While-drilling geochemical information testing method and system based on spectral analysis

Abstract

The invention discloses a method and system for testing geochemical information while drilling based on spectral analysis, and relates to the technical field of geochemical information testing while drilling, and the system comprises a drilling module which is used for drilling a tunnel face and flushing rock slag; the rock slag extraction module is used for automatically collecting flushing fluid in the drilling process, performing multi-layer filtering and drying on the flushing fluid, and extracting dried rock slag and rock powder; the geochemical information acquisition module is used for testing and acquiring spectral data and element data of the rock powder and the rock slag by using a probe; the geochemical information analysis module comprises a control unit and a data processing unit; wherein the control unit is used for controlling operation of the drilling module, the rock slag extraction module and the geochemical information acquisition module; and the data processing unit is used for performing fusion identification analysis on the acquired spectral data and element data by adopting a multi-modal fusion identification model based on MMCA-Net, and efficiently and accurately identifying the types and contents of elements and minerals of the tunnel face rock mass rock.

Description

Technical Field

[0001] The present invention relates to the technical field of geochemical information testing while drilling, and in particular to a method and system for geochemical information testing while drilling based on spectral analysis. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] During tunnel construction, it is of great significance to quickly and accurately obtain geochemical information (referred to as geochemical information) of underground rock formations to ensure the safety of tunnel construction. Traditional methods of obtaining geochemical information usually require sampling during the drilling process and sending the samples to the laboratory for manual identification and analysis or indoor testing. However, it usually takes a lot of time to get from the tunnel entrance to the tunnel face, and it is also time-consuming and laborious to send the samples to the laboratory for testing. This method also cannot provide real-time feedback on geological changes during drilling, affecting construction efficiency and timeliness of decision-making. Therefore, it is particularly important to develop a downhole testing system that can obtain geochemical information in real time.

[0004] At present, systems or methods for testing geochemical information while drilling have been proposed in the prior art, but there are still certain problems. On the one hand, the degree of automation of the existing system is still low, and it is impossible to automatically adjust and control according to the actual situation. Equipment failures often occur, affecting the test efficiency; on the other hand, the existing method needs to be identified based on a variety of data, and its data processing is complex. It usually only identifies whether there are altered minerals, and cannot effectively identify the various types and mineral contents of minerals, and the identification accuracy is low. Summary of the invention

[0005] In order to address the deficiencies of the above-mentioned prior art, the present invention provides a method and system for testing geochemical information while drilling based on spectral analysis. The method automatically collects the flushing fluid containing rock slag and rock powder generated while drilling in real time during the drilling process, obtains the spectrum and element data of the rock mass in front of the face after on-site processing and testing, and then performs spectral analysis in combination with a machine learning algorithm to obtain the types and contents of elements and minerals in the rock during the drilling process, thereby realizing the rapid acquisition of geochemical information while drilling and ensuring the accuracy of the acquired geochemical information. The method has a higher degree of automation and higher detection and testing efficiency.

[0006] In a first aspect, the present invention provides a geochemical information testing system while drilling based on spectral analysis.

[0007] A geochemical information testing system while drilling based on spectral analysis, comprising:

[0008] Drilling module, used for drilling and washing rock debris on the tunnel face;

[0009] The slag extraction module is used to automatically collect the flushing fluid during the drilling process, and extract the dry slag rock powder after multi-layer filtration and drying of the flushing fluid;

[0010] A geochemical information acquisition module, used to collect spectral data and elemental data of the rock powder and slag using a probe test, and transmit the data to a geochemical information analysis module;

[0011] The geochemical information analysis module includes a control unit and a data processing unit; wherein the control unit is used to control the operation of the drilling module, the slag extraction module and the geochemical information acquisition module; the data processing unit is used to use a multimodal fusion recognition model based on MMCA-Net to perform fusion recognition and analysis on the collected spectral data and element data, and identify the types and contents of elements and minerals in the rock mass at the tunnel face.

[0012] In a second aspect, the present invention provides a method for testing geochemical information while drilling based on spectral analysis.

