Coal mining machine positioning navigation error calibration method

Abstract

The invention relates to the technical field of coal mining, in particular to a coal mining machine positioning navigation error calibration method which comprises the following steps: acquiring three-dimensional pose, travel mileage and multi-axis vibration spectrum data of a coal mining machine in real time, and correcting an inertial navigation error by adopting a dynamic compensation model. A high-precision fusion positioning track is generated based on adaptive Kalman filtering fusing inertial navigation and odometer data, a geological disturbance error mapping model is constructed through three-dimensional geological model registration in combination with parameters such as coal seam hardness, fracture development indexes and dirt band layer density, the three-dimensional error compensation amount is calculated, the track is corrected by adopting a particle filtering algorithm, and the positioning precision is improved. Physical constraint verification is applied to the corrected track, and a calibrated navigation coordinate data packet is generated. The method can improve the navigation precision of the coal mining machine, reduce the inertial navigation drift error and enhance the adaptive capacity of the system, and is suitable for intelligent coal mining operation in a complex coal seam environment.

Description

Technical Field

[0001] This invention relates to the field of coal mining technology, and in particular to a method for calibrating the positioning and navigation errors of a coal mining machine. Background Technology

[0002] In modern intelligent coal mining, high-precision positioning and navigation of coal mining machines are key technologies for ensuring safe and efficient operation. Due to the complex underground environment, traditional positioning methods for coal mining machines based on inertial navigation and odometers accumulate errors over long periods of operation, leading to deviations in the navigation trajectory and affecting the cutting accuracy of the coal mining machine. Furthermore, the complexity of the coal seam geological environment, such as the seam dip angle, fracture development, and distribution of interbedded rock layers, also significantly interferes with navigation accuracy.

[0003] While current inertial navigation technology combines multiple sensors such as gyroscopes, accelerometers, and encoders, and incorporates Kalman filtering for data fusion, it still faces the following key technical challenges: Due to severe vibrations in coal mines, inertial navigation data is susceptible to shock interference, leading to further amplification of errors, making it difficult for existing compensation methods to adapt to different working conditions in real time; existing Kalman filtering algorithms primarily rely on inertial navigation and odometer data, failing to fully consider the impact of coal seam geological parameters on navigation errors, resulting in further accumulation of positioning errors in complex coal seam environments; and most existing error compensation methods rely on fixed model parameters, lacking adaptive online optimization capabilities, and unable to dynamically adjust compensation strategies based on actual working conditions during long-term operation. Summary of the Invention

[0004] This invention provides a method for calibrating the positioning and navigation errors of a coal mining machine. By integrating inertial navigation error correction, geological disturbance error compensation, and adaptive online optimization, the method improves the navigation accuracy and operational stability of the coal mining machine in complex coal seam environments.

[0005] A method for calibrating positioning and navigation errors of a coal mining machine includes the following steps: S1: Real-time acquisition of three-dimensional pose data output by the inertial navigation module of the coal mining machine, travel distance data output by the encoder, and multi-axis vibration spectrum data output by the vibration sensor; S2: Extract features from the vibration spectrum data to generate a set of vibration feature parameters corresponding to the current working state. The feature parameters include the main vibration frequency band, vibration intensity coefficient, and vibration direction vector. S3: Input the set of vibration characteristic parameters into the dynamic compensation model to generate a real-time compensation matrix for three-dimensional pose data. The dynamic compensation model is trained based on historical working condition data and the compensation matrix is ​​applied to the three-dimensional pose data collected in S1 to generate the compensated three-dimensional pose. S4: The compensated three-dimensional pose is spatiotemporally fused with the travel mileage data collected in S1 to generate a fused positioning trajectory; S5: Based on the three-dimensional geological model of the working face, the spatial coordinates in the fused positioning trajectory are corrected for geological disturbance, and the calibrated navigation coordinate data is output.

