Underground cavern geology intelligent prediction analysis method, system and device, storage medium and program product

Abstract

The invention discloses an underground cavern geology intelligent prediction analysis method, system and device, a storage medium and a program product, and the method comprises the steps: training an original recognition model based on geophysical detection result sample data obtained after geological interpretation, and obtaining an unfavorable geologic body prediction model, inputting the current geophysical exploration result data of the to-be-analyzed cavern section into the unfavorable geologic body prediction model for geological analysis to obtain unfavorable geologic body information, and performing feature quantization on the unfavorable geologic body information of the underground cavern, the tunnel face surrounding rock category and the excavation space extension information; advanced geological prediction analysis is carried out on the underground cavern on the basis of the unfavorable geologic body development level value, the tunnel face surrounding rock category value and the spatial extension feature value obtained after feature quantization, and accurate geological analysis of the underground cavern is achieved by predicting the unfavorable geologic body type and the distribution range. And the accuracy of advanced geological forecast is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of geological prediction, and in particular to an intelligent prediction and analysis method, system, equipment, storage medium and program product for underground cavern geology. Background Art

[0002] With the development of the energy industry and the need for urban construction, the construction of underground caverns is crucial for water conservancy and hydropower, railway and highway transportation, energy mining, and municipal engineering. Advanced geological prediction is an effective method to grasp the geological conditions in front of the face during the construction of underground caverns. It is an indispensable part of the current underground cavern construction. It plays an important role in safety assurance and disaster prevention and control, and is conducive to information-based construction. At present, the establishment of an advanced geological prediction system for underground caverns in the water conservancy and hydropower industry is still imperfect. There are limitations in the use of a single detection method, and the reliability of geological analysis based on manual interpretation is highly dependent on the experience of the interpreter, resulting in limited accuracy of advanced geological prediction of underground caverns. Summary of the invention

[0003] The main purpose of the present invention is to provide an intelligent prediction and analysis method, system, equipment, storage medium and program product for underground cavern geology, aiming to solve the technical problems that the existing technology uses a single detection method with limitations, and the reliability of geological analysis based on manual interpretation is highly dependent on the experience of the interpreter, resulting in limited accuracy of advanced geological prediction of underground caverns.

[0004] To achieve the above object, the present invention provides an intelligent prediction and analysis method for underground cavern geology, the method comprising the following steps: Obtain historical geophysical exploration data of underground caverns; Conduct geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data; Constructing an original recognition model, and training the original recognition model based on the sample data of the geophysical exploration results to obtain a poor geological body prediction model, wherein the original recognition model includes a U-Net network model; Inputting the current geophysical exploration results data of the to-be-analyzed cave section of the underground cave into the bad geological body prediction model for geological analysis, and obtaining the bad geological body information of the to-be-analyzed cave section, wherein the bad geological body information includes the bad geological body type and the spatial distribution information of the bad geological body type; Obtaining the type of surrounding rock and excavation space extension information of the underground cavern; Quantifying the characteristics of the bad geological body information, the tunnel face surrounding rock category and the excavation space extension information to obtain the bad geological body development level value, tunnel face surrounding rock category value and space extension characteristic value; Based on the unfavorable geological body development level value, the tunnel face surrounding rock category value and the spatial extension characteristic value, an advanced geological prediction analysis is performed on the underground cavern.

[0005] Optionally, the historical geophysical exploration achievement data includes seismic wave exploration achievement data and ground penetrating radar exploration achievement data, the seismic wave exploration achievement data includes wave velocity diagram and seismic wave profile diagram, and the ground penetrating radar exploration achievement data includes ground penetrating radar profile diagram; the acquisition of the historical geophysical exploration achievement data of underground caverns includes: Obtain surrounding rock geological conditions and geological structural conditions based on historical geological analysis data of underground caverns; Determining a cavern prediction strategy based on the surrounding rock geological conditions and the geological structural conditions; Based on the cavern prediction strategy, seismic wave detection data and ground penetrating radar detection data are collected for the underground cavern to obtain seismic wave detection data and ground penetrating radar detection data; The seismic wave detection data and the ground penetrating radar detection data are preprocessed to obtain historical geophysical detection results data, and the preprocessing includes static correction processing, denoising processing, dynamic correction processing, time-depth conversion processing and normalization processing.

[0006] Optionally, the geological interpretation of the historical geophysical exploration achievement data to obtain geophysical exploration achievement sample data includes: Performing geological interpretation feature analysis on the seismic wave detection result data and the ground penetrating radar detection result data to obtain parameter change feature information of seismic wave parameters and ground penetrating radar parameters, wherein the seismic wave parameters include seismic wave velocity, seismic wave frequency and seismic wave amplitude, and the ground penetrating radar parameters include ground penetrating radar frequency and ground penetrating radar amplitude; Extracting abnormal features from parameter change feature information of the seismic wave parameters and the ground penetrating radar parameters to obtain abnormal seismic wave feature information and abnormal ground penetrating radar feature information; Perform correlation analysis on the abnormal characteristic information of the seismic wave and the abnormal characteristic information of the ground penetrating radar to obtain the type of the sample bad geological body; Geophysical exploration achievement sample data is constructed based on the sample poor geological body type and the historical geophysical exploration achievement data.

[0007] Optionally, the constructing of geophysical exploration achievement sample data based on the sample poor geological body type and the historical geophysical exploration achievement data includes: Determine the seismic wave abnormal section and the ground penetrating radar abnormal section based on the seismic wave abnormal characteristic information and the ground penetrating radar abnormal characteristic information; Performing spatial superposition analysis on the abnormal seismic wave segment and the abnormal ground penetrating radar segment to obtain an abnormal overlapping area and an abnormal non-overlapping area; Generate geometric constraints based on the geological structure law information of the underground cavern and the type of the sample unfavorable geological body; Determine the pile number segment covered by the bad geological body of the sample bad geological body type according to the geometric constraint condition, the abnormal overlapping area and the abnormal non-overlapping area; Determine the sample spatial distribution information of the sample bad geological body type based on the spatial position of the pile number segment; Geophysical exploration achievement sample data is constructed based on the historical geophysical exploration achievement data, the sample poor geological body type and the sample spatial distribution information.

[0008] Optionally, the constructing of the original recognition model and training the original recognition model based on the sample data of the geophysical exploration results to obtain the bad geological body prediction model includes: Determine an encoder, a decoder and a skip connection layer, and construct an original recognition model based on the encoder, the decoder and the skip connection layer, and initialize weight parameters of the original recognition model; The encoder performs convolution and pooling processing on the sample data of the geophysical exploration results, and outputs global semantic feature information of the bad geological body type; The decoder performs transposition convolution on the global semantic feature information to output spatial distribution information and type labels of bad geological bodies; The global semantic feature information, spatial distribution information and type label of the bad geological body type are spatially aligned and multi-scale feature fused through the skip connection layer; A model weight parameter is obtained based on the output results of the encoder, the decoder and the skip connection layer, and a poor geological body prediction model is constructed based on the model weight parameter.