[0013] A method for testing geochemical information while drilling based on spectral analysis, comprising:

[0014] Drilling and rock washing of tunnel faces;

[0015] Automatically collect the flushing fluid during drilling, filter and dry the flushing fluid in multiple layers, and then extract the dry rock slag and rock powder;

[0016] Using a probe to test and collect spectral data and elemental data of the rock powder and slag;

[0017] The multimodal fusion recognition model based on MMCA-Net is used to fuse and identify the collected spectral data and elemental data to identify the types and contents of elements and minerals in the rock mass at the tunnel face.

[0018] In a third aspect, the present invention further provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the steps of the method described in the second aspect are completed.

[0019] In a fourth aspect, the present invention further provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps of the method described in the second aspect.

[0020] One or more of the above technical solutions have the following beneficial effects:

[0021] 1. The present invention provides a method and system for testing geochemical information while drilling based on spectral analysis. The flushing fluid containing rock slag and rock powder generated while drilling is automatically collected in real time during the drilling process. After on-site processing and testing, the spectrum and element data of the rock mass in front of the face are obtained. Then, spectral analysis is performed in combination with a machine learning algorithm to obtain the types and contents of elements and minerals in the rock during the drilling process, thereby achieving rapid and accurate identification of the types and contents of rock elements and minerals on the face, ensuring the accuracy of geochemical information. The method and system have the advantages of high recognition accuracy, fast recognition speed, strong engineering applicability, etc., and the system has a higher degree of automation, higher detection and testing efficiency, and better application prospects.

[0022] 2. According to the characteristics of drilling construction, the present invention sets up a flushing fluid collection mechanism and a geochemical information analysis mechanism, which can realize the effective collection of flushing fluid and quantitative analysis of its geochemical information during the drilling test process; based on the drilling test process, through the flushing fluid slag powder spectrum test, a large amount of geochemical characteristic information contained in the drilling process is interpreted, which can realize the timely and efficient acquisition of geochemical information, avoiding the problem that the traditional method requires a lot of manpower and material resources and has low identification efficiency.

[0023] 3. The present invention improves the existing recognition model and proposes a multimodal fusion recognition model based on MMCA-Net. The model is used to jointly analyze the extracted spectral data and elemental data. For spectral data and elemental data of different dimensions, different feature extraction networks are used to extract spectral features and elemental features according to the characteristics of data of different dimensions, so as to ensure the effectiveness of the extracted features and the accuracy of the final recognition. Then, a multi-head cross-attention mechanism and an adaptive weighting strategy are used to further extract the correlation between multi-dimensional features, and dynamically weighted fuse multi-dimensional features and their relationships. Through this fusion method, it is ensured that the model can still maintain high-precision recognition and content analysis effects under complex geological conditions, thereby providing accurate quantitative analysis of minerals. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0025] Figure 1 It is a structural schematic diagram of a geochemical information testing system while drilling based on spectral analysis according to an embodiment of the present invention;

[0026] Figure 2 The present invention is a flowchart of a method for testing geochemical information while drilling based on spectral analysis according to an embodiment of the present invention.

[0027] Among them, 1. crawler vehicle; 2. water tank; 3. water pump; 4. industrial control integrated computer; 5. water pipe; 6. base; 7. drilling rig; 8. ground feature spectrometer probe; 9. X-ray fluorescence analyzer probe; 10. rock slag drying device; 11. rock slag filtering device; 12. flushing liquid collection device. DETAILED DESCRIPTION

[0028] It should be noted that the following detailed descriptions are exemplary only, are intended to describe specific embodiments, are intended to provide further explanation of the present invention, and are not intended to limit exemplary embodiments according to the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those of ordinary skill in the art to which the present invention belongs. In addition, it should also be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0029] Embodiment 1

[0030] This embodiment provides a system for testing geochemical information while drilling based on spectral analysis, including:

[0031] Drilling module, used for drilling and washing the slag on the tunnel face;

[0032] The slag extraction module is used to automatically collect the flushing fluid during the drilling process, and extract the dry slag rock powder after multi-layer filtration and drying of the flushing fluid;

[0033] A geochemical information acquisition module, used to collect spectral data and elemental data of the rock powder and slag using a probe test, and transmit the data to a geochemical information analysis module;

[0034] The geochemical information analysis module includes a control unit and a data processing unit; wherein the control unit is used to control the operation of the drilling module, the slag extraction module and the geochemical information acquisition module; the data processing unit is used to use a multimodal fusion recognition model based on MMCA-Net to perform fusion recognition and analysis on the collected spectral data and element data, and identify the types and contents of elements and minerals in the rock mass at the tunnel face.

