Method and system for evaluating grouting effect of fractured surrounding rock of underground cavern

Abstract

The invention discloses a grouting effect evaluation method and system for fractured surrounding rock of an underground cavern. The method comprises the following steps: acquiring measurement-while-drilling parameters and surrounding rock image data in a construction investigation stage of the underground cavern; inputting the measurement-while-drilling parameters into a random forest model to obtain the uniaxial compressive strength of the rock mass; inputting the surrounding rock image data into an improved Unet model to obtain fracture quantization parameters; the improved Unet model integrates a CBAM attention mechanism and a KAN module; and obtaining grouting construction data, inputting the uniaxial compressive strength of the rock mass, the fracture quantification parameters and the grouting construction data into a gradient boosting tree model to obtain unit ash consumption, and evaluating the grouting effect of the fractured surrounding rock through the unit ash consumption. According to the method, the nonlinear relation processing capability of the tree element learner is improved in a gradient manner through the integrated learning model in combination with the random forest, and the accuracy of unit ash consumption and uniaxial compressive strength prediction can be effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of rock mass grouting construction in the project of groundwater-sealed caverns, and particularly relates to a method for evaluating the grouting effect of fractured surrounding rock in underground caverns. Background Art

[0002] The technology of groundwater-sealed oil caverns is a method of excavating underground spaces in rock masses below the groundwater level to store crude oil. This technology realizes water sealing by maintaining the water level in the cavern higher than the petroleum products, preventing petroleum leakage, and is widely popular for its high safety, environmental protection, economy, and concealment. In the design and construction of underground projects, the drill and blast method and the tunnel boring machine (TBM) are two main methods for excavating rock masses in underground projects. The drill and blast method is known for its flexibility and adaptability and is suitable for underground caverns of various shapes and sizes. Before construction, key geological information is obtained through two methods: contact acquisition (such as measurement parameters while drilling: drilling pressure, torque, rotation speed, and drilling speed, etc.) and non-contact acquisition (such as obtaining the fracture information of the surrounding rock and the tunnel face of the excavated cavern using digital images). In addition, with the development of artificial intelligence technology, machine learning and deep learning methods have been used to predict rock mass mechanical parameters and grouting effects, as well as the identification and quantification of rock mass fractures.

[0003] Although the existing technology has made certain progress in the design and construction of groundwater-sealed caverns, there are still some limitations. First, traditional methods for processing data while drilling rely on experience and qualitative judgment, lacking quantitative analysis means, resulting in limited data processing capabilities, making it difficult to achieve real-time update and feedback of data while drilling, thus limiting the accuracy and reliability of prediction results. Second, in non-contact acquisition methods, the efficiency of manually identifying and marking rock mass fracture information is low, the accuracy is poor, and the labor intensity is high. In addition, the grouting process is concealed, and it is difficult to directly observe the distribution of grout, making the evaluation of construction quality and effect complex. Although numerical simulation methods can evaluate the grouting effect, the modeling process is complex and time-consuming, and it does not meet the rapid requirements of on-site construction. In short, the existing technology has problems such as unclear transparent information of rock masses, difficult evaluation of rock mass grouting effects, limitations of numerical simulation methods, high complexity of prediction models, and lack of real-time prediction and decision support in the construction investigation stage. Summary of the Invention

[0004] To solve the above technical problems, the present invention proposes a method for evaluating the grouting effect of fractured surrounding rock in underground caverns to solve the problems existing in the above-mentioned existing technology.