[0009] Optionally, the constructing of the original recognition model and training the original recognition model based on the sample data of the geophysical exploration results to obtain the bad geological body prediction model includes: Constructing an original recognition model, wherein the original recognition model includes a ReLU activation function and a Sigmoid activation function, wherein the ReLU activation function is used for a convolutional layer of the original recognition model, and the Sigmoid function is used for an output layer of the original recognition model; Training the original recognition model based on the geophysical exploration achievement sample data to obtain a candidate recognition model; Construct a loss function and an optimizer, wherein the optimizer includes an AdaDelta adaptive learning rate optimizer, and the loss function includes a multi-classification cross entropy loss function, wherein the multi-classification cross entropy loss function includes: in, Indicates The actual bad geological body type label of samples, Represents the candidate recognition model for the The sample labels predicted by samples, represents the total number of samples, Indicates the number of bad geological body types, represents the total penalty parameter for prediction error; The candidate identification model is optimized and converged based on the loss function and the optimizer to obtain a poor geological body prediction model.

[0010] In addition, to achieve the above-mentioned purpose, the present invention also proposes an underground cavern geological intelligent prediction and analysis system, the underground cavern geological intelligent prediction and analysis system comprising: Detection data acquisition module, used to obtain historical geophysical detection results data of underground caverns; A geological interpretation module is used to perform geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data; A model building module is used to build an original recognition model and train the original recognition model based on the sample data of the geophysical exploration results to obtain a poor geological body prediction model, wherein the original recognition model includes a U-Net network model; A geological analysis module, used for inputting the current geophysical exploration results data of the to-be-analyzed cave section of the underground cave into the bad geological body prediction model for geological analysis, and obtaining the bad geological body information of the to-be-analyzed cave section, wherein the bad geological body information includes the bad geological body type and the spatial distribution information of the bad geological body type; A cavern information acquisition module, used to acquire the type of surrounding rock of the tunnel face and the extension information of the excavation space of the underground cavern; A feature quantification module is used to perform feature quantification on the bad geological body information, the tunnel face surrounding rock category and the excavation space extension information to obtain the bad geological body development level value, tunnel face surrounding rock category value and space extension feature value; The advanced geological prediction and analysis module is used to perform advanced geological prediction and analysis on the underground cavern based on the development level value of the unfavorable geological body, the category value of the surrounding rock of the tunnel face and the spatial extension characteristic value.

[0011] In addition, to achieve the above-mentioned purpose, the present application also proposes an intelligent prediction and analysis device for underground cavern geology, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the intelligent prediction and analysis method for underground cavern geology as described above.

[0012] In addition, to achieve the above-mentioned purpose, the present application also proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the intelligent prediction and analysis method for underground cavern geology as described above are implemented.

[0013] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the underground cavern geology intelligent prediction and analysis method as described above.

[0014] The present invention obtains historical geophysical exploration results data of underground caverns; performs geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data; constructs an original recognition model, and trains the original recognition model based on the geophysical exploration results sample data to obtain a poor geological body prediction model, wherein the original recognition model includes a U-Net network model; inputs the current geophysical exploration results data of the to-be-analyzed cave section of the underground cavern into the poor geological body prediction model for geological analysis to obtain poor geological body information of the to-be-analyzed cave section, wherein the poor geological body information includes the type of poor geological body and the spatial distribution information of the type of poor geological body; obtains the type of surrounding rock of the tunnel face and the excavation space extension information of the underground cavern; performs feature quantification on the poor geological body information, the type of surrounding rock of the tunnel face and the excavation space extension information to obtain the development of the poor geological body. level value, face surrounding rock category value and spatial extension characteristic value; based on the development level value of the adverse geological body, the face surrounding rock category value and the spatial extension characteristic value, the underground cavern is subjected to advanced geological prediction and analysis; since the present invention performs geological interpretation on historical geophysical exploration results data and constructs geophysical exploration results sample data based on the geological interpretation results, the sample quality and model training efficiency are effectively improved; the adverse geological body prediction model obtained after training is used to perform geological analysis on the current geophysical exploration results data of the tunnel section to be analyzed, thereby accurately obtaining the type and spatial distribution information of the adverse geological body of the tunnel section to be analyzed; by quantifying the characteristics of the adverse geological body information, face surrounding rock category, excavation progress information and spatial extension information, it is possible to accurately evaluate the changes in the geological conditions in front of the underground cavern, thereby realizing accurate geological analysis of the underground cavern and effectively improving the accuracy of advanced geological prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0016] Figure 1 It is a structural diagram of an underground cavern geological intelligent prediction and analysis device in a hardware operating environment involved in an embodiment of the present invention; Figure 2 A schematic diagram of the process of the first embodiment of the intelligent prediction and analysis method for underground cavern geology of the present invention; Figure 3 It is a schematic diagram of the flow chart of the second embodiment of the underground cavern geological intelligent prediction and analysis method of the present invention; Figure 4 It is a structural block diagram of the first embodiment of the underground cavern geology intelligent prediction and analysis system of the present invention.

[0017] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0018] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0019] Reference Figure 1 , Figure 1 The present invention is a schematic diagram of the structure of an intelligent prediction and analysis device for underground cavern geology in the hardware operating environment involved in the embodiment of the present invention.

[0020] like Figure 1As shown, the underground cavern geological intelligent prediction and analysis device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, WI-FI) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM), or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. The memory 1005 may also be a storage system independent of the aforementioned processor 1001.

[0021] Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation on the intelligent prediction and analysis equipment for underground cavern geology, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0022] like Figure 1 As shown, the memory 1005 as a computer-readable storage medium may include an operating system, a network communication module, a user interface module, and an intelligent prediction and analysis program for underground cavern geology.

[0023] exist Figure 1 In the underground cavern geology intelligent prediction and analysis device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the underground cavern geology intelligent prediction and analysis device of the present invention can be set in the underground cavern geology intelligent prediction and analysis device, and the underground cavern geology intelligent prediction and analysis device calls the underground cavern geology intelligent prediction and analysis program stored in the memory 1005 through the processor 1001, and executes the underground cavern geology intelligent prediction and analysis method provided by the embodiment of the present invention.

[0024] The embodiment of the present invention provides an intelligent prediction and analysis method for underground cavern geology, referring to Figure 2 , Figure 2 It is a schematic diagram of the flow chart of the first embodiment of the intelligent prediction and analysis method for underground cavern geology of the present invention.

[0025] In this embodiment, the underground cavern geological intelligent prediction and analysis method comprises the following steps: Step S10: Acquire historical geophysical exploration results data of underground caverns.

[0026] It should be noted that the present embodiment can be applied to the advanced geological forecast of underground caverns, which can promote the rapid development of the digitalization and refinement of geological analysis and geological prediction technologies in the construction of underground caverns. Among them, the advanced geological forecast can be an advanced forecast of the surrounding rock and stratum conditions in front of and around the heading face when excavating the underground cavern.

