CN115114860B - Data modeling amplification method for concrete pipeline damage identification - Google Patents
Data modeling amplification method for concrete pipeline damage identification Download PDFInfo
- Publication number
- CN115114860B CN115114860B CN202210872779.0A CN202210872779A CN115114860B CN 115114860 B CN115114860 B CN 115114860B CN 202210872779 A CN202210872779 A CN 202210872779A CN 115114860 B CN115114860 B CN 115114860B
- Authority
- CN
- China
- Prior art keywords
- concrete pipeline
- damage
- data
- model
- concrete
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 230000003321 amplification Effects 0.000 title abstract description 4
- 238000003199 nucleic acid amplification method Methods 0.000 title abstract description 4
- 238000012549 training Methods 0.000 claims abstract description 32
- 238000012360 testing method Methods 0.000 claims abstract description 28
- 238000001514 detection method Methods 0.000 claims abstract description 25
- 238000012545 processing Methods 0.000 claims abstract description 6
- 238000003062 neural network model Methods 0.000 claims description 33
- 238000012795 verification Methods 0.000 claims description 16
- 238000011156 evaluation Methods 0.000 claims description 15
- 230000009466 transformation Effects 0.000 claims description 15
- 201000010099 disease Diseases 0.000 claims description 10
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 10
- 238000011549 displacement method Methods 0.000 claims description 10
- 238000011176 pooling Methods 0.000 claims description 8
- 230000005653 Brownian motion process Effects 0.000 claims description 5
- 238000005537 brownian motion Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 238000011423 initialization method Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims description 3
- 230000003416 augmentation Effects 0.000 claims description 2
- 238000005070 sampling Methods 0.000 claims description 2
- 238000013519 translation Methods 0.000 claims description 2
- 230000002708 enhancing effect Effects 0.000 claims 1
- 238000013528 artificial neural network Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000003902 lesion Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 241000135164 Timea Species 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000010865 sewage Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
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
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- G—PHYSICS
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/14—Pipes
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Civil Engineering (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Architecture (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Structural Engineering (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a data modeling and amplifying method for concrete pipeline damage identification. The method comprises the following steps: building a concrete pipeline damage model, processing data, constructing a shape ContextNet model, training the model, adjusting parameters of the model, testing the model and detecting the model in real time. The data modeling and amplifying method for concrete pipeline damage identification provided by the invention develops a shape ContextNet neural network algorithm suitable for concrete pipeline damage detection, and performs data amplification by using the data modeling method under the condition that the real concrete pipeline damage data amount is rare, so that the robustness and generalization capability of the method are improved, and the automatic identification of concrete pipeline damage is realized.
Description
Technical Field
The invention relates to the technical field of data modeling, in particular to a data modeling and amplifying method for concrete pipeline damage identification.
Background
With the development of town, the urban drainage system is responsible for urban civilization construction and healthy life of human beings. With the development of social economy and the continuous expansion of cities, the total length of the drainage pipeline in China is rapidly increased, and accordingly, the maintenance work of the drainage pipeline is important. Each year, municipal departments need to spend a great deal of funds and resources to treat pavement collapse caused by drainage pipelines, urban waterlogging and environmental problems caused by sewage, and the aim of drainage pipeline detection is to discover defects of pipelines early, so that the importance of urban drainage pipeline treatment is self-evident.
At present, a detection method of a drainage pipeline is artificial disease identification and detection, and a Closed Circuit Television (CCTV) is used as one of visual inspection systems, and is an existing main flow drainage pipeline detection method, and the detection method is transmitted to an external display through a closed circuit video by means of a pipeline robot carrying high-definition cameras. Compared with the traditional detection method, the television detection does not need to manually enter a pipeline for detection, and can provide a more concise and obvious image result. However, this method of detection requires a large amount of specialized training personnel to resolve and evaluate the lesions over a long period of time, which can be time and resource intensive. In recent years, structural health analysis using three-dimensional information has become a new trend. The three-dimensional laser scanning has strong anti-interference capability, can be used in dim light and fog environments, is more effective in detecting complex environments in a drainage pipeline, and provides ideas for other detection due to the appearance of three-dimensional information, such as calculating depth, area and volume information of damage according to the three-dimensional information of a structural damage part.