[0005] To achieve the above object, in the first aspect, the present invention provides a method for evaluating the grouting effect of fractured surrounding rock in underground caverns, including:

[0006] Obtaining the measurement parameters while drilling and the surrounding rock image data during the construction investigation stage of the underground cavern;

[0007] Input the measurement-while-drilling parameters into a random forest model to obtain the uniaxial compressive strength of the rock mass;

[0008] Input the surrounding rock image data into an improved Unet model to obtain a binary fracture image; the improved Unet model integrates a CBAM attention mechanism and a KAN module; then, obtain fracture quantification parameters through a fracture geometric feature quantification integration algorithm;

[0009] Obtain grouting construction data, input the uniaxial compressive strength of the rock mass, the fracture quantification parameters, and the grouting construction data into a gradient boosting tree model to obtain the unit ash consumption, and evaluate the grouting effect of fractured surrounding rock through the unit ash consumption.

[0010] Preferably, the measurement-while-drilling parameters include the drilling speed, rotation speed, and rotary torque, and the surrounding rock image data includes the original surrounding rock image and its corresponding binary fracture image.

[0011] Preferably, integrate the measurement-while-drilling parameters and the uniaxial compressive strength of the rock mass into a first data set;

[0012] Integrate the surrounding rock image and its corresponding binary fracture image into a second data set;

[0013] Integrate the uniaxial compressive strength of the rock mass, the fracture quantification parameters, and the grouting construction data into a third data set.

[0014] Preferably, use the first data set to train the random forest model. The random forest model is based on decision trees as base learners and integrates multiple decision trees through the Bagging method;

[0015] Use the third data set to train the gradient boosting tree model. The gradient boosting tree model is based on decision trees as base learners and integrates multiple decision trees through the Boosting method.

[0016] Preferably, use the second data set to train the improved Unet model. The improved Unet model adds a CBAM module after each convolutional block and integrates a KAN module to identify fracture features;

[0017] The loss function of the improved Unet model is a combination of binary cross-entropy loss and Dice loss, and the model is trained through an optimizer.

[0018] Preferably, the process of inputting the surrounding rock image data into the improved Unet model includes:

[0019] Extract fracture features from the surrounding rock image data through the improved Unet model to obtain fracture quantification parameters, where the fracture quantification parameters include fracture length, width, area, and main direction;

[0020] Use morphological closing operation, contour extraction and skeleton extraction algorithms to process and analyze the fracture characteristics.

[0021] Preferably, the grouting construction data includes pre-grouting water permeability, grouting hole sequence, grouting pressure and elevation parameters.

[0022] In a second aspect, the present invention provides an evaluation system for the grouting effect of fractured surrounding rock in underground caverns, comprising:

[0023] A data acquisition module for acquiring the measurement parameters while drilling and the surrounding rock image data during the construction investigation stage of the underground cavern;

[0024] A first prediction module for inputting the measurement parameters while drilling into a random forest model to obtain the uniaxial compressive strength of the rock mass;

[0025] A second prediction module for inputting the surrounding rock image data into an improved Unet model to obtain a binary fracture image; the improved Unet model integrates a CBAM attention mechanism and a KAN module; and then obtains fracture quantification parameters through a fracture geometric feature quantification integration algorithm;

[0026] A third prediction module for acquiring grouting construction data, inputting the uniaxial compressive strength of the rock mass, the fracture quantification parameters and the grouting construction data into a gradient boosting tree model to obtain the unit ash consumption, and evaluating the grouting effect of the fractured surrounding rock through the unit ash consumption.

[0027] In a third aspect, the present invention also discloses a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0028] In a fourth aspect, the present invention also discloses a computer program product, comprising a computer program, and when the computer program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0029] Compared with the prior art, the present invention has the following advantages and technical effects:

[0030] The present invention provides a method for evaluating the grouting effect of fractured surrounding rock in underground chambers. First, the measurement parameters while drilling and the surrounding rock image data during the construction survey stage of the underground chamber are obtained. Secondly, the measurement parameters while drilling are input into a random forest model to obtain the uniaxial compressive strength of the rock mass. Then, the surrounding rock image data is input into an improved Unet model to obtain a binary image of fractures. The improved Unet model integrates the CBAM attention mechanism and the KAN module. Next, fracture quantification parameters are obtained through a fracture geometric feature quantification integration algorithm. Finally, the grouting construction data is obtained, and the uniaxial compressive strength of the rock mass, the fracture quantification parameters, and the grouting construction data are input into a gradient boosting tree model to obtain the unit ash consumption, and the grouting effect of the fractured surrounding rock is evaluated through the unit ash consumption.

