Underground coal mine intelligent rapid tunneling system

Abstract

The invention discloses a coal mine underground intelligent rapid tunneling system, which comprises an optical fiber inertial navigation measurement unit and a laser tracking measurement unit, a data acquisition module receives three-axis acceleration data, three-axis angular velocity data and three-dimensional space coordinate data, a central processing module inputs the data to an extended Kalman filter, and the extended Kalman filter outputs the data to the laser tracking measurement unit. The method comprises the following steps: generating a six-degree-of-freedom pose state data stream, inputting the six-degree-of-freedom pose state data stream into a long short-term memory neural network by an intelligent prediction module, generating a future pose wriggling trend track, generating a continuous micro hydraulic cylinder stroke compensation instruction by a control decision module, and driving a hydraulic cylinder to generate micro telescopic motion by an electro-hydraulic servo driving module. Therefore, the pose creep trend of the heading machine can be accurately pre-judged, continuous trace dynamic compensation is realized, the zigzag fluctuation of the heading track is effectively inhibited, the straightness and forming precision of roadway heading are improved, the compensation action is free of mechanical interference, and continuous heading operation is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of coal mine tunneling technology, specifically to an intelligent and rapid underground tunneling system for coal mines. Background Technology

[0002] This invention relates to the field of coal mine tunneling technology, specifically to an intelligent and rapid underground tunneling system for coal mines, particularly concerning the technology of tunneling machine position monitoring and precise control. Underground roadway tunneling is a core pre-process in coal mining. As the core operating equipment, the position control accuracy of the tunneling machine directly determines the straightness and forming quality of the roadway, and is also crucial to ensuring the efficiency and construction safety of subsequent continuous transportation, roadway support, and other processes. Therefore, precise position monitoring and correction of the tunneling machine is a core technical point of intelligent underground tunneling.

[0003] Currently, underground tunneling machines in coal mines generally adopt a posture control mode of deviation accumulation-threshold triggering-discrete correction. During tunneling operations, the tunneling machine body will undergo slow posture creep due to factors such as coal seam undulation, soft floor, and hydraulic system drift. Existing technologies mostly acquire posture data through single or simple combination of measurement methods, which lacks continuity, real-time performance, and accuracy in data acquisition. Furthermore, the control system lacks the ability to predict posture creep trends. It is necessary to wait until the deviation accumulates to a preset threshold before driving the hydraulic cylinder to perform a step-like discrete correction operation, correcting the accumulated deviation through a single large-amplitude extension.

[0004] Existing technologies lack the ability to fuse and process high-precision, continuous pose data and the ability to predict pose creep trends. They can only achieve discrete step-like correction triggered by thresholds, and cannot perform micro-compensation control for trend-based pose creep. This not only results in the tunneling trajectory being a sawtooth-shaped broken line around the design axis, making it difficult to achieve smooth and continuous trajectory control, but also interferes with the continuous advancement of the tunneling machine due to large discrete correction actions. At the same time, it cannot suppress the accumulation of deviations at the root, reducing the straightness and forming accuracy of the tunnel excavation. Summary of the Invention

[0005] The present invention aims to at least partially solve the technical problems in the above-mentioned technologies.

[0006] Therefore, this invention discloses an intelligent rapid tunneling system for underground coal mines, comprising:

[0007] The fiber optic inertial navigation measurement unit and the laser tracking measurement unit are respectively installed on the tunneling machine;

[0008] The data acquisition module is used to receive the three-axis acceleration data and three-axis angular velocity data of the tunneling machine output by the fiber optic inertial navigation measurement unit, and to receive the three-dimensional spatial coordinate data of the tunneling machine output by the laser tracking measurement unit;

[0009] The central processing module is used to input the triaxial acceleration data, the triaxial angular velocity data, and the three-dimensional spatial coordinate data into the extended Kalman filter. The extended Kalman filter uses the triaxial acceleration data and the triaxial angular velocity data as state predictions and the three-dimensional spatial coordinate data as observation updates, and iteratively executes state prediction and observation updates to generate a six-degree-of-freedom pose state data stream of the tunneling machine.