However, the existing damage data of the drainage pipeline is insufficient, and the intelligent detection progress of the drainage pipeline is seriously affected. Therefore, it is necessary to propose a data modeling and amplifying method for identifying damage of concrete pipelines so as to solve the above problems.
Disclosure of Invention
The invention aims to provide a data modeling and amplifying method for concrete pipeline damage identification, which aims to solve the problem that the existing drainage pipeline damage data is insufficient and the intelligent detection progress of the drainage pipeline is seriously affected.
The invention provides a data modeling and amplifying method for concrete pipeline damage identification, which comprises the following steps:
establishing a concrete pipeline damage rough inner surface based on a random midpoint displacement method in a fractional Brownian motion theory, and combining the concrete pipeline damage rough inner surface to a concrete pipeline model to obtain the concrete pipeline damage model;
combining real concrete pipeline damage data acquired by using a depth camera with the concrete pipeline damage model, classifying and labeling a concrete pipeline damage data set, amplifying the concrete pipeline damage data set by using a data enhancement method, and dividing the concrete pipeline damage data set into a training set, a verification set and a test set according to a proportion, so as to construct an amplified concrete pipeline damage data set;
building a shape context Net neural network model, wherein the shape context Net neural network model mainly comprises a shape context layer, an MLP layer and a maximum pooling layer;
initializing the shape ContextNet neural network model by adopting a Lyco initialization method, setting super parameters, importing a training set into the shape ContextNet neural network model, and training the model;
setting different super parameters, introducing a verification set to verify the shape context Net neural network model, comparing the loss value and the change curve of the accuracy of the shape context Net neural network model under different super parameters, and searching for the optimal super parameters;
testing the shape context Net neural network model under the optimal super parameters according to the test set, outputting each numerical evaluation index, and judging whether the expected value is reached;
and inputting an image obtained by shooting damage diseases of the concrete pipeline by using the pipeline robot into a shape ContextNet neural network model reaching an expected value, and judging whether an output evaluation index can reach a required requirement.
Further, based on a random midpoint displacement method in a fractional brownian motion theory, a rough inner surface of a concrete pipeline damage is established, and the rough inner surface of the concrete pipeline damage is combined to a concrete pipeline model to obtain the concrete pipeline damage model, which comprises the following steps:
according to a random midpoint displacement method, utilizing matlab to establish a damaged rough inner surface of the concrete pipeline;
building a concrete pipeline model by using comsol;
and embedding the generated point cloud of the damaged rough inner surface of the concrete pipeline into a concrete pipeline model.
Further, combining real concrete pipeline damage data acquired by using a depth camera with the concrete pipeline damage model, classifying and labeling a concrete pipeline damage data set, amplifying the concrete pipeline damage data set by using a data enhancement method, and dividing the concrete pipeline damage data set into a training set, a verification set and a test set according to a proportion, and constructing an amplified concrete pipeline damage data set, wherein the method comprises the following steps of:
uniformly sampling a concrete pipeline damage model by using CloudCompare to generate three-dimensional point cloud data;
establishing a data set by using a concrete pipeline damage model point cloud and a concrete pipeline real damage point cloud according to the ratio of 3:1;
classifying and marking concrete pipeline damage by using a CloudCompare marking program, wherein the concrete pipeline damage is classified into normal and damaged;
amplifying the marked data set by using a data enhancement method;
the dataset was classified as 6 using the program python: 3: the scale of 1 is divided into a training set, a validation set and a test set.
Further, in the step of amplifying the labeled dataset using a data enhancement method, the data enhancement method includes translation transformation, scaling, rotation transformation, symmetry transformation, stretching transformation, and dithering transformation.
Further, the dataset was written using the python classification program as 6:3:1 is divided into a training set, a verification set and a test set, wherein images of the training set, the verification set and the test set are not overlapped in a crossing way.
Further, a shape context net neural network model is built, the shape context net neural network model mainly comprises a shape context layer, an MLP layer and a maximum pooling layer, wherein the shape net layer is used for integrating local and whole information of the three-dimensional point cloud and mapping the characteristics to high dimensions, the MLP layer is used for extracting the high-dimensional characteristics and reducing dimensions, and the maximum pooling layer is used for carrying out dimension reduction processing on the three-dimensional point cloud characteristics under the condition that the symmetrical function is used for guaranteeing that the characteristics of the extracted three-dimensional point cloud are unchanged.