[0006] Optionally, S1 includes: S11: The original pose signal is acquired at a sampling frequency of 200Hz through the MEMS gyroscope and accelerometer group built into the inertial navigation module. The original pose signal is subjected to sliding window noise reduction processing to generate preprocessed three-dimensional pose data. The three-dimensional pose data includes pitch angle, roll angle, yaw angle and X / Y / Z axial displacement. S12: Set up a pulse signal redundancy verification channel at the encoder output end, synchronously collect A / B phase quadrature pulse signals and Z phase zero position signals, perform integrity verification on the pulse sequence based on the Hamming distance algorithm, and output the verified travel distance data, which includes travel speed, cumulative travel distance and motion direction indicator; S13: Acquire time-domain signals of X / Y / Z axial vibration acceleration through a triaxial vibration sensor, and generate multi-axis vibration spectrum data after performing Fourier transform on the time-domain signals. The spectrum coverage range is 0-10kHz, and the resolution is set to 5Hz / division. S14: Establish a multi-source data time synchronization mechanism. Using the inertial navigation module clock as a reference, timestamp align the preprocessed three-dimensional pose data, the verified travel mileage data, and the multi-axis vibration spectrum data. Use cubic spline interpolation to eliminate clock drift between sensors and generate time-synchronized multi-source data packets. S15: Perform a joint judgment on data validity and trigger data resampling when an anomaly is detected.

[0007] Optionally, S2 includes: S21: Perform wavelet packet decomposition on the multi-axis vibration spectrum data, perform energy calculation of eight-layer tree nodes in the 3-5kHz frequency band, extract continuous frequency bands with energy proportion of each node exceeding 15% as candidate main vibration frequency bands, and select the top three candidate frequency bands according to the frequency band energy integral value to form the main vibration frequency band set.

[0008] S22: Based on the set of main vibration frequency bands obtained in S21, calculate the vibration energy integral values ​​of the X / Y / Z axes in the selected frequency bands respectively, multiply the integral values ​​of each axis by the center frequency of its corresponding frequency band to generate the axial vibration intensity factor, and sum the triaxial factors after dynamic weight allocation to generate the vibration intensity coefficient.

[0009] S23: Construct the covariance matrix of the three-axis vibration acceleration, perform eigenvalue decomposition on it, extract the eigenvector corresponding to the largest eigenvalue as the vibration direction vector, and record the cosine values ​​of the angles between the eigenvectors and each axis of the coal mining machine body coordinate system.

[0010] S24: Normalize the set of main vibration frequency bands, vibration intensity coefficients and vibration direction vectors to synthesize a set of vibration characteristic parameters, and add timestamps and working condition type labels to generate a standardized characteristic parameter data package.

[0011] S25: Perform feature parameter validity verification and trigger feature recalculation when a specific anomaly is detected.

[0012] Optionally, S3 includes: S31: Initialize the dynamic compensation model, load the pre-trained basic LSTM network weights and the initial parameters of the working condition adaptation layer. The basic LSTM network is trained based on 12 typical vibration modes of a laboratory standard vibration table, and the input layer dimension matches the number of elements in the vibration feature parameter set. S32: Sort the current set of vibration characteristic parameters by time series, construct a sliding window time series data of length 10, input it into the dynamic compensation model for real-time inference, and output the basic compensation matrix and working condition correction factor; S33: Perform matrix fusion operation to synthesize the real-time compensation matrix by combining the basic compensation matrix and the operating condition correction factor.

[0013] Optionally, S3 further includes: S34: Convert the 3D pose data acquired by S1 into homogeneous coordinates, perform matrix multiplication with the real-time compensation matrix, and generate the compensated 3D pose.

[0014] S35: Perform online verification of the compensation effect, calculate the residual error of the 3D pose data before and after compensation, and trigger online update of the dynamic compensation model when the error exceeds the limit.