[0035] The geochemical information testing system based on spectral analysis while drilling proposed in this embodiment includes a drilling module, a slag extraction module, a geochemical information acquisition module, and a geochemical information analysis module. It can realize real-time processing of the flushing fluid containing slag generated during the drilling process during tunnel drilling, extract dry slag and rock powder from it, and then obtain the element type and content of the rock mass in front of the tunnel face and the mineral type and content of the rock mass through geochemical information testing. The equipment has a high degree of automation and rapid data acquisition, and has good application prospects.

[0036] The following content introduces the geochemical information testing system while drilling based on spectral analysis proposed in this embodiment in more detail.

[0037] like Figure 1 As shown, in the system proposed in this embodiment, the crawler vehicle 1 is used as the main body structure, and the crawler vehicle is provided with the above-mentioned drilling module, slag extraction module, geochemical information collection module, and geochemical information analysis module.

[0038] First, the drilling module carries out drilling work in the unexcavated section of the tunnel, and is used to drill the surrounding rock in front of the tunnel face, and produces drilling flushing fluid during the drilling process. Flushing with water during the drilling process can cool the drill bit and play a lubricating role, reduce the vibration of the drill tool, and increase the service life of the drill bit. On the other hand, it can clean the bottom of the hole and flush out the slag and rock powder. After processing, the geochemical information while drilling can be obtained. The drilling module includes a drilling rig 7 and a flushing mechanism, wherein the drilling rig 7 includes a drill rod and a drill bit, and the drill bit is arranged at the top of the drill rod and is used to drill the tunnel face; the flushing mechanism includes a water tank 2, a water pump 3, and a water pipe 5. One end of the water pipe 5 is connected to the water tank 2, and the other end is connected to the drilling rig 7. After passing through the drill rod of the drilling rig 7 (which can be arranged inside or outside the drill rod), the other end of the water pipe 5 is arranged at the drill bit. A water pump 3 is provided at one end of the water pipe 5 close to the water tank 2. By controlling the water pump 3, the flushing fluid in the water tank 2 is pumped out, thereby providing a certain pressure of water flow to the drill bit to assist drilling and slag flushing.

[0039] Secondly, the slag extraction module includes a flushing liquid collecting device 12 , a slag filtering device 11 , and a slag drying device 10 .

[0040] Specifically, the flushing fluid collecting device 12 is used to collect flushing fluid containing rock slag and rock powder during the drilling process. The end of the flushing fluid collecting device is connected to the tail of the drill rod of the drilling module through a spring. During use, the spring is always in a compressed state, so that the top of the flushing fluid collecting device is always against the palm face, so that the flushing fluid containing drilling rock slag and rock powder can be accurately obtained to prevent other impurities from entering.

[0041] Furthermore, in order to realize automatic control, the flushing liquid collection device 12 is equipped with a liquid level sensor, which is electrically connected to the control unit in the geochemical information analysis module, and is used to monitor the liquid level height in the flushing liquid collection device 12 in real time, and transmit the monitored liquid level height information to the control unit. The control unit controls the operation of the water pump in the drilling module to ensure that the system operates at the optimal liquid level and prevent the liquid level from being too high or too low and affecting the performance of the device.

[0042] Through the above settings, the liquid level sensor can accurately sense the liquid level of the flushing liquid and provide real-time feedback to the control system, so that the system can keep track of the liquid status in the collection device at any time. Specifically, based on the liquid level height feedback from the liquid level sensor, when the monitored liquid level reaches the preset upper and lower limits, the control unit controls the water pump 3 to automatically adjust the flow rate according to the feedback signal of the liquid level sensor, including: when the liquid level is higher than the upper limit, the control unit controls to reduce or suspend the inflow of flushing liquid to prevent overflow; and when the liquid level is lower than the lower limit, the control unit controls to increase the input of flushing liquid to maintain a normal collection process, wherein the change in the amount of flushing liquid inflow is proportional to the monitored liquid level over-limit value. Through this setting, automatic control of flushing liquid collection is achieved to avoid the problem of device failure caused by always using the same flow rate for collection during the flushing liquid collection process.