Further, a method of using a Lycopi initialization is adopted to initialize the ShapcontextNet neural network model, super parameters are set, a training set is imported into the ShapcontextNet neural network model, and in the step of training the model, the method of using a statistical principle to solve the problems that the variance of parameters is reduced and gradient disappears as layers deepen is adopted by the Lycopi initialization.
Further, setting different super parameters, introducing a verification set to verify the shape context neural network model, and comparing the loss value and the change curve of the accuracy of the shape context neural network model under different super parameters, wherein the super parameters comprise a learning rate, a total iteration number, a three-dimensional point cloud number and an attenuation rate in the step of searching the optimal super parameters.
Further, testing the shape context neural network model under the optimal super parameters according to the test set, and outputting various numerical evaluation indexes including precision, accuracy and recall rate in the step of judging whether the expected value is reached.
Further, inputting an image obtained by shooting damage diseases of a concrete pipeline by using a pipeline robot into a shape ContextNet neural network model reaching an expected value, and judging whether an output evaluation index can reach a required step, wherein the evaluation index comprises accuracy, detection efficiency and an IOU value.
The beneficial effects of the invention are as follows: the data modeling and amplifying method for concrete pipeline damage identification provided by the invention adopts a three-dimensional point cloud mode, can intuitively express the damage shape, provides more intuitive size information such as depth, area and volume information for engineers, and provides information for the subsequent repair of drainage pipelines; the method of combining the ShapcontextNet neural network with the data modeling and amplifying is adopted to automate the concrete pipeline damage detection and segmentation process, and the robustness of the network model is improved while the concrete pipeline damage detection efficiency is improved.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings that are needed in the embodiments will be briefly described below, and it will be obvious to those skilled in the art that other drawings can be obtained from these drawings without inventive effort.
FIG. 1 is a flow chart of a data modeling augmentation method for identifying damage to a concrete pipe.
Fig. 2 is a schematic diagram of the random midpoint displacement method of the present invention.
FIG. 3 is a general structure diagram of a data modeling and amplifying method for identifying damage of a concrete pipeline.
Fig. 4 is a schematic diagram of a processing result of a data modeling and amplifying method for identifying damage of a concrete pipeline.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to specific embodiments of the present invention and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention. The following describes in detail the technical solutions provided by the embodiments of the present invention with reference to the accompanying drawings.
Referring to fig. 1, the data modeling and amplifying method for identifying damage of a concrete pipeline of the present invention includes the following steps:
step one: and (6) building a concrete pipeline damage model.
The method comprises the steps of establishing a concrete pipeline damage rough inner surface based on random midpoint displacement in a fractional Brownian motion theory, and combining the rough inner surface to a concrete pipeline model, wherein the concrete pipeline damage rough inner surface comprises the following specific processes:
1.1 building a damaged rough inner surface of a concrete pipeline by using matlab according to a random mid-site displacement method.
1.2A concrete pipe model was built using comsol.
1.3 embedding the generated rough inner surface point cloud of the concrete pipeline damage into a concrete pipeline model.
Specifically, as shown in FIG. 2, the random midpoint displacement method is first established by 4 corner points A 0 ,B 0 ,C 0 D (D) 0 Determining the construction area of the fracture surface, wherein the height value obeys the initial coordinates of the fracture surface, the average value is 0, and the variance is delta 2 Is a gaussian distribution random function N (0, delta) 2 ) Linear interpolation is carried out according to the average value of 4 angular points to obtain a central point A 1 Then linearly interpolating the midpoints of the adjacent angular points according to the adjacent angular points, B 1 、C 1 、D 1 、E 1 At the same time add compliance toA random value. Wherein the variance is
Delta in 2 H is the Hurst index, H is more than or equal to 0 and less than or equal to 1, and is the initial standard deviation. Repeating the above process, and obtaining the second interpolation result as A 2 、B 2 、C 2 、D 2 、E 2 Interpolation is performed n times, one at a timeA random value. Wherein the variance is
n is the number of interpolations.
Specifically, by modifying the size control of H, a differently damaged roughened inner surface is generated.
Specifically, a concrete drainage pipeline model with the diameter of 120cm is arranged for matching with real pipeline damage data, and modeling is performed by using triangle, square and round drainage pipeline damage shapes, so that the robustness of the model is improved.