[0010] The intelligent prediction module is used to receive the six-degree-of-freedom pose state data stream of the tunneling machine, input the six-degree-of-freedom pose state data stream into the long short-term memory neural network, and the long short-term memory neural network generates predicted six-degree-of-freedom pose state data within the future time window based on the historical six-degree-of-freedom pose state data stream, so as to generate the future pose creep trend trajectory of the tunneling machine.

[0011] The control decision module is used to receive the future posture creep trend trajectory of the tunneling machine, calculate the difference between the future posture creep trend trajectory and the standard posture creep trend trajectory and compare it with a preset micro-fluctuation threshold. When the difference exceeds the preset micro-fluctuation threshold, the module calculates the ideal compensation displacement curve to offset the difference based on the kinematic model of the tunneling machine, so as to generate a continuous micro-hydraulic cylinder stroke compensation command.

[0012] The electro-hydraulic servo drive module is used to receive the continuous micro-hydraulic cylinder stroke compensation command, convert the continuous micro-hydraulic cylinder stroke compensation command into an electro-hydraulic servo control signal, and drive the hydraulic cylinder to generate a micro-extension and contraction motion synchronized with the ideal compensation displacement curve.

[0013] The intelligent rapid tunneling system for underground coal mines disclosed in this invention can accurately predict the creep trend of the tunneling machine's position and posture, achieve continuous micro-dynamic compensation, effectively suppress sawtooth fluctuations in the tunneling trajectory, improve the straightness and forming accuracy of the roadway, and ensure continuous tunneling operations without mechanical interference.

[0014] In addition, the intelligent rapid tunneling system for underground coal mines disclosed in this invention may also have the following additional technical features:

[0015] Furthermore, in the data acquisition module, the triaxial acceleration data is... The triaxial angular velocity data are The three-dimensional spatial coordinate data is ;

[0016] In the central processing module, the six-degree-of-freedom pose state data stream is ,in, This refers to the three-dimensional coordinate position in the geodetic coordinate system. These are roll angle, pitch angle, and yaw angle, respectively.

[0017] In the intelligent prediction module, the historical six-DOF pose state data stream is The predicted six-degree-of-freedom pose state data is The future pose creep trend trajectory is ,in, For the pace of history, The step size of the future time window;

[0018] In the control decision module, the standard pose creep trend trajectory The future pose creep trend trajectory and the standard pose creep trend trajectory The difference between them is The ideal compensation displacement curve is The continuous micro-hydraulic cylinder stroke compensation command is ,in, This refers to the number of hydraulic cylinders used for compensation in the tunneling machine;

[0019] In the electro-hydraulic servo drive module, the electro-hydraulic servo control signal is: .

[0020] Furthermore, in the central processing module, the extended Kalman filter specifically comprises:

[0021] The state vector is The state transition matrix is The observation matrix is The process noise covariance matrix is ​​as follows: diagonal matrix form The observed noise covariance matrix is ​​as follows: diagonal matrix form The error covariance matrix is of , wherein the observation matrix Only observe the state vector In .

[0022] Furthermore, the central processing module generates the six-degree-of-freedom pose state data stream of the tunneling machine. Specifically:

[0023] S1.1: Set the initial value for iteration When, initialize the state vector and the error covariance matrix ;

[0024] S1.2: When At that time, the triaxial acceleration data is... The triaxial angular velocity data are The input is fed into the extended Kalman filter, and a linear state equation is constructed based on the tunneling machine's kinematics. This equation is then processed through the state transition matrix. After linearization, the predicted state vector is calculated. and prediction error covariance matrix ;

[0025] S1.3: The first The three-dimensional spatial coordinate data mentioned in step [step] are The input is fed into the extended Kalman filter, the Kalman gain is calculated, and the predicted state vector is corrected using the Kalman gain. The updated optimal state vector is obtained. And update the error covariance matrix simultaneously. ;

[0026] S1.4: From the optimal state vector Extract the first six dimensions of data to generate the six-degree-of-freedom pose state data stream. vectors in ;

[0027] S1.5: Repeat steps S1.2 to S1.4 to finally generate the six-degree-of-freedom pose state data stream. .

