CN114972906A

CN114972906A - Soil quality type identification method for excavation surface of soil pressure balance shield

Info

Publication number: CN114972906A
Application number: CN202210479274.8A
Authority: CN
Inventors: 张东明; 傅蕾; 黄宏伟
Original assignee: Tongji University
Current assignee: Tongji University
Priority date: 2022-05-05
Filing date: 2022-05-05
Publication date: 2022-08-30

Abstract

The invention relates to the field of earth pressure balance shield tunnel construction. The method for identifying the soil property type of the excavation surface of the earth pressure balance shield is characterized by comprising the following steps of: step 1, establishing a muck classification system; step 2, establishing a muck identification database; and 3, constructing a muck identification model. The method can scientifically reflect the self properties of the soil body and is suitable for the actual requirements of shield construction, and the constructed residue soil eyesight identification criterion provides a basis for manufacturing a data set label; the deep learning algorithm is utilized to quickly and accurately obtain the front soil property type information of the excavation face by identifying the muck image, and the method has the characteristics of real time and low cost.

Description

Soil quality type identification method for excavation surface of soil pressure balance shield

Technical Field

The invention relates to the field of earth pressure balance shield tunnel construction.

Background

With the continuous acceleration of the urbanization process in China, the mileage and scale of the subway are rapidly expanded, which brings great challenges to the safety of tunnel construction and the control of the influence on the surrounding environment. The shield method is widely applied to the construction of subway tunnels due to the characteristics of high construction speed, small influence on the surrounding environment and high degree of mechanization. However, the construction decision during shield tunneling depends on the knowledge of the front stratum condition, and when the actual soil quality is not consistent with the known condition, great threat can be brought to the construction quality and the construction safety. At present, geological information along the tunnel engineering mastered by constructors is only derived from geological description provided by geological exploration before construction, and geological conditions among drill holes are only obtained by an interpolation method, so that the uncertainty is high. When the actual stratum does not match the stratum information provided in the geological survey report, the tunneling process may cause tunnel quality problems, and even safety accidents may occur in severe cases. Therefore, the system for identifying the soil quality of the excavation surface in real time by using the muck image can ensure that constructors can master the stratum change in time, adjust the construction decision and ensure the safe and stable tunneling of the tunnel.

The method mainly comprises the following steps of aiming at the problem of real-time identification of soil quality of an excavation surface of a shield tunneling machine at present:

the Chinese patent application with the application number of 202110183343.6 provides a stratum characteristic determination method based on shield real-time tunneling parameters, which comprises the steps of pre-classifying stratum conditions according to survey reports before tunnel construction, simultaneously transforming and processing parameters collected by a shield machine in real time, drawing processed indexes into a two-dimensional plane diagram, judging whether new stratum types are generated and updating the quantity of the stratum types. And finally, inputting the standardized parameters into a K-Means algorithm, and outputting the stratum types determined by the corresponding parameters.

The Chinese patent application with the application number of 201911421797.1 provides a method for driving and inverting geology based on earth pressure balance shield machine parameter data. And processing historical operating data of the earth pressure balance shield machine to prepare a data set, learning a random forest model on a training set, and verifying the model on a testing set. And extracting real-time tunneling data of the earth pressure balance shield machine, and inputting the real-time tunneling data into the random forest model to obtain geological condition information.

The chinese patent application with application number 202110152053.5 proposes a real-time soil classification recognition method based on vibration signals. Firstly, a sensor is arranged on a tunneling mechanism to collect vibration signals generated by excavation, then the average shear modulus of a contact soil body is analyzed according to the vibration signals, and finally the analysis result is compared with data in a soil body category database to judge the category of the soil body.

The Chinese patent application with the application number of 202110558201.3 provides a real-time prediction system and a real-time prediction method for the soil quality of an excavation surface in shield tunnel construction, wherein the prediction system comprises a similar engineering data acquisition module, a soil quality information processing module, a construction data processing module, a soil quality predictor construction module and a soil quality prediction module. The soil property predictor construction module adopts a convolutional neural network to construct a first soil property predictor and a second soil property predictor based on a soil property characteristic gray-scale image and a muck image which are obtained by construction data, and combines prediction results of the two predictors to obtain a soil property real-time detection result.

Disadvantages of the prior art

The shield construction parameters are related to factors such as soil property type, shield machine type selection, tunnel burial depth, muck improvement state and the like, so that a geological identification model established by the method for inverting the geological types by utilizing the real-time tunneling parameters can only be used in tunnel engineering with similar construction conditions, and otherwise, the identification effect is difficult to ensure; the method for analyzing soil parameters by using vibration signals collected by the sensors needs to arrange additional sensors, and soil can be classified only based on one parameter index of average shear modulus; in the method for identifying the soil property by utilizing the muck image, the soil property characteristic information corresponding to the image is obtained by analyzing and calculating based on the geological survey result before tunnel construction, the geological survey result has the characteristic that the information is not complete and inaccurate, the soil layer distribution condition among drill holes has great uncertainty, and the inaccuracy of a label directly influences the identification effect of the model.

Disclosure of Invention

Technical problem to be solved by the invention

A muck classification method and a system for identifying soil quality types in real time based on muck monitoring videos are designed, and the problem that a shield driver is difficult to acquire front stratum information is solved. The method can realize real-time and accurate identification of the soil type of the stratum in front of the cutter head, and provides information for subsequent construction decisions such as adjustment of construction parameters and muck improvement schemes for a shield driver.

Objects of the invention

The core of the invention is to develop a muck classification system which can combine the self engineering property and the engineering experience of the soil body and identify the front soil property type in real time according to the muck image shot by the monitoring camera of the shield tunneling machine. By using the method, a residue soil classification result based on shield construction characteristics can be obtained, and residue soil is positioned in real time and front geological types are judged in real time based on a residue video acquired by a monitoring camera. The method is characterized in that: the method is low in cost, automatic and suitable for tunnels tunneled in various soil strata.