Step two: and (5) preprocessing data.
The real concrete pipeline damage data acquired by using a depth camera are combined with the concrete pipeline damage model obtained in the step one, a concrete pipeline damage data set is classified and marked, the concrete pipeline damage data set is amplified by using a data enhancement method and is divided into a training set, a verification set and a test set according to a proportion, and the concrete pipeline damage data set is constructed by the following specific processes:
2.1 the concrete pipeline damage model was sampled uniformly using cloudcomputer software to generate three-dimensional point cloud data.
2.2, establishing a data set by using the concrete pipeline damage model point cloud and the concrete pipeline real damage point cloud according to the ratio of 3:1.
2.3 classification of concrete pipe damage using the cloudCompare labeling program, classified as normal and broken.
2.4 amplifying the marked data set by using a data enhancement technology.
2.5 data sets were classified as 6 using the write python classification program: 3: the scale of 1 is divided into a training set, a validation set and a test set.
Specifically, the acquired real lesion image should be provided with:
the images should be of a variety and the concrete pipe dataset should include images of light intensity, shadows, etc.
The image shooting distance has universality, the shooting distances are different, and the generalization capability is improved.
The image viewing angle should be diversified, including a front view image, a side view image, a skew view image, and the like.
Disease targets should be representative, with obvious disease characteristics.
In particular, the main methods of data enhancement are translational transformation, scaling, rotational transformation, symmetric transformation, stretching transformation, dithering transformation.
Specifically, the images of the training set, the verification set and the test set are not overlapped mutually, which is beneficial to the robustness and generalization capability of the test model.
Step three: and constructing a shape ContextNet model.
The model mainly comprises a shape context layer, an MLP layer and a maximum pooling layer.
Specifically, the shape net layer is used for integrating local and whole information of the three-dimensional point cloud, mapping the features to high dimensions, the MLP layer is used for extracting the high-dimensional features and reducing dimensions, and the maximum pooling layer performs dimension reduction processing on the three-dimensional point cloud features on the basis that the symmetric functions are used for guaranteeing that the features of the extracted three-dimensional point cloud are unchanged.
Specifically, as shown in fig. 3, the shape context layer mainly includes three modules of selecting, aggregating and converting, where the selecting module generates an affinity matrix according to the three-dimensional point cloud neighborhood information, thereby representing the relationship between a certain point in the three-dimensional point cloud and other points, and establishes a shape context kernel by means of spherical coordinates, and L bins introduced by the shape context kernel generate L affinity matrices, where each matrix corresponds to a specific bin. The result of the aggregation operation is a aggregated feature set, so the tensor output is of the shape nxl x D. (N points, L is the number of bins, and D is the vector dimension after each bin aggregation). The conversion operation is implemented by a convolutional layer having an L,1 kernel that aggregates the L feature point sets and projects them into the output (high-dimensional) feature vector.
Step four: and (5) model training.
Initializing a model by adopting a Lycopi initialization method, setting super parameters, importing a training set into a shape ContextNet neural network model, and training the model.
Specifically, the jersey initialization applies a large number of statistical principles to solve the problem that the parameter variance is reduced and the gradient disappears along with the deepening of the layer, so that the training efficiency can be accelerated and the detection precision can be improved.
Specifically, deep learning algorithms have to be performed on high performance computers based on the high bandwidth and multi-threaded parallel computing characteristics of GPUs.
Step five: and (5) model parameter adjustment.
Different super parameters are set, verification set data test models are introduced, the loss values and the change curves of accuracy of the models under different super parameters are compared, and the optimal super parameters are found.
Specifically, the main super parameters are learning rate, total iteration times, three-dimensional point cloud number, attenuation coefficient and the like.
Specifically, the learning rate controls the speed of updating the weight parameter. In the model training process, setting an excessive learning rate capacity can lead to the model missing the optimal solution, so that the model has an overfitting phenomenon. Setting too small a learning rate may slow down the learning rate. The learning rate needs to be set empirically and experimentally continuously.
Specifically, the total iteration number refers to the number of times that the whole training set is input to the neural network for training, and when the difference between the test error rate and the training error rate is smaller, the current iteration number can be considered to be appropriate. When the test error rate is reduced and then increased, the iteration number is excessively increased, the iteration number needs to be reduced, and if not, the fitting is easy to occur.
