Intelligent detection and identification method for unfavorable geology of vertical shaft heading machine under complex geology

Abstract

The invention discloses an intelligent unfavorable geology detection and identification method for a vertical shaft heading machine under complex geology, and the method achieves the high-precision real-time shooting of a geological structure in a well and the complete recording of whole-flow drilling data through a panoramic geology intelligent identification technology and a digital while-drilling monitoring system technology. Underground geological information and dynamic changes of a drilling process can be accurately captured, so that rich and reliable data support is provided for geological risk assessment; through the quantitative index construction and recognition classification system technology, quantitative evaluation of the strength and integrity of the rock mass in front of the tunnel face and intelligent recognition of the unfavorable geological conditions are achieved, and the unfavorable geological conditions possibly existing underground can be accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of poor geological conditions detection and identification, and in particular to a method for intelligently detecting and identifying poor geological conditions for a shaft boring machine under complex geological conditions. Background Art

[0002] During the shaft drilling process, due to the existence of faults and holes underground, the shaft boring machine may encounter adverse geological conditions during the drilling process, which may result in equipment damage, drilling delays and the risk of safety accidents.

[0003] The detection and identification of unfavorable geology during drilling is the key and difficulty of shaft drilling. How to detect and identify unfavorable geology in the complex geology during shaft boring has become an urgent problem to be solved. In view of the geological detection patent document CN117759156A, a horizontal borehole geological radar and a directional reflection imaging detection method of a borehole geological radar are proposed. By obtaining geological samples, the geological samples are imaged and processed using directional reflection of electromagnetic waves, and the images are analyzed and evaluated to determine the geological information, thereby completing the detection and identification of unfavorable geology.

[0004] However, the above detection methods have certain limitations. First, the geological structure of tunnels is complex and changeable. The structure, lithology, and fracture system of underground rock formations affect the propagation and reflection of geological radar signals, making radar detection results susceptible to interference and misunderstanding. Obtaining specific geological samples for analysis cannot reflect the overall geological characteristics of the shaft, and specific rock sampling cannot feedback shaft structural information such as shaft structural surface depth, occurrence, orientation and geometric characteristics, structural surface thickness, fillings and crushing characteristics, and cannot be applied to geological condition identification in shaft environments; at the same time, the tunnel boring machine in the shaft works in real time, and detection and identification are real-time, and sampling analysis makes detection efficiency low. Summary of the invention

[0005] Based on the problems raised by the above background technology, the purpose of the present invention is to provide an intelligent detection and identification method for poor geology of a shaft boring machine under complex geology, which collects real-time shooting and data records, uses geological feature hybrid segmentation technology to identify key geological information of real-time data, and feeds back the shaft structure information under the shaft environment, and realizes the assessment of rock strength and intelligent identification of poor geological conditions under the shaft environment through quantitative indicator construction and identification classification system technology, thereby improving the safety and efficiency of the drilling process, and solving the problem of unclear and inaccurate identification of poor geology in complex geology during the drilling process of the shaft boring machine, which leads to equipment damage, drilling delays and safety accident risks when the shaft boring machine encounters poor geology during the drilling process.

[0006] The present invention is achieved through the following technical solutions:

[0007] The first aspect of the present invention provides a method for intelligently detecting and identifying unfavorable geology of a shaft boring machine under complex geology, comprising the following steps:

[0008] Acquire geological image data of the well wall in real time, and extract key geological information of the geological image data through geological feature hybrid segmentation technology; wherein the key geological information is used to train the recognition classification system in combination with historical drilling data as a training set;

[0009] A digital monitoring while drilling system is used to record the full process data of the drilling process, and quantitative indicators are constructed based on the full process data;

[0010] Based on the quantitative indexes, an identification and classification system is used to evaluate the rock mass in front of the tunnel face, and an evaluation result of adverse geological conditions is obtained.

[0011] In the above technical solution, panoramic geological intelligent recognition technology is used to take real-time photos of the well, capture geological image data of the well wall in real time, and extract key geological information from the geological image data through geological feature hybrid segmentation technology; among which, key geological information includes wellbore structural surface depth, occurrence, orientation and geometric characteristics, structural surface thickness, filling and fragmentation feature information. The key geological information is subsequently combined with historical drilling data as a training set to train the recognition and classification system to enhance the recognition and analysis capabilities of specific address information.