[0028] Furthermore, in the intelligent prediction module, the long short-term memory neural network specifically comprises:

[0029] The input layer dimension is The output layer dimension is The hidden layers are 3, and the loss function is mean squared error.

[0030] Furthermore, in the intelligent prediction module, the future pose creep trend trajectory of the tunneling machine is generated. Specifically:

[0031] S2.1: From the six-degree-of-freedom pose state data stream Extract the most recent Step data, forming the historical six-DOF pose state data stream. ;

[0032] S2.2: For the historical six-degree-of-freedom pose state data stream Preprocessing is required;

[0033] S2.3: The preprocessed historical six-DOF pose state data stream The data is input into the long short-term memory neural network and forward inference is performed to ultimately generate the future pose creeping trend trajectory. .

[0034] Furthermore, in the control decision module, the kinematic model of the tunneling machine is specifically as follows:

[0035] The input is the future pose creep trend trajectory. and the standard pose creep trend trajectory The difference between The output is the ideal compensation displacement curve. In The input and output have a linear mapping relationship. ,in, This is the first relational mapping matrix.

[0036] Furthermore, in the control decision module, the ideal compensation displacement curve is generated as follows: Specifically:

[0037] S3.1: Calculate the future pose creep trend trajectory and the standard pose creep trend trajectory The difference between ;

[0038] S3.2: Calculate the difference at each step. Compare with the preset small fluctuation threshold, if the difference is... If the fluctuation does not exceed the preset micro-fluctuation threshold, no compensation is required; otherwise, compensation is based on a linear mapping relationship. Calculate the ideal compensation displacement curve In ;

[0039] S3.3: Repeat steps S3.1 and S3.2, calculating sequentially. The ideal compensation displacement curve for each step In .

[0040] Furthermore, in the control decision module, the continuous micro-hydraulic cylinder stroke compensation command is generated. Specifically:

[0041] for The ideal compensation displacement curve for each step In According to the linear mapping relationship The continuous micro-hydraulic cylinder stroke compensation commands are calculated sequentially. In ,in, This is the second relational mapping matrix.

[0042] Additional features and advantages of this invention will be set forth in the description which follows, or may be learned by practicing the invention. Attached Figure Description

[0043] The technical solution and beneficial effects of the present invention will become apparent and readily understood from the following description in conjunction with the accompanying drawings, wherein:

[0044] Figure 1 This is an architectural diagram of the intelligent rapid tunneling system for underground coal mines according to the present invention;

[0045] Figure 2 This is a flowchart of the central processing module of the present invention;

[0046] Figure 3 This is a flowchart illustrating the workflow of the intelligent prediction module of the present invention.

[0047] Figure 4 This is a flowchart of the control decision module of the present invention. Detailed Implementation

[0048] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0049] The intelligent rapid tunneling system for underground coal mines disclosed in this invention will now be described with reference to the accompanying drawings.

[0050] like Figure 1 As shown, an intelligent rapid tunneling system for underground coal mines includes:

[0051] The fiber optic inertial navigation measurement unit and the laser tracking measurement unit are respectively installed on the tunneling machine;

[0052] The data acquisition module is used to receive the three-axis acceleration data and three-axis angular velocity data of the tunneling machine output by the fiber optic inertial navigation measurement unit, and to receive the three-dimensional spatial coordinate data of the tunneling machine output by the laser tracking measurement unit;

[0053] The central processing module is used to input triaxial acceleration data, triaxial angular velocity data and three-dimensional spatial coordinate data into the extended Kalman filter. The extended Kalman filter uses triaxial acceleration data and triaxial angular velocity data as state prediction and three-dimensional spatial coordinate data as observation update, and iteratively executes state prediction and observation update to generate a six-degree-of-freedom pose state data stream of the tunneling machine.