[0015] Optionally, S4 includes: S41: Establish a time synchronization queue for the compensated 3D pose data and mileage data. Based on the time synchronization multi-source data packets generated in S14, a double buffering mechanism is used to align the data streams and generate a synchronized spatiotemporal data group.

[0016] S42: Based on the generated synchronous spatiotemporal data set, perform spatial coordinate system transformation to transform the travel mileage data from the local coordinate system of the traveling wheel to the coordinate system of the coal mining machine body, and generate travel mileage data in the body coordinate system; S43: Based on the compensated 3D pose data and the travel distance data in the body coordinate system, an adaptive Kalman filter is constructed, and the two data are fused to generate an optimized state vector.

[0017] Optionally, S4 further includes: S44: Based on the optimized state vector, perform a multi-cycle prediction-correction loop to generate a fused positioning trajectory and extract the nodes of the fused positioning trajectory; S45: Based on the fused positioning trajectory nodes, perform trajectory smoothing optimization, and then perform fused positioning trajectory optimization.

[0018] Optionally, S5 includes: S51: Perform spatial registration of the fused positioning trajectory and the 3D geological model to generate a geological-trajectory association data set; S52: Based on the geological-trajectory association data set, construct a geological disturbance error mapping model and generate a three-dimensional error compensation amount.

[0019] Optionally, S5 further includes: S53: Based on the three-dimensional error compensation amount, perform particle filter correction to generate geologically corrected trajectory nodes; S54: Based on the geologically corrected trajectory nodes, perform physical constraint verification of the corrected trajectory.

[0020] S55: Based on the verified trajectory, generate a calibrated navigation coordinate data packet and transmit it to the coal mining machine navigation control system.

[0021] The beneficial effects of this invention are: This invention achieves real-time correction of inertial navigation drift by collecting inertial navigation pose data, mileage data, and multi-axis vibration characteristics, and employing a dynamic compensation model to correct inertial navigation errors. Specifically, key parameters such as vibration intensity coefficient and vibration direction vector are extracted, and a real-time compensation matrix is ​​calculated based on the dynamic compensation model to dynamically compensate for inertial navigation pose, effectively reducing accumulated inertial navigation errors. Furthermore, an adaptive Kalman filter algorithm is used to fuse inertial navigation and odometer data, continuously optimizing positioning accuracy in multi-cycle prediction-correction loops, ultimately generating a high-precision fused positioning trajectory, thereby improving the navigation stability and accuracy of the coal mining machine under complex working conditions.

[0022] This invention utilizes a three-dimensional geological model for trajectory correction in S5, constructing a geological disturbance error mapping model. By combining geological parameters such as coal seam hardness coefficient, fracture development index, and interbedded rock density, the three-dimensional error compensation is calculated. A particle filtering algorithm is employed to fully consider the impact of coal seam structure on the coal mining machine's trajectory, optimizing the positioning results. Furthermore, after trajectory correction, physical constraints are applied for verification to ensure that the corrected trajectory meets conditions such as coal seam roof and floor elevation limits, vibration intensity, and cutting resistance matching requirements. This method effectively reduces the impact of geological disturbances on the navigation trajectory, improves the reliability and adaptability of the trajectory, and makes the navigation system more consistent with the actual coal seam environment.

[0023] This invention employs a multi-layered error compensation mechanism, combining inertial navigation error correction, fusion positioning trajectory generation, and geological disturbance error correction to establish a complete error correction system. Simultaneously, a residual error calculation mechanism is used, enabling the system to continuously optimize compensation parameters based on actual operating conditions and trigger dynamic updates when errors exceed limits. Furthermore, this invention constructs a final calibrated navigation data package and transmits it in real-time to the coal mining machine navigation control system via an underground industrial ring network, achieving online closed-loop adjustment of error compensation. This method can adapt to different coal seam geological conditions and mining conditions, improving the stability and intelligence level of the navigation system, reducing manual intervention, and increasing the automation level of coal mine production. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the S3 process in an embodiment of the present invention. Detailed Implementation