[0012] A digital monitoring while drilling system is used to record the full process data of the drilling process. The digital monitoring while drilling system is a system based on data acquisition and transmission technology and parameter feature extraction technology for real-time monitoring and recording of key parameters in the drilling process. The data acquisition and transmission technology uses a mean filter to replace each point in the signal with the average value of the points in its neighborhood to smooth the signal; an adaptive filter is used to automatically adjust the parameters that change with the input signal to improve data quality; the parameter feature extraction technology is based on the combination of the Gaussian process regression model and the first convolutional neural network, which is used to process drilling data with high uncertainty and complex patterns, capture complex geological change patterns, and record them.

[0013] Quantitative indicators including rock elastic modulus estimation, fracture density and drilling efficiency index are constructed by analyzing the characteristic data in the whole process data.

[0014] The identification and classification system is used to classify and identify potential adverse geological conditions according to the rock elastic modulus, fracture density and drilling efficiency indicators in the quantitative indicators to provide a scientific basis for drilling decisions, optimize drilling parameters, and improve safety and efficiency. Based on the identification and evaluation results of adverse geological conditions, suggestions for drilling parameter adjustment and drilling route modification are provided, and the adjustment effect is tracked in real time to further optimize the drilling strategy.

[0015] In an optional embodiment, the geological feature hybrid segmentation technology includes extracting features from the geological image data through a first convolutional neural network to obtain a loss function of the first convolutional neural network model; and segmenting the geological image data through a Graph Cut technology to obtain a total energy of geological image segmentation;

[0016] The loss function of the first convolutional neural network model and the total energy of the geological image segmentation are adjusted by geological feature weights to obtain a total loss function, and the total loss function is used to extract key geological information of the geological image data.

[0017] In an optional embodiment, the mathematical model of the first convolutional neural network is expressed as follows:

[0018]

[0019] Among them, L CNN represents the loss function of the first convolutional neural network model, which is used to represent the difference between the predicted geological image features and the real geological image features; y i represents the true label; represents the probability that the predicted geological image feature of the first convolutional neural network model belongs to the i-type geological image feature; C represents the category of geological image feature.

[0020] In an optional embodiment, the mathematical model of the Graph Cut technology is expressed as follows:

[0021] E(L)=∑ p∈P D p (L p )+λ∑ (p,q)∈N S pq [L p ≠L q ]

[0022] Where E(L) represents the total energy of the entire geological image segmentation; L represents the assigned geological feature label; D p (L p ) is the matching value, which is used to measure the actual geological feature label at the p pixel point and the geological feature label L assigned to the p pixel point p The degree of match between pq is a smoothing term; λ is a balancing coefficient; P represents the set of all pixels; N represents the set of adjacent pixel pairs; [L p ≠L q ] is an indicator function used to indicate the geological feature label L assigned to the p pixel point p and the geological feature label L assigned to the q pixel point q Are they equal?

[0023] In an optional embodiment, the loss function of the first convolutional neural network and the total energy of the geological image segmentation are adjusted by geological feature weights to obtain a total loss function including:

[0024]

[0025] Among them, L custom represents the total loss function, which combines the loss of the first convolutional neural network, the energy loss of graph segmentation, and the weight adjustment of geological features; w i represents the weight of geological feature i; represents the predicted value of the model for geological feature i; represents the true value of the model for geological feature i; α and β are adjustment parameters; It represents the difference between the predicted value and the true value, and is used to emphasize the accuracy of the model in identifying geological features.

[0026] In an optional embodiment, using a digital drilling monitoring system to record the full process data during the drilling process includes the following steps:

[0027] A second convolutional neural network is used to extract time series data features from time series data;

[0028] The time series data features are used as input of a Gaussian regression process model, and the Gaussian regression process model is used to perform data prediction on the time series data.

[0029] In an optional embodiment, the quantitative indicators include rock elastic modulus estimation, fracture density and drilling efficiency indicators;

[0030] Among them, constructing the rock elastic modulus estimation through the full-process data includes: calculating geological image features and drilling parameters through a regression analysis model to obtain an estimated value of the rock elastic modulus.

