Surrounding rock grading model training method, surrounding rock grading method and device

Abstract

The invention provides a surrounding rock grading model training method and a surrounding rock grading method and device, and relates to the technical field of tunneling machine construction. The surrounding rock grading model training method comprises the steps of obtaining multi-modal sample data and corresponding labeling data in a tunneling construction process, wherein the labeling data is a surrounding rock grading result; and training the surrounding rock grading model by using the multi-modal sample data and the annotation data so as to adjust parameters of the surrounding rock grading model.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel boring machine construction, and in particular to a surrounding rock classification model training method, a surrounding rock classification method and a surrounding rock classification device. Background Art

[0002] The applicability, construction efficiency and cost of tunnel construction by tunnel boring machines are significantly affected by surrounding rock conditions, such as the selection of support features, the selection of tunneling parameters, and the project budget. Surrounding rock classification is a commonly used method for characterizing surrounding rock characteristics in domestic drilling and blasting tunnel construction. Surrounding rock is divided into categories I to V according to different characteristics. Due to the narrow space between the cutter head and the face of the tunnel boring machine, the surrounding rock conditions of the face cannot be observed in time. Surrounding rock parameters generally have shortcomings such as high cost and cumbersome testing. Due to the influence of economy, convenience, hysteresis, etc., it is difficult to conduct large-scale, rapid, and real-time geological detection. Rapid evaluation of tunnel surrounding rock classification and then selecting appropriate construction methods and support measures are important methods to reduce the impact of poor geological bodies on tunnel boring machine construction. Summary of the invention

[0003] A technical problem to be solved by the present disclosure is to provide a surrounding rock classification model training method, a surrounding rock classification method and a surrounding rock classification device, which can improve the accuracy of surrounding rock classification prediction.

[0004] According to one aspect of the present disclosure, a surrounding rock classification model training method is proposed, including: obtaining multimodal sample data and corresponding labeled data during excavation construction, the labeled data being the surrounding rock classification result; and using the multimodal sample data and the labeled data to train the surrounding rock classification model to adjust the parameters of the surrounding rock classification model.

[0005] In some embodiments, the multimodal sample data includes excavation sample data, slag sample data and vibration sample data, and obtaining the multimodal sample data includes: obtaining the original excavation sample parameters, original slag sample images and original vibration sample data during the excavation construction process; and based on the first excavation time of the tunnel boring machine, fusing the original excavation sample parameters, original slag sample images and original vibration sample data to obtain the excavation sample data, slag sample data and vibration sample data.

[0006] In some embodiments, based on the excavation time of the tunnel boring machine, the original tunnel boring sample parameters, the original slag sample image and the original vibration sample data are fused and processed to obtain the tunnel boring sample data, the slag sample data and the vibration sample data, including: determining the tunnel boring state of the tunnel boring machine according to the original tunnel boring sample parameters; obtaining the first tunnel boring time of the tunnel boring machine according to the tunnel boring state; selecting the first tunnel boring parameter corresponding to the first tunnel boring time in the original tunnel boring sample parameters, selecting the first slag image corresponding to the first tunnel boring time in the original slag sample image, and selecting the first vibration parameter corresponding to the first tunnel boring time in the original vibration sample data; and determining the tunnel boring sample data according to the first tunnel boring parameter, determining the slag sample data according to the first slag image, and determining the vibration sample data according to the first vibration parameter.

[0007] In some embodiments, determining the excavation sample data based on the first excavation parameter includes: extracting the excavation feature data related to the surrounding rock classification in the first excavation parameter; and using the excavation feature data related to the surrounding rock classification in the first excavation parameter as the excavation sample data.

[0008] In some embodiments, determining the residue sample data according to the first residue image includes: extracting parameters characterizing the size characteristics of the residue in the first residue image; and using the parameters characterizing the size characteristics of the residue in the first residue image as the residue sample data.

[0009] In some embodiments, determining the vibration sample data according to the first vibration parameter includes: performing mathematical calculation on the first vibration parameter to obtain a first characteristic parameter characterizing the vibration data; and using the first characteristic parameter characterizing the vibration data as the vibration sample data.

[0010] In some embodiments, the multimodal sample data includes excavation sample data, slag sample data and vibration sample data, wherein obtaining the multimodal sample data includes: obtaining original excavation sample parameters, original slag sample images and original vibration sample data during the excavation construction process; and using the original excavation sample parameters as excavation sample data, the original slag sample images as slag sample data, and the original vibration sample data as vibration sample data.

[0011] According to another aspect of the present disclosure, a surrounding rock classification method is also proposed, including: acquiring multimodal data during the excavation construction process; and inputting the multimodal data into a surrounding rock classification model to perform surrounding rock classification prediction, wherein the surrounding rock classification model is trained based on the above-mentioned surrounding rock classification model training method.

[0012] In some embodiments, the multimodal data includes excavation data, slag data and vibration data, and obtaining the multimodal data includes: obtaining the original excavation parameters, original slag images and original vibration data during the excavation construction process; and based on the second excavation time of the tunnel boring machine, fusing the original excavation parameters, original slag images and original vibration data to obtain the excavation data, slag data and vibration data.