[0054] The intelligent prediction module is used to receive the six-degree-of-freedom pose state data stream of the tunneling machine, input the six-degree-of-freedom pose state data stream into the long short-term memory neural network, and generate the predicted six-degree-of-freedom pose state data within the future time window based on the historical six-degree-of-freedom pose state data stream, so as to generate the future pose creep trend trajectory of the tunneling machine.

[0055] The control decision module is used to receive the future posture creep trend trajectory of the tunneling machine, calculate the difference between the future posture creep trend trajectory and the standard posture creep trend trajectory and compare it with the preset micro-fluctuation threshold. When it exceeds the preset micro-fluctuation threshold, it calculates the ideal compensation displacement curve to offset the difference based on the kinematic model of the tunneling machine, so as to generate a continuous micro-hydraulic cylinder stroke compensation command.

[0056] The electro-hydraulic servo drive module is used to receive continuous micro-hydraulic cylinder stroke compensation commands, convert the continuous micro-hydraulic cylinder stroke compensation commands into electro-hydraulic servo control signals, and drive the hydraulic cylinder to produce micro-extension and contraction movements synchronized with the ideal compensation displacement curve.

[0057] In the data acquisition module, the triaxial acceleration data is: The three-axis angular velocity data are Three-dimensional spatial coordinate data are ;

[0058] In the central processing module, the six-DOF pose state data stream is... ,in, This refers to the three-dimensional coordinate position in the geodetic coordinate system. These are roll angle, pitch angle, and yaw angle, respectively.

[0059] In the intelligent prediction module, the historical six-DOF pose state data stream is... Predicted six-DOF pose state data is The future pose creep trend trajectory is as follows ,in, For the pace of history, The step size for the future time window;

[0060] In the control decision module, the standard pose creep trend trajectory Future posture creep trend trajectory and standard pose creep trend trajectory The difference between them is The ideal compensation displacement curve is The continuous micro-hydraulic cylinder stroke compensation command is: ,in, This refers to the number of hydraulic cylinders used for compensation in the tunneling machine;

[0061] In the electro-hydraulic servo drive module, the electro-hydraulic servo control signal is: .

[0062] Example 1

[0063] In the central processing module, the extended Kalman filter is specifically as follows:

[0064] The state vector is ;

[0065] State vector Includes three-dimensional coordinates 3D attitude angle 3D velocity and 3D coordinates Using the starting point of the underground tunneling face as the origin, and adopting the unified coordinate system for underground coal mines, the three-dimensional attitude angles are... The sampling frequency is 100Hz, which is used to characterize the spatial attitude of the tunneling equipment. The sampling interval can also be dynamically adjusted according to the underground tunneling speed to ensure the real-time performance of the data.

[0066] The state transition matrix is ;

[0067] State transition matrix Based on the kinematic model of the tunneling equipment, the Euler angle update equation is adopted and dynamically adjusted according to the attitude of the equipment.

[0068] The observation matrix is ;

[0069] Observation matrix The matrix is ​​sparse, and the non-zero elements correspond only to three-dimensional coordinates. All values ​​are 1, and the rest are 0, which can effectively reduce the amount of data processing.

[0070] The process noise covariance matrix is ​​as follows: diagonal matrix form ;

[0071] Process noise covariance matrix The diagonal elements take values ​​in the range of (1e-6, 1e-4), with the covariance values ​​corresponding to velocity and angular velocity being slightly larger, and the covariance values ​​corresponding to coordinates and attitude angles being smaller, to adapt to the noise characteristics of the downhole vibration environment.

[0072] The observation noise covariance matrix is ​​as follows: diagonal matrix form ;

[0073] Observation noise covariance matrix The diagonal elements are all 1e-5, corresponding to the observation error range of the downhole laser positioning sensor.

[0074] The error covariance matrix is of Among them, the observation matrix Only observe the state vector In .