[0031] In an optional embodiment, the mathematical model of the regression analysis model is expressed as follows:

[0032] E=f(Dcr,θ,ΔT)

[0033] Where E is the estimated value of rock elastic modulus; Dcr is the fracture density; θ is the structural surface inclination; ΔT is the torque change rate; and f is the regression analysis function.

[0034] In an optional embodiment, the calculation formula of the crack density is expressed as follows:

[0035]

[0036] Where Dcr is the crack density; L total is the total length of all fractures in the geological image; A is the total area of ​​the investigation area.

[0037] In an optional embodiment, the calculation formula of the drilling efficiency index is expressed as follows:

[0038]

[0039] Among them, E drill is the drilling efficiency; V is the drilling speed; P is the drilling pressure; T is the torque.

[0040] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0041] 1. In the method for intelligent detection and identification of adverse geology in complex geological shaft boring machines, the panoramic geological intelligent identification technology and the digital drilling monitoring system technology are used to achieve high-precision real-time shooting of the geological structure in the well and complete recording of the whole process drilling data. It can accurately capture the geological information underground and the dynamic changes of the drilling process, thus providing rich and reliable data support for geological risk assessment;

[0042] 2. In the method of intelligent detection and identification of unfavorable geological conditions for the complex geological shaft boring machine, quantitative indicator construction and identification classification system technology are used to achieve quantitative evaluation of the strength and integrity of the rock mass in front of the heading face and intelligent identification of unfavorable geological conditions, and can accurately predict possible unfavorable geological conditions in the well, such as faults and cavities. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0044] Figure 1 A schematic diagram of a flow chart of a method for intelligently detecting and identifying unfavorable geology of a shaft boring machine in complex geology provided in Example 1 of the present invention;

[0045] Figure 2 A schematic diagram of the structure of an electronic device provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0046] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0047] Example 1

[0048] Figure 1 The flowchart of the method for intelligently detecting and identifying unfavorable geology of a shaft boring machine under complex geology provided in Example 1 of the present invention is as follows: Figure 1 As shown, the method for intelligent detection and identification of bad geology of shaft boring machine under complex geology includes the following steps:

[0049] Acquire geological image data of the well wall in real time, and extract key geological information of the geological image data through geological feature hybrid segmentation technology; wherein the key geological information is used to train the recognition classification system in combination with historical drilling data as a training set;

[0050] A digital monitoring while drilling system is used to record the full process data of the drilling process, and quantitative indicators are constructed based on the full process data;

[0051] Based on the quantitative indexes, an identification and classification system is used to evaluate the rock mass in front of the tunnel face, and an evaluation result of adverse geological conditions is obtained.

[0052] It should be noted that the present invention uses panoramic geological intelligent recognition technology to take real-time photos of the well, capture geological image data of the well wall in real time, and extract key geological information from the geological image data through geological feature hybrid segmentation technology; wherein, key geological information includes wellbore structural surface depth, occurrence, orientation and geometric characteristics, structural surface thickness, filling and crushing feature information. The key geological information is used to combine with historical drilling data in the future as a training set to train the recognition and classification system, and enhance the recognition and analysis capabilities of specific address information.

[0053] A digital monitoring while drilling system is used to record the full process data of the drilling process. The digital monitoring while drilling system is a system based on data acquisition and transmission technology and parameter feature extraction technology for real-time monitoring and recording of key parameters in the drilling process. The data acquisition and transmission technology uses a mean filter to replace each point in the signal with the average value of the points in its neighborhood to smooth the signal; an adaptive filter is used to automatically adjust the parameters that change with the input signal to improve data quality; the parameter feature extraction technology is based on the combination of the Gaussian process regression model and the first convolutional neural network, which is used to process drilling data with high uncertainty and complex patterns, capture complex geological change patterns, and record them.

[0054] Quantitative indicators including rock elastic modulus estimation, fracture density and drilling efficiency index are constructed by analyzing the characteristic data in the whole process data.