[0031] The present invention adopts a prediction model based on ensemble learning. By combining the nonlinear relationship processing capabilities of random forests and gradient boosting tree meta-learners, it can effectively improve the accuracy of predicting the unit ash consumption and the uniaxial compressive strength. This ensemble method makes full use of the advantages of different learners. The random forest has good robustness due to the randomness of its construction process and is not easily overfitted. The gradient boosting tree can adapt to various loss functions through iterative optimization and has high accuracy.

[0032] The present invention provides an efficient and accurate solution for the transparency of rock mass information and the evaluation of grouting effects in the field of underground engineering, significantly enhancing the accuracy and real-time performance of predictions, and providing strong scientific support for engineering decision-making and construction safety. The present invention significantly improves the performance of evaluating grouting effects, can quickly and accurately guide on-site construction decisions, and ensure the economic benefits and engineering stability of grouting. In actual grouting projects, the present invention can provide a reliable method support for estimating the grouting volume in the area to be grouted, has important engineering application value, and due to the high requirements for controlling the seepage water volume of the rock mass in the groundwater-sealed cavern project, the model is not only applicable to specific projects but also expected to be extended to other underground grouting projects. Description of the Drawings

[0033] The drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation of this application. In the drawings:

[0034] Figure 1 is the flowchart of the method according to the embodiment of the present invention;

[0035] Figure 2 is the technical roadmap according to the embodiment of the present invention;

[0036] Figure 3 is the architecture diagram of the fracture recognition model according to the embodiment of the present invention. Detailed Embodiments

[0037] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0038] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0039] Embodiment 1

[0040] As Figure 1-2 shown, a method for evaluating the grouting effect of fractured surrounding rock in an underground chamber is provided in this embodiment, including:

[0041] S1. Obtain the measurement parameters while drilling and the surrounding rock image data during the construction investigation stage of the underground chamber;

[0042] Furthermore, the measurement parameters while drilling include the drilling speed, rotation speed, and rotary torque, and the surrounding rock image data includes the original surrounding rock image and its corresponding binary fracture image.

[0043] Specifically, parameters such as the drilling speed, rotation speed, and rotary torque are obtained through the measurement technology while drilling.

[0044] S2. Input the measurement parameters while drilling into the random forest model to obtain the uniaxial compressive strength of the rock mass;

[0045] As an innovative implementation method, integrate the measurement parameters while drilling and the uniaxial compressive strength of the rock mass into a first data set; use the first data set to train the random forest model, and the random forest model is based on decision trees as the base learners and integrates multiple decision trees through the Bagging method;

[0046] Specifically, the random forest model first constructs multiple decision trees, and each tree is trained based on different random samples and feature subsets to reduce the correlation between the trees. For a single decision tree, its output is obtained by assigning the input data x to the corresponding leaf node and returning the average value of this node, that is

[0047]

[0048] where, w i is the weight of the leaf node, R i is the region of the leaf node, and I(x ∈ R i ) is the indicator function.

[0049] The overall prediction result of the random forest is the average of the prediction values of all decision trees, that is

[0050]

[0051] Where M is the number of decision trees.

[0052] S3. Input the surrounding rock image data into the improved Unet model to obtain a fracture binary image; the improved Unet model integrates the CBAM attention mechanism and the KAN module; then obtain fracture quantization parameters through the fracture geometric feature quantization integration algorithm;

[0053] As an innovative implementation, integrate the surrounding rock image and its corresponding binary fracture image into a second data set; use the second data set to train the improved Unet model, and the improved Unet model adds a CBAM module after each convolutional block and integrates the KAN module to identify fracture features;

[0054] Specifically, the improved Unet model integrates the CBAM and KAN modules to enhance the model's ability to identify rock mass fracture features. Specifically, the CBAM module is added after each convolutional block of the Unet model, and the model's recognition ability is enhanced through the channel and spatial attention mechanisms; the KAN module is further integrated on the basis of the CBAM module, and the recognition accuracy of the model for fracture features under complex backgrounds is improved through the kernel attention mechanism.