[0013] In some embodiments, based on the second excavation time of the tunnel boring machine, the original excavation parameters, the original slag image and the original vibration data are fused and processed to obtain the excavation data, the slag data and the vibration data, including: determining the excavation state of the tunnel boring machine according to the original excavation parameters; determining the second excavation time of the tunnel boring machine according to the excavation state of the tunnel boring machine; selecting the second excavation parameter corresponding to the second excavation time in the original excavation parameters, selecting the second slag image corresponding to the second excavation time in the original slag image, and selecting the second vibration parameter corresponding to the second excavation time in the original vibration data; and determining the excavation data according to the second excavation parameter, determining the slag data according to the second slag image, and determining the vibration data according to the second vibration parameter.

[0014] In some embodiments, determining the excavation data according to the second excavation parameter includes: extracting the excavation characteristic data related to the surrounding rock classification in the second excavation parameter; and using the excavation characteristic data related to the surrounding rock classification in the second excavation parameter as the excavation data.

[0015] In some embodiments, determining the residue piece data according to the second residue piece image includes: extracting parameters characterizing the size characteristics of the residue piece in the second residue piece image; and using the parameters characterizing the size characteristics of the residue piece in the second residue piece image as the residue piece data.

[0016] In some embodiments, determining the vibration data according to the second vibration parameter includes: performing mathematical calculation on the second vibration parameter to obtain a second characteristic parameter characterizing the vibration data; and using the second characteristic parameter characterizing the vibration data as the vibration data.

[0017] In some embodiments, the multimodal data includes excavation data, slag data and vibration data, and obtaining the multimodal data includes: obtaining the original excavation parameters, original slag images and original vibration data during the excavation construction process; and using the original excavation parameters as excavation data, the original slag images as slag data, and the original vibration data as vibration data.

[0018] According to another aspect of the present disclosure, a surrounding rock classification model training device is also proposed, including: a first data processing module, configured to obtain multimodal sample data and corresponding labeled data during the excavation construction process, the labeled data being the surrounding rock classification result; and a training module, configured to use the multimodal sample data and the labeled data to train the surrounding rock classification model to adjust the parameters of the surrounding rock classification model.

[0019] According to another aspect of the present disclosure, a surrounding rock classification model training device is also proposed, including: a memory; and a processor coupled to the memory, the processor being configured to execute the surrounding rock classification model training method as described above based on instructions stored in the memory.

[0020] According to another aspect of the present disclosure, a surrounding rock classification device is also proposed, including: a second data processing module, configured to obtain multimodal data during the excavation construction process; a prediction module, configured to input the multimodal data into a surrounding rock classification model to perform surrounding rock classification prediction, wherein the surrounding rock classification model is trained based on the above-mentioned surrounding rock classification model training method.

[0021] According to another aspect of the present disclosure, a surrounding rock classification device is also provided, comprising:

[0022] A processor is coupled to the memory, and the processor is configured to execute the surrounding rock classification method as described above based on instructions stored in the memory.

[0023] According to another aspect of the present disclosure, a computer-readable storage medium is further provided, on which computer program instructions are stored, and when the instructions are executed by a processor, the above-mentioned surrounding rock classification model training method or the above-mentioned surrounding rock classification method is implemented.

[0024] According to another aspect of the present disclosure, a computer program product is also proposed, including a computer program or instructions, and when the computer program or instructions are executed by a processor, the above-mentioned surrounding rock classification model training method, or the above-mentioned surrounding rock classification method is implemented.

[0025] Other features and advantages of the present disclosure will become apparent from the following detailed description of exemplary embodiments of the present disclosure with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings, which constitute a part of the specification, illustrate embodiments of the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0027] The present disclosure may be more clearly understood from the following detailed description with reference to the accompanying drawings, in which:

[0028] Figure 1 It is a flowchart of some embodiments of the surrounding rock classification model training method disclosed in the present invention;

[0029] Figure 2 Schematic diagram of the flow chart of other embodiments of the surrounding rock classification model training method disclosed in the present invention;

[0030] Figure 3 Schematic diagram of some embodiments of the surrounding rock classification method disclosed in the present invention;

[0031] Figure 4 Schematic diagrams of the flow charts of other embodiments of the surrounding rock classification method disclosed in the present invention;

[0032] Figure 5 A schematic diagram of the flow chart of some embodiments of surrounding rock classification model training and surrounding rock classification disclosed in the present invention;

[0033] Figure 6 It is a schematic diagram of the structure of some embodiments of the surrounding rock classification model training device disclosed in the present invention;

[0034] Figure 7 It is a schematic diagram of the structure of some other embodiments of the surrounding rock classification model training device disclosed in the present invention;

[0035] Figure 8 It is a schematic diagram of the structure of some embodiments of the surrounding rock classification device disclosed in the present invention;

[0036] Fig. 9 Schematic diagrams of the structures of other embodiments of the surrounding rock grading device disclosed in the present invention. DETAILED DESCRIPTION

[0037] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. It should be noted that the relative arrangement of components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present disclosure unless otherwise specifically stated.

[0038] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0039] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present disclosure, its application, or uses.

[0040] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered as part of the specification.

[0041] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0042] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0043] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0044] In the related art, one modal data is usually used to determine the surrounding rock classification prediction, but the prediction result is not accurate. The present disclosure provides a surrounding rock classification model training method and surrounding rock classification scheme and a corresponding surrounding rock classification model training scheme, which uses multi-modal data to complement the information between them to improve the accuracy of data analysis, and can obtain richer feature information, thereby achieving accurate prediction of surrounding rock classification. The following will introduce the scheme of the present disclosure with a specific embodiment.

