Method and system for calibrating position of tunneling equipment in coal mine tunnel

Abstract

The invention relates to the technical field of equipment positioning, in particular to a method and system for calibrating the position of tunneling equipment in a coal mine roadway. The reflection signals are analyzed, and reflection components of different geological stratums are separated; constructing a neural network model, and generating a three-dimensional reflection intensity distribution diagram with a lithologic label; geological boundary point cloud data are extracted based on the distribution map, and a geological interface curved surface is constructed; acquiring multi-source motion trail data of the tunneling equipment, and calculating geological constraint deviation; and the geological constraint deviation and the multi-source motion trail data are fused, the optimal pose estimation is output, the pose estimation value is optimized, and the final coordinate position of the tunneling equipment is obtained. According to the method, the dynamic geologic layer reflection map is constructed as the standard reference for calibration by utilizing the reflection characteristic difference of different geologic layers in the coal mine tunnel, and the physical characteristics of the geologic layers are directly utilized as the calibration basis, so that the method adapts to complex geologic changes.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment positioning, and in particular to a method and system for calibrating the position of tunneling equipment in a coal mine tunnel. Background Art

[0002] The precise positioning of coal mine tunneling equipment is crucial to the realization of intelligent mining. However, compared with the rapid development and application of intelligent technology in fully mechanized mining faces, the tunneling face has a complex construction process, a low level of intelligence, and a prominent "tunneling imbalance" contradiction, which seriously restricts the improvement of production efficiency. Among them, the autonomous measurement and dynamic positioning of the posture of coal mine underground tunneling equipment has become the primary problem facing the intelligent development of tunneling equipment.

[0003] At present, a lot of research has been carried out in the fields of inertial navigation, laser targets, UWB, iGPS and total stations to solve the problem of posture measurement of tunneling equipment. For example, some studies have used laser line targets and industrial cameras to build visual measurement systems, and obtained the position and posture parameters of the tunneling machine body by solving the laser beam image. However, the underground tunnel environment of coal mines is extremely complex, with problems such as high dust and water mist, low illumination, and large stray light interference, which makes the existing positioning technology face many challenges. For example, positioning methods based on optical signals such as lasers and infrared rays are seriously affected by signal transmission in high dust environments, resulting in a significant reduction in measurement accuracy. Although the inertial navigation system can theoretically achieve autonomous positioning, it has the problem of error accumulation over time. After long-term operation, the positioning deviation increases significantly and cannot meet actual production needs.

[0004] At the same time, many existing positioning methods require manual establishment of a benchmark behind the tunnel. When the excavation distance exceeds the effective measurement range of the system, the benchmark needs to be manually moved and calibrated again, which is cumbersome and inefficient. Therefore, it is urgent to develop a high-precision and stable excavation equipment position calibration method that can adapt to the complex environment of coal mines. Summary of the Invention

[0005] In view of the above-mentioned shortcomings of the prior art, the present invention provides a method and system for calibrating the position of tunneling equipment in a coal mine tunnel, which can effectively solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0007] The present invention provides a method for calibrating the position of tunneling equipment in a coal mine tunnel, comprising the following steps:

[0008] S1, transmit composite detection signals to the surrounding rock of the tunnel through a multi-band detection device, synchronously collect reflected signals, and extract signal characteristic parameters from the reflected signals;

[0009] S2. Perform wavelet transform denoising and time-frequency joint analysis on the reflected signal to separate the reflection components of different geological layers;

[0010] S3. Build a neural network model to identify lithologic characteristics based on the separated geological layer reflection components and match them with the geological database to generate a three-dimensional reflection intensity distribution map with lithologic labels;

[0011] S4. Extracting geological boundary point cloud data based on the three-dimensional reflection intensity distribution map with lithology labels and constructing the geological interface surface;

[0012] S5. Acquire multi-source motion trajectory data of the tunneling equipment, project the motion trajectory onto the geological interface surface, and calculate the shortest Euclidean distance between the trajectory point and the geological interface surface as the geological constraint deviation;

[0013] S6. Integrate geological constraint deviations with multi-source motion trajectory data to output the optimal pose estimate, and optimize the pose estimate to obtain the final coordinate position of the tunneling equipment.