Technical scheme of the invention

The method comprises the steps of classifying earth pressure balance shield muck on the basis of shield construction characteristics and carrying out image recognition on the muck on the basis of a deep learning algorithm.

The method determines the soil quality type of the excavation surface according to the identification of the muck, so that the classification of the muck type is determined based on the classification of the soil quality type. Firstly, a muck classification system which simultaneously considers the self engineering characteristics of soil bodies and the shield construction experience is established. And then, collecting muck data, and selecting and processing the data to obtain a muck identification data set of the earth pressure balance shield. And then constructing and training to obtain a muck target detection model. And finally, inputting a muck monitoring video during construction of the earth pressure balance shield to realize real-time identification of muck and obtain soil quality type information of the excavation surface.

The method comprises the following steps:

step 1, establishing a muck classification system

In order to realize the identification of the soil property type during shield tunneling, the first step is to know the property of the shallow stratum engineering in the area where the shield engineering is located. The static Cone Penetration Test (CPT) is an in-situ test method capable of comprehensively reflecting the physical and mechanical properties of soil. By analyzing the statistical characteristic values of the penetration resistance of different stratums, the engineering characteristics of soil bodies of all the stratums can be mastered, and a foundation is laid for residue soil classification.

1.1 firstly collecting geological survey reports of the area where the shield construction is located, arranging a static sounding layering parameter table in the geological survey reports, counting penetration resistance values of all drill holes passing through the stratum, and eliminating abnormal values, namely, obvious unreasonable data.

And 1.2, calculating the statistical characteristic values of the injection resistance indexes of the formations, including the mean value, the standard deviation and the coefficient of variation.

Firstly, carrying out large-class classification according to soil layer names of all layers, such as: clay (CL), Silt (SI), Sand (SA).

And then, performing detailed division according to the average value of the penetration resistance which can reflect the soil engineering properties, such as dividing Clay (CL) into Soft Clay (SC), common clay (NC) and Hard Clay (HC).

According to the method, the strata in a certain area can be combined and classified according to the engineering properties of the strata, and soil property type division suitable for adjusting the soil pressure balance shield parameters is obtained.

1.3 after the soil property types of a certain area are divided, determining the classification of the muck on the basis.

First, whether special soil or confined water which has a large influence on the earth pressure balance shield construction exists or not is considered, and if such a stratum with special properties exists, and the stratum is combined with other strata into a whole when classification is performed according to a penetration resistance value, the stratum should be divided separately. In addition, considering the situation that the excavation surface passes through multiple strata frequently in shield construction, the discharged residue soil can be determined as 'mixed soil'. Therefore, the classification result of the muck classification considering the engineering property of the soil body and the construction experience can be obtained by adjusting on the basis of the classification of the soil property.

1.4 in the process of establishing the data set later, the muck type of each muck picture is marked, so that a criterion for identifying the muck type with attention is constructed based on the visual characteristics of the muck, and the subsequent label making is facilitated. According to the residue soil identification criterion, the residue soil type is determined by adopting the apparent characteristics of four types of residue soil, namely residue soil shape (MS), residue soil section shape (CS), residue soil surface flatness (SC) and residue soil color (MC).

Step 2, establishing a muck identification database

2.1 firstly, widely collecting the muck monitoring video of the area to which the method is applied.

Typically, the surveillance video resolution is 1920 x 1080, preferably not less than 1280 x 720.

The muck monitoring camera is arranged right above or obliquely above the belt conveyor and is aligned with the muck outlet of the spiral muck discharging machine, so that the complete muck discharging process and muck form are guaranteed to be shot. And the lighting equipment is arranged above the soil outlet, and the surface of the lens is kept clean, so that the shot image is bright and clear and is easy to identify.

2.2 the data is preprocessed after the video data is collected.

Firstly, an effective video of a stable unearthing stage is intercepted, and then video frames are extracted at the frequency of one frame per 30 frames (the specific frame extraction frequency can be adjusted according to the advancing speed of the shield tunneling machine). And selecting images according to the principle that the shape of the muck and the image background are changed as much as possible, and removing images shot under the condition that a lens is stained and mosaic bad images and redundant images caused by video blockage.

And 2.3, manually marking the muck image by adopting LabelImg image marking software, framing out the muck body target in each image by using a rectangular frame, and inputting the muck type of each target frame. The label information is in a PASCAL VOC format, and the real target frame position information and the corresponding muck type information of each picture are stored as XML files. In the training set and the verification set, each image must be labeled, and the information recorded therein is called a true value (GT). The true value is provided to the training process of 3.3.4.

2.4 the acquired muck image file and the corresponding tag file are muck data sets, and the muck image file and the corresponding tag file are calculated according to the following steps of 8: 1: the scale of 1 is divided into a training set, a validation set, and a test set.

Step 3, construction of muck recognition model

The construction process of the muck identification model is also a code component and a writing sequence. The invention adopts a convolutional neural network to establish a muck identification model, and the task belongs to a target detection task, so that a target detection network needs to be established. The model is built on a deep learning frame Pyorch and is realized by adopting a python language.

The muck identification model comprises: the system comprises a user-defined data set module, a muck detection network building module, a training process building module, a model testing and evaluation index module and a muck identification result visualization module.

3.1 in custom data set Module

3.1.1 the image data of the dregs and the corresponding label information are required to be uploaded from the data set storage path and are respectively stored in a list.

3.1.2 certain processing and conversion is then performed on the image data and the label data. The image data is converted from PIL format to Tensor format in the shape of (C, H, W), where C represents the number of image channels, H represents the picture height, W represents the picture width, and all pixel values are normalized to between 0-1 divided by 255.

3.1.3 then, before the image is input into the network, certain random image enhancement operation is carried out on the image, five kinds of data enhancement including horizontal turning, grid mask, random cutting, color dithering and fuzzy are carried out on the image with the probability of 0.5, the diversity of database pictures is improved, and the robustness of the model is enhanced.

3.1.4, calculating the mean value and the standard deviation of the three-channel pixel values of the picture in the whole data set, standardizing the picture so as to facilitate the learning of the model, and providing the standard deviation and the mean value for the step 3.3.