[0045] Figure 1 It is a flowchart of some embodiments of the surrounding rock classification model training method disclosed in the present invention, and the embodiment includes steps S11-S12.

[0046] In step S11, multimodal sample data and corresponding annotated data in the excavation construction process are obtained, and the annotated data is the surrounding rock classification result.

[0047] Some clues about the state of the surrounding rock only exist in certain modal data. Different modal data are expressed differently. There is some cross-information redundancy in each modal data, and there may even be multiple different information interactions between modal data. For example, the data information contained in different information sources has both similarities and differences. For example, the excavation parameters reflect the strength of the surrounding rock and equipment load during excavation, the vibration data reflects abnormal impact, surrounding rock strength, equipment operation status, etc., and the slag information reflects the particle size distribution, surrounding rock crushing, etc.

[0048] In some embodiments, the multimodal sample data includes excavation sample data, slag sample data and vibration sample data. This step also requires obtaining surrounding rock classification results at the same time, and then using the surrounding rock classification results as annotation data, and the annotation data is characterized by the exposed formation surrounding rock classification results. The surrounding rock classification results include, for example, Class I surrounding rock, Class II surrounding rock, Class III surrounding rock, Class IV surrounding rock, and Class V surrounding rock. Of course, the surrounding rock can also be divided into more or fewer categories according to actual conditions.

[0049] In step S12, the surrounding rock classification model is trained using the multimodal sample data and the labeled data to adjust the parameters of the surrounding rock classification model.

[0050] For example, deep learning networks such as CNN (Convolutional Neural Networks) and BP (Backpropagation Algorithm) neural networks are used, with multimodal sample data as input and surrounding rock classification as output, to establish a surrounding rock classification model. The training of the surrounding rock classification model is achieved through the optimization of CNN weight parameters and model tuning.

[0051] After the surrounding rock classification model is trained, the multimodal data acquired in real time during the excavation construction process is input into the surrounding rock classification model to obtain accurate surrounding rock classification.

[0052] In the above embodiment, compared with single modality sample data, the multimodal sample data contains richer information, effectively integrates the information advantages between different modality sample data, realizes information complementarity, and thus can train a surrounding rock classification model that can more accurately predict surrounding rock classification.

[0053] In step S11, multimodal sample data is obtained, for example, by Figure 2 The steps shown are implemented, Figure 2 It is a flowchart diagram of other embodiments of the surrounding rock classification model training method disclosed in the present invention, including step S111 and step S112.

[0054] In step S111, original excavation sample parameters, original slag sample images and original vibration sample data in the excavation construction process are obtained.

[0055] In some embodiments, excavation data of the tunnel boring machine that has been constructed is collected, and the excavation data is obtained by using the equipment's own monitoring system; slag images are collected, and the slag images are mainly obtained by collecting the real-time data stream on the belt conveyor by the monitoring system; host vibration monitoring data is collected, which mainly uses vibration sensors deployed at the cutter head, main drive and other positions to obtain real-time vibration data of the equipment.

[0056] The tunnel boring machine is, for example, a TBM (Tunnel Boring Machine).

[0057] The original excavation sample parameters comprehensively reflect the average hardness information of the surrounding rock of the tunnel face, the original slag sample image can reflect the uneven hardness information of the tunnel face, and the original vibration sample data can reflect the strength information of the tunnel face.

[0058] In step S112, based on the first excavation time of the tunnel boring machine, the original excavation sample parameters, the original slag sample image and the original vibration sample data are fused to obtain excavation sample data, slag sample data and vibration sample data.

[0059] The real-time data stream records all data including shutdown, excavation and other processes. When analyzing the surrounding rock classification, the data during excavation needs to be selected.

[0060] In some embodiments, the excavation state of the tunnel boring machine is determined based on the original excavation sample parameters; based on the excavation state, a first excavation time of the tunnel boring machine is obtained; a first excavation parameter corresponding to the first excavation time is selected from the original excavation sample parameters, a first slag image corresponding to the first excavation time is selected from the original slag sample image, and a first vibration parameter corresponding to the first excavation time is selected from the original vibration sample data; and based on the first excavation parameter, the excavation sample data is determined, based on the first slag image, the slag sample data is determined, and based on the first vibration parameter, the vibration sample data is determined.

[0061] For example, an effective excavation process is selected by a data segmentation method, wherein the data segmentation method can be expressed as

[0062] , where N is the cutterhead speed, F is the cutterhead thrust, and D represents the state of the tunneling machine. First, the tunneling process data is screened out, and then the slag data and vibration data corresponding to the tunneling data are selected according to time, thereby realizing time-based data fusion and obtaining effective sample data.

[0063] In some embodiments, excavation characteristic data related to surrounding rock classification in the first excavation parameter is extracted; and the excavation characteristic data related to surrounding rock classification in the first excavation parameter is used as excavation sample data.

[0064] For example, a correlation analysis is performed on the first excavation parameter, and with the surrounding rock classification as the target, the Pearson correlation coefficient method is used to select excavation data that has a greater impact on the surrounding rock classification, so as to achieve feature screening of the excavation data. Alternatively, the selection method based on expert experience can also be used to select excavation feature data. The excavation feature data includes at least one of the cutterhead thrust sample data, the cutterhead torque sample data, and the penetration sample data. Parameters with a high correlation with the surrounding rock classification are selected from a large number of excavation databases, and several groups of data are selected to establish an excavation parameter matrix.

