of each time period and the focused B-scan data are combined to form a two-dimensional B-scan data pair corresponding to each time period of the tunnel lining area to be detected;
the dielectric constant model acquisition module is used for respectively inputting the two-dimensional B-scan data pairs corresponding to all the time periods into the trained neural network model; the neural network model is formed by cascading a time phase slice inversion module and a phase slice inversion module through a time phase and phase inversion module; the neural network model outputs a dielectric constant model corresponding to each time period;
the change characteristic acquisition module is used for acquiring the positions and shapes of the defects of the area to be detected in different time periods through the dielectric constant models corresponding to the time periods to obtain the change characteristics of the defects of the tunnel lining area to be detected.
10. A computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the tunnel lining disease change feature intelligent inversion method steps of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211246860.4A CN115630492B (en) | 2022-10-12 | 2022-10-12 | Tunnel lining defect change characteristic intelligent inversion method, system and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211246860.4A CN115630492B (en) | 2022-10-12 | 2022-10-12 | Tunnel lining defect change characteristic intelligent inversion method, system and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115630492A true CN115630492A (en) | 2023-01-20 |
| CN115630492B CN115630492B (en) | 2023-12-19 |
Family
ID=84905323
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211246860.4A Active CN115630492B (en) | 2022-10-12 | 2022-10-12 | Tunnel lining defect change characteristic intelligent inversion method, system and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115630492B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107132539A (en) * | 2017-05-03 | 2017-09-05 | 中国地质科学院探矿工艺研究所 | Landslide early-stage identification method of time sequence InSAR (interferometric synthetic Aperture Radar) based on small baseline set |
| CN110361453A (en) * | 2019-07-24 | 2019-10-22 | 四川幔壳科技有限公司 | Array dystopy based on non-destructive testing receives ultrasound synthesis aperture focus method |
| CN112213724A (en) * | 2020-09-16 | 2021-01-12 | 深圳航天科技创新研究院 | Backward projection imaging method and system based on ground penetrating radar data and storage medium |
| WO2021068848A1 (en) * | 2019-10-09 | 2021-04-15 | 山东大学 | Tunnel structure disease multi-scale measurement and intelligent diagnosis system and method |
| CN113552571A (en) * | 2020-09-25 | 2021-10-26 | 哈尔滨工业大学（威海） | PSM (pulse position modulation) algorithm-based underwater laser induced acoustic SAFT (safety and safety transform) imaging method |
| CN114353876A (en) * | 2022-01-07 | 2022-04-15 | 兰州大学 | Loess highway slope health monitoring method |
| 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 |
| CN114707532A (en) * | 2022-01-11 | 2022-07-05 | 中铁隧道局集团有限公司 | Ground penetrating radar tunnel disease target detection method based on improved Cascade R-CNN |
| WO2022147969A1 (en) * | 2021-01-08 | 2022-07-14 | 中国民航大学 | Airport pavement underground structure disease automatic detection method based on deep learning |
| CN114897804A (en) * | 2022-04-26 | 2022-08-12 | 云南航天工程物探检测股份有限公司 | Ground penetrating radar tunnel lining quality detection method based on self-supervision learning |
| CN114942436A (en) * | 2022-05-31 | 2022-08-26 | 西安建筑科技大学 | Tunnel lining disease identification method and system based on geological radar and deep learning |
| CN115009310A (en) * | 2022-06-24 | 2022-09-06 | 中国铁道科学研究院集团有限公司 | Self-walking railway tunnel evaluation and maintenance platform and method |
-
2022
- 2022-10-12 CN CN202211246860.4A patent/CN115630492B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107132539A (en) * | 2017-05-03 | 2017-09-05 | 中国地质科学院探矿工艺研究所 | Landslide early-stage identification method of time sequence InSAR (interferometric synthetic Aperture Radar) based on small baseline set |
| CN110361453A (en) * | 2019-07-24 | 2019-10-22 | 四川幔壳科技有限公司 | Array dystopy based on non-destructive testing receives ultrasound synthesis aperture focus method |
| WO2021068848A1 (en) * | 2019-10-09 | 2021-04-15 | 山东大学 | Tunnel structure disease multi-scale measurement and intelligent diagnosis system and method |
| CN112213724A (en) * | 2020-09-16 | 2021-01-12 | 深圳航天科技创新研究院 | Backward projection imaging method and system based on ground penetrating radar data and storage medium |