[0026] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0027] like Figures 1-2 As shown, a method for calibrating the positioning and navigation errors of a coal mining machine includes the following steps: S1: Real-time acquisition of three-dimensional pose data output by the inertial navigation module of the coal mining machine, travel distance data output by the encoder, and multi-axis vibration spectrum data output by the vibration sensor; S2: Extract features from the vibration spectrum data to generate a set of vibration feature parameters corresponding to the current working state. The feature parameters include the main vibration frequency band, vibration intensity coefficient, and vibration direction vector. S3: Input the set of vibration characteristic parameters into the dynamic compensation model to generate a real-time compensation matrix for the three-dimensional pose data. The dynamic compensation model is trained based on historical working condition data and the compensation matrix is ​​applied to the three-dimensional pose data collected by S1 to generate the compensated three-dimensional pose. S4: The compensated 3D pose is spatiotemporally fused with the travel distance data collected in S1 to generate a fused positioning trajectory. S5: Based on the three-dimensional geological model of the working face, the spatial coordinates in the fused positioning trajectory are corrected for geological disturbance, and the calibrated navigation coordinate data is output.

[0028] S1 includes: S11: The original pose signal is acquired at a sampling frequency of 200Hz through the MEMS gyroscope and accelerometer group built into the inertial navigation module. The original pose signal is subjected to sliding window noise reduction processing to generate preprocessed three-dimensional pose data. The three-dimensional pose data includes pitch angle, roll angle, yaw angle and X / Y / Z axial displacement. S12: A pulse signal redundancy verification channel is set at the encoder output. The A / B phase quadrature pulse signals and the Z phase zero-position signal are acquired synchronously. The integrity of the pulse sequence is verified based on the Hamming distance algorithm. The verified travel distance data is output, which includes travel speed, cumulative travel distance, and motion direction indicator, as follows: (1) Pulse signal encoding: Assume the encoder's A-phase and B-phase outputs are binary sequences. and The Z phase is a zero-position signal sequence. The sampling window length is set to N pulse cycles, and the encoding state within each pulse cycle is denoted as: ; in These represent the A / B phase states (0 or 1) of the i-th pulse, respectively.

[0029] (2) Ideal coding pattern (reference standard sequence): Ideally, A / B should exhibit a standard orthogonal coding pattern, and its state changes should follow the Gray code pattern, represented as: ; (3) The Hamming distance is calculated as follows: ; ; The total Hamming distance is calculated as follows: ; If Hamming distance Exceeding the threshold (as set) If the value is 2, then the verification is considered to have failed. S13: Acquire time-domain signals of X / Y / Z axial vibration acceleration through a triaxial vibration sensor, and generate multi-axis vibration spectrum data after performing Fourier transform on the time-domain signals. The spectrum coverage range is 0-10kHz, and the resolution is set to 5Hz / division. S14: Establish a multi-source data time synchronization mechanism. Using the inertial navigation module clock as a reference, timestamps are aligned on preprocessed three-dimensional pose data, verified travel mileage data, and multi-axis vibration spectrum data. Cubic spline interpolation is used to eliminate clock drift between sensors and generate time-synchronized multi-source data packets. S15: Perform a joint data validity check and trigger data resampling if any of the following anomalies are detected: In the preprocessed 3D pose data, the absolute value of the pitch angle is >30° and the duration of exceeding the limit is ≥500ms; The Hamming distance of the travel distance data satisfies the following conditions in three consecutive sampling periods. > If so, data resampling will be triggered; The difference in spectral energy along the X / Y / Z axes in the multi-axis vibration spectrum data exceeds 200% of the preset threshold.