[0065] In some embodiments, parameters characterizing the size characteristics of the residue pieces in the first residue piece image are extracted; and the parameters characterizing the size characteristics of the residue pieces in the first residue piece image are used as the residue piece sample data.

[0066] For example, by using image recognition methods such as YOLO (You Only Look Once, a real-time target detection system) and CNN to identify the particle size, unevenness coefficient, curvature coefficient, etc. of the slag flakes, the size characteristics of the slag flakes are obtained, which can effectively reduce the complexity of the data calculation process. Therefore, the parameters characterizing the size characteristics of the slag flakes in the first slag flake image include at least one of the slag flake particle size sample data, the unevenness coefficient sample data, and the curvature coefficient sample data. This embodiment can perform morphological methods such as corrosion, expansion, opening and closing on the slag flake image to establish high-quality image data.

[0067] In some embodiments, a mathematical calculation is performed on the first vibration parameter to obtain a first characteristic parameter representing the vibration data; and the first characteristic parameter representing the vibration data is used as the vibration sample data.

[0068] For example, the vibration data is first filtered and denoised, and then the mean square amplitude, peak value, mean value, standard deviation, etc. are calculated to establish characteristic parameters characterizing the vibration data, thereby establishing a high-quality vibration database. Therefore, the first characteristic parameter characterizing the vibration data includes at least one of vibration mean square amplitude sample data, vibration peak value sample data, vibration mean value sample data, and vibration standard deviation sample data.

[0069] In the above embodiment, based on the first excavation time of the tunnel boring machine, the original excavation sample parameters, the original slag sample image and the original vibration sample data are fused to obtain excavation sample data, slag sample data and vibration sample data, so as to reflect the comprehensive characteristics of the surrounding rock through the characteristic information between multiple modal data, and provide data support for the training of the surrounding rock classification model.

[0070] In some embodiments, original excavation sample parameters, original slag sample images and original vibration sample data are obtained during the excavation construction process; and the original excavation sample parameters are used as excavation sample data, the original slag sample images are used as slag sample data, and the original vibration sample data are used as vibration sample data, and then the excavation sample data, slag sample data, vibration sample data and labeled data are used to train the surrounding rock classification model to adjust the parameters of the surrounding rock classification model.

[0071] This embodiment uses directly acquired massive data to perform model training. Although the training process is time-consuming, the accuracy of model prediction is also improved because the data is more comprehensive.

[0072] Figure 3 Schematic diagram of some embodiments of the surrounding rock classification method disclosed in the present invention, which includes steps S31-S32.

[0073] In step S31, multimodal data during the excavation construction process is acquired.

[0074] In some embodiments, the multimodal data includes excavation data, slag data and vibration data. The excavation data comprehensively reflects the average hardness information of the surrounding rock of the face, and the excavation data is obtained by the monitoring system of the equipment itself. The slag data reflects the uneven hardness information of the face, and the slag data is a real-time data stream on the belt conveyor collected by the monitoring system. The vibration data reflects the strength information of the face, and the vibration data is the real-time vibration data of the equipment obtained by the vibration sensors deployed at the cutter head, main drive and other positions.

[0075] In step S32, the multimodal data is input into a surrounding rock classification model to perform surrounding rock classification prediction, wherein the surrounding rock classification model is trained based on the surrounding rock classification model training method described above.

[0076] For example, inputting multimodal data such as excavation data, slag data and vibration data into a trained surrounding rock classification model can quickly predict the classification of the surrounding rock.

[0077] In the above embodiment, the surrounding rock classification model is used as a basis, the rich information of multimodal data is utilized, the information advantages between different modal data are effectively integrated, information complementarity is achieved, and the accuracy of surrounding rock classification prediction is achieved. According to the surrounding rock classification prediction results, a basis can be provided for tunnel support and tunnel boring machine parameter control, major geological disasters of tunnel boring machines can be prevented, and construction efficiency can be improved.

[0078] In step S31, multimodal data of the excavation process is obtained, for example, by Figure 4 The steps shown are implemented, Figure 4 Schematic diagram of the flow chart of other embodiments of the surrounding rock classification method disclosed in the present invention, including steps S311-S312.

[0079] In step S311, original excavation parameters, original slag images and original vibration data during the excavation construction process are obtained.

[0080] In some embodiments, the excavation parameters reflect information such as the surrounding rock strength and equipment load during excavation, the vibration data reflects information such as abnormal impact, surrounding rock strength, and equipment operating conditions, and the slag image reflects information such as particle size distribution and surrounding rock crushing conditions.

[0081] In step S312, based on the second excavation time of the tunnel boring machine, the original excavation parameters, the original slag image and the original vibration data are fused to obtain excavation data, slag data and vibration data.

[0082] The real-time data stream records all data including shutdown, excavation and other processes. When analyzing the surrounding rock classification, the data during excavation needs to be selected.

[0083] In some embodiments, the excavation state of the tunnel boring machine is determined based on the original excavation parameters; the second excavation time of the tunnel boring machine is determined based on the excavation state of the tunnel boring machine; the second excavation parameter corresponding to the second excavation time is selected in the original excavation parameters, the second slag image corresponding to the second excavation time is selected in the original slag image, and the second vibration parameter corresponding to the second excavation time is selected in the original vibration data; and the excavation data is determined based on the second excavation parameter, the slag data is determined based on the second slag image, and the vibration data is determined based on the second vibration parameter.