[0043] Furthermore, the flushing liquid collection device 12 is connected to the slag filter device 11 through a sample delivery pipeline, and the slag and rock powder in the flushing liquid are separated by the slag filter device 11. The slag filter device 11 adopts a multi-layer filtration design, and the filtration degree of the multi-layer filtration increases successively, and a particle monitoring sensor is provided between each filtration layer. The particle monitoring sensor is installed after each filtration layer and is electrically connected to the control unit, and is used to monitor the particle concentration in the slag water after filtration of the current filtration layer in real time.

[0044] Through the above settings, the control unit can judge the cleanliness of the slag water after filtration. Specifically, the judgment is made based on the particle concentration of the filtered slag water fed back by the particle monitoring sensor. When the particle concentration exceeds the set threshold, it means that the filtering effect of the current filter layer is poor. At this time, the filtered flushing liquid is introduced into the next filter layer for further filtration. Otherwise, the filtered flushing liquid is directly discharged. Through the above-mentioned step-by-step filtration method from coarse to fine, different filtration methods can be used for filtration according to the size of the slag particles collected during the current drilling, thereby improving the separation efficiency of the slag rock powder and the flushing liquid. Preferably, the slag filtering device 11 can adopt a modular design, and can flexibly adjust and replace suitable filter elements such as detachable filter screens or filter tubes according to different drilling environments.

[0045] As another embodiment, the slag filter device 11 is also equipped with a flow rate sensor and a backwash mechanism. Specifically, the flow rate sensor is installed at the inlet and outlet of the slag filter device 11, connected to the control unit, and monitors the water flow rate before and after filtration in real time. According to the difference in water flow rate before and after filtration fed back by the flow rate sensor, it is judged whether there is blockage in the slag filter device 11, and the start of the backwash mechanism is controlled according to the blockage situation. Specifically, when the difference between the monitored outlet water flow rate and the inlet water flow rate exceeds the set threshold, the control unit controls the start of the backwash mechanism to flush and clean the filter device to ensure the continuous and efficient operation of the filter device, thereby improving the automation level and work efficiency of the device.

[0046] Furthermore, the slag filtering device 11 is connected to the slag drying device 10 through a conveying device, and the slag rock powder obtained by filtering the slag filtering device 11 enters a conveying device (such as a conveyor belt), and the conveying device sends the collected slag rock powder into the drying device for drying; after drying, it is transported to a set position by the conveying device, and a spectral and elemental data detection device is provided at the set position.

[0047] Then, the geochemical information acquisition module includes a spectrum and element data detection device, which includes an X-ray fluorescence analyzer probe 9 and a ground object spectrometer probe 8, and an X-ray fluorescence analyzer and a ground object spectrometer electrically connected to the probes respectively.

[0048] Through the above arrangement, the flushing fluid collecting device 12 collects the flushing fluid generated during the drilling process, and transports it to the slag filtering device 11 through the sample input pipeline for filtration. After filtration, the slag is transported to the slag drying device 10 for drying, and the dried slag rock powder is transported to the set position for geochemical testing through the conveying device. During the test, the slag is scanned using the X-ray fluorescence analyzer probe 9 and the ground feature spectrometer probe 8, and the scanned data is transmitted to the X-ray fluorescence analyzer and the ground feature spectrometer, so as to test the dried sample and obtain the rock elements and spectral data of the slag rock powder.

[0049] Furthermore, the above-mentioned flushing liquid collecting device 12, slag filtering device 11, slag drying device 10, X-ray fluorescence analyzer probe 9, ground object spectrometer probe 8, X-ray fluorescence analyzer and ground object spectrometer are all installed on the drilling rig body platform and base 6.

[0050] Finally, the geochemical information analysis module includes a control unit and a data processing unit. The control unit includes an industrial control integrated machine 4, which is electrically connected to the water pump 3, the drilling rig 7, the X-ray fluorescence analyzer, the ground feature spectrometer, and the controllers of the flushing liquid collection device, the slag filtering device, the slag drying device, and the conveying device. The automatic operation of the entire system is controlled by the industrial control integrated machine.

[0051] The data processing unit is equipped with a pre-trained multimodal fusion recognition model based on MMCA-Net. The model serves as a rock spectral mineral recognition model. The spectral data and element data are input into the model to identify the types and contents of elements and minerals in the rock in front of the tunnel face.