[0027] In some embodiments, the advance geological prediction of underground caverns can be carried out by three methods: engineering geological analysis, geophysical detection and geological exploration. The engineering geological analysis method includes the surface geological body projection method and the tunnel face catalog prediction method. The geological exploration method includes the advanced pilot hole method and the advanced drilling method. The engineering geological analysis method and the geological exploration method are to identify key high-risk areas by analyzing the geological conditions, classify the surrounding rock categories, and provide guidance for the excavation of underground caverns. However, they need to be carried out by professional geologists, the prediction distance is short, and the prediction accuracy is closely related to the experience of the forecasters.

[0028] This embodiment aims at the advanced geological prediction work based on the geophysical detection method, and realizes intelligent prediction analysis with the help of artificial intelligence technology. It analyzes the unfavorable geological body by combining seismic wave detection and ground penetrating radar detection, and analyzes the unfavorable geological body from the detection results of two dimensions, which effectively improves the prediction accuracy. It performs advanced geological prediction based on the unfavorable geological body information, the type of surrounding rock of the face and the extension information of the excavation space, thereby effectively improving the accuracy of the advanced geological prediction and providing reliable data support for the construction of underground caverns.

[0029] It should be understood that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or a terminal electronic device capable of realizing the above functions, etc. The following takes the underground cavern geological intelligent prediction and analysis device (hereinafter referred to as the analysis device) as an example to illustrate this embodiment and the following embodiments.

[0030] It should be noted that the historical geophysical exploration results data may be geophysical exploration results data of other cave sections that have been historically explored in the underground cavern. In some embodiments, the historical geophysical exploration results data may be seismic wave detection data and ground penetrating radar detection data that have been verified through excavation.

[0031] It can be understood that this embodiment obtains seismic wave detection data and ground-penetrating radar detection data through seismic wave detection and ground-penetrating radar detection, and obtains historical geophysical detection results data of underground caverns based on the seismic wave detection data and ground-penetrating radar detection data.

[0032] Furthermore, in order to improve the data accuracy and data quality of geophysical exploration results data, thereby improving the accuracy of geological analysis and prediction, in some embodiments, the historical geophysical exploration results data includes seismic wave exploration results data and ground penetrating radar exploration results data, the seismic wave exploration results data includes wave velocity diagrams and seismic wave profile diagrams, and the ground penetrating radar exploration results data includes ground penetrating radar profile diagrams; the above step S10 may include: Step S101: Obtaining surrounding rock geological conditions and geological structural conditions based on historical geological analysis data of underground caverns; Step S102: determining a cavern prediction strategy based on the surrounding rock geological conditions and the geological structural conditions; Step S103: collecting seismic wave detection data and ground penetrating radar detection data of the underground cavern based on the cavern prediction strategy to obtain seismic wave detection data and ground penetrating radar detection data; Step S104: pre-processing the seismic wave detection data and the ground penetrating radar detection data to obtain historical geophysical detection results data.

[0033] It should be noted that the analysis equipment can obtain historical geophysical exploration results data through geophysical exploration methods. The geophysical exploration methods used in this embodiment include seismic wave method (a method of long-distance advance exploration) and ground penetrating radar method (a method of short-distance advance exploration). Therefore, the geophysical exploration results involved include seismic wave method exploration results and ground penetrating radar exploration results. Seismic wave method exploration results include: longitudinal wave velocity diagram, shear wave velocity diagram, seismic profile diagram (time-distance two-dimensional image); ground penetrating radar method exploration results include: radar profile diagram (time-distance two-dimensional image).

[0034] It should be noted that the preprocessing includes static correction processing, denoising processing, dynamic correction processing, time-to-depth conversion processing and normalization processing; among them, static correction is used to eliminate the time difference caused by terrain and velocity differences; dynamic correction and superposition processing are used to eliminate propagation time differences and enhance effective wave energy; depth conversion processing can be time-to-depth conversion, which is used to calculate the depth based on the electromagnetic wave velocity.

[0035] In some embodiments, the analysis equipment selects a suitable observation method and related observation parameters according to the geological conditions and structural conditions of the surrounding rock; collects data according to the design plan of the observation system; and performs pre-processing such as static correction, denoising, dynamic correction, time-depth conversion, and normalization on the collected data due to the complexity of the on-site construction environment and the different types, formats, and time dimensions of the collected data, to obtain geophysical exploration results data.

[0036] Step S20: Performing geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data.

[0037] It should be noted that the geophysical exploration achievement sample data may be sample data used to train the bad geological body network model. The geophysical exploration achievement sample data may include geophysical exploration achievements and the bad geological body type and spatial distribution information corresponding to the geophysical exploration achievements.

[0038] In a specific implementation, the analysis equipment can perform geological interpretation on the geophysical exploration results data to obtain the types of unfavorable geological bodies and spatial distribution information, and construct geophysical exploration results sample data based on the geophysical exploration results, types of unfavorable geological bodies and spatial distribution information.

[0039] In some embodiments, the analysis equipment marks the types and spatial distribution information of unfavorable geological bodies verified by excavation on the detection results to form one or more samples for machine learning, so that the latter can realize the function of intelligent identification of unfavorable geological bodies by learning sample data.

[0040] Furthermore, in order to accurately perform geological interpretation on historical geophysical exploration results data and construct high-quality geophysical exploration results sample data, thereby improving model training efficiency and performance, in some embodiments, the above step S20 may include: Step S201: performing geological interpretation feature analysis on the seismic wave detection result data and the ground penetrating radar detection result data to obtain parameter change feature information of seismic wave parameters and ground penetrating radar parameters; Step S202: extracting abnormal features from parameter change feature information of the seismic wave parameters and the ground penetrating radar parameters to obtain abnormal seismic wave feature information and abnormal ground penetrating radar feature information; Step S203: performing correlation analysis on the seismic wave abnormal characteristic information and the ground penetrating radar abnormal characteristic information to obtain the type of sample bad geological body; Step S204: constructing geophysical exploration achievement sample data based on the sample poor geological body type and the historical geophysical exploration achievement data.

[0041] It should be noted that the seismic wave parameters include seismic wave velocity, seismic wave frequency and seismic wave amplitude, and the ground penetrating radar parameters include ground penetrating radar frequency and ground penetrating radar amplitude. The ground penetrating radar abnormal feature information may include abnormal frequency features and abnormal amplitude features in the ground penetrating radar features; the seismic wave abnormal feature information may include abnormal wave velocity features, abnormal frequency features and abnormal amplitude features in the seismic wave features.

[0042] It can be understood that the analysis equipment can analyze the signal characteristics of the geophysical exploration data, and explain the changes in various parameters. For seismic exploration results, the focus is on: wave velocity, frequency, amplitude, etc., and for ground penetrating radar detection results, the focus is on: frequency, amplitude, etc.; secondly, combined with the signal propagation law, the geological body characteristics that may cause the above-mentioned signal characteristics are analyzed, including: whether there are unfavorable geological bodies, the types of unfavorable geological bodies, and the spatial distribution of unfavorable geological bodies, and the geological conditions in front of the face are interpreted; finally, the surrounding rock classification results are given.

