Machine vision-based tunnel face surrounding rock classification method and system

Abstract

The invention discloses a tunnel face surrounding rock classification method and system based on machine vision, and relates to the technical field of tunnel construction, and the method comprises the steps: firstly collecting surrounding rock image data, carrying out the preprocessing of the surrounding rock image data, including feature calibration and data expansion, and obtaining a surrounding rock image data set with a classification label; inputting the surrounding rock image data set with the label into a Swin Transform model, and performing training to obtain a trained surrounding rock classification model; and inputting a to-be-classified tunnel face surrounding rock image into the surrounding rock classification model to obtain an output result of the model, namely a classification result of the to-be-classified tunnel face surrounding rock. According to the method, the stratum information, the joint characteristics, the rock strength and other multi-dimensional data of the surrounding rock can be rapidly extracted, an objective and quantitative basis is provided for judgment of the level of the surrounding rock, the classification accuracy and consistency are remarkably improved, and the safety and efficiency of a tunnel construction interface can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel construction, and more particularly to a method and system for classifying surrounding rocks of a tunnel face based on machine vision. Background Art

[0002] In the complex and critical field of tunnel engineering, the stability and classification of the surrounding rock of the tunnel face are always indispensable elements to ensure construction safety and efficiency. These two aspects are not only directly related to the stability and long-term durability of the tunnel structure, but also have a profound impact on the life safety of construction workers and the progress of the project. However, facing this core issue, the traditional surrounding rock classification method has exposed obvious shortcomings.

[0003] These methods rely too much on manual observation and empirical judgment, which not only means a lot of manpower investment and time consumption, but more importantly, people's subjective judgment is often difficult to avoid differences. Each engineer or technician may have different knowledge background, work experience and judgment criteria, which directly leads to different classification results of the surrounding rock of the same tunnel face. This inaccuracy and inconsistency of classification undoubtedly brings unpredictable safety risks to tunnel construction and hinders the improvement of construction efficiency.

[0004] Therefore, how to extract the multi-dimensional information of surrounding rock more efficiently and accurately is a problem that technical personnel in this field need to solve urgently. Summary of the invention

[0005] In view of this, the present invention provides a method and system for classifying surrounding rock of a tunnel face based on machine vision, which aims to classify the surrounding rock grade, weathering degree and hardness of the surrounding rock.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for classifying surrounding rock of a tunnel face based on machine vision comprises the following steps:

[0008] Step 1, collecting surrounding rock image data;

[0009] Step 2: preprocessing the surrounding rock image data, including feature calibration and data expansion, to obtain a surrounding rock image data set with classification labels;

[0010] Step 3: input the labeled surrounding rock image dataset into the Swin Transformer model for training to obtain a trained surrounding rock classification model;

[0011] Step 4: input the tunnel face surrounding rock image to be classified into the surrounding rock classification model to obtain the output result of the model, which is the classification result of the tunnel face surrounding rock to be classified.

[0012] Optionally, in step 1, the collected surrounding rock image data includes surrounding rock images of different rock properties, surrounding rock grades, weathering degrees, and hardness degrees.

[0013] Optionally, in step 2, the feature calibration process includes determining the rock properties, surrounding rock grade, weathering degree, and hardness of the collected surrounding rock image data.

[0014] Optionally, in step 2, the surrounding rock image data is expanded by rotation.

[0015] Optionally, the rock lithology includes 8 types of rock lithology.

[0016] Optionally, the weathering degree includes four weathering degree labels: "fresh", "slightly weathered", "moderately weathered" and "strongly weathered".

[0017] Optionally, in step 3, the SwinTransformer model is trained using image features as input information and classification labels as output information.

[0018] Optionally, step 5 is also included, generating reports and charts based on the classification results of step 4 to present the analysis results in a visual manner.

[0019] A machine vision-based tunnel face surrounding rock classification system, including:

[0020] An image acquisition module, used for acquiring surrounding rock image data;

[0021] A preprocessing module, used for preprocessing the surrounding rock image data, including feature calibration and data expansion, to obtain a surrounding rock image data set with classification labels;

[0022] A model training module, used for inputting the labeled surrounding rock image data set into the Swin Transformer model for training to obtain a trained surrounding rock classification model;

[0023] The image classification module is used to input the tunnel face surrounding rock image to be classified into the surrounding rock classification model to obtain the output result of the model, which is the classification result of the tunnel face surrounding rock to be classified.

[0024] It can be seen from the above technical solutions that the present invention provides a method and system for classifying surrounding rock at a tunnel face based on machine vision, which has the following beneficial effects compared with the prior art:

[0025] Compared with traditional methods, the present invention significantly reduces the need for manual observation and judgment, and reduces classification inaccuracies and inconsistencies caused by human factors, which not only improves construction efficiency but also greatly reduces the safety risks of construction workers.

[0026] The present invention adopts the Swin Transformer model for training. The model has powerful image feature extraction capabilities and can accurately identify and classify surrounding rock images with different characteristics. This ensures the consistency and accuracy of the classification results and provides a more reliable decision-making basis for tunnel construction.

[0027] Through feature calibration and data expansion, the present invention improves the generalization ability of the model, enabling it to cope with more diverse surrounding rock images, which enhances the practicability of the model and enables it to be applicable to tunnel projects in different geological conditions and construction environments.