[0055] The parameter configuration step includes defining the key parameters for model training to ensure the stability and tunability of model training. Specific parameters include the number of training epochs (Epochs), batch size (Batch_size), optimizer type (Optimizer), learning rate (Lr), and weight decay (Weight_decay), etc. To ensure the repeatability of the experiment, set a random seed to ensure the consistency of data loading and model initialization. This step includes specifying a fixed random seed value for initializing the random number generator.

[0056] Furthermore, the loss function of the improved Unet model is a combination of binary cross-entropy loss and Dice loss, and the model is trained through an optimizer.

[0057] Specifically, according to the parameter configuration, define the loss function to quantify the difference between the model prediction and the actual result. Specifically, select the binary cross-entropy loss with logits (BCEWithLogitsLoss), which is a loss function that combines the binary cross-entropy loss (BCE Loss) and the Dice loss (Dice Loss), and is commonly used in image segmentation tasks. The purpose of this loss function is to combine the advantages of the two loss functions to improve the performance of the model. BCE Loss is a loss function for binary classification problems, and its mathematical formula is:

[0058]

[0059] where N is the number of samples, and y i is the actual label of the i-th sample, taking values of 0 or 1, is the predicted value of the i-th sample, with the value range between (0, 1). This loss function measures the difference between the probability distribution predicted by the model and the true label.

[0060] Dice Loss converts the Dice coefficient into a loss form:

[0061]

[0062] where X and Y represent the prediction result and the true label respectively; |X∩Y| is the intersection area of the predicted bounding box X and the true bounding box Y; |X| is the area of the predicted bounding box X; |Y| is the area of the true bounding box Y. This loss function focuses on the overlapping area between the predicted region and the true region.

[0063] BCEDiceLoss combines the above two loss functions, and its calculation method is usually the weighted sum of the two losses:

[0064] BCEDiceLoss = α × BCE + β × DiceLoss

[0065] where α and β are weight parameters used to balance the influence of the two loss functions. In practical applications, these weights can be adjusted according to the requirements of specific tasks.

[0066] According to the parameter configuration, an image segmentation model is created. The model structure includes an encoder, a CBAM attention mechanism, a KAN module, and a decoder. Specifically, it includes convolutional layers for extracting image features, a self-attention mechanism module for further extracting and fusing features, and a decoder for mapping the features extracted by the encoder and the KAN module back to the image space for pixel-level prediction.

[0067] Furthermore, according to the parameter configuration, an optimizer and a learning rate scheduler are set to dynamically adjust the learning rate and optimize the model training process. Specifically, it includes selecting optimizers such as Adam or SGD, and setting learning rate schedulers such as CosineAnnealingLR or ReduceLROnPlateau.

[0068] Furthermore, the Albumentations library is used to preprocess the data, including random rotation, flipping, and normalization. The dataset is divided into a training set and a validation set. Specifically, it includes enhancing and normalizing the images to improve the generalization ability of the model, and dividing the dataset into a training set and a validation set for model training and performance evaluation.

[0069] Furthermore, BCEDiceLoss is used to calculate the loss and update the model parameters.

[0070] Furthermore, for each training epoch, the following steps are performed: training the model using the training data, calculating the loss and updating the model parameters; evaluating the model performance on the validation set and recording the loss and IoU metrics; adjusting the learning rate according to the strategy of the learning rate scheduler; if the IoU metric on the validation set improves, saving the current best model. The calculation formula of IOU is as follows:

[0071]

[0072] In the formula, |X∩Y| represents the intersection area of the predicted bounding box X and the ground truth bounding box Y; |X∪Y| represents the union area of the two bounding boxes.