[0043] Taking the seismic wave method as an example: the longitudinal wave velocity of the seismic wave in the section of pile number J0+400~J0+500 is between 2600m / s and 3200m / s, and the overall surrounding rock conditions are equivalent to the current tunnel face conditions; the reflection signal of the section of pile number J0+438~J0+463 is strong, the frequency is low, and the amplitude is large, which indicates that there is rich water in this area. The rock mass quality of the section of pile number J0+400~J0+500 is Grade III, and the rock mass quality of the section of pile number J0+438~J0+463 is Grade IV, so attention should be paid during excavation.

[0044] In some embodiments, the analysis device can analyze whether there are places in the result data where the signal characteristics differ from the surrounding rock signal characteristics; if so, analyze the possible causes of the signal change and determine the type of bad geological body that caused the signal change; delineate the contours of the bad geological body: for the seismic wave method result data, the wave velocity reduction, the reflection wave event axis misalignment, and the amplitude mutation area may be the boundary of the bad geological body; for the ground penetrating radar method result data, the reflection wave event axis misalignment and the amplitude mutation area may be the boundary of the bad geological body. Finally, based on the geological interpretation results of the geophysical exploration result data, the preliminary conclusions of the geophysical exploration analysis are obtained, such as the possible types of bad geological bodies and spatial distribution information, and the surrounding rock categories of the geophysical exploration analysis.

[0045] Step S30: constructing an original recognition model, and training the original recognition model based on the sample data of the geophysical exploration results to obtain a poor geological body prediction model.

[0046] It should be noted that the original recognition model includes a U-Net network model.

[0047] In some embodiments, the analysis device can determine the encoder, decoder and skip connection layers of the U-Net network model, construct an original recognition model based on the encoder, decoder and skip connection layers, generate a training set, a test set and a validation set based on the sample data of geophysical exploration results, train, test and validate the original recognition model based on the sample data of geophysical exploration results, and obtain a poor geological body prediction model.

[0048] Furthermore, in order to improve the model prediction efficiency and accuracy, in some embodiments, the above step S30 may include: Step S301: determining an encoder, a decoder and a skip connection layer, and constructing an original recognition model based on the encoder, the decoder and the skip connection layer, and initializing weight parameters of the original recognition model; Step S302: performing convolution and pooling processing on the sample data of the geophysical exploration results through the encoder to output global semantic feature information of the bad geological body type; Step S303: performing transposition convolution on the global semantic feature information through the decoder to output spatial distribution information and type labels of bad geological bodies; Step S304: performing spatial alignment and multi-scale feature fusion on the global semantic feature information, spatial distribution information and type label of the bad geological body type through the skip connection layer; Step S305: obtaining model weight parameters based on the output results of the encoder, the decoder and the skip connection layer, and constructing a poor geological body prediction model based on the model weight parameters.

[0049] It should be noted that the encoder includes five stages of convolutional layers and four stages of downsampling layers, wherein the first convolutional layer of the first stage includes a 3×3 convolution kernel and 1 input channel, the second convolutional layer and the third convolutional layer both include 3×3 convolutional kernels and 32 input channels, the first convolutional layer and the second convolutional layer of the second stage both include 3×3 convolutional kernels and 64 input channels, the first convolutional layer and the second convolutional layer of the third stage both include 3×3 convolutional kernels and 128 input channels, the first convolutional layer and the second convolutional layer of the fourth stage both include 3×3 convolutional kernels and 256 input channels, the first convolutional layer and the second convolutional layer of the fifth stage both include 3×3 convolutional kernels and 512 input channels, and the downsampling layers of the four stages all include 2×2 maximum pooling layers.

[0050] The decoder includes five stages of convolutional layers and four stages of upsampling layers, wherein the first convolutional layer and the second convolutional layer of the first stage both include 3×3 convolutional kernels and 256 input channels, the first convolutional layer and the second convolutional layer of the second stage both include 3×3 convolutional kernels and 128 input channels, the first convolutional layer and the second convolutional layer of the third stage both include 3×3 convolutional kernels and 64 input channels, the first convolutional layer and the second convolutional layer of the fourth stage both include 3×3 convolutional kernels and 32 input channels, the first convolutional layer of the fifth stage includes a 1×1 convolutional kernel and 1 input channel, and the upsampling layers of the four stages all include 2×2 transposed convolutional layers.

[0051] The jump connection layer includes four connection layers, wherein the first connection layer connects the 32-channel feature map obtained by the first-stage convolution layer in the decoding module with the 32-channel feature map obtained by the fifth-stage convolution layer in the encoding module to obtain a 64-channel fused feature map, the second connection layer connects the 64-channel feature map obtained by the second-stage convolution layer in the decoding module with the 64-channel feature map obtained by the fourth-stage convolution layer in the encoding module to obtain a 128-channel fused feature map, the third connection layer connects the 128-channel feature map obtained by the third-stage convolution layer in the decoding module with the 128-channel feature map obtained by the third-stage convolution layer in the encoding module to obtain a 256-channel fused feature map, and the fourth connection layer connects the 256-channel feature map obtained by the fourth-stage convolution layer in the decoding module with the 256-channel feature map obtained by the second-stage convolution layer in the encoding module to obtain a 512-channel fused feature map.

[0052] It should be noted that the encoder is used to abstract geological features, and the processing operation is to perform convolution and pooling on the sample data of geophysical exploration results; the output data is the global semantics of the output of the poor geological body type, for example: water-rich confidence is 90%.

[0053] The decoder is used to restore spatial details, and the processing operation is to perform transposed convolution on the encoder output features; the output data is the spatial distribution information and type label of the output bad geological body, for example: pile number xx~xx section, rich in water, confidence level 90%.

[0054] The skip connection layer is used for multi-scale information fusion. The processing operation is to connect the signal features extracted by the encoder with the signal features extracted by the decoder. The processing operation includes spatial alignment. The output data is the output water content estimation to improve the forecast accuracy.

[0055] Furthermore, in order to improve the optimization efficiency and convergence speed of the model, effectively avoid the problems of local optimum and overfitting, and thus improve the prediction accuracy of poor geological bodies, in some embodiments, the above step S30 may include: Step S3001: constructing an original recognition model; Step S3002: training the original recognition model based on the geophysical exploration achievement sample data to obtain a candidate recognition model; Step S3003: construct a loss function and an optimizer; Step S3004: Optimizing and converging the candidate identification model based on the loss function and the optimizer to obtain a poor geological body prediction model.

[0056] It should be noted that the original recognition model includes a ReLU activation function and a Sigmoid activation function. The ReLU activation function is used after the convolution layer of the original recognition model, and the Sigmoid function is used for the output layer of the original recognition model.

[0057] It should be noted that the optimizer includes an AdaDelta adaptive learning rate optimizer. The AdaDelta adaptive learning rate optimizer can effectively solve the problem of too fast a decrease in learning rate by adaptively adjusting the learning rate for the second-order matrix estimate of the gradient update.

[0058] The loss function includes a multi-classification cross entropy loss function, and the multi-classification cross entropy loss function includes: in, Indicates The actual bad geological body type label of samples, Represents the candidate recognition model for the The sample labels predicted by samples, represents the total number of samples, Indicates the number of bad geological body types, Represents the total penalty parameter for prediction error.