[0014] Furthermore, the composite detection signal includes at least one of an acoustic wave signal, an electromagnetic wave signal, and a laser signal; the signal characteristic parameters include a reflection intensity gradient, a time delay difference, a frequency attenuation rate, and a phase offset; and the specific steps of extracting the signal characteristic parameters from the reflection signal include:

[0015] S11. Extract the reflection intensity gradient G of the acoustic wave signal. The calculation formula is:

[0016]

[0017] in, is the change in the intensity of the sound wave signal reflection, is the distance interval along the roadway direction, and the time delay difference of the acoustic wave signal is extracted at the same time. The formula is:

[0018]

[0019] in, and are the arrival times of the reflected signals from different geological layers respectively;

[0020] S12. Extracting the frequency attenuation rate of the reflected electromagnetic wave signal , the formula is:

[0021]

[0022] in, is the frequency change, and the phase offset of the electromagnetic wave signal is extracted at the same time , the formula is:

[0023]

[0024] in, and are the phases of the reflection signals from two different geological layers;

[0025] S13. Extract the reflection intensity gradient and surface roughness parameters of the laser signal. The reflection intensity gradient calculation method is the same as the acoustic signal feature extraction, and the surface roughness parameters are obtained by calculating the intensity fluctuation characteristics of the laser reflection signal.

[0026] Furthermore, the specific formula of the wavelet transform in step S2 is:

[0027]

[0028] in, is the time offset of the reflected signal and scale parameters The energy distribution intensity under This is the time domain reflection waveform collected by the multi-band detection device. is the wavelet basis function;

[0029] The output is the separated reflection components of each geological layer .

[0030] Furthermore, the specific steps of step S3 include:

[0031] S21, separate the reflection components of each geological layer Convert to time-spectrogram;

[0032] S22, taking the time-frequency spectrum as input, extracting deep features through the pre-trained neural network model;

[0033] The neural network model is constructed based on a convolutional neural network, and the specific construction steps include:

[0034] S201, the network architecture of the neural network model adopts ResNet-18, including 4 residual blocks and 18 layers in total, and the size of the convolution kernel of the first layer is 7×7;

[0035] S202. During the training process, the mean square error is used as the loss function, and the formula is:

[0036]

[0037] in, is the loss function, is the number of training samples, is the true geological layer characteristic value of the i-th sample, is the characteristic value of the geological layer predicted by the neural network;

[0038] S203: Training data is enhanced by random time shifting and frequency band masking, with the masking ratio not exceeding 15% of the area of ​​the time-spectrogram.

[0039] S23. Calculate the similarity between the depth feature and the lithologic standard feature vector in the geological database, and assign a lithologic identifier to each geological layer based on the similarity matching result.

[0040] Furthermore, the step of constructing the geological interface surface includes:

[0041] S31, using the Delaunay triangulation algorithm to interpolate the geological layer boundary point cloud data to generate a triangular mesh model of the geological layer boundary;

[0042] S32, fitting the triangular mesh model based on the moving least squares method to obtain a continuous geological interface surface, and outputting the geological interface surface equation and normal vector. The geological interface surface equation is:

[0043]

[0044] Among them, a, b, c and d are fitting parameters, and the normal vector is .

[0045] Furthermore, in step S5, the multi-source motion trajectory data includes the inertial measurement unit trajectory, ultra-wideband positioning data, and odometer displacement. The motion trajectory points of the tunneling equipment are projected onto the geological boundary surface, and the shortest Euclidean distance between the motion trajectory points and the geological interface surface is calculated as the deviation. The deviation calculation formula is:

[0046]

[0047] in, is the geological constraint deviation, is the coordinate of the motion trajectory point of the tunneling equipment, are the coordinates of the nearest point on the geological interface surface.

[0048] Furthermore, in step S7, the extended Kalman filter is used to fuse the geological constraint deviation and the multi-source motion trajectory data. The state vector of the extended Kalman filter is defined as , the state vector Including the three-dimensional position of the tunneling equipment in the tunnel coordinate system , three-dimensional velocity and attitude angle , the specific expression is:

[0049] ;

[0050] The state update of the extended Kalman filter adopts the Jacobian matrix.

[0051] Furthermore, the iterative optimization algorithm adopts Newton's method, and the position update formula is:

[0052]

[0053] in, is the Jacobian matrix of the pose residual to position.