3.2 in the muck detection network building module, the network is mainly divided into three parts: backbone network, neck network and top network, provided to step 3.3.

The main function of the backbone network is to extract image features, the basic structure of the backbone network is the combination of a convolutional layer, a batch normalization layer and an activation layer, and the basic structure is stacked to complete the construction of the backbone network. The depth of the network can be increased by adding residual connection in the backbone network, and the capability of extracting the features of the network is effectively improved.

The main functions of the neck network are feature enhancement and fusion, and by way of example and not limitation, the Spatial Pyramid Pooling (SPP) and the path aggregation network (PANet) adopted by the present invention are common modules of the neck network. The SPP module can effectively expand the receptive field and separate the most important context information, and the PANet can enhance the extracted image characteristics by fusing parameters of different levels of the backbone network. The neck network is greatly helpful to the performance improvement of the whole target detection network.

The top network is a detector, and the main function of the detector is to perform final regression prediction on the position and the type information of the muck. The higher resolution output feature map contains more detailed features of the input image that are good for detection of small objects, and the lower resolution output feature map contains coarser features of the input image that are good for detection of large objects.

As an example, the backbone network employs Darknet53, the neck network employs SPP and PANet, and the top network consists of three output scales of target detectors. The pictures are first input into a backbone network consisting of 53 layers of convolution combinations, where a certain number of residual connections are inserted. The residual connection is to solve the problem that when the deep neural network is trained, the network may degenerate after a certain number of layers of the network are reached, that is, the expression capability of the model cannot be improved, but the effect of the model is deteriorated. Each convolution combination can be regarded as a function F, and the input observation value can be used to obtain an output predicted value y ═ F (x). The residual connection is divided into two lines, one line is F (x) in the convolution combination of the input representation function F of the observation value x, the other line directly transmits the observation value x, and the final predicted value is the sum of the output results of the two lines, namely F (x) + x. When the entire residual concatenation is regarded as a function H and the input observation value is x, the predicted value y ═ H (x) ═ f (x) + x. F (x) ═ h (x) -x is the residual, i.e., the difference between the predicted value y and the observed value x. In this way, the next layer connected by the residual error not only contains the information of the previous layer after nonlinear change (convolution combination), but also contains the original information of the previous layer, so that the information can only increase layer by processing, and the performance of the model cannot be reduced due to the increase of the network depth.

After layer-by-layer feature extraction of a backbone network, the feature map already contains high-level semantic information capable of identifying the muck, but many detailed features are not lost, so that the detection of small targets is not facilitated. Therefore, the feature map obtained by backbone network training is input into the neck network for further feature fusion and enhancement.

The feature map at the top end of the backbone network is firstly input into an SPP structure of the neck network, the SPP network performs three kinds of pooling operation of different scales on the feature map and then splices the feature map together in channel dimensions, the problem of repeated extraction of features by a convolutional network can be solved, the speed of generating a detection candidate frame is greatly improved, and the calculation cost is saved. And then, inputting the feature map at the topmost end of the backbone network and the feature maps at two different levels in the middle of the backbone network into the PANet network of the neck network to perform bidirectional fusion from top to bottom and from bottom to top, so that the feature maps have not only deep semantic information but also basic information such as shallow texture, color and the like, the integrity and diversity of features are ensured, and the final prediction effect is improved.

Three feature graphs (a feature graph 1, a feature graph 2 and a feature graph 3) of different levels in the backbone network are respectively input into a final top network detector after fusion enhancement of a neck network, and are respectively subjected to simple convolution combination and regression to obtain final prediction results, namely a prediction output 1, a prediction output 2 and a prediction output 3. The three feature maps with different sizes respectively contain features with different scales, and the targets with different sizes are correspondingly predicted. The feature map of the prediction output 1 is large, contains the most detail information, and is responsible for predicting small target objects; the feature map of the prediction output 3 is small, so that the overall information is easier to distinguish, and the prediction output 3 is responsible for predicting large target objects. The three prediction output results are merged together to obtain the prediction result of the whole network on the picture.

3.3 in building the training Process Module

3.3.1 firstly, a user-defined data set module and a muck detection network building module are imported, a muck data set and a muck detection network are instantiated, and a data loader is built to input a user-defined picture and tag data into a network.

3.3.2 designing the loss function of the network, wherein the loss function of the target detection network used by the method is divided into three parts: the muck Localization loss (Localization loss), the target frame Confidence loss (Confidence loss), and the muck Classification loss (Classification loss) are shown in formula (1).

Loss＝Localization loss+Confidence loss+Classification loss

(1)

The confidence of the target frame represents whether the target frame contains the slag soil body or not and the intersection ratio of the target frame and the real frame when the target frame contains the slag soil body.

3.3.3 during network training, many hyper-parameters need to be set, including the initial value of the learning rate and the changing mode of the learning rate along with the increase of the training cycle number, the parameters of the optimizer and the optimizer, the batch size of the input pictures, the training cycle number, the weight of each item of the loss function, and the like. After the loss function and the hyperparameter are set, the training can be started in step 3.3.4.

3.3.4 each batch of images is input into the network to obtain a prediction result, and the current loss value, that is, the distance between the prediction value and the real value, can be calculated by inputting the prediction value and the real value (GT) in the label of step 2.3 into a loss function. And calculating the derivative of the loss value to all the network parameters, and optimizing and updating the network parameters by using an optimizer, namely a round of training iteration. When all pictures in the training set are input into the network for training in turn, a training cycle is obtained. After finishing a training cycle, inputting the pictures of the verification set into the network in turn to calculate the network precision, and observing the network training condition according to the network precision and using the network training condition as a basis for adjusting the hyper-parameters in the next network training. And when the loss is reduced and converged to a certain stable value, the training can be finished, and the trained network parameters are stored, so that the identification model capable of accurately positioning the muck target and judging the muck type is obtained.

And 3.4, loading the trained model and parameters in the model testing and evaluating index module, and inputting the muck picture to be tested to obtain regressed muck positioning information and classification information. And compiling an evaluation index calculation code, carrying out quantitative evaluation on the detection result of the test set, and evaluating the effect of the detection model.