[0059] In a specific implementation, the analysis device uses the geophysical detection data to perform model training on the U-Net network model, optimizes and converges the model using the loss function and the AdaDelta adaptive learning rate optimizer, and obtains the poor geological body network model.

[0060] In some embodiments, the training and optimization convergence process of the bad geological body prediction model may include: Step A1: Data construction, for example, using a measured data set of an underground cavern project of a pumped storage power station, including: The TGS360Pro system was used to collect seismic longitudinal wave velocity maps (50 images), shear wave velocity maps (50 images), and seismic profiles (50 images). The CO730 system was used to collect ground penetrating radar profiles. The collected seismic wave data and ground penetrating radar data were jointly annotated using pre-trained expert models and manual work, including four types of semantic labels (fault fracture zone, water-rich, soft rock, and cave) and pixel-level spatial coordinates.

[0061] Step A2: Model training, including: (1) Network parameter initialization: Convolutional layer initialization: use He normal initialization; Upsampling layer initialization: bilinear interpolation initialization; Skip connection layer initialization: Xavier uniform initialization is used; Output layer initialization: The semantic segmentation head is initialized with zero-mean Gaussian, and the spatial regression head is initialized with LeCun normal; (2) Construct multi-classification cross entropy loss as the loss function.

[0062] The input data is the geophysical exploration results data; according to the dual output requirements of classification task and positioning task, a weighted composite loss function is constructed to calculate the loss; back propagation and parameter update are performed, referring to the following formula: in, and is the weight coefficient, .

[0063] (3) The AdaDelta adaptive learning rate optimizer is used to optimize the model for convergence.

[0064] Due to the complex data collection environment on site, the geophysical exploration data is noisy. AdaDelta dynamically adjusts the learning rate to show stronger stability in the noise gradient generation process. At the same time, in response to the task output requirements, the type and spatial position of the unfavorable geological body need to be predicted at the same time. AdaDelta can optimize the learning rate for each parameter separately to avoid manual parameter adjustment. AdaDelta automatically completes the following operations: gradient square accumulation, parameter update step, and update cumulative variables.

[0065] Step A3: Convergence process analysis, including: By tracking the convergence index, parameters can be adjusted. The convergence index can be the following three types: (1) Classification: Comprehensively evaluate the recognition accuracy of four types of unfavorable geological bodies, referring to the following formula: Among them, TP represents the number of samples of this category correctly predicted by the model, FP represents the number of samples of this category that are not correctly predicted by the model, FN represents the number of samples of this category that are not predicted in the annotation, and mIoU represents the mean of the intersection over union ratio of each category.

[0066] (2) Positioning, which is used to measure the mean absolute error of spatial distribution information prediction, according to the following formula: in, N represents the number of samples; y true Indicates iThe actual pile number segment and spatial distribution position of the samples, where the pile number segment parameter is the sum of the starting pile number and the ending pile number, and the spatial distribution position is the lateral development position (relative position from the axis of the underground cavern, cross-sectional view x axial mapping) and longitudinal developmental position (burial depth, profile y axis mapping); y pred Indicates i The predicted pile number segment and spatial distribution position of each sample; MAE represents the spatial position error of the adverse geological body.

[0067] (3) Loss reduction rate is used to determine whether the model has entered the stable convergence stage, refer to the following formula: in, L represents the total penalty for prediction error; L t Indicates the current stage loss; L t-10 Represents the loss in the previous period; DR represents the loss decline rate.

[0068] Step A4: Use the test data set to verify the prediction accuracy. The prediction accuracy of the model for fault fracture zones, water-rich areas, soft rocks, and caves is between 85% and 90%. Compared with manual interpretation, the accuracy is increased by 5 to 10 percentage points, and the interpretation rate is increased by 70%.

[0069] Step S40: inputting the current geophysical exploration results data of the to-be-analyzed cave section of the underground cavern into the unfavorable geological body prediction model for geological analysis to obtain the unfavorable geological body information of the to-be-analyzed cave section.

[0070] It should be noted that the bad geological body information includes the bad geological body type and the spatial distribution information of the bad geological body type. For example, the bad geological body type may include fault fracture zone, water-rich, soft rock, and karst cave. The cave section to be analyzed may be an unexplored (unknown) cave section in front of the tunnel face.

[0071] It should be noted that the above-mentioned spatial distribution information can be the distribution range of the unfavorable geological body, specifically including the pile number segment and spatial orientation. The spatial orientation mainly refers to when the unfavorable geological body is only locally developed, not only the pile number segment should be marked, but also the spatial position should be indicated, such as the top / bottom / left / right side of the xx pile number~xx pile number.

[0072] In some embodiments, the underground cavern advanced geological prediction can be applied to water conservancy and hydropower projects. According to the characteristics and requirements of water conservancy and hydropower projects, the main types of unfavorable geological bodies that the analysis equipment focuses on include: fault fracture zone, water-rich, soft rock, and karst cave. The above four categories are the types of unfavorable geological bodies we expect to output. In the actual detection process, the input data of the unfavorable geological body prediction model includes: geophysical detection results data.

[0073] Step S50: Obtaining the type of surrounding rock and excavation space extension information of the tunnel face of the underground cavern.

[0074] It should be noted that the type of surrounding rock of the tunnel face can be the rock or soil around the working face of the underground cavern and the unexcavated part in front. The type of surrounding rock of the tunnel face mainly describes the properties of this part of the geological body, such as its solidity, stability, whether there are cracks, whether it contains water, etc.

[0075] It should be noted that the above excavation space extension information may be the spatial extension information of the structure predicted and revealed by the previous excavation. The excavation space extension information may include the excavated length, the time point when a specific geological layer is reached, the spatial distribution information of the adverse geological bodies revealed by the excavation, etc.

[0076] In some embodiments, the analysis equipment may classify the surrounding rock within a certain distance in front of the tunnel face based on the preliminary conclusions of the engineering geological analysis, the results of the geophysical advanced detection analysis, and the results of the geological exploration analysis to obtain the surrounding rock category of the tunnel face.

[0077] Step S60: quantifying the characteristics of the bad geological body information, the tunnel face surrounding rock type and the excavation space extension information to obtain the bad geological body development level value, tunnel face surrounding rock type value and space extension characteristic value.

[0078] In some embodiments, the analysis equipment can convert the surrounding rock categories of the face obtained by engineering geological analysis into corresponding numerical values ​​as the input layer values ​​of the neural network. The input layer values ​​of the neural network corresponding to surrounding rock categories Ⅰ, Ⅱ, Ⅲ1, Ⅲ2, Ⅳ1, Ⅳ2, and Ⅴ are: 1, 2, 3, 3.5, 4, 4.5, and 5, respectively.

[0079] In some embodiments, the value of the development level of poor geological bodies is divided into the following categories according to the prediction confidence: 0-25% corresponds to the non-existence of poor geological bodies, 25-50% (including 25%) corresponds to a low possibility of the existence of poor geological bodies, 50-75% (including 50%) corresponds to a high possibility of the existence of poor geological bodies, and 75-100% (including 75%) corresponds to the existence of poor geological bodies. The value of the development level of poor geological bodies is taken with reference to the following Table 1, which is a quantitative table of the development level of poor geological bodies.