[0084] For example, an effective excavation process is selected by a data segmentation method, wherein the data segmentation method can be expressed as

[0085] , where N is the cutterhead speed, F is the cutterhead thrust, and D represents the state of the tunneling machine. First, the tunneling process data is screened out, and then the slag data and vibration data corresponding to the tunneling data are selected according to time, thereby realizing time-based data fusion and obtaining effective monitoring data.

[0086] In some embodiments, excavation characteristic data related to surrounding rock classification in the second excavation parameter is extracted; and the excavation characteristic data related to surrounding rock classification in the second excavation parameter is used as excavation data.

[0087] For example, a correlation analysis is performed on the first excavation parameter, and with the surrounding rock classification as the target, the excavation data with a greater impact on the surrounding rock classification is selected using the Pearson correlation coefficient method to achieve feature screening of the excavation data. Alternatively, the selection method using expert experience can also be used to achieve the selection of excavation feature data. The excavation feature data related to the surrounding rock classification in the second excavation parameter includes at least one of the cutterhead thrust, cutterhead torque, and penetration.

[0088] In some embodiments, a parameter characterizing the size characteristics of the residue pieces in the second residue piece image is extracted; and the parameter characterizing the size characteristics of the residue pieces in the second residue piece image is used as the residue piece data.

[0089] For example, morphological methods such as corrosion, expansion, opening and closing are performed on the slag image to create high-quality image data. The parameter characterizing the slag size characteristics includes, for example, at least one of slag particle size sample data, non-uniformity coefficient sample data, and curvature coefficient sample data.

[0090] In some embodiments, a mathematical calculation is performed on the second vibration parameter to obtain a second characteristic parameter representing the vibration data; and the second characteristic parameter representing the vibration data is used as the vibration data.

[0091] For example, after filtering and denoising the second vibration parameter, the mean square amplitude, peak value, mean value, standard deviation, etc. are obtained to establish characteristic parameters characterizing the vibration data, thereby establishing a high-quality vibration database. The vibration data, for example, includes at least one of vibration mean square amplitude data, vibration peak value data, vibration mean value data, and vibration standard deviation data.

[0092] In the above embodiment, the rich information of multimodal data is utilized to effectively integrate the information advantages between different modal data and achieve information complementarity, so that the surrounding rock classification can be predicted more accurately.

[0093] In some embodiments, original excavation parameters, original slag images and original vibration data are obtained during the excavation construction process; and the original excavation parameters are used as excavation data, the original slag images are used as slag data, and the original vibration data are used as vibration data. The obtained original multimodal data is input into the surrounding rock classification model to perform surrounding rock classification prediction. Due to the rich data information, the accuracy of surrounding rock classification prediction can be improved.

[0094] The following will introduce the solution of the present disclosure by taking a specific embodiment as an example. Figure 5 As shown, Figure 5 It is a flowchart of some embodiments of surrounding rock classification model training and surrounding rock classification disclosed in the present invention, and the embodiment includes steps S51-S52.

[0095] In step S51, a multimodal database is established.

[0096] For example, the detection system of the tunnel boring machine itself is used to obtain tunneling data that can comprehensively reflect the average hardness information of the surrounding rock of the tunnel face; the monitoring system is used to collect the real-time data stream on the belt conveyor to obtain the slag image; the vibration sensors deployed at the cutter head, main drive and other positions are used to obtain the real-time vibration data of the tunnel boring machine. In addition, the surrounding rock classification data is established based on the exposed formation surrounding rock classification results.

[0097] In step S52, data fusion processing is performed.

[0098] For example, select the excavation data, slag image and vibration data of the roadheader during excavation.

[0099] In step S53, data feature screening is performed.

[0100] The excavation parameters are used to establish a correlation analysis model. With the surrounding rock classification as the goal, parameters with high correlation with surrounding rock classification are selected from numerous excavation databases, and several groups of data are selected to establish an excavation parameter matrix. Morphological methods such as corrosion, expansion, opening and closing are used to operate the slag image to establish high-quality image data, such as slag particle size, unevenness coefficient, curvature coefficient, etc. Through filtering, noise reduction and other methods, the effective vibration data of the excavation process is extracted, and a high-quality vibration database is established, such as mean square amplitude vibration data, peak vibration data, mean vibration data, standard deviation vibration data, etc.

[0101] In step S54, a surrounding rock classification model is trained.

[0102] For example, based on the deep neural network algorithm, with feature data as input and surrounding rock classification as output, a surrounding rock classification model is established, and parameter tuning is achieved through iterative parameter optimization.

[0103] In step S55, the surrounding rock classification model is used to predict the surrounding rock classification online.

[0104] During the construction of the tunnel boring machine, multi-modal data such as tunneling parameters, slag images, and vibration data are obtained in real time. After data fusion processing and data feature screening, the multi-modal data is input into the surrounding rock classification model, which can quickly predict the current surrounding rock classification in real time. This process relies on computer programs to achieve automated analysis.

[0105] In the above embodiment, the characteristic information between multimodal data can further reflect the comprehensive characteristics of the surrounding rock. Compared with the analysis results of single modal data, this embodiment uses richer characteristic information, and therefore can further improve the prediction accuracy, achieve accurate prediction of surrounding rock classification, and effectively ensure the safety of tunnel boring machines in disaster geological construction.