[0054] A system for calibrating the position of tunneling equipment in a coal mine tunnel, comprising:

[0055] Multi-band detection module, used to transmit composite detection signals to the surrounding rock of the tunnel and simultaneously collect reflected signals;

[0056] The signal processing module is used to extract the signal characteristic parameters from the reflection signal, and perform wavelet transform denoising and time-frequency joint analysis on the reflection signal to separate the reflection components of different geological layers;

[0057] The lithology recognition module includes a pre-trained neural network model for identifying lithology characteristics based on the separated geological layer reflection components and matching them with the geological database to generate a three-dimensional reflection intensity distribution map with lithology labels;

[0058] The geological modeling module is used to extract geological boundary point cloud data based on the three-dimensional reflection intensity distribution map with lithology labels and construct the geological interface surface;

[0059] A trajectory acquisition module is used to obtain multi-source motion trajectory data of the tunneling equipment, wherein the multi-source motion trajectory data includes an inertial measurement unit trajectory, ultra-wideband positioning data, and odometer displacement;

[0060] The geological constraint module is used to project the motion trajectory onto the geological interface surface and calculate the shortest Euclidean distance between the trajectory point and the geological interface surface as the geological constraint deviation;

[0061] The pose optimization module is used to fuse geological constraint deviations and multi-source motion trajectory data through an extended Kalman filter to output the optimal pose estimate, and uses an iterative optimization algorithm based on the Jacobian matrix to optimize the pose estimate to obtain the final coordinate position of the tunneling equipment.

[0062] Compared with the prior art, the technical solution provided by the present invention has the following beneficial effects:

[0063] 1. The present invention utilizes the differences in the reflection characteristics of different geological layers in coal mine tunnels to sound waves, electromagnetic waves or lasers to construct a dynamic geological layer reflection map, and uses it as a reference for calibration. It does not rely on a fixed tunnel structure, but directly uses the physical properties of the geological layer itself as the basis for calibration, adapting to complex geological changes.

[0064] 2. The present invention transmits a composite detection signal through a multi-band detection device, and performs comprehensive feature parameter extraction, denoising, and time-frequency joint analysis on the reflected signal, which can accurately separate the reflection components of different geological layers. A neural network model is used to identify lithologic characteristics and match the geological database to generate a three-dimensional reflection intensity distribution map with lithologic labels. On this basis, a geological interface surface is constructed, providing high-precision geological constraints for the position calibration of tunneling equipment. By projecting the motion trajectory of the tunneling equipment onto the geological interface surface to calculate the geological constraint deviation and fusing it with multi-source motion trajectory data, the positioning accuracy is effectively improved. Compared with traditional methods, the positioning error can be controlled within a smaller range, meeting the demand for high-precision positioning in coal mine tunneling.

[0065] 3. The entire process, from detection signal transmission and reflection signal processing to lithology identification and geological modeling to tunneling equipment position calculation and optimization, is fully automated, eliminating the need for frequent human intervention to establish benchmarks or perform complex parameter adjustments. This reduces errors and operational workload caused by human factors. The system automatically updates the estimated position of the tunneling equipment in real time based on its motion state and changing geological conditions, improving the continuity and efficiency of tunneling operations and laying a solid foundation for unmanned, intelligent tunneling in coal mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0067] Figure 1 Schematic diagram of the method steps of the present invention;

[0068] Figure 2 It is a schematic diagram of the system module flow of the present invention. DETAILED DESCRIPTION

[0069] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0070] The present invention will be further described below with reference to the embodiments.

[0071] Example 1: Reference Figure 1 A method for calibrating the position of tunneling equipment in a coal mine tunnel comprises the following steps:

[0072] S1, transmit composite detection signals to the surrounding rock of the tunnel through a multi-band detection device, synchronously collect reflected signals, and extract signal characteristic parameters from the reflected signals;

[0073] S2. Perform wavelet transform denoising and time-frequency joint analysis on the reflected signal to separate the reflection components of different geological layers;

[0074] S3. Build a neural network model to identify lithologic characteristics based on the separated geological layer reflection components and match them with the geological database to generate a three-dimensional reflection intensity distribution map with lithologic labels;

[0075] S4. Extracting geological boundary point cloud data based on the three-dimensional reflection intensity distribution map with lithology labels and constructing the geological interface surface;

[0076] S5. Acquire multi-source motion trajectory data of the tunneling equipment, project the motion trajectory onto the geological interface surface, and calculate the shortest Euclidean distance between the trajectory point and the geological interface surface as the geological constraint deviation;

[0077] S6. Integrate geological constraint deviations with multi-source motion trajectory data to output the optimal pose estimate, and optimize the pose estimate to obtain the final coordinate position of the tunneling equipment.