[0055] An identification and classification system is used to classify and identify potential adverse geological conditions according to the rock elastic modulus, fracture density and drilling efficiency indicators in the quantitative indicators to provide a scientific basis for drilling decisions, optimize drilling parameters, and improve safety and efficiency.

[0056] In an optional embodiment, the geological feature hybrid segmentation technology includes extracting features from the geological image data through a first convolutional neural network to obtain a loss function of the first convolutional neural network; and segmenting the geological image data through a GraphCut technology to obtain a total energy of geological image segmentation;

[0057] The loss function of the first convolutional neural network and the total energy of the geological image segmentation are adjusted by geological feature weights to obtain a total loss function, and the total loss function is used to extract key geological information of the geological image data.

[0058] It should be noted that the geological feature hybrid segmentation technology is implemented based on the first convolutional neural network and Graph Cut technology, which is used to accurately identify and segment key geological information in geological images. Among them, the first convolutional neural network is used to extract features from geological images. Graph Cut is a classic graph cutting algorithm. In the present invention, Graph Cut technology is used to segment geological images.

[0059] In an optional embodiment, the mathematical model of the first convolutional neural network is expressed as follows:

[0060]

[0061] Among them, L CNN represents the loss function of the first convolutional neural network model, which is used to represent the difference between the predicted geological image features and the real geological image features; y i represents the true label; represents the probability that the predicted geological image feature of the first convolutional neural network model belongs to the i-type geological image feature; C represents the category of geological image feature.

[0062] It should be noted that the geological image features in the wellbore can be identified and analyzed through a variety of logging technologies. These features are of great significance for understanding oil and gas reservoirs, rock structure, and geological guidance. Among them, geological image features include fracture-type geological image features, lithology-type geological image features, texture-type geological image features, color-type geological image features, and other geological image features. The first convolutional neural network is used to identify various geological image features to provide a basis for subsequent geological image segmentation.

[0063] In an optional embodiment, the mathematical model of the Graph Cut technology is expressed as follows:

[0064] E(L)=∑ p∈P D p (L p )+λ∑ (p,q)∈N S pq [L p ≠L q ]

[0065] Where E(L) represents the total energy of the entire geological image segmentation; L represents the assigned geological feature label; D p (L p ) is the matching value, which is used to measure the actual geological feature label at the p pixel point and the geological feature label L assigned to the p pixel point p The degree of match between pq is a smoothing term; λ is a balancing coefficient; P represents the set of all pixels; N represents the set of adjacent pixel pairs; [L p ≠L q ] is an indicator function used to indicate the geological feature label L assigned to the p pixel point p and the geological feature label L assigned to the q pixel point q Are they equal?

[0066] In an optional embodiment, the loss function of the first convolutional neural network and the total energy of the geological image segmentation are adjusted by geological feature weights to obtain a total loss function including:

[0067]

[0068] Among them, L custom represents the total loss function, which combines the loss of the first convolutional neural network, the energy loss of graph segmentation, and the weight adjustment of geological features; w i represents the weight of geological feature i; represents the predicted value of the model for geological feature i; represents the true value of the model for geological feature i; α and β are adjustment parameters; It represents the difference between the predicted value and the true value, and is used to emphasize the accuracy of the model in identifying geological features.

[0069] In an optional embodiment, using a digital drilling monitoring system to record the full process data during the drilling process includes the following steps:

[0070] A second convolutional neural network is used to extract time series data features from time series data;

[0071] The time series data features are used as input of a Gaussian regression process model, and the Gaussian regression process model is used to perform data prediction on the time series data.

[0072] It should be noted that the digital monitoring while drilling system is based on data acquisition and transmission technology and parameter feature extraction technology, and is used to monitor and record key parameters during the drilling process in real time.

[0073] Among them, data acquisition and transmission are based on replacing each point in the input signal with the average value of the points in the field of the point through a mean filter to smooth the signal, and using an adaptive filter to automatically adjust the signal parameters as the input signal changes;

[0074] The parameter feature extraction technology is built based on the combination of the Gaussian process regression model and the second convolutional neural network model. It is used to process drilling data with high uncertainty and complex patterns to capture complex geological change patterns and record them.