And 3.5 in a muck identification result visualization module, drawing the positioning information and the category information of the muck body target frame obtained by network regression on an original image to visualize the identification result. And drawing the target frames in the original image by using the position information x0 and y0 of the central points of the target frames and the length and width information h and w obtained by network regression, writing the regressed muck type information and the corresponding confidence coefficient in the upper left corner of each target frame in a character form, and simultaneously drawing and writing the target frames and the type information of different muck types by adopting different colors.

Advantageous effects

1. The invention provides a muck classification method based on soil engineering properties and shield construction experience and a muck visual identification criterion.

2. The muck is the most direct data reflecting the front soil quality type, and the determining factor of the apparent characteristics is the front soil quality condition, so that the front soil quality type information of the excavation surface can be quickly and accurately obtained by identifying muck images by utilizing a deep learning algorithm. And the application range of the muck recognition model updated along with the expansion of the muck image database is wider and wider, and the limit that the model which can only be trained by similar engineering can be used for predicting the soil quality type of a certain engineering does not exist.

3. The method provided by the invention only needs to utilize the original muck monitoring video data on site in the data collection and practical application processes, and does not need to additionally install a sensor or other equipment. The self-reasoning speed of the adopted algorithm is higher than the frame rate of the monitoring video, so the method has the characteristics of real time and low cost.

Drawings

FIG. 1 is a flow chart of the method of the present invention

FIG. 2 is a process for establishing a muck classification system according to the present invention

FIG. 3 is a process of establishing a muck recognition database according to the present invention

FIG. 4 is a schematic view of the image data acquisition of the muck of the present invention

FIG. 5 is a process of constructing a muck recognition model according to the present invention

FIG. 6 is a schematic diagram of a network structure according to an embodiment of the present invention

Detailed Description

The technical solutions provided in the present application will be further described with reference to the following specific embodiments and accompanying drawings. The advantages and features of the present application will become more apparent in conjunction with the following description.

Examples

The method comprises the steps of classifying earth pressure balance shield muck based on shield construction characteristics and carrying out image recognition on the muck based on a deep learning algorithm.

The overall process of the invention is shown in figure 1.

Step 1, establishing a muck classification system

The process of establishing the muck classification system is shown in fig. 2. In order to realize the identification of the soil property type during shield tunneling, the first step is to know the property of the shallow stratum engineering in the area where the shield engineering is located. The static Cone Penetration Test (CPT) is an in-situ test method capable of comprehensively reflecting the physical and mechanical properties of soil. By analyzing the penetration resistance statistical characteristic values of different strata, the engineering characteristics of soil bodies of all the strata can be mastered, and a foundation is laid for residue soil classification.

1.1, firstly, collecting a geological survey report of an area where shield construction is located, arranging a static sounding layering parameter table in the geological survey report, counting penetration resistance values of all drill holes passing through a stratum in excel, and rejecting abnormal values, namely obvious unreasonable data.

And 1.2, calculating the statistical characteristic values of the injection resistance indexes of the formations, including the mean value, the standard deviation and the coefficient of variation. Firstly, carrying out large-class classification according to soil layer names of all layers, such as: clay (CL), Silt (SI), Sand (SA). And then, performing detailed division according to the average value of the penetration resistance which can reflect the soil engineering properties, such as dividing Clay (CL) into Soft Clay (SC), common clay (NC) and Hard Clay (HC). According to the method, the strata in a certain area can be combined and classified according to the engineering properties of the strata, and soil property type division suitable for adjusting the soil pressure balance shield parameters is obtained.

1.3 after the soil property types of a certain area are divided, determining the classification of the muck on the basis. First, whether special soil or confined water which has a large influence on the earth pressure balance shield construction exists or not is considered, and if such a stratum with special properties exists, and the stratum is combined with other strata into a whole when classification is performed according to a penetration resistance value, the stratum should be divided separately. In addition, considering the situation that the excavation surface passes through multiple strata frequently in shield construction, the discharged residue soil can be determined as 'mixed soil'. Therefore, the classification result of the muck classification considering the engineering property of the soil body and the construction experience can be obtained by adjusting on the basis of the classification of the soil property.

Step 2, establishing a muck identification database

The process of establishing the muck identification database is shown in fig. 3.

2.1 the method is applied to the area of the residue soil monitoring video is firstly collected widely, the resolution of the monitoring video is 1920 x 1080, and preferably not less than 1280 x 720. The image data acquisition schematic diagram of the slag soil is shown in fig. 4, and the slag soil monitoring camera is arranged right above or obliquely above the belt conveyor and is aligned with the soil outlet of the spiral soil discharging machine, so that the complete slag discharging process and the slag soil form are ensured to be shot. And the lighting equipment is arranged above the soil outlet, and the surface of the lens is kept clean, so that the shot image is bright and clear and is easy to identify.

2.2 the data is preprocessed after the video data is collected. Firstly, an effective video of a stable unearthing stage is intercepted, and then video frames are extracted at the frequency of one frame per 30 frames (the specific frame extraction frequency can be adjusted according to the advancing speed of the shield tunneling machine). And selecting images according to the principle that the shape of the muck and the image background are changed as much as possible, and removing images shot under the condition that a lens is stained and mosaic bad images and redundant images caused by video blockage.

Step 3, construction of muck recognition model

The construction process of the muck identification model is shown in fig. 5, and is also a code component and a writing sequence. The invention adopts a convolutional neural network to establish a muck identification model, and the task of the model belongs to a target detection task, so that a target detection network needs to be established. The model is built on a deep learning frame Pyorch and is realized by adopting a python language.

3.1 in custom data set Module

3.1.2 certain processing and conversion is then performed on the image data and the label data. The image data is converted from the PIL format to a Tensor format having a shape of (C, H, W), where C represents the number of image channels, H represents the picture height, W represents the picture width, and all pixel values are normalized to between 0-1 by dividing 255.