[0030] S2 includes: S21: Perform wavelet packet decomposition on the multi-axis vibration spectrum data, calculate the energy of eight-layer tree nodes in the 3-5kHz frequency band, extract continuous frequency bands with an energy ratio exceeding 15% for each node as candidate main vibration frequency bands, and select the top three candidate frequency bands according to the frequency band energy integral value to form the main vibration frequency band set. The specific steps are as follows: Wavelet packet decomposition was performed on the multi-axis vibration spectrum data to construct an eight-layer tree-like node energy distribution in the 3-5kHz frequency band; Calculate the energy percentage of each node and select continuous frequency bands with an energy percentage exceeding 15% as candidate main resonant frequency bands; Calculate the energy integral value for each frequency band and sort them according to energy level; The three candidate frequency bands with the highest energy are selected to form the main resonant frequency band set.

[0031] S22: Based on the set of main vibration frequency bands obtained in S21, calculate the vibration energy integral values ​​of the X / Y / Z axes within the selected frequency bands respectively. Multiply each axial integral value by its corresponding frequency band center frequency to generate axial vibration intensity factors. After dynamically weighting the triaxial factors, sum them to generate vibration intensity coefficients. The specific steps are as follows: Within the dominant frequency band, calculate the integral values ​​of vibration energy along the X / Y / Z axes; The axial vibration intensity factor is generated by multiplying the integral value of each axial vibration energy by the center frequency of its corresponding frequency band. The weight of the triaxial vibration intensity factor is dynamically adjusted based on the current load rate of the coal mining machine's cutting motor. The vibration intensity coefficient is calculated using a weighted summation method and used as the input parameter for subsequent compensation calculations.

[0032] S23: Construct the covariance matrix of the triaxial vibration acceleration, perform eigenvalue decomposition on it, extract the eigenvector corresponding to the largest eigenvalue as the vibration direction vector, and record the cosine values ​​of the angles between the eigenvectors and each axis of the coal mining machine's coordinate system. The specific steps are as follows: Calculate the covariance matrix of the triaxial vibration acceleration data; Perform eigenvalue decomposition on the covariance matrix and extract the eigenvector corresponding to the largest eigenvalue; The eigenvector is taken as the vibration direction vector, and its projection in the coordinate system of the coal mining machine is calculated; Calculate the cosine values ​​of the angles between the vibration direction vector and each axis of the coal mining machine's coordinate system, and record them for subsequent analysis.

[0033] S24: Normalize the main vibration frequency band set, vibration intensity coefficient, and vibration direction vector to synthesize a set of vibration characteristic parameters, and add timestamps and working condition type labels to generate a standardized characteristic parameter data package. The specific steps are as follows: The frequency range of the main oscillation frequency band set is normalized to make it suitable for comparison under different working conditions; The vibration intensity coefficient is standardized to eliminate the influence of different measurement environments; Normalize the vibration direction vector to make its direction information independent of the amplitude; The normalized data is combined, timestamps and operating condition type labels are added, and it is then packaged into a standardized feature parameter data package.

[0034] S25: Perform feature parameter validity verification. When a specific anomaly is detected, feature recalculation is triggered. The specific steps are as follows: Determine whether the highest energy frequency band in the main vibration frequency band set accounts for less than 40% of the total vibration energy. If so, trigger a recalculation. Calculate the angle deviation between the vibration direction vector and the coal mining machine's movement direction. If it is greater than 45° and continues to exceed the limit for ≥2s, then trigger a recalculation. The amplitude of the sudden change in vibration intensity coefficient between two adjacent calculation cycles is calculated. If it exceeds 300% of the preset safety threshold, a recalculation is triggered. When a recalculation is triggered, historical vibration spectrum data from the previous three calculation cycles are automatically retrieved for sliding window correction to ensure stable and reliable calculation results.