[0075] Among them, the Gaussian process regression model is used to predict time series data with high uncertainty, and its mathematical model is expressed as follows:

[0076] y(x)~GP(m(x),k(x,x′))

[0077] Where y(x) is the target value at position x; m(x) is the mean function; k(x, x′) is the kernel function; and GP stands for Gaussian process.

[0078] Among them, the mathematical model of the second convolutional neural network is expressed as follows:

[0079] f CNN (X) = Softmax(W c *ReLU(W b *ReLU(W a *X+b a )+b b )+b c )

[0080] Among them, X is the time series data of the input; W a ,W b, W cis the weight of the convolutional layer; b a, b b ,b c is the bias term; ReLU is the activation function; Softmax is the activation function of the output layer.

[0081] The digital monitoring while drilling system uses the second convolutional neural network to automatically extract features from time series data, and uses the features as the input data of the Gaussian process regression model. The second convolutional neural network is combined with the Gaussian process regression model to obtain the mathematical model of the digital monitoring while drilling system. The mathematical model after the combination of the two is expressed as follows:

[0082] y final (X) = GP(f CNN (X),k(f CNN (X),f CNN (X′)))

[0083] Among them, y final (X) represents the final prediction output after combining the features extracted by the convolutional neural network and the prediction model of the Gaussian process regression model; f CNN (X) represents the features of the time series data X extracted by the convolutional neural network; it represents the kernel function or covariance function between the two inputs X and X′.

[0084] In an optional embodiment, the quantitative indicators include rock elastic modulus estimation, fracture density and drilling efficiency indicators;

[0085] Among them, constructing the rock elastic modulus estimation through the full-process data includes: calculating geological image features and drilling parameters through a regression analysis model to obtain an estimated value of the rock elastic modulus.

[0086] In an optional embodiment, the mathematical model of the regression analysis model is expressed as follows:

[0087] E=f(Dcr,θ,ΔT)

[0088] Where E is the estimated value of rock elastic modulus; Dcr is the fracture density; θ is the structural surface inclination; ΔT is the torque change rate; and f is the regression analysis function.

[0089] In an optional embodiment, the calculation formula of the crack density is expressed as follows:

[0090]

[0091] Where Dcr is the crack density; L total is the total length of all fractures in the geological image; A is the total area of ​​the investigation area.

[0092] In an optional embodiment, the calculation formula of the drilling efficiency index is expressed as follows:

[0093]

[0094] Among them, E drill is the drilling efficiency; V is the drilling speed; P is the drilling pressure; T is the torque.

[0095] It should be noted that fracture density refers to the total length of fractures per unit area, which is used to reflect the degree of rock fragmentation. Fracture density is used as a quantitative indicator of this method to evaluate geological conditions. The drilling efficiency index is used to analyze the relationship between drilling speed and drilling pressure and torque parameters, and to evaluate the drilling efficiency under different rock conditions.

[0096] In this embodiment, the identification and classification system is used to identify adverse geological conditions underground, which includes:

[0097] Build support vector machine models for classification and analysis to identify adverse geological conditions;

[0098] Use historical drilling data and known geological conditions as training sets to improve the accuracy and reliability of training models;

[0099] The newly collected data is input into the training model to automatically identify and classify the current geological conditions and determine whether there are adverse geological conditions;

[0100] When the adverse geological conditions predicted by the model exceed the predetermined threshold, the early warning mechanism is automatically triggered and warning signs and suggestions are directly displayed.

[0101] The rock assessment is specifically to classify the rock according to the calculated rock elastic modulus, fracture density and drilling efficiency index, identify potential adverse geological conditions to provide a scientific basis for drilling decisions, optimize drilling parameters, and improve safety and efficiency.

[0102] Furthermore, the support vector machine model is used to map the geological data x into a feature space, and then find an optimal segmentation hyperplane in the space. The specific formula involved is as follows:

[0103] w T x+b=0

[0104] Constraints: y i (w T x i +b)≥1

[0105] Among them, x i For the collected geological data; T is the weight vector; b is the bias term; y iis the class label of each sample.