[0075] In the central processing module, a six-degree-of-freedom pose state data stream of the tunneling machine is generated. Specifically:

[0076] S1.1: Set the initial value for iteration When initializing the state vector And error covariance matrix ;

[0077] Iteration initial value When initializing the state vector And error covariance matrix During the initialization phase, the initial position coordinates of the tunneling equipment are obtained through an underground laser positioning system. The attitude angle is calibrated using a gyroscope. The initial values ​​of velocity and angular velocity are both set to 0, that is... Error covariance matrix The diagonal elements are all set to 1e-3 to represent the uncertainty of the initial state and to fit the error range of the initial positioning in the well.

[0078] S1.2: When At that time, the triaxial acceleration data were... and triaxial angular velocity data are The input is fed into an extended Kalman filter, and a linear state equation is constructed based on the tunneling machine's kinematics. The state transition matrix is ​​then used to... After linearization, the predicted state vector is calculated. and prediction error covariance matrix ;

[0079] when At that time, triaxial acceleration Data is collected by downhole accelerometers and preprocessed by Kalman filtering to eliminate vibration noise. The sampling frequency is consistent with the state vector. The linear state equation adopts the Newton-Euler equation of motion and is linearized by first-order Taylor expansion. Downhole vibration compensation coefficients are added during the prediction process and dynamically adjusted according to the stability of the downhole surrounding rock to avoid prediction deviations caused by vibration.

[0080] S1.3: The first Step 3D spatial coordinate data The input is fed into an extended Kalman filter, the Kalman gain is calculated, and the predicted state vector is corrected using the Kalman gain. The updated optimal state vector is obtained. And update the error covariance matrix simultaneously. ;

[0081] Three-dimensional spatial coordinate data is The Kalman gain is obtained by fusing downhole laser positioning sensors and inertial measurement units. The matrix inversion optimization algorithm is used in the calculation process to reduce computing power consumption.

[0082] After the state is updated, the optimal state vector is... The attitude angles in the data are normalized.

[0083] S1.4: From the optimal state vector Extract the first six dimensions of data to generate a six-DOF pose state data stream. vectors in ;

[0084] From the optimal state vector The first 6 dimensions of data extracted are, in order, three-dimensional coordinates. 3D attitude angle After truncating, the data is smoothed by using a moving average filtering algorithm to eliminate data fluctuations caused by instantaneous noise.

[0085] S1.5: Repeat steps S1.2 to S1.4 to finally generate a six-DOF pose state data stream. .

[0086] The remaining technical details of this embodiment are described above and will not be repeated here.

[0087] Example 2

[0088] In the intelligent prediction module, the Long Short-Term Memory neural network is specifically as follows:

[0089] The input layer dimension is The output layer dimension is The hidden layers are 3, and the loss function is mean squared error;

[0090] Based on the underground tunneling speed in coal mines, historical step length Take 60, future time window step size Take 30;

[0091] Each hidden layer has 128 neurons and uses the ReLU activation function;

[0092] The network training batch size is 32, the number of iterations is 1000, the initial learning rate is 0.001, Adam optimization is used, and the training convergence condition is that the loss function is less than 1e-4;

[0093] The intelligent prediction module generates the future pose creep trend trajectory of the tunnel boring machine. Specifically:

[0094] S2.1: From the six-DOF pose state data stream Extract the most recent Step data to form a historical six-degree-of-freedom pose state data stream. ;

[0095] The extraction process is synchronized with the data stream storage format of the central processing module, and the data is extracted in the order of timestamps to ensure the temporal continuity of the data; after extraction, the data is time-aligned to correct the time difference caused by sensor sampling delay;

[0096] S2.2: Historical six-DOF pose state data stream Preprocessing is required;

[0097] The min-max normalization method is used to map the parameters of each dimension in the historical data to the range (0,1). After normalization, the data is detrended by using linear regression to remove long-term trend terms and retain short-term fluctuation characteristics, thus avoiding the interference of trend terms with the prediction results. After preprocessing, the data is dimensionally compressed by using principal component analysis (PCA) to compress the 6-dimensional data into 4-dimensional data, thereby reducing the computational load of the network.