[0078] Furthermore, the composite detection signal includes at least one of an acoustic wave signal, an electromagnetic wave signal, and a laser signal; the signal characteristic parameters include a reflection intensity gradient, a time delay difference, a frequency attenuation rate, and a phase offset; and the specific steps of extracting the signal characteristic parameters from the reflection signal include:

[0079] S11. Extract the reflection intensity gradient G of the acoustic wave signal. The calculation formula is:

[0080]

[0081] in, is the change in the intensity of the sound wave signal reflection, is the distance interval along the roadway direction, and the time delay difference of the acoustic wave signal is extracted at the same time. The formula is:

[0082]

[0083] in, and are the arrival times of the reflected signals from different geological layers respectively;

[0084] S12. Extracting the frequency attenuation rate of the reflected electromagnetic wave signal , the formula is:

[0085]

[0086] in, is the frequency change, and the phase offset of the electromagnetic wave signal is extracted at the same time , the formula is:

[0087]

[0088] in, and are the phases of the reflection signals from two different geological layers;

[0089] S13. Extract the reflection intensity gradient and surface roughness parameters of the laser signal. The reflection intensity gradient calculation method is the same as the acoustic signal feature extraction method, and the surface roughness parameters are obtained by calculating the intensity fluctuation characteristics of the laser reflection signal.

[0090] In one specific embodiment, a multi-band detection device is installed in a coal mine tunnel, emitting a composite detection signal consisting of acoustic, electromagnetic, and laser signals into the tunnel's surrounding rock. The acoustic signal's transmission frequency range is set to 20-20,000 Hz, the electromagnetic signal's frequency range is set to 1-10 GHz, and the laser signal uses a red laser with a wavelength of 650 nm. The reflected signal is then collected synchronously, ensuring that the sampling frequency of the acquisition device reaches 100 kHz to accurately capture the detailed features of the reflected signal.

[0091] Signal characteristic parameters are extracted from the reflected signals. For example, for acoustic signals, the measured change in reflected intensity within a 10m tunnel section is 0.5 Pa, with a 1m interval along the tunnel. The reflection intensity gradient G is calculated as 0.5 Pa / m. The arrival times of reflected signals from two different geological layers are measured to be 0.005s and 0.008s, respectively, and the time delay difference is calculated to be 0.003s. For reflected electromagnetic signals, the measured frequency change is 0.5 GHz, and the frequency attenuation rate is calculated using the formula. The phases of reflected signals from two different geological layers are measured to be 0.2π and 0.5π, respectively, and the phase offset is calculated using the formula. For laser signals, the reflection intensity gradient is calculated using a similar method to that for acoustic signals. Simultaneously, by analyzing the intensity fluctuation characteristics of the laser reflected signal, a surface roughness parameter of 0.05 μm is obtained.

[0092] The collected reflection signals were subjected to wavelet transform denoising and time-frequency analysis. Based on the wavelet transform formula, an appropriate wavelet basis function, such as the db4 wavelet basis function, was selected. The time offset range was set to 0-0.01s, and the scale parameter range was set to 1-10. The reflection signals were processed to successfully separate the reflection components from different geological layers.

[0093] The separated reflection components of each geological layer were converted into time-frequency spectra, which served as input for a pre-trained neural network model. This neural network model was built based on a convolutional neural network, employing a ResNet-18 network architecture. It consists of 18 layers with four residual blocks, and the first convolution kernel size is 7×7. During training, the mean squared error (MSE) loss function was used, and the number of training samples was 1000. The true geological layer characteristic values ​​of the i-th sample were obtained through annotation by geological experts, and the neural network predicted geological layer characteristic values ​​were output by the model. Training data was augmented using random time shifting and frequency band masking, with the masking ratio controlled at 10%. By calculating the similarity between depth features and the standard lithologic feature vectors in the geological database, lithologic labels were assigned to each geological layer, generating a 3D reflection intensity distribution map with lithologic labels.

[0094] Based on the lithologic-labeled 3D reflection intensity distribution map, we extracted geological boundary point cloud data. We interpolated the geological layer boundary point cloud data using the Delaunay triangulation algorithm to generate a triangular mesh model of the geological layer boundary. We then fitted the triangular mesh model using the moving least squares method to obtain a continuous geological interface surface. We assumed that the fitted surface equation was 3x + 2y - 5z + 10 = 0, with a normal vector of (3, 2, -5).