Fig. 6 is a schematic structural diagram of an embodiment network, the backbone network employs Darknet53, the neck network employs SPP and PANET, and the top network is composed of target detectors with three output scales. The pictures are first input into a backbone network consisting of 53 layers of convolution combinations, where a certain number of residual connections are inserted. The residual error connection is to solve the problem that when the deep neural network is trained, the network may be degraded after the network reaches a certain number of layers, that is, not only the expression capability of the model cannot be improved, but also the effect of the model is deteriorated. Each convolution combination can be regarded as a function F, and the input observation value can be used to obtain an output predicted value y ═ F (x). The residual connection is divided into two lines, one line is F (x) in the convolution combination of the input representation function F of the observation value x, the other line directly transmits the observation value x, and the final predicted value is the sum of the output results of the two lines, namely F (x) + x. When the entire residual concatenation is regarded as a function H and the input observation value is x, the predicted value y ═ H (x) ═ f (x) + x. F (x) ═ h (x) -x is the residual, i.e., the difference between the predicted value y and the observed value x. In this way, the next layer connected by the residual error not only contains the information of the previous layer after nonlinear change (convolution combination), but also contains the original information of the previous layer, so that the information can only increase layer by processing, and the performance of the model cannot be reduced due to the increase of the network depth.

The feature map at the top end of the backbone network is firstly input into an SPP structure of the neck network, the SPP network performs three kinds of pooling operation of different scales on the feature map and then splices the feature map together in channel dimensions, the problem of repeated extraction of features by a convolutional network can be solved, the speed of generating a detection candidate frame is greatly improved, and the calculation cost is saved. And then, inputting the feature map at the top end of the backbone network and the feature maps at two different levels in the middle of the backbone network into the PANET network of the neck network for bidirectional fusion from top to bottom and from bottom to top, so that the feature map has both deep semantic information and basic information such as shallow texture and color, the integrity and diversity of the features are ensured, and the final prediction effect is improved.

Three feature graphs (a feature graph 1, a feature graph 2 and a feature graph 3) of different levels in the backbone network are respectively input into a final top network detector after fusion enhancement of a neck network, and are respectively subjected to simple convolution combination and regression to obtain final prediction results, namely a prediction output 1, a prediction output 2 and a prediction output 3. The three feature maps with different sizes respectively contain features with different scales, and objects with different sizes are correspondingly predicted. The feature map of the prediction output 1 is large, contains the most detail information, and is responsible for predicting small target objects; the feature map of the prediction output 3 is small, so that the overall information is easier to distinguish, and the prediction output 3 is responsible for predicting large target objects. The three results output by prediction are merged together to obtain the prediction result of the whole network on the picture.

3.3 in building modules of the training Process

3.3.2 designing loss function of network, the loss function of target detection network used in the method is mainly divided into three parts: the muck Localization loss (Localization loss), the target frame Confidence loss (Confidence loss), and the muck Classification loss (Classification loss) are shown in the formula (1).

Loss＝Localization loss+Confidence loss+Classificaltion loss (1)

3.3.4 each batch of images is input into the network to obtain a prediction result, and the current loss value, that is, the distance between the prediction value and the real value, can be calculated by inputting the prediction value and the real value (GT) in the label of step 2.3 into a loss function. And calculating the derivative of the loss value to all the network parameters, and optimizing and updating the network parameters by using an optimizer, namely a round of training iteration. When all pictures in the training set are input into the network for training in turn, a training cycle is obtained. After finishing a training cycle, inputting the picture turns of the verification set into the network to calculate the network precision, and observing the network training condition according to the network precision and taking the network training condition as the basis for adjusting the hyper-parameters in the next network training. And when the loss is reduced and converged to a certain stable value, finishing the training, and storing the trained network parameters to obtain the recognition model capable of accurately positioning the muck target and judging the type of the muck.

And 3.4, loading the trained model and parameters in a model testing and evaluation index module, and inputting the muck image to be tested to obtain regressed muck positioning information and classification information. And compiling an evaluation index calculation code, carrying out quantitative evaluation on the detection result of the test set, and evaluating the effect of the detection model.

And 3.5 in a muck identification result visualization module, drawing the positioning information and the category information of the muck body target frame obtained by network regression on an original image to visualize the identification result. Using network regressionPosition information x of center point of output target frame ₀ 、y ₀ And drawing target frames by the length and width information h and w in the original image, writing the regressed muck type information and corresponding confidence coefficient in the upper left corner of each target frame in a character form, and simultaneously drawing and writing the target frames and the type information of different muck types by adopting different colors.

The muck classification system and the muck visual discrimination criterion obtained by the method provided by the invention are updated along with the continuous expansion of the collection range of the geological survey report and the continuous expansion of the muck database, so that the muck classification system and the visual discrimination criterion with a correspondingly wider application range are obtained, and meanwhile, the muck identification model with a correspondingly wider application range can be obtained by updating and training the muck identification model. Therefore, the method provided by the invention is theoretically suitable for identifying the soil property type in front of the excavation face of the soil pressure balance shield in all regions.

Innovation point of the invention

1. The invention provides a soil pressure balance shield muck classification method based on soil engineering properties and shield construction experience. The muck type division result obtained by the classification method can be identified in the image through a deep learning algorithm, the actual engineering requirement during shield construction can be met, and a classification basis is laid for subsequent muck image identification.

2. The invention provides a method for identifying the eye force of slag discharging soil. The identification criteria of different muck types are formulated according to four important appearance characteristics of the muck, namely the shape of the muck, the section shape of the muck, the surface flatness of the muck and the color of the muck, so that identification bases are provided for subsequently manufacturing data set labels.

3. The invention provides a method for identifying soil property types of an excavation surface of an earth pressure balance shield, namely, muck image identification is carried out based on deep learning to obtain front soil property type information. The method can carry out model training and real-time identification in application by utilizing the monitoring video data, and has the characteristics of accuracy, rapidness and low cost.