[0052] Embodiment 2

[0053] This embodiment provides a method for testing geochemical information while drilling based on spectral analysis. Figure 2 As shown, the method is implemented based on a geochemical information testing system while drilling, and specifically includes the following steps:

[0054] Drilling and rock washing of tunnel faces;

[0055] Automatically collect the flushing fluid during drilling, filter and dry the flushing fluid in multiple layers, and then extract the dry rock slag and rock powder;

[0056] Using a probe to test and collect spectral data and elemental data of the rock powder and slag;

[0057] The multimodal fusion recognition model based on MMCA-Net is used to fuse and identify the collected spectral data and elemental data to identify the types and contents of elements and minerals in the rock mass at the tunnel face.

[0058] In this embodiment, the flushing fluid during the drilling process is automatically collected through the geochemical information testing system, and rock slag and rock powder are extracted therefrom. The spectral data and elemental data of the rock powder and rock slag are collected through probe testing. Then, the mineral identification and quantitative analysis are carried out based on the collected spectral and elemental data using a multimodal fusion recognition model.

[0059] Specifically, the above-mentioned multimodal fusion recognition model (MMCA-Net) adopts a multimodal convolution and attention fusion network (Multimodal Convolutional-Attention Network) architecture. In geochemical information analysis, spectral data and elemental data belong to two different modalities. The multimodal convolution and attention fusion network can fuse these two features to achieve more accurate mineral identification and elemental analysis, and can obtain the final accurate identification of geochemical information results in a timely manner.

[0060] Specifically, the multimodal fusion recognition model (MMCA-Net) consists of an input layer, a feature extraction module, an attention module, a feature fusion module and a classification and regression output layer, wherein the feature extraction module includes a multi-layer convolutional layer for spectral data and a multi-layer perceptron for elemental data, which are respectively used to extract spectral features and elemental features; the spectral features and elemental features are automatically selected through the multi-head attention mechanism in the attention module through adaptive attention allocation, and the spectral features and elemental features that are more relevant and more distinctive for mineral identification are automatically selected; the extracted features are analyzed through the multi-head cross-attention mechanism and adaptive weighting strategy of the feature fusion module to extract the correlation between multimodal features to extract the fusion features containing the spectral and elemental interaction information, and the fusion features are dynamically weighted fused with the input features to extract the joint features; finally, the joint features are output through the classification and regression layer. The mineral identification result includes the mineral type and mineral content.

[0061] The specific structure of the above multimodal fusion recognition model (MMCA-Net) is:

[0062] (1) Input layer: The input layer is used to receive raw data of different modalities, including spectral data (sequence format) and element data (vector format of specific dimensions). By inputting data of different modalities into their own independent feature extraction modules, it is ensured that the features of each modality can be fully preserved in subsequent processing to avoid information loss.

[0063] (2) Convolutional layer: A one-dimensional convolutional layer (1D CNN) is used for spectral data to extract local band features and capture local dependencies between bands; a multi-layer perceptron (MLP) is used for elemental data to extract the relationships and deep-level features between elements.

[0064] (2.1) Spectral data contain the optical reflection and absorption characteristics of rocks or minerals in different bands. For feature extraction of spectral data, the MMCA-Net model uses a multi-layer one-dimensional convolutional neural network (1DCNN) to extract features from spectral data in order to fully capture the dependencies and key features between spectral bands, that is, to extract spectral features.

[0065] (2.1.1) One-dimensional convolution layer: First, the spectral data is processed by a one-dimensional convolution operation. The convolution kernel of the one-dimensional convolution slides in the band direction, which can identify the local features of adjacent bands and capture the interdependence between bands. In this way, the correlation information between the features of different bands can be effectively extracted, thereby identifying the spectral features with discrimination.

[0066] (2.1.2) Multi-layer convolution structure: In order to gradually extract the high-order features of spectral data, multiple layers of convolution are stacked on the basis of the initial convolution layer. The multi-layer convolution structure can expand the receptive field of the model, so that it can capture longer-range band dependencies, thereby identifying the spectral characteristics of different minerals in a larger band range. As the number of convolution layers increases, multi-level spectral information from low to high levels can be gradually extracted.

[0067] (2.1.3) Pooling layer: After each convolution operation, a pooling layer (such as maximum pooling or average pooling) is used to reduce the dimension and retain important feature information. The pooling operation can remove redundant information, reduce the amount of calculation, and enhance the expressiveness of features to avoid the risk of overfitting.