| CN113552571A (en) * | 2020-09-25 | 2021-10-26 | 哈尔滨工业大学（威海） | PSM (pulse position modulation) algorithm-based underwater laser induced acoustic SAFT (safety and safety transform) imaging method |
| WO2022147969A1 (en) * | 2021-01-08 | 2022-07-14 | 中国民航大学 | Airport pavement underground structure disease automatic detection method based on deep learning |
| CN114353876A (en) * | 2022-01-07 | 2022-04-15 | 兰州大学 | Loess highway slope health monitoring method |
| CN114707532A (en) * | 2022-01-11 | 2022-07-05 | 中铁隧道局集团有限公司 | Ground penetrating radar tunnel disease target detection method based on improved Cascade R-CNN |
| 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 |
| CN114897804A (en) * | 2022-04-26 | 2022-08-12 | 云南航天工程物探检测股份有限公司 | Ground penetrating radar tunnel lining quality detection method based on self-supervision learning |
| CN114942436A (en) * | 2022-05-31 | 2022-08-26 | 西安建筑科技大学 | Tunnel lining disease identification method and system based on geological radar and deep learning |
| CN115009310A (en) * | 2022-06-24 | 2022-09-06 | 中国铁道科学研究院集团有限公司 | Self-walking railway tunnel evaluation and maintenance platform and method |
Non-Patent Citations (3)
| Title |
|---|
| 王石磊 等: "铁路运营隧道检测技术综述", 交通运输工程学报, vol. 20, no. 05 * |
| 耿大新;李洪梅;郭俊;梁国卿;: "地质雷达检测隧道衬砌常见病害模型实验研究", 华东交通大学学报, no. 06 * |
| 胡永占;: "InSAR技术对采空区铁路隧道病害分析研究", 铁道工程学报, no. 02 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115630492B (en) | 2023-12-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Automatic detection of moisture damages in asphalt pavements from GPR data with deep CNN and IRS method | |
| Liu et al. | Detection and localization of rebar in concrete by deep learning using ground penetrating radar | |
| Dinh et al. | Automated visualization of concrete bridge deck condition from GPR data | |
| Ji et al. | Deep neural network-based permittivity inversions for ground penetrating radar data | |
| Xiang et al. | Robust cascaded frequency filters to recognize rebar in GPR data with complex signal interference | |
| Wang et al. | GPRI2Net: A deep-neural-network-based ground penetrating radar data inversion and object identification framework for consecutive and long survey lines | |
| CN107390213B (en) | A kind of time lag curve extracting method of the Ground Penetrating Radar record section based on sliding window | |
| Persico et al. | Microwave tomography for processing of GPR data at Ballachulish | |
| Wang et al. | Deep learning-based rebar clutters removal and defect echoes enhancement in GPR images | |
| Chen et al. | A novel image-based approach for interactive characterization of rock fracture spacing in a tunnel face | |
| Sun et al. | Estimating parameters of the tree root in heterogeneous soil environments via mask-guided multi-polarimetric integration neural network | |
| Feng et al. | Subsurface pipes detection using DNN-based back projection on GPR data | |
| CN116778329A (en) | Urban road underground shallow disease detection method, device, equipment and medium | |
| US6980482B2 (en) | Seismic survey system | |
| Xiong et al. | Automatic detection and location of pavement internal distresses from ground penetrating radar images based on deep learning | |
| Vy et al. | Damage localization using acoustic emission sensors via convolutional neural network and continuous wavelet transform | |
| CN107358162B (en) | Polarization SAR image change detection method based on depth stack network | |
| RU2746691C1 (en) | System for intelligent identification of tectonic faults with the help of seismic data on the basis of convolutional neural networks | |
| Abdel-Qader et al. | Comparative study of deconvolution algorithms for GPR bridge deck imaging | |
| Hu et al. | Defects identification and location of underground space for ground penetrating radar based on deep learning | |
| CN115630492B (en) | Tunnel lining defect change characteristic intelligent inversion method, system and storage medium | |
| Zhang et al. | A deep learning method for rough surface clutter reduction in GPR image | |
| Li et al. | 3D ground penetrating radar road underground target identification algorithm using time-frequency statistical features of data | |
| Xiong et al. | Automatic defect detection in operational high-speed railway tunnels guided by train-mounted ground penetrating radar data | |
| Wang et al. | Seafloor classification based on deep-sea multibeam data—Application to the southwest Indian Ridge at 50.47° E |
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 |