[0080] Table 1. Quantitative table of development level of unfavorable geological bodies In some embodiments, the spatial extension characteristic value may be the spatial extension characteristic value of the structure inferred from the early excavation and revealed by the excavation. The analysis device may determine the spatial extension characteristic value by analyzing the spatial extension characteristics of the structure inferred from the early excavation and revealed by the excavation, and comparing it with the following Table 2, which is a quantification table of spatial extension characteristics.

[0081] Table 2, Quantization table of spatial extension features Step S70: performing advanced geological prediction analysis on the underground cavern based on the unfavorable geological body development level value, the tunnel face surrounding rock category value and the spatial extension characteristic value.

[0082] In some embodiments, the analysis device can carry out comprehensive advanced geological prediction analysis based on multi-information fusion on the basis of advanced geological prediction data collection, data processing and data analysis. The analysis conclusions of different prediction methods are integrated, and the types and spatial distribution information of unfavorable geological bodies and surrounding rock categories are cross-validated. Finally, a comprehensive advanced geological prediction conclusion (unfavorable geological body types and spatial distribution, surrounding rock categories, and excavation spatial extension characteristics) is obtained.

[0083] In some embodiments, the analysis device may input the unfavorable geological body development level value, the tunnel face surrounding rock category value and the spatial extension characteristic value into a pre-established geological prediction model, which is constructed by combining historical data and expert knowledge and can output the risk level of the geological conditions ahead. For example, if the unfavorable geological body development level value is high and the tunnel face surrounding rock category value indicates that the surrounding rock is very fragile, and the spatial extension characteristic shows that there is a fault extending in the direction of the current predicted tunnel section, it is predicted that there is a high risk of collapse ahead.

[0084] This embodiment obtains historical geophysical exploration results data of underground caverns; performs geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data; constructs an original recognition model, and trains the original recognition model based on the geophysical exploration results sample data to obtain a poor geological body prediction model, wherein the original recognition model includes a U-Net network model; inputs the current geophysical exploration results data of the to-be-analyzed tunnel section of the underground cavern into the poor geological body prediction model for geological analysis to obtain poor geological body information of the to-be-analyzed tunnel section, wherein the poor geological body information includes poor geological body types and spatial distribution information of the poor geological body types; obtains the surrounding rock type and excavation space extension information of the tunnel face of the underground cavern; performs feature quantification on the poor geological body information, the surrounding rock type of the tunnel face and the excavation space extension information to obtain the development of the poor geological body. level value, face surrounding rock category value and spatial extension characteristic value; based on the development level value of the adverse geological body, the face surrounding rock category value and the spatial extension characteristic value, the underground cavern is subjected to advanced geological prediction analysis; since this embodiment performs geological interpretation on historical geophysical exploration results data, and constructs geophysical exploration results sample data based on the geological interpretation results, the sample quality and model training efficiency are effectively improved; the adverse geological body prediction model obtained after training is used to perform geological analysis on the current geophysical exploration results data of the tunnel section to be analyzed, thereby accurately obtaining the adverse geological body type and spatial distribution information of the tunnel section to be analyzed; by quantifying the characteristics of the adverse geological body information, face surrounding rock category, excavation progress information and spatial extension information, it is possible to accurately evaluate the changes in the geological conditions in front of the underground cavern, thereby realizing accurate geological analysis of the underground cavern, and effectively improving the accuracy of advanced geological prediction.

[0085] refer to Figure 3 , Figure 3 It is a flow chart of the second embodiment of the intelligent prediction and analysis method for underground cavern geology of the present invention.

[0086] Based on the first embodiment above, in this embodiment, step S204 further includes: Step S2041: Determine the seismic wave anomaly segment and the ground penetrating radar anomaly segment based on the seismic wave anomaly characteristic information and the ground penetrating radar anomaly characteristic information.

[0087] It should be noted that the analysis equipment can analyze the seismic wave profile and identify abnormal areas in parameters such as seismic wave velocity and amplitude. These anomalies may indicate the existence of underground geological structures such as caves, faults or aquifers; signal processing technology (such as filtering and deconvolution) is used to identify abnormal areas in characteristics such as reflection signal intensity and phase changes, indicating the possible existence of cracks, cavities or other geological anomalies.

[0088] Step S2042: performing spatial superposition analysis on the abnormal segment of the seismic wave and the abnormal segment of the ground penetrating radar to obtain an abnormal overlapping area and an abnormal non-overlapping area.

[0089] In some embodiments, the analysis device can perform spatial overlay analysis on the seismic wave abnormal segment and the ground penetrating radar abnormal segment on the GIS platform to clarify the overlapping area between the two and the independent abnormal area. This helps to accurately locate the potential adverse geological body location and its impact range.

[0090] Step S2043: Generate geometric constraints based on the geological structure law information of the underground cavern and the type of the sample poor geological body.

[0091] In some embodiments, the analysis equipment can analyze the underground caverns, collect and analyze the face data under similar geological structures in past projects, understand the geological structure laws (such as fault direction, rock formation inclination), and set corresponding geometric constraints based on the above laws and the types of unfavorable geological bodies in the samples (such as broken zones and water-rich areas) (for example, the width of the fault zone does not exceed 1 meter and the inclination is 30°-60° northwest).

[0092] Step S2044: determining the pile number segment covered by the bad geological body of the sample bad geological body type according to the geometric constraint condition, the abnormal overlapping area and the abnormal non-overlapping area.

[0093] In some embodiments, the analysis equipment can further narrow down the specific pile number segments where the poor geological body may exist based on the geometric constraints and the specific conditions of the abnormal overlapping areas and non-overlapping areas; use drilling verification to refine the initially delineated pile number segments and ultimately determine the exact coverage range of the poor geological body.

[0094] Step S2045: Determine the sample spatial distribution information of the sample bad geological body type based on the spatial position of the pile number segment.

[0095] It should be understood that the analysis equipment can draw a spatial distribution map of the adverse geological body, including its length, width, depth and other information, based on the finally confirmed pile number segment and its corresponding geological structure characteristics.

[0096] Step S2046: constructing geophysical exploration achievement sample data based on the historical geophysical exploration achievement data, the sample poor geological body type and the sample spatial distribution information.

[0097] It should be noted that the analysis equipment can summarize historical geophysical exploration results data (such as previous seismic waves, ground penetrating radar detection reports), sample poor geological body types and sample spatial distribution information, standardize all data to ensure format consistency and readability, and create a geophysical exploration results sample database, including but not limited to geological descriptions, detection methods, abnormal characteristics, sample spatial distribution and other contents. Based on the geophysical exploration results sample database, geophysical exploration results sample data are obtained to train, test and verify the poor geological body analysis model of the analyzed tunnel section in a targeted manner.