[0106] It should be noted that geological data include estimated rock elastic modulus, crack density, and drilling efficiency. The estimated rock elastic modulus is an indicator to measure the ability of rock to resist deformation when subjected to force. It is the ratio of stress to strain of rock under compression. The crack density is used to describe the degree of rock fragmentation and the development of cracks. It is the proportion of cracks per unit volume or unit area. The drilling efficiency refers to the distance drilled by the drill bit per unit time during the drilling process. It is usually affected by drilling pressure, torque and geological conditions, and is presented as a percentage.

[0107] The estimated value of rock elastic modulus can help evaluate the hardness and stability of the rock, the crack density can indicate the integrity of the rock and the degree of crack development that may affect the drilling process, and the drilling efficiency can reflect the difficulty of drilling under specific geological conditions and the efficiency of the drilling equipment. The three aspects are used to comprehensively evaluate the complex geological conditions and assign different weight vectors to them according to the actual situation for calculation, thereby classifying and analyzing the geological conditions in the wellbore to identify adverse geological conditions.

[0108] Based on the identification and evaluation of adverse geological conditions, we provide suggestions for drilling parameter adjustment and drilling route modification, and track the adjustment effects in real time to further optimize the drilling strategy, including:

[0109] receiving input parameters through a drilling control system and automatically adjusting drilling operations;

[0110] Use sensors and digital while-drilling monitoring systems to collect real-time data on drill bit position, penetration rate, drilling pressure and torque;

[0111] Compare the changes in drilling speed before and after the warning to measure the efficiency improvement;

[0112] Document drilling experiences, case studies, adjustment strategies and impact evaluations in a centralized knowledge management system.

[0113] Among them, the drilling control system is an integrated software and hardware solution that uses advanced sensors, data collection technology and real-time data processing capabilities to improve drilling efficiency, safety and cost-effectiveness, monitor the status and performance of drilling equipment in real time, and automatically adjust drilling parameters according to geological conditions and actual drilling data.

[0114] Example 2

[0115] Figure 2 A schematic diagram of the structure of an electronic device provided in Embodiment 2 of the present invention, such as Figure 2As shown, the electronic device includes a processor 21, a memory 22, an input device 23 and an output device 24; the number of processors 21 in the computer device can be one or more. Figure 2 A processor 21 is taken as an example; the processor 21, the memory 22, the input device 23 and the output device 24 in the electronic device can be connected by a bus or other means. Figure 2 The example of connecting through bus is taken in the following.

[0116] The memory 22 is a computer-readable storage medium that can be used to store software programs, computer executable programs and modules. The processor 21 executes various functional applications and data processing of the electronic device by running the software programs, instructions and modules stored in the memory 22, that is, the method for intelligent detection and identification of poor geology of a shaft boring machine under complex geology of Example 1 is implemented.

[0117] The memory 22 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory 22 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 22 may further include a memory remotely arranged relative to the processor 21, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0118] The input device 23 can be used to receive the ID and password input by the user, etc. The output device 24 is used to output the network configuration page.

[0119] Example 3

[0120] Embodiment 3 of the present invention further provides a computer-readable storage medium, wherein the computer-executable instructions, when executed by a computer processor, are used to implement the method for intelligent detection and identification of poor geology of a shaft boring machine under complex geology as provided in Embodiment 1.

[0121] An embodiment of the present invention provides a storage medium containing computer executable instructions, and its computer executable instructions are not limited to the method operations provided in Example 1, but can also execute related operations in the method for intelligent detection and identification of poor geology of a shaft boring machine under complex geology provided in any embodiment of the present invention.

[0122] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An intelligent detection and identification method for poor geology of shaft boring machines under complex geology, characterized in that: The steps include: Acquire geological image data of the well wall in real time, and extract key geological information of the geological image data through geological feature hybrid segmentation technology; wherein the key geological information is used to train the recognition classification system in combination with historical drilling data as a training set; A digital monitoring while drilling system is used to record the full process data of the drilling process, and quantitative indicators are constructed based on the full process data; Based on the quantitative indexes, an identification and classification system is used to evaluate the rock mass in front of the tunnel face to obtain an evaluation result of adverse geological conditions.