[0095] Acquire multi-source trajectory data for the tunneling equipment, including inertial measurement unit (IMU) trajectory data, ultra-wideband positioning data, and odometry displacement. At a specific moment, the coordinates of the tunneling equipment's trajectory point are (5, 3, 2). Project the trajectory point onto the geological boundary surface, and using the deviation calculation formula, calculate the geological constraint deviation to be 0.5 m.

[0096] An extended Kalman filter is used to fuse geological constraint deviations with multi-source motion trajectory data. The state vector of the extended Kalman filter is defined as follows: the three-dimensional position of the tunneling equipment in the roadway coordinate system is (5, 3, 2), the three-dimensional velocity is (0.1, 0.2, 0.1) m / s, and the attitude angle is (0.1, 0.2, 0.1) rad. The state is updated using the Jacobian matrix, and the optimal pose estimate is output through iterative calculation.

[0097] The Newton method was used as an iterative optimization algorithm. Based on the position update formula, the Jacobian matrix of the pose residual versus position was calculated. After multiple iterations of optimization, the final coordinate position of the tunneling equipment was (5.1, 3.05, 2.03), meeting the equipment position accuracy requirements for tunneling in the coal mine.

[0098] Furthermore, the specific formula of wavelet transform in step S2 is:

[0099]

[0100] in, is the time offset of the reflected signal and scale parameters The energy distribution intensity under This is the time domain reflection waveform collected by the multi-band detection device. is the wavelet basis function;

[0101] The output is the separated reflection components of each geological layer .

[0102] Furthermore, the specific steps of step S3 include:

[0103] S21, separate the reflection components of each geological layer Convert to time-spectrogram;

[0104] S22, taking the time-frequency spectrum as input, extracting deep features through the pre-trained neural network model;

[0105] The neural network model is built based on the convolutional neural network. The specific construction steps include:

[0106] S201,The network architecture of the neural network model adopts ResNet-18, which includes 4 residual blocks and 18 layers in total. The size of the first convolution kernel is 7×7;

[0107] S202. During the training process, the mean square error is used as the loss function, and the formula is:

[0108]

[0109] in, is the loss function, is the number of training samples, is the true geological layer characteristic value of the i-th sample, is the characteristic value of the geological layer predicted by the neural network;

[0110] S203: Training data is enhanced by random time shifting and frequency band masking, with the masking ratio not exceeding 15% of the area of ​​the time-spectrogram.

[0111] S23. Calculate the similarity between the depth feature and the lithologic standard feature vector in the geological database, and assign a lithologic identifier to each geological layer based on the similarity matching result.

[0112] Furthermore, the steps of constructing the geological interface surface include:

[0113] S31, using the Delaunay triangulation algorithm to interpolate the geological layer boundary point cloud data to generate a triangular mesh model of the geological layer boundary;

[0114] S32. Fit the triangular mesh model based on the moving least squares method to obtain a continuous geological interface surface, and output the geological interface surface equation and normal vector. The geological interface surface equation is:

[0115]

[0116] Among them, a, b, c and d are fitting parameters, and the normal vector is .

[0117] Furthermore, in step S5, the multi-source motion trajectory data includes the inertial measurement unit trajectory, ultra-wideband positioning data, and odometer displacement. The motion trajectory points of the tunneling equipment are projected onto the geological boundary surface, and the shortest Euclidean distance between the motion trajectory points and the geological interface surface is calculated as the deviation. The deviation calculation formula is:

[0118]

[0119] in, is the geological constraint deviation, is the coordinate of the motion trajectory point of the tunneling equipment, are the coordinates of the nearest point on the geological interface surface.

[0120] Furthermore, in step S7, the extended Kalman filter is used to fuse the geological constraint deviation and the multi-source motion trajectory data. The state vector of the extended Kalman filter is defined as , the state vector Including the three-dimensional position of the tunneling equipment in the tunnel coordinate system , three-dimensional velocity and attitude angle , the specific expression is:

[0121] ;

[0122] The state update of the extended Kalman filter uses the Jacobian matrix.

[0123] Furthermore, the iterative optimization algorithm adopts the Newton method, and the position update formula is:

[0124]

[0125] in, is the Jacobian matrix of the pose residual to position.