[0028] In summary, the present invention can quickly extract the formation information, joint characteristics, rock strength and other multi-dimensional data of the surrounding rock by automatically collecting, processing and analyzing the surrounding rock images of the tunnel face. These data provide an objective and quantitative basis for the determination of the surrounding rock level, significantly improving the accuracy and consistency of classification. In addition, the classification method based on machine vision can also significantly improve the safety and efficiency of the tunnel construction interface, provide important reference data for the design, construction and later maintenance of tunnel engineering, and help to achieve more scientific and reasonable construction decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0030] Figure 1 It is a schematic diagram of the method steps of the present invention;

[0031] Figure 2 This is the structure diagram of the Swin Transformer-Tiny model;

[0032] Figure 3 This is a test example diagram. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] The embodiment of the present invention discloses a method for classifying surrounding rocks of a tunnel face based on machine vision, see Figure 1 , including the following steps:

[0035] Step 1: Collect surrounding rock image data. The collected surrounding rock image data includes surrounding rock images of different rock properties, surrounding rock grades, weathering degrees, and hardness degrees.

[0036] Step 2: preprocess the surrounding rock image data, including feature calibration and data expansion, to obtain a surrounding rock image data set with classification labels.

[0037] The feature calibration process includes determining the rock lithology, rock grade, weathering degree, and hardness of the collected surrounding rock image data, and putting them into the corresponding folder. The rock lithology includes 8 types of rock lithology, and the weathering degree includes four weathering degree labels: "fresh", "slightly weathered", "moderately weathered" and "strongly weathered".

[0038] Data augmentation refers to the expansion of surrounding rock image data through rotation method.

[0039] During the specific implementation process, 882 rock feature sample images were collected, covering 8 types of rock types. The images were expanded to 31,752 images through the rotation method, and a self-made rock feature sample image classification dataset was created in accordance with the ImageNet dataset format. The rock specimens in this dataset have different shapes, diverse perspectives, and are either bright or dark, ensuring the generalization ability of the classification model.

[0040] Eight self-made rock weathering feature sample image classification datasets were created for the eight lithologies studied. The rock images of each lithology contain four weathering degree labels: "fresh", "slightly weathered", "moderately weathered" and "strongly weathered".

[0041] Step 3: Input the labeled surrounding rock image dataset into the Swin Transformer model, see Figure 2 , using image features as input information and classification labels as output information, the Swin Transformer model is trained to obtain a trained surrounding rock classification model;

[0042] Step 4: Input the tunnel face surrounding rock image to be classified into the surrounding rock classification model to obtain the output result of the model, which is the classification result of the tunnel face surrounding rock to be classified. Figure 3 .

[0043] Furthermore, the method further includes step 5, generating reports and charts based on the classification results of step 4 to present the analysis results in a visual manner.

[0044] The embodiment of the present invention further discloses a machine vision-based tunnel face surrounding rock classification system, comprising:

[0045] An image acquisition module, used for acquiring surrounding rock image data;

[0046] A preprocessing module, used for preprocessing the surrounding rock image data, including feature calibration and data expansion, to obtain a surrounding rock image data set with classification labels;

[0047] A model training module, used for inputting the labeled surrounding rock image data set into the Swin Transformer model for training to obtain a trained surrounding rock classification model;

[0048] The image classification module is used to input the tunnel face surrounding rock image to be classified into the surrounding rock classification model to obtain the output result of the model, which is the classification result of the tunnel face surrounding rock to be classified.

[0049] As for the system module disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0050] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0051] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for classifying surrounding rock at a tunnel face based on machine vision, characterized in that: The following steps are involved: Step 1, collecting surrounding rock image data; Step 2: preprocessing the surrounding rock image data, including feature calibration and data expansion, to obtain a surrounding rock image data set with classification labels; Step 3: input the labeled surrounding rock image dataset into the Swin Transformer model for training to obtain a trained surrounding rock classification model; Step 4: input the tunnel face surrounding rock image to be classified into the surrounding rock classification model to obtain the output result of the model, which is the classification result of the tunnel face surrounding rock to be classified.

2. The method for classifying surrounding rock of a tunnel face based on machine vision according to claim 1, characterized in that: In step 1, the collected surrounding rock image data includes surrounding rock images of different rock properties, surrounding rock grades, weathering degrees, and hardness degrees.

3. The method for classifying surrounding rock of a tunnel face based on machine vision according to claim 1, characterized in that: In step 2, the feature calibration process includes determining the rock properties, surrounding rock grade, weathering degree, and hardness of the collected surrounding rock image data.

4. The method for classifying surrounding rock of a tunnel face based on machine vision according to claim 1, characterized in that: In the step 2, the surrounding rock image data is expanded by a rotation method.

5. The method for classifying surrounding rock of a tunnel face based on machine vision according to claim 3, characterized in that: The rock lithology includes 8 types of rock lithology.

6. The method for classifying surrounding rock of a tunnel face based on machine vision according to claim 3, characterized in that: The weathering degree includes four weathering degree labels: "fresh", "slightly weathered", "moderately weathered" and "strongly weathered".

7. The method for classifying surrounding rock of a tunnel face based on machine vision according to claim 1, characterized in that: In step 3, the Swin Transformer model is trained using image features as input information and classification labels as output information.

8. The method for classifying surrounding rock of a tunnel face based on machine vision according to claim 1, characterized in that: The method further includes step 5, generating reports and charts according to the classification results of step 4 to present the analysis results in a visual manner.

9. A machine vision-based tunnel face surrounding rock classification system, characterized in that: include: An image acquisition module, used for acquiring surrounding rock image data; A preprocessing module, used for preprocessing the surrounding rock image data, including feature calibration and data expansion, to obtain a surrounding rock image data set with classification labels; A model training module, used for inputting the labeled surrounding rock image data set into the Swin Transformer model for training to obtain a trained surrounding rock classification model; The image classification module is used to input the tunnel face surrounding rock image to be classified into the surrounding rock classification model to obtain the output result of the model, which is the classification result of the tunnel face surrounding rock to be classified.

Images