[0098] S2.3: Stream the preprocessed historical six-DOF pose state data. The data is input into a long short-term memory neural network and subjected to forward inference, ultimately generating a future pose creeping trajectory. ;

[0099] The preprocessed historical data is input into the network to generate future... The predicted six-DOF pose data are stitched together to form a future pose creeping trend trajectory. During the forward inference process, the dropout regularization method is used. After the prediction is completed, the output data is denormalized to restore the actual coordinates and attitude angle data.

[0100] The remaining technical details of this embodiment are described above and will not be repeated here.

[0101] Example 3

[0102] In the control decision module, the kinematic model of the tunneling machine is as follows:

[0103] The input is the future pose creep trend trajectory. and standard pose creep trend trajectory The difference between The output is the ideal compensation displacement curve. In The input and output have a linear mapping relationship. ,in, This is the first relational mapping matrix;

[0104] First relational mapping matrix The element values ​​range from [0.01, 0.1], corresponding to the influence weight of different pose deviations on the compensation displacement of each hydraulic cylinder. The weight corresponding to the coordinate deviation is slightly larger, and the weight corresponding to the attitude angle deviation is slightly smaller. It can be dynamically adjusted according to the structural parameters of the underground tunneling equipment.

[0105] Standard pose creeping trend trajectory Pre-set by the underground tunneling planning system, generated according to parameters such as tunneling direction and slope, each step corresponding Includes three-dimensional coordinates and three-dimensional attitude angles;

[0106] Difference Corresponding to three-dimensional coordinates and three-dimensional attitude angle The deviation is calculated using absolute value operations.

[0107] In the control decision module, the ideal compensation displacement curve is generated as follows: Specifically:

[0108] S3.1: Calculate the future pose creep trend trajectory and standard pose creep trend trajectory The difference between ;

[0109] S3.2: Calculate the difference at each step. Compare with a preset small fluctuation threshold; if the difference is... If the fluctuation does not exceed the preset small fluctuation threshold, no compensation is required; otherwise, compensation is based on the linear mapping relationship. Calculate the ideal compensation displacement curve In ;

[0110] S3.3: Repeat steps S3.1 and S3.2, calculating sequentially. Ideal compensation displacement curve for each step In .

[0111] In the control decision module, continuous micro-hydraulic cylinder stroke compensation commands are generated. Specifically:

[0112] for Ideal compensation displacement curve for each step In According to the linear mapping relationship Calculate the continuous micro-hydraulic cylinder stroke compensation commands sequentially. In ,in, This is the second relational mapping matrix;

[0113] Second relational mapping matrix It is a diagonal matrix, and the diagonal elements are the displacement-command conversion coefficients of the hydraulic cylinder, which are the control command values ​​corresponding to each millimeter displacement of the hydraulic cylinder.

[0114] The calculation process is based on future time steps. The process is executed sequentially. After each step of the calculation is completed, the difference is smoothed using an exponential moving average algorithm to eliminate instantaneous fluctuations caused by prediction bias.

[0115] To address sudden deviations caused by downhole vibrations, a deviation threshold judgment is added. When the difference value at a certain step exceeds a temporary threshold, it is judged as a sudden deviation, the current calculation is paused, and the average historical difference value is used as a substitute. Normal calculation is resumed after the deviation stabilizes.

[0116] After the calculation is completed, the difference is generated. sequence.

[0117] Difference Compared with the preset micro-fluctuation threshold, which is set according to the accuracy requirements of underground tunneling, the threshold in the coordinate direction is ±0.02 meters, and the threshold in the attitude angle direction is ±0.03 rad. The threshold can be manually adjusted through the underground control terminal to adapt to the accuracy requirements of different tunneling conditions (such as hard rock tunneling and soft rock tunneling). The threshold judgment adopts a dimension-by-dimensional judgment method. When the difference is... When the deviation of any dimension exceeds the corresponding threshold, compensation calculation is triggered. If none of the dimensions exceed the threshold, a no-compensation command is output to control the hydraulic cylinder to maintain the current state. At the same time, a hysteresis judgment mechanism is added. Compensation calculation is only performed when the deviation exceeds the threshold for three consecutive steps to avoid frequent compensation caused by instantaneous deviation and protect the hydraulic cylinder equipment.