[0106] Figure 6 Schematic diagram of the structure of some embodiments of the surrounding rock classification model training device disclosed in the present invention, the surrounding rock classification model training device includes a first data processing module 610 and a training module 620.

[0107] The first data processing module 610 is configured to obtain multimodal sample data and corresponding annotated data during the excavation construction process, where the annotated data is the surrounding rock classification result.

[0108] In some embodiments, the multimodal sample data includes excavation sample data, slag sample data and vibration sample data. The first data processing module 610 obtains the original excavation sample parameters, original slag sample images and original vibration sample data during the excavation construction process; based on the first excavation time of the tunnel boring machine, the original excavation sample parameters, the original slag sample images and the original vibration sample data are fused to obtain the excavation sample data, slag sample data and vibration sample data.

[0109] For example, the excavation state of the tunnel boring machine is determined based on the original excavation sample parameters; the first excavation time of the tunnel boring machine is obtained based on the excavation state; the first excavation parameter corresponding to the first excavation time is selected in the original excavation sample parameters, the first slag image corresponding to the first excavation time is selected in the original slag sample image, and the first vibration parameter corresponding to the first excavation time is selected in the original vibration sample data; and the excavation sample data is determined based on the first excavation parameter, the slag sample data is determined based on the first slag image, and the vibration sample data is determined based on the first vibration parameter.

[0110] Determining the excavation sample data according to the first excavation parameter includes: extracting excavation characteristic data related to surrounding rock classification in the first excavation parameter; and using the excavation characteristic data related to surrounding rock classification in the first excavation parameter as excavation sample data. The excavation characteristic data includes at least one of cutterhead thrust sample data, cutterhead torque sample data and penetration sample data.

[0111] Determining the slag sample data according to the first slag image includes: extracting parameters characterizing the slag size characteristics in the first slag image; and using the parameters characterizing the slag size characteristics in the first slag image as the slag sample data. The parameters characterizing the slag size characteristics in the first slag image include at least one of slag particle size sample data, non-uniformity coefficient sample data, and curvature coefficient sample data.

[0112] According to the first vibration parameter, determining the vibration sample data includes: performing mathematical calculation on the first vibration parameter to obtain a first characteristic parameter characterizing the vibration data; and using the first characteristic parameter characterizing the vibration data as the vibration sample data. The first characteristic parameter characterizing the vibration data includes at least one of vibration mean square amplitude sample data, vibration peak value sample data, vibration mean value sample data, and vibration standard deviation sample data.

[0113] In some embodiments, the first data processing module 610 obtains original excavation sample parameters, original slag sample images and original vibration sample data during the excavation construction process; and uses the original excavation sample parameters as excavation sample data, the original slag sample images as slag sample data, and the original vibration sample data as vibration sample data.

[0114] The training module 620 is configured to train the surrounding rock classification model using the multimodal sample data and the labeled data to adjust the parameters of the surrounding rock classification model.

[0115] In the above embodiment, compared with single modality sample data, the multimodal sample data contains richer information, effectively integrates the information advantages between different modality sample data, realizes information complementarity, and thus can train a surrounding rock classification model that can more accurately predict surrounding rock classification.

[0116] Figure 7 Schematic diagram of the structure of some other embodiments of the surrounding rock classification model training device disclosed in the present invention, the surrounding rock classification model training device 700 includes a memory 710 and a processor 720. Among them: the memory 710 can be a disk, a flash memory or any other non-volatile storage medium. The memory 710 is used to store the instructions in the above embodiments. The processor 720 is coupled to the memory 710 and can be implemented as one or more integrated circuits, such as a microprocessor or a microcontroller. The processor 720 is used to execute the instructions stored in the memory.

[0117] In some embodiments, the processor 720 is coupled to the memory 710 via a BUS bus 730. The surrounding rock classification model training device 700 can also be connected to an external storage device 750 via a storage interface 740 to call external data, and can also be connected to a network or another computer system (not shown) via a network interface 760. No further detailed description will be given here.

[0118] Figure 8 Schematic diagram of the structure of some embodiments of the surrounding rock classification device disclosed in the present invention, the surrounding rock classification device includes a second data processing module 810 and a prediction module 820.

[0119] The second data processing module 810 is configured to acquire multimodal data during the excavation construction process.

[0120] In some embodiments, original excavation parameters, original slag images and original vibration data are obtained during the excavation construction process; and based on the second excavation time of the tunnel boring machine, the original excavation parameters, original slag images and original vibration data are fused to obtain excavation data, slag data and vibration data.

[0121] For example, the excavation state of the tunnel boring machine is determined according to the original excavation parameters; the second excavation time of the tunnel boring machine is determined according to the excavation state of the tunnel boring machine; the second excavation parameter corresponding to the second excavation time is selected in the original excavation parameters, the second slag image corresponding to the second excavation time is selected in the original slag image, and the second vibration parameter corresponding to the second excavation time is selected in the original vibration data; and the excavation data is determined according to the second excavation parameter, the slag data is determined according to the second slag image, and the vibration data is determined according to the second vibration parameter.

[0122] Determining the excavation data according to the second excavation parameter includes: extracting excavation characteristic data related to surrounding rock classification in the second excavation parameter; and using the excavation characteristic data related to surrounding rock classification in the second excavation parameter as the excavation data. The excavation characteristic data related to surrounding rock classification in the second excavation parameter includes at least one of cutterhead thrust, cutterhead torque and penetration.