Specifically, the number of the three-dimensional point clouds refers to the number of the point clouds in the input training model, and is determined by counting the number of the three-dimensional point clouds in the data set.
Specifically, the attenuation coefficient is set up to solve the problem that the learning rate is too large or too small, and the learning rate is continuously adjusted in the iterative process.
Specifically, the main basis for the present example to adjust the superparameter is that, when different superparameters are set, the detection performance of the model on the verification set, that is, the recognition efficiency, the maximum value of the accuracy rate rising curve and the smoothness and convergence of the loss value falling curve.
Step six: and (5) model testing.
And testing the optimal model according to the test set data, outputting each numerical evaluation index, and judging whether the expected value is reached.
Specifically, each numerical evaluation index in the step six comprises precision, accuracy and recall rate.
Step seven: and (5) detecting in the field.
And shooting damage diseases of the concrete pipeline by using a robot of the pipeline, transmitting the damage diseases back to a high-performance computer terminal for detection, and judging whether an evaluation index can reach the required requirement.
Specifically, the PhoXi3DScanner dynamic 3D camera uses invisible infrared laser with specific wavelength as a light source, the emitted light is projected on an object through a certain code, the distortion of a returned code pattern is calculated through a certain algorithm to obtain the position and depth information of the object, the three-dimensional point throughput of the instrument is 1600 ten thousand points per second, the resolution reaches 320 ten thousand pixels, and the detection requirement of an underground drainage pipeline is met.
Specifically, the evaluation index mainly includes accuracy, detection efficiency, and IOU value.
Specifically, correct and incorrect disease identification statistics should be performed, whether the accuracy rate reaches more than 85% is calculated, if the accuracy rate does not reach the set requirement, false detection disease data needs to be counted again, and retraining is performed again until each index reaches the expected requirement.
The processing result of the method is shown in fig. 4, wherein fig. 4 (a) is a rough inner surface generated by a random mid-site displacement method, fig. 4 (b) is a concrete pipeline damage modeling diagram, and fig. 4 (c) is a shape context algorithm recognition result diagram.
The rock roughness is represented by using a fractal theory by means of reference to some fluid-solid coupling researches, and a rock rough surface is generated by modeling, so that three-dimensional model data are obtained, and data support can be provided for a deep learning model. The invention combines data modeling and three-dimensional point cloud deep learning, and provides a novel automatic detection method for damage of a concrete drainage pipeline, which can detect the damage under the condition that the number of the three-dimensional point clouds of the damage of the concrete drainage pipeline is small.
According to the embodiment, the data modeling and amplifying method for concrete pipeline damage identification is provided, a shape ContextNet neural network algorithm suitable for concrete pipeline damage detection is developed, the data modeling method is used for data amplification under the condition that the real concrete pipeline damage data amount is rare, the robustness and generalization capability of the method are improved, and therefore automatic concrete pipeline damage identification is achieved.
The same or similar parts between the various embodiments in this specification are referred to each other. The embodiments of the present invention described above do not limit the scope of the present invention.
Claims (10)
1. The data modeling and amplifying method for concrete pipeline damage identification is characterized by comprising the following steps of:
establishing a concrete pipeline damage rough inner surface based on a random midpoint displacement method in a fractional Brownian motion theory, and combining the concrete pipeline damage rough inner surface to a concrete pipeline model to obtain the concrete pipeline damage model;
combining real concrete pipeline damage data acquired by using a depth camera with the concrete pipeline damage model, classifying and labeling a concrete pipeline damage data set, amplifying the concrete pipeline damage data set by using a data enhancement method, and dividing the concrete pipeline damage data set into a training set, a verification set and a test set according to a proportion, so as to construct an amplified concrete pipeline damage data set;
building a shape context Net neural network model, wherein the shape context Net neural network model mainly comprises a shape context layer, an MLP layer and a maximum pooling layer;
initializing the shape ContextNet neural network model by adopting a Lyco initialization method, setting super parameters, importing a training set into the shape ContextNet neural network model, and training the model;
setting different super parameters, introducing a verification set to verify the shape context Net neural network model, comparing the loss value and the change curve of the accuracy of the shape context Net neural network model under different super parameters, and searching for the optimal super parameters;
testing the shape context Net neural network model under the optimal super parameters according to the test set, outputting each numerical evaluation index, and judging whether the expected value is reached;
and inputting an image obtained by shooting damage diseases of the concrete pipeline by using the pipeline robot into a shape ContextNet neural network model reaching an expected value, and judging whether an output evaluation index can reach a required requirement.