[0098] This embodiment determines the seismic wave anomaly segment and the ground penetrating radar anomaly segment based on the seismic wave anomaly feature information and the ground penetrating radar anomaly feature information, performs spatial superposition analysis on the seismic wave anomaly segment and the ground penetrating radar anomaly segment, obtains anomaly overlapping areas and anomaly non-overlapping areas, generates geometric constraints based on the geological structure law information of the underground cavern and the sample poor geological body type, determines the pile number segment covered by the poor geological body of the sample poor geological body type according to the geometric constraints, the abnormal overlapping areas and the abnormal non-overlapping areas, determines the sample spatial distribution information of the sample poor geological body type based on the spatial position of the pile number segment, and constructs geophysical exploration results sample data based on the historical geophysical exploration results data, the sample poor geological body type and the sample spatial distribution information, so as to accurately combine the ground penetrating radar characteristics and the seismic wave characteristics for analysis, comprehensively analyze the poor geological body information from two dimensions, and improve the accuracy of poor geological analysis.

[0099] In addition, an embodiment of the present invention also proposes a computer-readable storage medium, on which an underground cavern geology intelligent prediction and analysis program is stored. When the underground cavern geology intelligent prediction and analysis program is executed by a processor, the steps of the underground cavern geology intelligent prediction and analysis method described above are implemented.

[0100] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0101] The above-mentioned computer-readable storage medium may be included in the underground cavern geology intelligent prediction and analysis device; or it may exist independently without being assembled into the underground cavern geology intelligent prediction and analysis device.

[0102] In addition, an embodiment of the present invention also proposes a computer program product, including an underground cavern geology intelligent prediction and analysis program, which implements the steps of the underground cavern geology intelligent prediction and analysis method described above when executed by a processor.

[0103] The specific implementation methods of the computer program product of the present invention are basically the same as the above-mentioned embodiments of the intelligent prediction and analysis method for underground cavern geology, and will not be repeated here.

[0104] Reference Figure 4 , Figure 4 It is a structural block diagram of the first embodiment of the underground cavern geology intelligent prediction and analysis system of the present invention.

[0105] like Figure 4 As shown, the underground cavern geological intelligent prediction and analysis system proposed in the embodiment of the present invention includes: The detection data acquisition module 10 is used to obtain historical geophysical detection results data of underground caverns; A geological interpretation module 20 is used to perform geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data; A model building module 30 is used to build an original recognition model and train the original recognition model based on the sample data of the geophysical exploration results to obtain a poor geological body prediction model, wherein the original recognition model includes a U-Net network model; The geological analysis module 40 is used to input the current geophysical exploration results data of the to-be-analyzed cave section of the underground cave into the bad geological body prediction model for geological analysis, and obtain the bad geological body information of the to-be-analyzed cave section, wherein the bad geological body information includes the bad geological body type and the spatial distribution information of the bad geological body type; The cavern information acquisition module 50 is used to acquire the type of surrounding rock and excavation space extension information of the tunnel face of the underground cavern; A feature quantification module 60 is used to perform feature quantification on the bad geological body information, the tunnel face surrounding rock category and the excavation space extension information to obtain a bad geological body development level value, a tunnel face surrounding rock category value and a space extension feature value; The advanced geological prediction and analysis module 70 is used to perform advanced geological prediction and analysis on the underground cavern based on the unfavorable geological body development level value, the tunnel face surrounding rock category value and the spatial extension characteristic value.

[0106] This embodiment obtains historical geophysical exploration results data of underground caverns; performs geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data; constructs an original recognition model, and trains the original recognition model based on the geophysical exploration results sample data to obtain a poor geological body prediction model, wherein the original recognition model includes a U-Net network model; inputs the current geophysical exploration results data of the to-be-analyzed tunnel section of the underground cavern into the poor geological body prediction model for geological analysis to obtain poor geological body information of the to-be-analyzed tunnel section, wherein the poor geological body information includes poor geological body types and spatial distribution information of the poor geological body types; obtains the surrounding rock type and excavation space extension information of the tunnel face of the underground cavern; performs feature quantification on the poor geological body information, the surrounding rock type of the tunnel face and the excavation space extension information to obtain the development of the poor geological body. level value, face surrounding rock category value and spatial extension characteristic value; based on the development level value of the adverse geological body, the face surrounding rock category value and the spatial extension characteristic value, the underground cavern is subjected to advanced geological prediction analysis; since this embodiment performs geological interpretation on historical geophysical exploration results data, and constructs geophysical exploration results sample data based on the geological interpretation results, the sample quality and model training efficiency are effectively improved; the adverse geological body prediction model obtained after training is used to perform geological analysis on the current geophysical exploration results data of the tunnel section to be analyzed, thereby accurately obtaining the adverse geological body type and spatial distribution information of the tunnel section to be analyzed; by quantifying the characteristics of the adverse geological body information, face surrounding rock category, excavation progress information and spatial extension information, it is possible to accurately evaluate the changes in the geological conditions in front of the underground cavern, thereby realizing accurate geological analysis of the underground cavern, and effectively improving the accuracy of advanced geological prediction.

[0107] The underground cavern geology intelligent prediction and analysis system provided by the present application adopts the underground cavern geology intelligent prediction and analysis method in the above-mentioned embodiment, and can solve the technical problems of underground cavern geology intelligent prediction and analysis. Compared with the prior art, the beneficial effects of the underground cavern geology intelligent prediction and analysis system provided by the present application are the same as the beneficial effects of the underground cavern geology intelligent prediction and analysis method provided by the above-mentioned embodiment, and other technical features of the underground cavern geology intelligent prediction and analysis system are the same as the features disclosed in the above-mentioned embodiment method, which will not be repeated here.

[0108] It should be understood that the above is only an example and does not constitute any limitation on the technical solution of the present invention. In specific applications, technicians in this field can make settings as needed, and the present invention does not limit this.

[0109] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, technicians in this field can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0110] In addition, for technical details not fully described in this embodiment, reference can be made to the intelligent prediction and analysis method for underground cavern geology provided in any embodiment of the present invention, and will not be repeated here.

[0111] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0112] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0113] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0114] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An intelligent prediction and analysis method for underground cavern geology, characterized in that: The underground cavern geological intelligent prediction and analysis method comprises: Obtain historical geophysical exploration data of underground caverns; Conduct geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data; Constructing an original recognition model, and training the original recognition model based on the sample data of the geophysical exploration results to obtain a poor geological body prediction model, wherein the original recognition model includes a U-Net network model; Inputting the current geophysical exploration results data of the to-be-analyzed cave section of the underground cave into the bad geological body prediction model for geological analysis, and obtaining the bad geological body information of the to-be-analyzed cave section, wherein the bad geological body information includes the bad geological body type and the spatial distribution information of the bad geological body type; Obtaining the type of surrounding rock and excavation space extension information of the underground cavern; Quantifying the characteristics of the bad geological body information, the tunnel face surrounding rock category and the excavation space extension information to obtain the bad geological body development level value, tunnel face surrounding rock category value and space extension characteristic value; Based on the unfavorable geological body development level value, the tunnel face surrounding rock category value and the spatial extension characteristic value, an advanced geological prediction analysis is performed on the underground cavern.