2. The method for intelligent detection and identification of bad geology of shaft boring machines under complex geology according to claim 1 is characterized in that: The geological feature hybrid segmentation technology includes extracting features from the geological image data through a first convolutional neural network to obtain a loss function of the first convolutional neural network model; and segmenting the geological image data through a Graph Cut technology to obtain a total energy of geological image segmentation; The loss function of the first convolutional neural network model and the total energy of the geological image segmentation are adjusted by geological feature weights to obtain a total loss function, and the total loss function is used to extract key geological information of the geological image data.

3. The method for intelligent detection and identification of poor geology of shaft boring machines under complex geology according to claim 2 is characterized in that: The mathematical model of the first convolutional neural network is expressed as follows: Among them, L CNN represents the loss function of the first convolutional neural network model, which is used to represent the difference between the predicted geological image features and the real geological image features; y i represents the true label; represents the probability that the predicted geological image feature of the first convolutional neural network model belongs to the i-type geological image feature; C represents the category of geological image feature.

4. The method for intelligent detection and identification of poor geology of shaft boring machines under complex geology according to claim 2 is characterized in that: The mathematical model of the Graph Cut technology is expressed as follows: Where E(L) represents the total energy of the entire geological image segmentation; L represents the assigned geological feature label; D p (L p ) is the matching value, which is used to measure the actual geological feature label at the p pixel point and the geological feature label L assigned to the p pixel point p The degree of match between pq is a smoothing term; λ is a balancing coefficient; P represents the set of all pixels; N represents the set of adjacent pixel pairs; [L p ≠L q ] is an indicator function used to indicate the geological feature label L assigned to the p pixel point p and the geological feature label L assigned to the q pixel point q Are they equal? 5. The method for intelligent detection and identification of bad geology of shaft boring machines under complex geology according to claim 2 is characterized in that: The loss function of the first convolutional neural network and the total energy of the geological image segmentation are adjusted by geological feature weights to obtain a total loss function including: Among them, L custom represents the total loss function, which combines the loss of the first convolutional neural network, the energy loss of graph segmentation, and the weight adjustment of geological features; w i represents the weight of geological feature i; represents the predicted value of the model for geological feature i; represents the true value of the model for geological feature i; α is the adjustment function of the total energy of the entire geological image segmentation; β is the adjustment parameter of the predicted difference; It represents the difference between the predicted value and the true value, which is used to emphasize the accuracy of the model in identifying geological features; E(L) represents the total energy of the entire geological image segmentation; L CNN Represents the loss function of the first convolutional neural network model, which is used to represent the difference between the predicted geological image features and the real geological image features.

6. The method for intelligent detection and identification of bad geology of shaft boring machines under complex geology according to claim 1 is characterized in that: The use of a digital monitoring while drilling system to record the full process data during drilling includes the following steps: A second convolutional neural network is used to extract time series data features from time series data; The time series data features are used as input of a Gaussian regression process model, and the Gaussian regression process model is used to perform data prediction on the time series data.

7. The method for intelligent detection and identification of bad geology of shaft boring machines in complex geology according to claim 1 is characterized in that: The quantitative indicators include rock elastic modulus estimation, fracture density and drilling efficiency indicators; Among them, constructing the rock elastic modulus estimation through the full-process data includes: calculating geological image features and drilling parameters through a regression analysis model to obtain an estimated value of the rock elastic modulus.

8. The method for intelligent detection and identification of poor geology of shaft boring machines in complex geology according to claim 7 is characterized in that: The mathematical model of the regression analysis model is expressed as follows: E=f(Dcr,θ,ΔT) Where, E is the estimated value of rock elastic modulus; Dcr is the fracture density; θ is the structural surface inclination; ΔT is the torque change rate; and f is the regression analysis function.

9. The method for intelligent detection and identification of bad geology of shaft boring machines in complex geology according to claim 7, characterized in that: The calculation formula of the crack density is expressed as follows: Where Dcr is the crack density; L total is the total length of all fractures in the geological image; A is the total area of ​​the investigation area.

10. The method for intelligent detection and identification of bad geology of shaft boring machines in complex geology according to claim 7, characterized in that: The calculation formula of the drilling efficiency index is as follows: Among them, E drill is the drilling efficiency; V is the drilling speed; P is the drilling pressure; T is the torque.

Images