[0123] Determining the slag data according to the second slag image includes: extracting parameters characterizing the slag size characteristics in the second slag image; and using the parameters characterizing the slag size characteristics in the second slag image as the slag data. The parameters characterizing the slag size characteristics include, for example, at least one of slag particle size sample data, non-uniformity coefficient sample data, and curvature coefficient sample data.

[0124] According to the second vibration parameter, determining the vibration data includes: performing mathematical calculation on the second vibration parameter to obtain a second characteristic parameter characterizing the vibration data; and using the second characteristic parameter characterizing the vibration data as the vibration data. The vibration data includes, for example, at least one of vibration mean square amplitude data, vibration peak value data, vibration mean data, and vibration standard deviation data.

[0125] In some embodiments, original excavation parameters, original slag images and original vibration data are obtained during the excavation construction process; and the original excavation parameters are used as excavation data, the original slag images are used as slag data, and the original vibration data are used as vibration data.

[0126] The prediction module 820 is configured to input the multimodal data into the surrounding rock classification model to perform surrounding rock classification prediction, wherein the surrounding rock classification model is trained based on the surrounding rock classification model training method mentioned above.

[0127] In the above embodiment, the surrounding rock classification model is used as a basis, the rich information of multimodal data is utilized, the information advantages between different modal data are effectively integrated, information complementarity is achieved, and the accuracy of surrounding rock classification prediction is achieved. According to the surrounding rock classification prediction results, a basis can be provided for tunnel support and tunnel boring machine parameter control, major geological disasters of tunnel boring machines can be prevented, and construction efficiency can be improved.

[0128] Fig. 9Schematic diagram of the structure of other embodiments of the surrounding rock classification device of the present disclosure. The surrounding rock classification device 900 includes a memory 910 and a processor 920. Among them: the memory 910 can be a disk, a flash memory or any other non-volatile storage medium. The memory 910 is used to store the instructions in the above embodiments. The processor 920 is coupled to the memory 910 and can be implemented as one or more integrated circuits, such as a microprocessor or a microcontroller. The processor 920 is used to execute the instructions stored in the memory.

[0129] In some embodiments, the processor 920 is coupled to the memory 910 via a BUS 930. The surrounding rock classification device 900 can also be connected to an external storage device 950 via a storage interface 940 to call external data, and can also be connected to a network or another computer system (not shown) via a network interface 960. No further detailed description will be given here.

[0130] In other embodiments, a computer-readable storage medium stores computer program instructions, which implement the steps of the methods in the above-mentioned embodiments when executed by a processor. It should be understood by those skilled in the art that the embodiments of the present disclosure may be provided as methods, devices, or computer program products. Therefore, the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-usable non-transient storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0131] In some embodiments, a computer program product is protected, including a computer program or an instruction, which implements the above method when the computer program or instruction is executed by a processor. The computer program product includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by the CPU, the above functions defined in the method of the embodiment of the present disclosure are executed.

[0132] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram and the combination of processes and / or blocks in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0133] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0134] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0135] So far, the present disclosure has been described in detail. In order to avoid obscuring the concept of the present disclosure, some details known in the art are not described. Based on the above description, those skilled in the art can fully understand how to implement the technical solution disclosed here.

[0136] The method and apparatus of the present disclosure may be implemented in many ways. For example, the method and apparatus of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present disclosure may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present disclosure. Therefore, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

[0137] Although some specific embodiments of the present disclosure have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present disclosure. It should be understood by those skilled in the art that the above embodiments may be modified without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A surrounding rock classification model training method, comprising: Acquire multimodal sample data and corresponding annotated data during the excavation construction process, wherein the annotated data is the surrounding rock classification result; as well as The surrounding rock classification model is trained using the multimodal sample data and the labeled data to adjust the parameters of the surrounding rock classification model.

2. The surrounding rock classification model training method according to claim 1, wherein: The multimodal sample data includes excavation sample data, slag sample data and vibration sample data, and the acquiring of the multimodal sample data includes: Acquiring original excavation sample parameters, original slag sample images and original vibration sample data during the excavation construction process; and Based on the first excavation time of the tunnel boring machine, the original excavation sample parameters, the original slag sample image and the original vibration sample data are fused to obtain the excavation sample data, the slag sample data and the vibration sample data.

3. The surrounding rock classification model training method according to claim 2, wherein: Based on the excavation time of the roadheader, the original excavation sample parameters, the original slag sample image and the original vibration sample data are fused to obtain the excavation sample data, the slag sample data and the vibration sample data, including: Determining the excavation state of the tunnel boring machine according to the original excavation sample parameters; According to the excavation state, obtaining a first excavation time of the excavator; Selecting a first excavation parameter corresponding to the first excavation time from the original excavation sample parameters, selecting a first slag image corresponding to the first excavation time from the original slag sample images, and selecting a first vibration parameter corresponding to the first excavation time from the original vibration sample data; and The excavation sample data is determined according to the first excavation parameter, the slag sample data is determined according to the first slag image, and the vibration sample data is determined according to the first vibration parameter.