[0068] (2.1.4) Output spectral feature vector: Through a series of convolution and pooling operations, a representative spectral feature vector is generated. This spectral feature vector contains the key information between different bands and provides a high-quality spectral feature foundation for subsequent multimodal fusion.

[0069] (2.2) Element data contains the concentration information of different elements in the sample, which reflects the chemical composition characteristics of the mineral. MMCA-Net extracts features from element data through a multi-layer perceptron (MLP) network to mine the relationships and hierarchical features between elements.

[0070] (2.2.1) Multilayer perceptron structure: Element data is input into the multilayer perceptron. Each layer of the perceptron consists of a fully connected layer, an activation function, and a normalization layer. The multilayer perceptron network can capture the nonlinear relationship between the concentrations of different elements through layer-by-layer transformation. After each fully connected layer, a nonlinear activation function (such as ReLU) is used to enhance the expressive power of the model, enabling it to learn more complex relationships between elements.

[0071] (2.2.2) Deep feature extraction of elemental relationships: Different elements in elemental data may have certain relationships or proportions, and these relationships are crucial to the identification of mineral types and contents. The multi-layer structure of multi-layer perceptrons can enable the model to gradually extract deep information of elemental characteristics.

[0072] (2.2.3) Normalization layer: To improve the stability of training and the generalization ability of the model, a normalization layer is added after each fully connected layer. The normalization operation can maintain the stability of feature distribution, help reduce fluctuations during training, and improve the recognition accuracy of the model on different samples.

[0073] (2.2.4) Output element feature vector: After being processed by the multi-layer perceptron, a high-dimensional element feature vector is generated. The element feature vector contains the deep relationship between the elements. This relationship is a concentrated expression of the chemical composition of the mineral, laying a solid foundation for the subsequent fusion with spectral features.

[0074] (3) Attention module: To further enhance the model’s attention to key features, MMCA-Net introduces a multi-head attention mechanism after the convolutional layer. Through adaptive attention allocation, it automatically selects the band features in the spectral features that are critical for mineral identification, while focusing on the distinguishing element features in the element data. This module can effectively reduce the interference of redundant information and ensure that the model gives priority to features with strong correlation when analyzing the content, thereby more accurately inferring the content distribution of minerals.

[0075] (4) Feature fusion layer: The feature fusion process of the MMCA-Net model combines the multi-head cross-attention mechanism and the adaptive weighting strategy to solve the problems of poor modal alignment and information loss in traditional methods. The multi-head cross-attention mechanism ensures that the model can flexibly focus on the subtle correlation between multi-modal features during the fusion process, while the weighted fusion strategy enables the model to dynamically adjust the importance of modalities under different conditions. Through this fusion design, MMCA-Net not only has higher accuracy in mineral type identification, but also can achieve accurate analysis of mineral content.

[0076] Specifically, first, through the multi-head cross-attention mechanism, the spectral features and elemental features are correlated to achieve deep fusion between modalities. This cross-attention mechanism can automatically adjust the attention according to the similarity between features, so that the features of different modalities can be accurately aligned, so as to better capture the complex relationship between them. Among them, in the multi-head attention mechanism, the spectral features are used as "queries" and the elemental features are used as "keys" and "values". Based on the attention mechanism, the similarity between the spectral features and the elemental features (i.e., the inner product) is calculated to determine the degree of attention of the spectral features to different elemental features, and the attention weight is generated; on this basis, based on the attention weight, the fused feature vector is weightedly spliced ​​so that the spectral and elemental features can be combined in the same feature space to obtain the fused feature vector. Through the above multi-head attention mechanism, a fused feature vector is obtained, which contains the interactive information of the spectral and elemental data.

[0077] Secondly, after completing the multi-head cross-attention calculation, the model concatenates the fused feature vector with the spectral feature and elemental feature vector of the original input to form a joint feature vector containing multimodal information. At this time, the model assigns different weights to each feature vector to ensure that the contribution of each modality in the fused feature is reasonable. Specifically, MMCA-Net uses an adaptive weighted generator to dynamically generate weight assignments based on the characteristics of the input data, which means that under different geological conditions or different mineral combinations, the model can automatically adjust the weighted ratio of spectral and elemental features. The weighted fused feature representation contains complete information of spectral and elemental data, and through adaptive weighting, the model can automatically focus on the most discriminative features in the current task. This fusion method ensures that the model can still maintain high-precision recognition and content analysis effects under complex geological conditions.