[0118] Ideal compensation displacement curve according to The calculations are performed step by step. After the calculations are completed, the curves are smoothed by using cubic spline interpolation to make the compensation displacement curve continuous and smooth, thus avoiding equipment vibration caused by sudden changes in the hydraulic cylinder's movement.

[0119] Compensation Instructions Depend on The conversion process includes instruction limiting, and after instruction generation, the output is generated using pulse width modulation (PWM).

[0120] Finally, the compensation command is transmitted to each hydraulic cylinder controller in real time to complete the compensation action, and the compensation result is fed back to the central processing module to form a closed-loop control.

[0121] The remaining technical details of this embodiment are described above and will not be repeated here.

[0122] In summary, the intelligent rapid tunneling system for underground coal mines disclosed in this invention can accurately predict the creep trend of the tunneling machine's position and posture, achieve continuous micro-dynamic compensation, effectively suppress sawtooth fluctuations in the tunneling trajectory, improve the straightness and forming accuracy of the roadway, and ensure continuous tunneling operations without mechanical interference.

[0123] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. An intelligent rapid tunneling system for underground coal mines, characterized in that, include: The fiber optic inertial navigation measurement unit and the laser tracking measurement unit are respectively installed on the tunneling machine; The data acquisition module is used to receive the three-axis acceleration data and three-axis angular velocity data of the tunneling machine output by the fiber optic inertial navigation measurement unit, and to receive the three-dimensional spatial coordinate data of the tunneling machine output by the laser tracking measurement unit; The central processing module is used to input the triaxial acceleration data, the triaxial angular velocity data, and the three-dimensional spatial coordinate data into the extended Kalman filter. The extended Kalman filter uses the triaxial acceleration data and the triaxial angular velocity data as state predictions and the three-dimensional spatial coordinate data as observation updates, and iteratively executes state prediction and observation updates to generate a six-degree-of-freedom pose state data stream of the tunneling machine. The intelligent prediction module is used to receive the six-degree-of-freedom pose state data stream of the tunneling machine, input the six-degree-of-freedom pose state data stream into the long short-term memory neural network, and the long short-term memory neural network generates predicted six-degree-of-freedom pose state data within the future time window based on the historical six-degree-of-freedom pose state data stream, so as to generate the future pose creep trend trajectory of the tunneling machine. The control decision module is used to receive the future posture creep trend trajectory of the tunneling machine, calculate the difference between the future posture creep trend trajectory and the standard posture creep trend trajectory and compare it with a preset micro-fluctuation threshold. When the difference exceeds the preset micro-fluctuation threshold, the module calculates the ideal compensation displacement curve to offset the difference based on the kinematic model of the tunneling machine, so as to generate a continuous micro-hydraulic cylinder stroke compensation command. The electro-hydraulic servo drive module is used to receive the continuous micro-hydraulic cylinder stroke compensation command, convert the continuous micro-hydraulic cylinder stroke compensation command into an electro-hydraulic servo control signal, and drive the hydraulic cylinder to generate a micro-extension and contraction motion synchronized with the ideal compensation displacement curve.

2. The intelligent rapid tunneling system for underground coal mines as described in claim 1, characterized in that, In the data acquisition module, the triaxial acceleration data is The triaxial angular velocity data are The three-dimensional spatial coordinate data is ; In the central processing module, the six-degree-of-freedom pose state data stream is ,in, This refers to the three-dimensional coordinate position in the geodetic coordinate system. These are roll angle, pitch angle, and yaw angle, respectively. In the intelligent prediction module, the historical six-DOF pose state data stream is The predicted six-degree-of-freedom pose state data is The future pose creep trend trajectory is ,in, For the pace of history, The step size of the future time window; In the control decision module, the standard pose creep trend trajectory The future pose creep trend trajectory and the standard pose creep trend trajectory The difference between them is The ideal compensation displacement curve is The continuous micro-hydraulic cylinder stroke compensation command is ,in, This refers to the number of hydraulic cylinders used for compensation in the tunneling machine; In the electro-hydraulic servo drive module, the electro-hydraulic servo control signal is: .