The above description is only illustrative of the preferred embodiments of the present application and is not intended to limit the scope of the present application in any way. Any changes or modifications made by those skilled in the art based on the above disclosure should be considered as equivalent effective embodiments, and all the changes or modifications should fall within the protection scope of the technical solution of the present application.

Attached: interpretation of terms:

formation: the term "rock layer" is used to refer to either consolidated rock or unconsolidated sediment, i.e., soil, as a layer or group of rock (soil) layers having certain uniform characteristics and properties and distinct from the upper and lower layers.

Soil texture: refers to the structure and properties of the soil. Also refers to the quality and structure of the soil properties.

Geological exploration: the geological conditions of rocks, stratum structures, mineral products, underground water, landforms and the like in a certain area are investigated and researched by applying geological exploration methods such as mapping, geophysical exploration, geochemical prospecting, drilling, pit exploration, sampling test, geological remote sensing and the like.

Static cone penetration test: the static cone penetration test is that a conical probe is pressed into the soil at a certain speed by static pressure, the penetration resistance (including cone head resistance and side wall friction resistance or friction resistance ratio) is measured, and the soil layer is divided according to the resistance to determine the engineering property of the soil.

Statistical characteristic value: the method refers to representative quantity characteristics which are obtained by sorting the original data of statistical investigation and can accurately describe the distribution of statistical data.

Tunneling: an engineering building buried in the ground is a form in which people utilize underground space.

Subway: the rail transit which is built in cities and has high speed, large traffic volume and electric traction is built in tunnels.

A shield method: a tunnel construction method is characterized in that a shield machine is used for excavating stratums and splicing tunnel segments.

The shield machine: a construction machine comprises a shell, a cutter head, pushing equipment, assembling equipment and other matched equipment, wherein the shell is a cylinder and plays a role in protection, and the other equipment is arranged inside the shell.

Earth pressure balance shield: the shield machine is characterized in that a soil body cut by a front end cutter head in the shield propelling process is used for filling a soil cabin, and the passive soil pressure of the soil body is basically balanced with the soil pressure and the water pressure on a digging and cutting surface, so that the digging and cutting surface and the shield surface are in a balanced state.

Earth pressure balance shield muck: digging waste soil when the earth pressure balance shield is pushed.

Cutter head: the shield machine is used for cutting the equipment of stratum, is located the shield machine front end, cuts off soil through rotatory extrusion.

A palm surface: and excavating a working surface with the gallery continuously pushed forwards.

Spiral unearthing machine: the spiral member rotates to convey soil cut by the cutterhead to a machine on the belt conveyor, and the rotation speed of the spiral member determines the soil discharging speed.

A belt conveyor: and a belt device for conveying the soil from the spiral soil discharging machine to the residue soil vehicle.

Shield construction parameters: various parameters required to be set in the construction process of the shield machine, such as the rotating speed of a cutter head and the like, and whether the parameter setting is reasonable or not determines the safety and the quality of the shield construction.

Artificial intelligence: a new technology science for researching and developing theories, methods, technologies and application systems for simulating, extending and expanding human intelligence.

Machine learning: a multi-field cross discipline relates to a plurality of disciplines such as probability theory, statistics, approximation theory, algorithm complexity theory and the like, is the core of artificial intelligence, and is a fundamental approach for enabling a computer to have intelligence.

Deep learning: the method is a new research direction in the field of machine learning, the intrinsic rules and the expression levels of sample data are learned, and the obtained information is greatly helpful for explaining data such as characters, images and sound.

Target detection: target detection is an important application of deep learning, namely, objects in pictures are identified, and the positions of the objects are marked.

Sample preparation: the basic unit of the data used in the machine learning method may be a one-dimensional matrix or a high-dimensional matrix.

Labeling: the actual information of the data.

LabelImg: and the target detection task is a tool for labeling the data set.

Data set: the collection of training, validation, and test sets is referred to as a data set, which includes data and labels.

Training set: are data samples of model fitting for tuning the neural network.

And (4) verification set: the method is a sample set which is set aside in the model training process and can be used for adjusting the hyper-parameters of the model and carrying out preliminary evaluation on the capability of the model so as to check the training effect.

And (3) test set: used to evaluate the generalization ability of the final model. But not as a basis for algorithm-related selection of parameters, selection features, and the like. It is used to test the actual learning capabilities of the network.

A convolutional neural network: the method is a feedforward neural network which comprises convolution calculation and has a deep structure, and is one of representative algorithms of deep learning.

An image channel: in the RGB color mode, three channels of red, green and blue are indicated.

Image enhancement: the image is matched to the visual response characteristics by adding some information or transforming data to the original image by some means to selectively highlight features of interest in the image or to suppress (mask) some unwanted features in the image.

Grid mask: and generating a grid with the same resolution as the original image, wherein the gray area value is 1, the black area value is 0, and multiplying the grid with the original image to obtain an enhanced image, so that the information deletion of a specific area is realized, and the method can be essentially understood as a regularization method.

Color dithering: the new image is generated by randomly adjusting the saturation, brightness and contrast of the original image.

Robustness: the system can still maintain certain performance characteristics under the condition of disturbance or uncertainty.

And (3) rolling layers: the convolutional layer is composed of a plurality of convolution units, and the parameters of each convolution unit are optimized through a back propagation algorithm. The convolution operation aims to extract different input features, the first layer of convolution layer can only extract some low-level features such as edges, lines, angles and other levels, and more layers of networks can iteratively extract more complex features from the low-level features.

Batch standardization layer: batch Normalization (BN) is a technique used to improve the performance and stability of artificial neural networks. It can normalize the input of any layer in the neural network, fixing the mean and variance of the input signal of each layer.

An active layer: and performing activation operation, namely nonlinear transformation on input data, mapping the features to a high-dimensional nonlinear interval for interpretation, and solving the problem which cannot be solved by a linear model.

Receptive field: the feature vector of a certain position in a certain layer of characteristic diagram is calculated by inputting a fixed area of a certain layer in front, and the area is the receptive field of the position. It is common that a certain layer of feature map corresponds to the receptive field of the input image.