[0078] (5) Classification and regression output layer: In the final output layer, the model uses the joint feature vector to perform mineral identification and classification and content prediction. In the classification task, the model outputs the probability of each mineral category through the Softmax activation function; in the content analysis task, the model uses the regression output to calculate the specific content ratio of each mineral and provide accurate mineral quantitative analysis.

[0079] Furthermore, during the detection while drilling process, the spectral and elemental data obtained by testing the excavated section samples collected in advance using the previous survey information and the excavated section data are used to continuously train and update the multimodal fusion recognition model to ensure the performance and recognition accuracy of the model.

[0080] The above method proposed in this embodiment uses the relationship between rock spectral data and rock mineral types and contents to establish a rock spectral mineral identification model, and combines rock element data to achieve accurate acquisition of geochemical information of the rock mass in front of the tunnel face, thereby improving the speed and accuracy of geochemical information testing.

[0081] Embodiment 3

[0082] This embodiment provides an electronic device, including a memory and a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps in the method for testing geochemical information while drilling based on spectral analysis as described above are completed.

[0083] Embodiment 4

[0084] This embodiment further provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the steps in the geochemical information testing method while drilling based on spectral analysis as described above are completed.

[0085] The steps involved in the above embodiments 2 to 4 correspond to the method embodiment 1, and the specific implementation methods can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0086] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0087] The above description is only a preferred embodiment of the present invention. Although the specific implementation mode of the present invention is described in conjunction with the accompanying drawings, it is not a limitation of the protection scope of the present invention. Those skilled in the art should understand that on the basis of the technical solution of the present invention, various modifications or variations that can be made by those skilled in the art without creative work are still within the protection scope of the present invention.

Claims

1. A geochemical information testing system while drilling based on spectral analysis, characterized in that: include: Drilling module, used for drilling and washing the slag on the tunnel face; The slag extraction module is used to automatically collect the flushing fluid during the drilling process, and extract the dry slag rock powder after multi-layer filtration and drying of the flushing fluid; A geochemical information acquisition module, used to collect spectral data and elemental data of the rock powder and slag using a probe test, and transmit the data to a geochemical information analysis module; The geochemical information analysis module includes a control unit and a data processing unit; wherein the control unit is used to control the operation of the drilling module, the slag extraction module and the geochemical information acquisition module; the data processing unit is used to use a multimodal fusion recognition model based on MMCA-Net to perform fusion recognition and analysis on the collected spectral data and element data, and identify the types and contents of elements and minerals in the rock mass at the tunnel face.

2. A spectral analysis-based geochemical information testing system while drilling as claimed in claim 1, characterized in that: The drilling module includes a drilling rig and a flushing mechanism; The drilling rig comprises a drill rod and a drill bit, wherein the drill bit is arranged on the top of the drill rod and is used to drill holes in the tunnel face; The flushing mechanism includes a water tank, a water pump, and a water pipe. One end of the water pipe is connected to the water tank, and the other end is connected to the drilling rig. After passing through the drill rod of the drilling rig, the outlet of the other end of the water pipe is arranged at the drill bit. A water pump is arranged at the end of the water pipe close to the water tank. By controlling the water pump, the flushing liquid in the water tank is pumped out to provide a water flow with a set pressure to the drill bit to assist drilling and rock slag flushing.

3. The system for testing geochemical information while drilling based on spectral analysis according to claim 1, characterized in that: The slag extraction module includes a flushing liquid collection device, a slag filtering device, a slag drying device, and a conveying device; The flushing liquid collecting device is used to collect flushing liquid containing rock slag and rock powder during drilling. The end of the flushing liquid collecting device is connected to the tail of the drill pipe of the drilling module through a spring. During use, the spring is always in a compressed state, so that the top of the flushing liquid collecting device is always against the tunnel face; the flushing liquid collecting device is connected to the rock slag filtering device through a sample delivery pipeline, and the flushing liquid collecting device is equipped with a liquid level sensor for real-time monitoring of the liquid level height in the flushing liquid collecting device; The rock slag filtering device adopts a multi-layer filtering design, the filtering degree of the multi-layer filtering increases successively, and a particle monitoring sensor is provided between each filtering layer, which is used to monitor the particle concentration in the slag water after filtering by the current filtering layer in real time; the rock slag filtering device is also equipped with a flow rate sensor and a backwashing mechanism, and the flow rate sensor is installed at the inlet and outlet of the rock slag filtering device, which is used to monitor the water flow rate before and after filtering in real time; The slag filtering device is connected to the slag drying device through a conveying device, and the slag drying device is used to dry the slag and rock powder sent by the conveying device; the conveying device is used to convey the dry slag and rock powder to a set position.