2. The data modeling and amplifying method for identifying damage to a concrete pipeline according to claim 1, wherein the method is characterized by establishing a rough inner surface of the damage to the concrete pipeline based on a random midpoint displacement method in a fractional brownian motion theory, combining the rough inner surface of the damage to the concrete pipeline model to obtain the damage model of the concrete pipeline, and comprises the following steps:
according to a random midpoint displacement method, utilizing matlab to establish a damaged rough inner surface of the concrete pipeline;
building a concrete pipeline model by using comsol;
and embedding the generated point cloud of the damaged rough inner surface of the concrete pipeline into a concrete pipeline model.
3. The data modeling and amplifying method for identifying damage to a concrete pipeline according to claim 1, wherein real concrete pipeline damage data acquired by using a depth camera are combined with the concrete pipeline damage model, concrete pipeline damage data sets are labeled in a classified manner, the concrete pipeline damage data sets are amplified by using a data enhancing method and are proportionally divided into a training set, a verification set and a test set, and the amplified concrete pipeline damage data sets are constructed and comprise:
uniformly sampling a concrete pipeline damage model by using CloudCompare to generate three-dimensional point cloud data;
establishing a data set by using a concrete pipeline damage model point cloud and a concrete pipeline real damage point cloud according to the ratio of 3:1;
classifying and marking concrete pipeline damage by using a CloudCompare marking program, wherein the concrete pipeline damage is classified into normal and damaged;
amplifying the marked data set by using a data enhancement method;
the dataset was classified as 6 using the program python: 3: the scale of 1 is divided into a training set, a validation set and a test set.
4. A method of modeling and amplifying data for identifying damage to concrete pipes as defined in claim 3, wherein the step of amplifying the marked data set using a data enhancement method includes translation transformation, scaling, rotation transformation, symmetry transformation, stretching transformation, and shaking transformation.
5. A data modeling augmentation method for concrete pipe damage identification according to claim 3, wherein the data set is formulated using a python classification program according to 6:3:1 is divided into a training set, a verification set and a test set, wherein images of the training set, the verification set and the test set are not overlapped in a crossing way.
6. The data modeling and amplifying method for concrete pipeline damage identification is characterized by comprising the steps of constructing a shape context net neural network model, wherein the shape context net neural network model mainly comprises a shape context layer, an MLP layer and a maximum pooling layer, the shape net layer is used for integrating local and whole information of a three-dimensional point cloud and mapping features to high dimensions, the MLP layer is used for extracting the high-dimensional features and reducing dimensions, and the maximum pooling layer is used for carrying out dimension reduction processing on the three-dimensional point cloud features on the basis that the symmetric functions are used for guaranteeing that the features of the extracted three-dimensional point cloud are unchanged.
7. The data modeling and amplifying method for concrete pipeline damage identification according to claim 1, wherein a method of using a Lycopi initialization is adopted to initialize the ShapcontextNet neural network model, super parameters are set, a training set is imported into the ShapcontextNet neural network model, and in the step of training the model, the method of using a statistical principle to solve the problems that parameter variance is reduced and gradient disappears with layer deepening.
8. The data modeling and amplifying method for concrete pipeline damage identification according to claim 1, wherein different super parameters are set, a verification set is introduced to verify a shape context neural network model, and the loss value and the change curve of accuracy of the shape context neural network model under different super parameters are compared, wherein in the step of searching the optimal super parameters, the super parameters comprise learning rate, total iteration times, three-dimensional point cloud number and attenuation rate.
9. The data modeling and amplifying method for identifying damage to concrete pipelines according to claim 1, wherein the method is characterized in that the shapcontnet neural network model under the optimal super parameters is tested according to a test set, and each numerical evaluation index is output to judge whether the expected value is reached, and in the step of judging whether the expected value is reached, the each numerical evaluation index comprises precision, accuracy and recall.