Characteristic diagram: and obtaining a layer after the image is subjected to feature extraction of the convolution layer.

Loss function: and the method is used for evaluating the degree of difference between the predicted value and the actual value of the model.

And (3) hyper-parameter: the hyper-parameter is a parameter that is set before the learning process is started, and is not parameter data obtained by training. In general, the hyper-parameters need to be optimized, and a set of optimal hyper-parameters is selected for the model, so as to improve the learning performance and effect.

Learning rate: is the step size at which the model parameters are updated each time.

Batch size: number of samples input to the network per training.

Claims

1. A method for identifying soil property types of an excavation surface of a soil pressure balance shield is characterized by comprising the following steps:

step 1, establishing a muck classification system

1.1 collecting geological survey reports of the area where shield construction is located, sorting a static sounding layering parameter table in the geological survey reports, counting penetration resistance values of all drill holes passing through a stratum, and eliminating abnormal values, namely obvious and unreasonable data;

1.2 calculating the statistical characteristic values of the injection resistance indexes of each stratum, including a mean value, a standard deviation and a variation coefficient; firstly, carrying out large-class classification according to soil layer names of all layers; dividing in detail according to the average value of the penetration resistance which can reflect the engineering property of the soil body;

according to the method, strata in a certain area are combined and classified according to the engineering properties of the strata, and soil property type division suitable for adjusting the soil pressure balance shield parameters is obtained;

1.3 after the soil property types of a certain area are divided, determining the classification of the muck on the basis;

firstly, considering whether special soil or confined water and other conditions which have great influence on the construction of the earth pressure balance shield exist, if the stratum with special properties exists, and the stratum and other strata are combined into one when the stratum is classified according to the penetration resistance value, the stratum should be divided separately; in addition, considering the condition that the excavation surface passes through a plurality of layers of stratums when shield construction is carried out, the discharged muck can be determined as a special type of 'mixed soil'; therefore, the classification result of the muck classification considering the engineering property of the soil body and the construction experience can be obtained by adjusting on the basis of the classification of the soil property;

1.4 constructing a criterion for identifying the type of the muck with the eyesight based on the visual characteristics of the muck, so as to facilitate the manufacture of subsequent labels;

step 2, establishing a muck identification database

2.1, widely collecting muck monitoring videos of an area to which the method is applied;

2.2 preprocessing the data after collecting the video data;

intercepting effective videos at a stable unearthing stage, and extracting video frames at a frequency of one frame per 30 frames, wherein the specific frame extraction frequency can be adjusted according to the propelling speed of the shield tunneling machine; selecting images according to the principle that the shape of the muck and the image background are changed as much as possible, and removing images shot under the condition that a lens is stained and mosaic poor images and redundant images caused by video blockage;

2.3 manually marking the muck image by adopting LabelImg image marking software, framing out a muck body target in each image by using a rectangular frame, and inputting the muck type of each target frame;

the label information is in a PASCAL VOC format, and the real target frame position information and the corresponding muck type information of each picture are stored as XML files;

in the training set and the verification set, each image must be provided with a label, and the information recorded in the image is called a true value which is provided for the training process of 3.3.4;

2.4, the obtained muck image file and the corresponding label file are the muck data set, and the muck image file and the corresponding label file are calculated according to the following steps of 8: 1: 1, dividing the ratio into a training set, a verification set and a test set;

step 3, construction of muck recognition model

The muck identification model comprises: the system comprises a user-defined data set module, a muck detection network building module, a construction training process module, a model testing and evaluation index module and a muck identification result visualization module;

3.1 in custom data set Module

3.1.1 loading the muck image data and corresponding label information from the data set storage path, and respectively storing the data and the corresponding label information in a list;

3.1.2 then processing and converting the image data and the label data; converting the image data from a PIL format into a Tensor format with the shape of (C, H, W), wherein C represents the number of image channels, H represents the height of the picture, W represents the width of the picture, and dividing all pixel values by 255 to be normalized to be between 0 and 1;

3.1.3, performing certain random image enhancement operation on the image before the image is input into the network, and performing five kinds of data enhancement of horizontal turning, grid mask, random cutting, color dithering and blurring on the image with the probability of 0.5, so as to improve the diversity of database pictures and enhance the robustness of a model;

3.1.4, calculating the mean value and the standard deviation of three-channel pixel values of the picture in the whole data set, standardizing the picture so as to facilitate the learning of the model, and providing the standard deviation and the mean value for the three-channel pixel values of the picture in the whole data set to the step 3.3;

3.2 in the muck detection network building module, the network is divided into three parts: a backbone network, a neck network and a top network, which are provided for step 3.3;

the main function of the backbone network is to extract image features, the basic structure of the backbone network is the combination of a convolutional layer, a batch normalization layer and an activation layer, and the basic structure is stacked to complete the construction of the backbone network; the depth of the network can be increased by adding residual connection in the backbone network, and the capability of extracting the features of the network is effectively improved;

the main function of the neck network is feature enhancement and fusion;

the top network is a detector, and the main function of the detector is to perform final regression prediction on the position and the category information of the residue soil; the output characteristic diagram with higher resolution contains more detailed characteristics of the input image and is good for detecting small targets, and the output characteristic diagram with lower resolution contains coarser characteristics of the input image and is good for detecting large targets;

3.3 in building the training Process Module

3.3.1, firstly, importing a custom data set module and a muck detection network building module, instantiating a muck data set and a muck detection network, and building a data loader to input a custom picture and tag data into a network;

3.3.2 designing the loss function of the network, wherein the loss function of the target detection network used by the method is divided into three parts: the muck positioning loss, the target frame confidence loss and the muck classification loss are shown in formula (1),

Loss＝Localization loss+Confidence loss+Classification loss (1)

the confidence coefficient of the target frame represents whether the target frame contains the slag soil body or not and the size of the intersection ratio of the target frame and the real frame when the target frame contains the slag soil body;

3.3.3 need set up the hyper-parameter when the network trains, include: the initial value of the learning rate and the change mode of the initial value along with the increase of the training cycle number, the parameters of an optimizer and the optimizer, the batch size of input pictures, the training cycle number, the weight of each item of a loss function and the like; after the loss function and the over-parameters are set, the step 3.3.4 can be entered for training;