[0126] Example 2: Reference Figure 2 , a calibration system for the position of tunneling equipment in a coal mine tunnel, comprising:

[0127] Multi-band detection module, used to transmit composite detection signals to the surrounding rock of the tunnel and simultaneously collect reflected signals;

[0128] The signal processing module is used to extract the signal characteristic parameters from the reflection signal, and perform wavelet transform denoising and time-frequency joint analysis on the reflection signal to separate the reflection components of different geological layers;

[0129] The lithology recognition module includes a pre-trained neural network model for identifying lithology characteristics based on the separated geological layer reflection components and matching them with the geological database to generate a three-dimensional reflection intensity distribution map with lithology labels;

[0130] The geological modeling module is used to extract geological boundary point cloud data based on the three-dimensional reflection intensity distribution map with lithology labels and construct the geological interface surface;

[0131] The trajectory acquisition module is used to obtain multi-source motion trajectory data of the tunneling equipment, including inertial measurement unit trajectory, ultra-wideband positioning data, and odometer displacement;

[0132] The geological constraint module is used to project the motion trajectory onto the geological interface surface and calculate the shortest Euclidean distance between the trajectory point and the geological interface surface as the geological constraint deviation;

[0133] The pose optimization module is used to fuse geological constraint deviations and multi-source motion trajectory data through an extended Kalman filter to output the optimal pose estimate, and uses an iterative optimization algorithm based on the Jacobian matrix to optimize the pose estimate to obtain the final coordinate position of the tunneling equipment.

[0134] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for calibrating the position of tunneling equipment in a coal mine tunnel, characterized in that: The following steps are involved: S1, transmit composite detection signals to the surrounding rock of the tunnel through a multi-band detection device, synchronously collect reflected signals, and extract signal characteristic parameters from the reflected signals; S2. Perform wavelet transform denoising and time-frequency joint analysis on the reflected signal to separate the reflection components of different geological layers; S3. Build a neural network model to identify lithologic characteristics based on the separated geological layer reflection components and match them with the geological database to generate a three-dimensional reflection intensity distribution map with lithologic labels; S4. Extracting geological boundary point cloud data based on the three-dimensional reflection intensity distribution map with lithology labels and constructing the geological interface surface; S5. Acquire multi-source motion trajectory data of the tunneling equipment, project the motion trajectory onto the geological interface surface, and calculate the shortest Euclidean distance between the trajectory point and the geological interface surface as the geological constraint deviation; S6. Fusing geological constraint deviations with multi-source motion trajectory data, outputting an optimal pose estimate, and optimizing the pose estimate to obtain the final coordinate position of the tunneling equipment, wherein the optimization of the pose estimate is performed using an iterative optimization algorithm.

2. A method for calibrating the position of tunneling equipment in a coal mine tunnel according to claim 1, characterized in that: The composite detection signal includes at least one of an acoustic wave signal, an electromagnetic wave signal, and a laser signal; the signal characteristic parameters include a reflection intensity gradient, a time delay difference, a frequency attenuation rate, and a phase offset; and the specific steps of extracting the signal characteristic parameters from the reflection signal include: S11. Extract the reflection intensity gradient G of the acoustic wave signal. The calculation formula is: ； in, is the change in the intensity of the sound wave signal reflection, is the distance interval along the roadway direction, and the time delay difference of the acoustic wave signal is extracted at the same time. The formula is: ； in, and are the arrival times of the reflected signals from different geological layers respectively; S12. Extracting the frequency attenuation rate of the reflected electromagnetic wave signal , the formula is: ； in, is the frequency change, and the phase offset of the electromagnetic wave signal is extracted at the same time , the formula is: ； in, and are the phases of the reflection signals from two different geological layers; S13. Extract the reflection intensity gradient and surface roughness parameters of the laser signal. The reflection intensity gradient calculation method is the same as the acoustic signal feature extraction method, and the surface roughness parameters are obtained by calculating the intensity fluctuation characteristics of the laser reflection signal.

3. The method for calibrating the position of tunneling equipment in a coal mine tunnel according to claim 1, characterized in that: The specific formula of wavelet transform in step S2 is: ； in, is the time offset of the reflected signal and scale parameters The energy distribution intensity under This is the time domain reflection waveform collected by the multi-band detection device. is the wavelet basis function; The output is the separated reflection components of each geological layer .