4. The surrounding rock classification model training method according to claim 3, wherein: The determining the excavation sample data according to the first excavation parameter comprises: extracting excavation characteristic data related to surrounding rock classification from the first excavation parameter; and The excavation characteristic data related to the surrounding rock classification in the first excavation parameter is used as the excavation sample data.

5. The surrounding rock classification model training method according to claim 3, wherein: The determining the slag piece sample data according to the first slag piece image comprises: Extracting parameters representing size characteristics of the slag pieces from the first slag piece image; and The parameters representing the size characteristics of the residue pieces in the first residue piece image are used as the residue piece sample data.

6. The surrounding rock classification model training method according to any one of claims 3 to 5, wherein: Determining the vibration sample data according to the first vibration parameter includes: Performing mathematical calculation on the first vibration parameter to obtain a first characteristic parameter representing the vibration data; and The first characteristic parameter characterizing the vibration data is used as the vibration sample data.

7. The surrounding rock classification model training method according to claim 1, wherein: The multimodal sample data includes excavation sample data, slag sample data and vibration sample data, wherein the acquiring of the multimodal sample data includes: Acquiring original excavation sample parameters, original slag sample images and original vibration sample data during the excavation construction process; and The original excavation sample parameters are used as the excavation sample data, the original slag sample image is used as the slag sample data, and the original vibration sample data is used as the vibration sample data.

8. A surrounding rock classification method, comprising: Acquire multimodal data during tunneling construction; as well as The multimodal data is input into a surrounding rock classification model to perform surrounding rock classification prediction, wherein the surrounding rock classification model is trained based on the surrounding rock classification model training method according to any one of claims 1 to 10.

9. The surrounding rock classification method according to claim 8, wherein: The multimodal data includes excavation data, slag data and vibration data, and the acquiring of the multimodal data includes: Acquiring original excavation parameters, original slag images and original vibration data during the excavation construction process; and Based on the second excavation time of the tunnel boring machine, the original excavation parameters, the original slag image and the original vibration data are fused to obtain the excavation data, the slag data and the vibration data.

10. The surrounding rock classification method according to claim 9, wherein: Based on the second excavation time of the roadheader, the original excavation parameters, the original slag image and the original vibration data are fused to obtain the excavation data, the slag data and the vibration data, including: Determining the tunneling state of the tunnel boring machine according to the original tunneling parameters; a second excavation time of the tunnel boring machine according to the excavation state of the tunnel boring machine; Selecting a second excavation parameter corresponding to the second excavation time from the original excavation parameters, selecting a second slag image corresponding to the second excavation time from the original slag image, and selecting a second vibration parameter corresponding to the second excavation time from the original vibration data; and The excavation data is determined according to the second excavation parameter, the slag data is determined according to the second slag image, and the vibration data is determined according to the second vibration parameter.

11. The surrounding rock classification method according to claim 10, wherein: Determining the excavation data according to the second excavation parameter comprises: extracting excavation characteristic data related to surrounding rock classification from the second excavation parameter; and The excavation characteristic data related to the surrounding rock classification in the second excavation parameter is used as the excavation data.

12. The surrounding rock classification method according to claim 10, wherein: The determining the residue piece data according to the second residue piece image includes: Extracting parameters representing size characteristics of the residue pieces from the second residue piece image; and The parameters representing the size characteristics of the residue pieces in the second residue piece image are used as the residue piece data.

13. The surrounding rock classification method according to any one of claims 10 to 12, wherein: The determining the vibration data according to the second vibration parameter comprises: Performing mathematical calculation on the second vibration parameter to obtain a second characteristic parameter representing the vibration data; and The second characteristic parameter characterizing the vibration data is used as the vibration data.

14. The surrounding rock classification method according to claim 8, wherein: The multimodal data includes excavation data, slag data and vibration data, and the acquiring of the multimodal data includes: Acquiring original excavation parameters, original slag images and original vibration data during the excavation construction process; and The original excavation parameters are used as the excavation data, the original slag image is used as the slag data, and the original vibration data is used as the vibration data.

15. A surrounding rock classification model training device, comprising: A first data processing module is configured to obtain multimodal sample data and corresponding annotated data during the tunneling construction process, wherein the annotated data is a surrounding rock classification result; as well as The training module is configured to train the surrounding rock classification model using the multimodal sample data and the labeled data to adjust the parameters of the surrounding rock classification model.

16. A surrounding rock classification model training device, comprising: Memory; as well as A processor coupled to the memory, wherein the processor is configured to execute the surrounding rock classification model training method according to any one of claims 1 to 7 based on instructions stored in the memory.

17. A surrounding rock classification device, comprising: A second data processing module is configured to obtain multimodal data during the excavation construction process; The prediction module is configured to input the multimodal data into a surrounding rock classification model to perform surrounding rock classification prediction, wherein the surrounding rock classification model is trained based on the surrounding rock classification model training method described in any one of claims 1 to 7.

18. A surrounding rock classification device, comprising: A processor coupled to the memory, wherein the processor is configured to execute the surrounding rock classification method according to any one of claims 8 to 14 based on instructions stored in the memory.

19. A computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, implement the surrounding rock classification model training method described in any one of claims 1 to 7, or the surrounding rock classification method described in any one of claims 8 to 14.

20. A computer program product, comprising a computer program or instructions, wherein when the computer program or instructions are executed by a processor, the surrounding rock classification model training method described in any one of claims 1 to 7, or the surrounding rock classification method described in any one of claims 8 to 14 is implemented.

Images