10. The data modeling and amplifying method for identifying damage to a concrete pipeline according to claim 1, wherein an image obtained by shooting damage to the concrete pipeline by using a pipeline robot is input into a shape context neural network model reaching an expected value, and whether an output evaluation index can reach a required requirement is judged, wherein the evaluation index comprises accuracy, detection efficiency and IOU value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210872779.0A CN115114860B (en) | 2022-07-21 | 2022-07-21 | Data modeling amplification method for concrete pipeline damage identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210872779.0A CN115114860B (en) | 2022-07-21 | 2022-07-21 | Data modeling amplification method for concrete pipeline damage identification |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115114860A CN115114860A (en) | 2022-09-27 |
| CN115114860B true CN115114860B (en) | 2024-03-01 |
Family
ID=83333415
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210872779.0A Active CN115114860B (en) | 2022-07-21 | 2022-07-21 | Data modeling amplification method for concrete pipeline damage identification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115114860B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020119103A1 (en) * | 2018-12-13 | 2020-06-18 | 程琳 | Aero-engine hole detection image damage intelligent identification method based on deep learning |
| CN114758222A (en) * | 2022-03-09 | 2022-07-15 | 哈尔滨工业大学水资源国家工程研究中心有限公司 | Concrete pipeline damage identification and volume quantification method based on PointNet ++ neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018165753A1 (en) * | 2017-03-14 | 2018-09-20 | University Of Manitoba | Structure defect detection using machine learning algorithms |
-
2022
- 2022-07-21 CN CN202210872779.0A patent/CN115114860B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020119103A1 (en) * | 2018-12-13 | 2020-06-18 | 程琳 | Aero-engine hole detection image damage intelligent identification method based on deep learning |
| CN114758222A (en) * | 2022-03-09 | 2022-07-15 | 哈尔滨工业大学水资源国家工程研究中心有限公司 | Concrete pipeline damage identification and volume quantification method based on PointNet ++ neural network |
Non-Patent Citations (1)
| Title |
|---|
| 谢坤明 ; .基于随机森林算法的结构损伤识别.福建交通科技.2020,(05),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115114860A (en) | 2022-09-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109840556B (en) | Image classification and identification method based on twin network | |
| CN111611861B (en) | Image change detection method based on multi-scale feature association | |
| CN114758222B (en) | Concrete pipeline damage identification and volume quantification method based on PointNet ++ neural network | |
| CN101140216A (en) | Gas-liquid two-phase flow type recognition method based on digital graphic processing technique | |
| Hu et al. | Efficient and automatic plane detection approach for 3-D rock mass point clouds | |
| CN110909623B (en) | Three-dimensional target detection method and three-dimensional target detector | |
| CN113159232A (en) | Three-dimensional target classification and segmentation method | |
| CN113569979A (en) | Three-dimensional object point cloud classification method based on attention mechanism | |
| CN114998566A (en) | Interpretable multi-scale infrared small and weak target detection network design method | |
| Wang et al. | Object verification based on deep learning point feature comparison for scan-to-BIM | |
| Deng et al. | Fusing geometrical and visual information via superpoints for the semantic segmentation of 3D road scenes | |
| CN115512247A (en) | Regional building damage grade assessment method based on image multi-parameter extraction | |
| CN115424237A (en) | Forward vehicle identification and distance detection method based on deep learning | |
| CN113822825B (en) | Optical building target three-dimensional reconstruction method based on 3D-R2N2 | |
| EP4174792A1 (en) | Method for scene understanding and semantic analysis of objects | |
| CN115114860B (en) | Data modeling amplification method for concrete pipeline damage identification | |
| Pang et al. | Multi-Scale Feature Fusion Model for Bridge Appearance Defect Detection | |
| CN117011274A (en) | Automatic glass bottle detection system and method thereof | |
| CN109934151B (en) | Face detection method based on movidius computing chip and Yolo face | |
| CN108010076B (en) | End face appearance modeling method for intensive industrial bar image detection | |
| CN114445726B (en) | Sample library establishing method and device based on deep learning | |
| CN113657438B (en) | Drainage pipeline disease detection method of VGG neural network under thermal infrared mode | |
| KR102129060B1 (en) | Content-based 3d model retrieval method using a single depth image, 3d model retrieval server for performing the methods and computer readable recording medium thereof | |
| Jia et al. | Research on Industrial Production Defect Detection Method Based on Machine Vision Technology in Industrial Internet of Things. | |
| CN113205465B (en) | Point cloud data set segmentation method and system |
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 |