4. The method for calibrating the position of tunneling equipment in a coal mine tunnel according to claim 1, characterized in that: The specific steps of step S3 include: S21, separate the reflection components of each geological layer Convert to time-spectrogram; S22, taking the time-frequency spectrum as input, extracting deep features through the pre-trained neural network model; The neural network model is constructed based on a convolutional neural network, and the specific construction steps include: S201, the network architecture of the neural network model adopts ResNet-18, including 4 residual blocks and 18 layers in total, and the size of the convolution kernel of the first layer is 7×7; S202. During the training process, the mean square error is used as the loss function, and the formula is: ； in, is the loss function, is the number of training samples, is the true geological layer characteristic value of the i-th sample, is the characteristic value of the geological layer predicted by the neural network; S203: Training data is enhanced by random time shifting and frequency band masking, with the masking ratio not exceeding 15% of the area of ​​the time-spectrogram. S23. Calculate the similarity between the depth feature and the lithologic standard feature vector in the geological database, and assign a lithologic identifier to each geological layer based on the similarity matching result.

5. The method for calibrating the position of tunneling equipment in a coal mine tunnel according to claim 1, characterized in that: The steps of constructing the geological interface surface include: S31, using the Delaunay triangulation algorithm to interpolate the geological layer boundary point cloud data to generate a triangular mesh model of the geological layer boundary; S32, fitting the triangular mesh model based on the moving least squares method to obtain a continuous geological interface surface, and outputting the geological interface surface equation and normal vector. The geological interface surface equation is: ； Among them, a, b, c and d are fitting parameters, and the normal vector is .

6. The method for calibrating the position of tunneling equipment in a coal mine tunnel according to claim 1, characterized in that: In step S5, the multi-source motion trajectory data includes the inertial measurement unit trajectory, ultra-wideband positioning data, and odometer displacement. The motion trajectory points of the tunneling equipment are projected onto the geological boundary surface, and the shortest Euclidean distance between the motion trajectory points and the geological interface surface is calculated as the deviation. The deviation calculation formula is: ； in, is the geological constraint deviation, is the coordinate of the motion trajectory point of the tunneling equipment, are the coordinates of the nearest point on the geological interface surface.

7. The method for calibrating the position of tunneling equipment in a coal mine tunnel according to claim 1, characterized in that: In step S6, the extended Kalman filter is used to fuse the geological constraint deviation and the multi-source motion trajectory data. The state vector of the extended Kalman filter is defined as , the state vector Including the three-dimensional position of the tunneling equipment in the tunnel coordinate system , three-dimensional velocity and attitude angle , the specific expression is: ； The state update of the extended Kalman filter adopts the Jacobian matrix.

8. The method for calibrating the position of tunneling equipment in a coal mine tunnel according to claim 1, characterized in that: The iterative optimization algorithm adopts Newton's method, and the position update formula is: ； in, is the Jacobian matrix of the pose residual to position.

9. A system for calibrating the position of tunneling equipment in a coal mine tunnel, characterized in that: include: Multi-band detection module, used to transmit composite detection signals to the surrounding rock of the tunnel and simultaneously collect reflected signals; The signal processing module is used to extract the signal characteristic parameters from the reflection signal, and perform wavelet transform denoising and time-frequency joint analysis on the reflection signal to separate the reflection components of different geological layers; The lithology recognition module includes a pre-trained neural network model for identifying lithology characteristics based on the separated geological layer reflection components and matching them with the geological database to generate a three-dimensional reflection intensity distribution map with lithology labels; The geological modeling module is used to extract geological boundary point cloud data based on the three-dimensional reflection intensity distribution map with lithology labels and construct the geological interface surface; A trajectory acquisition module is used to obtain multi-source motion trajectory data of the tunneling equipment, wherein the multi-source motion trajectory data includes an inertial measurement unit trajectory, ultra-wideband positioning data, and odometer displacement; The geological constraint module is used to project the motion trajectory onto the geological interface surface and calculate the shortest Euclidean distance between the trajectory point and the geological interface surface as the geological constraint deviation; The pose optimization module is used to fuse geological constraint deviations and multi-source motion trajectory data through an extended Kalman filter to output the optimal pose estimate, and uses an iterative optimization algorithm based on the Jacobian matrix to optimize the pose estimate to obtain the final coordinate position of the tunneling equipment.

Images