[0073] Furthermore, if the IoU metric on the validation set does not improve for multiple consecutive epochs, the early stopping mechanism is triggered to end the training process. Specifically, a threshold is set to determine when to trigger the early stopping mechanism.

[0074] Furthermore, the loss, IoU metric, and performance on the validation set during the training process are recorded in a log file and visualized using TensorboardX. Specifically, key metrics during the training process are recorded for subsequent analysis and model selection, and tools such as TensorboardX are used to visualize the training process to intuitively show the changes in model performance.

[0075] Furthermore, the process of inputting the surrounding rock image data into the improved Unet model includes:

[0076] Extracting fracture features from the surrounding rock image data through the improved Unet model to obtain fracture quantification parameters, where the fracture quantification parameters include fracture length, width, area, and main direction;

[0077] Using morphological closing operation, contour extraction, and skeleton extraction algorithms to process and analyze the fracture features.

[0078] As an innovative implementation, this embodiment realizes the skeleton extraction step through the following algorithm: morphological processing, including erosion and dilation operations, to process the image segmentation result to remove noise points and highlight the fracture continuity, and then applying a skeletonization algorithm such as the iterative erosion method to extract the fracture centerline and using a fracture contour extraction algorithm to extract the contour.

[0079] As an innovative implementation method, the geometric feature quantification steps include: length measurement, where the total length of the crack is obtained by calculating the distances between points on the crack skeleton and accumulating them; dip angle calculation, where the dip angle of the crack relative to the horizontal plane is calculated using the endpoint coordinates of the crack skeleton; area and width statistics, where the area is obtained by pixel counting of the crack area, and the average width is calculated by the width variation of the skeleton.

[0080] In this embodiment, an innovative technical solution for automatically extracting and quantifying cracks is formed by combining the skeleton extraction step, geometric feature quantification step with the CBAM-UKAN model.

[0081] In this embodiment, by adopting advanced deep learning technology, automatic extraction of crack features is achieved, the model training process is optimized, and the repeatability of experimental results is ensured; through detailed performance evaluation and intuitive visualization tools, the efficiency of model training is not only improved, but also overfitting is effectively avoided.

[0082] In this embodiment, rock mass cracks are its key structural features, which are manifested as measurable line segments between discontinuous surfaces on the rock mass surface. Accurately grasping the quantitative characteristics of rock joints and cracks is crucial for surrounding rock grade determination and construction parameter setting. These features include trace length, dip angle, width, area, etc. Traditionally, the drawing of crack trace maps relies on on-site visual observation by engineers and contact tools, which is not only inefficient, but also poses safety risks and is easily affected by personal experience and perspective, resulting in discrimination deviations.

[0083] With the progress of computer vision and deep learning technologies, data sample collection and information analysis methods have been significantly improved. Deep learning networks can achieve end-to-end feature extraction. Compared with traditional image processing methods (such as Otsu algorithm, Canny algorithm, Laplacian algorithm, etc.), the cumbersome steps of manually adjusting pixel thresholds are omitted, the operation is more concise, and the accuracy is higher.

[0084] Please refer to Figure 3 , this embodiment provides an architecture diagram of a crack recognition model to improve the accuracy and efficiency of crack detection in diverse rock masses and complex background environments. The following are the specific implementation steps:

[0085] 1. Model initialization

[0086] Create an instance of the CBAM-UKAN class and set the parameters as follows:

[0087] Input_channels: 3, because the input is an RGB image.

[0088] Img_size: 512, which is the same as the size of the input image.

[0089] Patch_size: 16, determined according to the design of Patch Embedding.

[0090] Embed_dims: [256, 320, 512], defining the embedding dimensions at different stages.

[0091] Drop_rate: 0.1, setting the dropout rate to prevent overfitting.

[0092] Drop_path_rate: 0.1, setting the drop path rate for regularization.