2. The intelligent prediction and analysis method for underground cavern geology according to claim 1, characterized in that: The historical geophysical exploration results data include seismic wave exploration results data and ground-penetrating radar exploration results data, the seismic wave exploration results data include wave velocity diagrams and seismic wave profile diagrams, and the ground-penetrating radar exploration results data include ground-penetrating radar profile diagrams; The historical geophysical exploration data of underground caverns are obtained, including: Obtain surrounding rock geological conditions and geological structural conditions based on historical geological analysis data of underground caverns; Determining a cavern prediction strategy based on the surrounding rock geological conditions and the geological structural conditions; Based on the cavern prediction strategy, seismic wave detection data and ground penetrating radar detection data are collected for the underground cavern to obtain seismic wave detection data and ground penetrating radar detection data; The seismic wave detection data and the ground penetrating radar detection data are preprocessed to obtain historical geophysical detection results data, and the preprocessing includes static correction processing, denoising processing, dynamic correction processing, time-depth conversion processing and normalization processing.

3. The intelligent prediction and analysis method for underground cavern geology according to claim 2, characterized in that: The geological interpretation of the historical geophysical exploration results data to obtain geophysical exploration results sample data includes: Performing geological interpretation feature analysis on the seismic wave detection result data and the ground penetrating radar detection result data to obtain parameter change feature information of seismic wave parameters and ground penetrating radar parameters, wherein the seismic wave parameters include seismic wave velocity, seismic wave frequency and seismic wave amplitude, and the ground penetrating radar parameters include ground penetrating radar frequency and ground penetrating radar amplitude; Extracting abnormal features from parameter change feature information of the seismic wave parameters and the ground penetrating radar parameters to obtain abnormal seismic wave feature information and abnormal ground penetrating radar feature information; Perform correlation analysis on the abnormal characteristic information of the seismic wave and the abnormal characteristic information of the ground penetrating radar to obtain the type of the sample bad geological body; Geophysical exploration achievement sample data is constructed based on the sample poor geological body type and the historical geophysical exploration achievement data.

4. The intelligent prediction and analysis method for underground cavern geology according to claim 3, characterized in that: The constructing of geophysical exploration achievement sample data based on the sample poor geological body type and the historical geophysical exploration achievement data includes: Determine the seismic wave abnormal section and the ground penetrating radar abnormal section based on the seismic wave abnormal characteristic information and the ground penetrating radar abnormal characteristic information; Performing spatial superposition analysis on the abnormal seismic wave segment and the abnormal ground penetrating radar segment to obtain an abnormal overlapping area and an abnormal non-overlapping area; Generate geometric constraints based on the geological structure law information of the underground cavern and the type of the sample unfavorable geological body; Determine the pile number segment covered by the bad geological body of the sample bad geological body type according to the geometric constraint condition, the abnormal overlapping area and the abnormal non-overlapping area; Determine the sample spatial distribution information of the sample bad geological body type based on the spatial position of the pile number segment; Geophysical exploration achievement sample data is constructed based on the historical geophysical exploration achievement data, the sample poor geological body type and the sample spatial distribution information.

5. The intelligent prediction and analysis method for underground cavern geology according to any one of claims 1 to 4, characterized in that: The constructing of the original recognition model and training the original recognition model based on the sample data of the geophysical exploration results to obtain a poor geological body prediction model includes: Determine an encoder, a decoder and a skip connection layer, and construct an original recognition model based on the encoder, the decoder and the skip connection layer, and initialize weight parameters of the original recognition model; The encoder performs convolution and pooling processing on the sample data of the geophysical exploration results, and outputs global semantic feature information of the bad geological body type; The decoder performs transposition convolution on the global semantic feature information to output spatial distribution information and type labels of bad geological bodies; The global semantic feature information, spatial distribution information and type label of the bad geological body type are spatially aligned and multi-scale feature fused through the skip connection layer; A model weight parameter is obtained based on the output results of the encoder, the decoder and the skip connection layer, and a poor geological body prediction model is constructed based on the model weight parameter.

6. The intelligent prediction and analysis method for underground cavern geology according to claim 5, characterized in that: The constructing of the original recognition model and training the original recognition model based on the sample data of the geophysical exploration results to obtain a poor geological body prediction model includes: Constructing an original recognition model, wherein the original recognition model includes a ReLU activation function and a Sigmoid activation function, wherein the ReLU activation function is used for a convolutional layer of the original recognition model, and the Sigmoid function is used for an output layer of the original recognition model; Training the original recognition model based on the geophysical exploration achievement sample data to obtain a candidate recognition model; Construct a loss function and an optimizer, wherein the optimizer includes an AdaDelta adaptive learning rate optimizer, and the loss function includes a multi-classification cross entropy loss function, wherein the multi-classification cross entropy loss function includes: in, Indicates The actual bad geological body type label of samples, Represents the candidate recognition model for the The sample labels predicted by samples, represents the total number of samples, Indicates the number of bad geological body types, represents the total penalty parameter for prediction error; The candidate identification model is optimized and converged based on the loss function and the optimizer to obtain a poor geological body prediction model.

7. An intelligent prediction and analysis system for underground cavern geology, characterized in that: The underground cavern geological intelligent prediction and analysis system comprises: Detection data acquisition module, used to obtain historical geophysical detection results data of underground caverns; A geological interpretation module is used to perform geological interpretation on the historical geophysical exploration results data to obtain geophysical exploration results sample data; A model building module is used to build an original recognition model and train the original recognition model based on the sample data of the geophysical exploration results to obtain a poor geological body prediction model, wherein the original recognition model includes a U-Net network model; A geological analysis module, used for inputting the current geophysical exploration results data of the to-be-analyzed cave section of the underground cave into the bad geological body prediction model for geological analysis, and obtaining the bad geological body information of the to-be-analyzed cave section, wherein the bad geological body information includes the bad geological body type and the spatial distribution information of the bad geological body type; A cavern information acquisition module, used to acquire the type of surrounding rock of the tunnel face and the extension information of the excavation space of the underground cavern; A feature quantification module is used to perform feature quantification on the bad geological body information, the tunnel face surrounding rock category and the excavation space extension information to obtain the bad geological body development level value, tunnel face surrounding rock category value and space extension feature value; The advanced geological prediction and analysis module is used to perform advanced geological prediction and analysis on the underground cavern based on the development level value of the unfavorable geological body, the category value of the surrounding rock of the tunnel face and the spatial extension characteristic value.

8. An intelligent prediction and analysis device for underground cavern geology, characterized in that: The underground cavern geology intelligent prediction and analysis device includes: a memory, a processor, and an underground cavern geology intelligent prediction and analysis program stored in the memory and executable on the processor, wherein the underground cavern geology intelligent prediction and analysis program is configured to implement the underground cavern geology intelligent prediction and analysis method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an underground cavern geology intelligent prediction and analysis program, and when the underground cavern geology intelligent prediction and analysis program is executed by the processor, the underground cavern geology intelligent prediction and analysis method according to any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that The computer program product includes an underground cavern geology intelligent prediction and analysis program, and when the underground cavern geology intelligent prediction and analysis program is executed by a processor, the steps of the underground cavern geology intelligent prediction and analysis method as described in any one of claims 1 to 6 are implemented.

Images