4. The system for testing geochemical information while drilling based on spectral analysis according to claim 1, characterized in that: The geochemical information acquisition module includes a spectrum and element data detection device arranged at a set position, and the spectrum and element data detection device includes an X-ray fluorescence analyzer probe and a ground object spectrometer probe, and an X-ray fluorescence analyzer and a ground object spectrometer electrically connected to the probes respectively.

5. The system for testing geochemical information while drilling based on spectral analysis according to claim 1, characterized in that: The control unit includes an industrial control all-in-one computer, which is electrically connected to the controllers of a water pump, a drilling rig, an X-ray fluorescence analyzer, a ground feature spectrometer, a flushing liquid collection device, a slag filtering device, a slag drying device, and a conveying device to control the automatic operation of the entire system.

6. A spectral analysis-based geochemical information testing system while drilling as claimed in claim 5, characterized in that: The control includes: Based on the liquid level height feedback from the liquid level sensor, when the monitored liquid level reaches the preset upper and lower limits, the water pump is controlled to automatically adjust the flow rate, including: when the monitored liquid level is higher than the upper limit, the flow of flushing liquid is controlled to be reduced or suspended; when the monitored liquid level is lower than the lower limit, the flow of flushing liquid is controlled to be increased; wherein the change in the flow of flushing liquid is proportional to the monitored liquid level exceeding the limit value; The particle concentration of the filter residue water is judged based on the feedback from the particle monitoring sensor. When the particle concentration exceeds the set threshold, the filtered flushing liquid is introduced into the next filter layer for further filtration. Otherwise, the filtered flushing liquid is directly discharged. The blockage is judged based on the water flow rate before and after filtration fed back by the flow rate sensor, and the start-up of the backwash mechanism is controlled according to the blockage situation.

7. A method for testing geochemical information while drilling based on spectral analysis, characterized in that: include: Drilling and rock washing of the tunnel face; Automatically collect the flushing fluid during drilling, filter and dry the flushing fluid in multiple layers, and then extract the dry rock slag and rock powder; Using a probe to test and collect spectral data and elemental data of the rock powder and slag; The multimodal fusion recognition model based on MMCA-Net is used to fuse and identify the collected spectral data and elemental data to identify the types and contents of elements and minerals in the rock mass at the tunnel face.

8. The method for testing geochemical information while drilling based on spectral analysis according to claim 7, characterized in that: A multimodal fusion recognition model based on MMCA-Net is used as a rock spectrum mineral recognition model, and spectral data and element data are input into the model to identify the types and contents of elements and minerals in the rock in front of the face. The multimodal fusion recognition model consists of an input layer, a feature extraction module, an attention module, a feature fusion module, and a classification and regression output layer. Among them, the feature extraction module includes a multi-layer convolution layer for spectral data and a multi-layer perceptron for elemental data, which are used to extract spectral features and elemental features respectively; The spectral features and elemental features are automatically selected through the multi-head attention mechanism in the attention module through adaptive attention allocation, which is more relevant and more distinctive for mineral identification. The extracted features are subjected to the multi-head cross-attention mechanism and adaptive weighting strategy of the feature fusion module to analyze the correlation between multimodal features to extract fusion features containing spectral and elemental interaction information, and the fusion features are dynamically weighted fused with the input features to extract joint features; The combined features are used in the classification and regression layers to output the mineral identification results, which include the mineral type and mineral content.

9. An electronic device, characterized in that: The method comprises a memory and a processor and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps of a method for testing geochemical information while drilling based on spectral analysis as described in any one of claims 7 to 8 are completed.

10. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the steps of a method for testing geochemical information while drilling based on spectral analysis as described in any one of claims 7 to 8.

Images