[0093] Norm_layer: Use nn.LayerNorm as the normalization layer.

[0094] 2. Data Preparation

[0095] Collect image data of the rock tunnel working face, ensuring that the size of each image is 512×512 pixels. Preprocess the images, including: Normalization: Scale the pixel values to the range [0, 1]. Data augmentation: Include random rotation, flipping, etc. to increase the generalization ability of the model.

[0096] 3. Data Loading

[0097] Use DataLoader to batch load the preprocessed image data, setting the appropriate batch size Batchsize = 32.

[0098] 4. Model Training

[0099] Define the loss function as cross-entropy loss, select the optimizer as Adam, and set the learning rate to 0.001.

[0100] Train the model for the specified number of epochs, which is 500 times. In each epoch: Input the image data into the UKAN model. Calculate the output of the loss function and perform backpropagation. Update the model weights. Evaluate the model performance on the validation set and save the model weights with the best performance.

[0101] 5. Feature Extraction and Attention Application

[0102] In the encoder stage, extract features and enhance the feature representation through the ConvLayer and CBAM module. In the KAN module, further extract features through the KAN module, where the KANLayer uses the characteristics of Kolmogorov - Arnold Networks for non-linear transformation.

[0103] 6. Decoding and Classification

[0104] In the decoder stage, the spatial resolution of the image is gradually restored through the D_ConvLayer and CBAM modules, and the feature representation is enhanced. Finally, classification is performed through the final layer, and 1x1 convolution is performed using nn.Conv2d to convert the feature map into a class probability map.

[0105] 7. Model Evaluation

[0106] The performance of the model is evaluated on an independent test set, using metrics such as accuracy, recall, and IoU. The prediction results are visualized, and the predicted fracture traces are compared with the actual fracture traces for qualitative analysis of the model performance.

[0107] 8. Closing Operation Filling:

[0108] After fracture identification, the close_skeleton_contour function is used to perform a closing operation on the image to fill small holes and connect small pieces, enhancing the continuity of the fracture traces. Closing operation is a type of morphological operation used to fill small holes inside foreground objects and connect adjacent foreground objects. The closing operation is an operation that first performs dilation and then erosion. Mathematically, the closing operation can be expressed as:

[0109]

[0110] where: A is the original image; ⊕ represents the dilation operation; C is the structuring element, usually a matrix that defines the shape and size of the dilation and erosion operations; represents the erosion operation; B is the image after the closing operation.

[0111] 9. Contour Extraction:

[0112] The processed image contours are extracted using the Find_contours function. Contour extraction is the process of identifying the boundaries of objects in an image. Mathematically, a contour can be regarded as the local maximum of the image gradient (image intensity change). The formula for contour extraction can be expressed as:

[0113] Contours = Find_contours(I, level)

[0114] where: I is the processed image; level is the threshold used to determine which gradient values are considered part of the contour; Contours is the set of extracted contours. The Find_contours function returns the coordinates of all contours in the image.

[0115] 10. Skeleton Extraction:

[0116] The Skeletonize function is used to skeletonize the image and extract the skeleton of the fracture trace. Skeleton extraction simplifies the binary objects in the image (usually the foreground in a binary image) to a single-pixel-wide skeleton. The mathematical principle of skeleton extraction is based on iteratively removing pixels on the boundary of the object until the object completely disappears or reaches a single-pixel width. The Zhang-Suen algorithm is a popular skeleton extraction algorithm, and its iterative process can be expressed as:

[0117] S = {p ∈ A | Thinning(p) = 1}

[0118] Where: A is the original binary image; S is the skeleton image; Thinning(p) is a function that determines whether to remove p based on pixel p and its neighborhood.

[0119] 11. Fracture trace feature extraction:

[0120] The length of the fracture trace is calculated by the Calculate_crack_length function.

[0121] The width of the fracture trace is calculated by the Calculate_crack_width function.