3.3.4 each batch of images is input into the network to obtain a prediction result, and the current loss value, namely the distance between the prediction value and the true value can be calculated by inputting the prediction value and the true value in the label in the step 2.3 into a loss function; calculating the derivative of the loss value to all the network parameters, and optimizing and updating the network parameters by using an optimizer, namely a round of training iteration; when all pictures in the training set are input into the network for training in turn, a training cycle is formed; after finishing a training cycle, inputting the pictures of the verification set into the network in turn to calculate the network precision, observing the network training condition according to the network precision, and taking the observed network training condition as a basis for adjusting the hyper-parameters in the next network training; when the loss is reduced and converged to a certain stable value, the training can be finished, and the trained network parameters are stored, so that an identification model capable of accurately positioning the muck target and judging the muck type is obtained;

3.4 loading the trained model and parameters in a model testing and evaluating index module, and inputting a muck picture to be tested to obtain regressed muck positioning information and classification information; compiling an evaluation index calculation code, carrying out quantitative evaluation on the detection result of the test set, and evaluating the effect of the detection model;

3.5 in a muck recognition result visualization module, drawing the positioning information and the category information of the muck body target frame obtained by network regression on an original image to visualize the recognition result; and drawing the target frames in the original image by using the position information x0 and y0 of the central points of the target frames and the length and width information h and w obtained by network regression, writing the regressed muck type information and the corresponding confidence coefficient in the upper left corner of each target frame in a character form, and simultaneously drawing and writing the target frames and the type information of different muck types by adopting different colors.

2. The method for identifying the soil property type of the excavation face of the earth pressure balance shield as claimed in claim 1, wherein the muck identification criterion determines the type of muck by using the apparent characteristics of four types of muck, which are respectively the shape of muck, the shape of the cross section of muck, the surface flatness of muck and the color of muck.

3. The method for identifying the soil property type of the excavation face of the earth pressure balance shield as claimed in claim 1, wherein the resolution of the muck monitoring video is 1920 x 1080 and is not lower than 1280 x 720;

the muck monitoring camera is arranged right above or obliquely above the belt conveyor and is aligned with the muck outlet of the spiral muck remover, so that the complete muck removing process and muck form are ensured to be shot; and the lighting equipment is arranged above the soil outlet, and the surface of the lens is kept clean, so that the shot image is bright and clear and is easy to identify.

4. The method for identifying the soil property of the excavation surface of the earth pressure balance shield as claimed in claim 1, wherein the muck identification model is built on a deep learning frame Pythrch and is implemented by adopting python language.

5. The method for identifying the soil property type of the excavation surface of the earth pressure balance shield as claimed in claim 1, wherein in the step 3.2, a space pyramid pooling and path aggregation network adopted by the residue soil detection network building module is a neck network module; the SPP module can effectively expand the receptive field and separate the most important context information, and the PANet can enhance the extracted image characteristics by fusing parameters of different levels of a backbone network; the neck network is greatly helpful for improving the performance of the whole target detection network;

the backbone network adopts Darknet53, the neck network adopts SPP and PANet, and the top network consists of three target detectors with output scales; firstly, inputting pictures into a backbone network, wherein the backbone network is formed by 53 layers of convolution combination, and a certain number of residual error connections are inserted; the residual error connection is used for solving the problem that when the deep neural network is trained, the network is degraded after a certain number of layers of the network is reached, namely, the expression capability of the model cannot be improved, and the effect of the model is poor; each convolution combination can be regarded as a function F, and an output predicted value y is obtained by inputting an observation value; the residual error is connected and divided into two lines, one line is F (x) in the convolution combination of the observation value x input representation function F, the other line directly transmits the observation value x, and the final predicted value is the addition of the output results of the two lines, namely F (x) + x; if the whole residual join is regarded as a function H and the input observation value is x, the predicted value y ═ H (x) ═ f (x) + x; f (x) h (x) -x is the residual, i.e. the difference between the predicted value y and the observed value x; therefore, the next layer connected by the residual errors not only contains the information of the previous layer after nonlinear change (convolution combination), but also contains the original information of the previous layer, so that the information can only increase gradually layer by processing, and the performance of the model cannot be reduced due to the increase of the network depth;

after layer-by-layer feature extraction of a backbone network, the feature map already contains high-level semantic information capable of identifying the muck, but many detailed features are not lost, so that the detection of small targets is not facilitated; therefore, the feature map obtained by backbone network training is input into the neck network for further feature fusion and enhancement;

the feature graph at the top end of the backbone network is firstly input into an SPP structure of a neck network, the SPP network performs three kinds of pooling operations with different scales on the feature graph and then splices the feature graph with channel dimensions, so that the problem of repeated extraction of features by a convolutional network can be solved, the speed of generating a detection candidate frame is greatly improved, and the calculation cost is saved; secondly, inputting the feature map at the topmost end of the backbone network and the feature maps at two different levels in the middle of the backbone network into a PANet network of a neck network for bidirectional fusion from top to bottom and from bottom to top, so that the feature maps have not only deep semantic information but also basic information such as shallow texture and color, the integrity and diversity of features are ensured, and the final prediction effect is improved;

three feature graphs of different levels in a backbone network, namely a feature graph 1, a feature graph 2 and a feature graph 3, are respectively input into a final top network detector after fusion enhancement of a neck network, and are respectively subjected to simple convolution combination and regression to obtain final prediction results, namely a prediction output 1, a prediction output 2 and a prediction output 3; the three feature maps with different sizes respectively contain features with different scales, and the targets with different sizes are correspondingly predicted; the feature map of the prediction output 1 is large, contains most detail information and is responsible for predicting small target objects; the feature map of the prediction output 3 is smaller, so that the overall information is easier to distinguish, and the prediction output 3 is responsible for predicting a large target object; the three results output by prediction are merged together to obtain the prediction result of the whole network on the picture.