3. The intelligent rapid tunneling system for underground coal mines as described in claim 1, characterized in that, In the central processing module, the extended Kalman filter specifically comprises: The state vector is The state transition matrix is The observation matrix is The process noise covariance matrix is ​​as follows: diagonal matrix form The observed noise covariance matrix is ​​as follows: diagonal matrix form The error covariance matrix is of , wherein the observation matrix Only observe the state vector In .

4. The intelligent rapid tunneling system for underground coal mines as described in claim 3, characterized in that, The central processing module generates the six-degree-of-freedom pose state data stream of the tunneling machine. Specifically: S1.1: Set the initial value for iteration When, initialize the state vector and the error covariance matrix ; S1.2: When At that time, the triaxial acceleration data and the triaxial angular velocity data The input is fed into the extended Kalman filter, and a linear state equation is constructed based on the tunneling machine's kinematics. This equation is then processed through the state transition matrix. After linearization, the predicted state vector is calculated. and prediction error covariance matrix ; S1.3: The first The three-dimensional spatial coordinate data mentioned in step [step] are The input is fed into the extended Kalman filter, the Kalman gain is calculated, and the predicted state vector is corrected using the Kalman gain. The updated optimal state vector is obtained. And update the error covariance matrix simultaneously. ; S1.4: From the optimal state vector Extract the first six dimensions of data to generate the six-degree-of-freedom pose state data stream. vectors in ; S1.5: Repeat steps S1.2 to S1.4 to finally generate the six-degree-of-freedom pose state data stream. .

5. The intelligent rapid tunneling system for underground coal mines as described in claim 2, characterized in that, In the intelligent prediction module, the long short-term memory neural network specifically comprises: The input layer dimension is The output layer dimension is The hidden layers are 3, and the loss function is mean squared error.

6. The intelligent rapid tunneling system for underground coal mines as described in claim 5, characterized in that, In the intelligent prediction module, the future pose creep trend trajectory of the tunneling machine is generated. Specifically: S2.1: From the six-degree-of-freedom pose state data stream Extract the most recent Step data, forming the historical six-DOF pose state data stream. ; S2.2: For the historical six-degree-of-freedom pose state data stream Preprocessing is required; S2.3: The preprocessed historical six-DOF pose state data stream The data is input into the long short-term memory neural network and forward inference is performed to ultimately generate the future pose creeping trend trajectory. .

7. The intelligent rapid tunneling system for underground coal mines as described in claim 2, characterized in that, In the control decision module, the kinematic model of the tunneling machine is specifically as follows: The input is the future pose creep trend trajectory. and the standard pose creep trend trajectory The difference between The output is the ideal compensation displacement curve. In The input and output have a linear mapping relationship. ,in, This is the first relational mapping matrix.

8. The intelligent rapid tunneling system for underground coal mines as described in claim 7, characterized in that, In the control decision module, the ideal compensation displacement curve is generated as follows: Specifically: S3.1: Calculate the future pose creep trend trajectory and the standard pose creep trend trajectory The difference between ; S3.2: Calculate the difference at each step. Compare with the preset small fluctuation threshold, if the difference is... If the fluctuation does not exceed the preset micro-fluctuation threshold, no compensation is required; otherwise, compensation is based on a linear mapping relationship. Calculate the ideal compensation displacement curve In ; S3.3: Repeat steps S3.1 and S3.2, calculating sequentially. The ideal compensation displacement curve for each step In .

9. The intelligent rapid tunneling system for underground coal mines as described in claim 8, characterized in that, In the control decision module, the continuous micro-hydraulic cylinder stroke compensation command is generated. Specifically: for The ideal compensation displacement curve for each step In According to the linear mapping relationship The continuous micro-hydraulic cylinder stroke compensation commands are calculated sequentially. In ,in, This is the second relational mapping matrix.

Images