[0122] The direction of the fracture trace is estimated by the Estimate_crack_direction function.

[0123] The area of the fracture trace is extracted by the Extract_crack_areas function.

[0124] S4. Obtain the grouting construction data, input the uniaxial compressive strength of the rock mass, the fracture quantification parameters, and the grouting construction data into the gradient boosting tree model to obtain the unit ash consumption, and evaluate the grouting effect of the fractured surrounding rock through the unit ash consumption.

[0125] As an innovative implementation, integrate the uniaxial compressive strength of the rock mass, the fracture quantification parameters, and the grouting construction data into a third data set. Use the third data set to train the gradient boosting tree model. The gradient boosting tree model is based on decision trees as the base learners and integrates multiple decision trees through the Boosting method.

[0126] The grouting construction data includes the pre-grouting permeability, the grouting hole sequence, the grouting pressure, and the elevation parameters.

[0127] Specifically, the parameters for constructing the grouting effect evaluation dataset include: summarizing the parameters while drilling, surrounding rock images, and grouting construction record sheets through machine learning and deep learning methods to obtain the corresponding uniaxial compressive strength, fracture trace length, width, dip angle, area, pre-grouting permeability, grouting pressure, hole sequence, and elevation parameters. The output parameter is the unit ash consumption, and the established dataset is randomly divided into a training set and a validation set according to a determined ratio.

[0128] Specifically, the gradient boosting tree model adopts a step-by-step optimization method. Starting from the initial model F0(x), it gradually fits the residuals by iteratively training decision trees. At the m-th step, the model F m (x) is the previous step model F m-1 (x) plus the newly trained decision tree h m (x) multiplied by the learning rate γ m , that is

[0129] F m (x) = F m-1 (x) + γ m h m (x)

[0130] After M iterations, the final model is

[0131]

[0132] In each iteration, the parameters θ of the next decision tree are determined by minimizing the loss function L m , that is

[0133]

[0134] where L is the mean square error.

[0135] Furthermore, analyze the performance of the optimized ensemble learning model in evaluating the grouting effect. Through the performance evaluation indicators of the model, including the coefficient of determination (R 2 ), root mean square error (RMSE), and mean absolute percentage error (MAE).

[0136]

[0137] This embodiment provides a systematic and repeatable method for underground engineering rock mass information transparency and rock mass grouting effect evaluation, which can effectively improve the accuracy and real-time performance of rock mass feature prediction.

[0138] This embodiment extracts and evaluates the fracture trace information of the rock tunnel working face, improves the detection accuracy and efficiency of fractures in different rock masses and complex environments, and makes the rock mass information clear and transparent from the inside out by combining the parameters while drilling. After the implementation of this embodiment, the fracture characteristics of the rock tunnel working face can be accurately and efficiently identified and analyzed without manual intervention, providing a scientific basis for the optimization of the grouting construction technology.

[0139] Embodiment 2

[0140] Based on the same inventive concept, this embodiment also provides an evaluation system for the grouting effect of fractured surrounding rock in underground chambers, including:

[0141] A data acquisition module for acquiring the parameters measured while drilling and the surrounding rock image data during the construction survey stage of the underground chamber;

[0142] A first prediction module for inputting the parameters measured while drilling into a random forest model to obtain the uniaxial compressive strength of the rock mass;

[0143] A second prediction module for inputting the surrounding rock image data into an improved Unet model to obtain a binary fracture image; the improved Unet model integrates the CBAM attention mechanism and the KAN module; and then obtains the fracture quantification parameters through a fracture geometric feature quantification integration algorithm;

[0144] A third prediction module for acquiring the grouting construction data, inputting the uniaxial compressive strength of the rock mass, the fracture quantification parameters, and the grouting construction data into a gradient boosting tree model to obtain the unit ash consumption, and evaluating the grouting effect of the fractured surrounding rock through the unit ash consumption.

[0145] The evaluation system for the grouting effect of fractured surrounding rock in underground chambers provided by this embodiment has all the advantages of the evaluation method for the grouting effect of fractured surrounding rock in underground chambers provided by Embodiment 1.

[0146] Embodiment 3

[0147] This embodiment also discloses a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in Embodiment 1 are implemented.

[0148] Embodiment 4

[0149] This embodiment also discloses a computer program product, including a computer program, and when the computer program is executed by a processor, the steps of the method described in Embodiment 1 are implemented.

[0150] The above are only the preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for evaluating the grouting effect of fractured surrounding rock in underground chambers, characterized in that, It includes the following steps: Obtain the measurement parameters while drilling and the surrounding rock image data during the construction survey stage of the underground chamber; Input the measurement parameters while drilling into the random forest model to obtain the uniaxial compressive strength of the rock mass; Input the surrounding rock image data into the improved Unet model to obtain the binary fissure image; the improved Unet model integrates the CBAM attention mechanism and the KAN module; and then obtain the fissure quantification parameters through the fissure geometric feature quantification integration algorithm; Obtain the grouting construction data, input the uniaxial compressive strength of the rock mass, the fissure quantification parameters and the grouting construction data into the gradient boosting tree model to obtain the unit ash consumption, and evaluate the grouting effect of the fissured surrounding rock through the unit ash consumption.

2. The method according to claim 1, wherein: The measurement parameters while drilling include the drilling speed, rotation speed, and rotational torque, and the surrounding rock image data includes the original surrounding rock image and its corresponding binary fissure image.

3. The method according to claim 1, wherein: Integrate the measurement parameters while drilling and the uniaxial compressive strength of the rock mass into the first data set; Integrate the surrounding rock image and its corresponding binary fissure image into the second data set; Integrate the uniaxial compressive strength of the rock mass, the fissure quantification parameters, and the grouting construction data into the third data set.

4. The method according to claim 3, wherein: Use the first data set to train the random forest model, and the random forest model is based on decision trees as base learners and integrates multiple decision trees through the Bagging method; Use the third data set to train the gradient boosting tree model, and the gradient boosting tree model is based on decision trees as base learners and integrates multiple decision trees through the Boosting method.

5. The method according to claim 3, wherein: Use the second data set to train the improved Unet model, and the improved Unet model adds the CBAM module after each convolutional block and integrates the KAN module to identify fissure features; The loss function of the improved Unet model is the combination of binary cross-entropy loss and Dice loss, and the model is trained through an optimizer.

6. The method according to claim 1, wherein: The process of inputting the surrounding rock image data into the improved Unet model includes: Extract the fissure features from the surrounding rock image data through the improved Unet model to obtain the fissure quantification parameters, where the fissure quantification parameters include the fissure length, width, area, and main direction; Use the morphological closing operation, contour extraction, and skeleton extraction algorithms to process and analyze the fissure features.

7. The method according to claim 1, wherein: The grouting construction data includes the pre-grouting water permeability, grouting hole sequence, grouting pressure, and elevation parameters.

8. An evaluation system for the grouting effect of fractured surrounding rock in underground chambers, characterized in that, It includes: A data acquisition module for obtaining the measurement parameters while drilling and the surrounding rock image data during the construction survey stage of the underground chamber; A first prediction module for inputting the measurement parameters while drilling into the random forest model to obtain the uniaxial compressive strength of the rock mass; The second prediction module is used to input the surrounding rock image data into an improved Unet model to obtain a fracture binary image; the improved Unet model integrates a CBAM attention mechanism and a KAN module; and then fracture quantization parameters are obtained through a fracture geometric feature quantization integration algorithm. The third prediction module is used to obtain grouting construction data, input the uniaxial compressive strength of the rock mass, the fracture quantization parameters and the grouting construction data into a gradient boosting tree model to obtain the ash consumption per unit, and evaluate the grouting effect of fractured surrounding rock through the ash consumption per unit.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.

Images