[0035] S3 includes: S31: Initialize the dynamic compensation model, load the pre-trained basic LSTM network weights and the initial parameters of the working condition adaptation layer. The basic LSTM network is trained based on 12 typical vibration modes of a laboratory standard vibration table. The input layer dimension matches the number of elements in the vibration feature parameter set. The specific steps are as follows: Load the pre-trained base LSTM network weights, which were trained on 12 typical vibration modes on a standard laboratory shaking table; Initialize the initial parameters of the working condition adaptation layer to adapt it to different working condition characteristics; Ensure that the dimension of the input layer is consistent with the number of elements in the set of vibration feature parameters to guarantee the integrity of the network input data; S32: Sort the current set of vibration characteristic parameters by time series, construct a sliding window time series data of length 10, input it into the dynamic compensation model for real-time inference, and output the basic compensation matrix and working condition correction factor. The specific steps are as follows: The set of vibration characteristic parameters is sorted according to the time series to form continuous time series data; A sliding window input is constructed by selecting data from the 10 most recent time steps to ensure that short-term dynamic characteristics are captured by the model; The dynamic compensation model is input for real-time inference, generating two key outputs: Basic compensation matrix: Used for basic correction of inertial navigation errors, with a dimension of 4×4, including rotation and translation components; Operating condition correction factor: To adapt to error changes under different operating conditions, it is set as a 3×3 diagonal matrix; S33: Perform matrix fusion operation to synthesize the real-time compensation matrix from the basic compensation matrix and the operating condition correction factor. The specific steps are as follows: The real-time compensation matrix is ​​calculated using the following fusion formula: ; in, The basic compensation matrix, This is the extended matrix of the operating condition correction factor. The environmental adaptability coefficient is dynamically calculated based on the frequency of abnormal events recorded in S15.

[0036] Calculate and store the real-time compensation matrix for use in subsequent pose correction calculations.

[0037] S3 also includes: S34: Convert the 3D pose data acquired in S1 into homogeneous coordinate form, perform matrix multiplication with the real-time compensation matrix to generate the compensated 3D pose. The specific steps are as follows: The three-dimensional pose data acquired by S1 is expressed in homogeneous coordinate form as follows: ; The matrix operation expression for calculating the compensated 3D pose is as follows: ; Calculate and store attitude angle corrections. This is for subsequent attitude error analysis.

[0038] S35: Perform online verification of the compensation effect, calculate the residual error of the 3D pose data before and after compensation, and trigger online update of the dynamic compensation model when the error exceeds the limit. The specific steps are as follows: Calculate the displacement residuals before and after compensation: ; Calculate the attitude residuals before and after compensation: ; Determine if the error exceeds the limit: If displacement residual or posture residual This triggers the online update of the dynamic compensation model. When an online update is triggered, perform the following steps: Pack the current set of vibration characteristic parameters and residual error into a new training sample; A sliding window mechanism is used to retain the most recent 100 samples to ensure that sufficient historical information is preserved when updating the model. Incremental learning is performed on the operating condition adaptation layer, and an update rate is set. It adaptively adjusts between 0.01 and 0.1 to balance learning speed and stability.

[0039] S4 includes: S41: Establish a time synchronization queue for the compensated 3D pose data and mileage data. Based on the time-synchronized multi-source data packets generated in S14, a double-buffering mechanism is used to align the data streams and generate synchronized spatiotemporal data groups. The specific steps are as follows: A 200Hz circular buffer is established for the compensated 3D pose data to ensure real-time data storage and access. A pulse-triggered event buffer is established for the mileage data, and the encoder pulse-triggered data is stored in the buffer; An interpolation resampling method is used to unify the data of both to a 100Hz time base, generating a synchronized spatiotemporal data set to ensure synchronized processing of inertial navigation and odometer data.

[0040] S42: Based on the generated synchronous spatiotemporal data set, perform a spatial coordinate system transformation to convert the travel mileage data from the local coordinate system of the traveling wheels to the coordinate system of the coal mining machine body, generating travel mileage data in the body coordinate system. The specific steps are as follows: (1) Based on the heading angle in the compensated three-dimensional pose The coordinate rotation matrix is ​​calculated as follows: ; (2) Transform the travel speed v and displacement s in the mileage data to the body coordinate system, and express them as follows; ; (3) Generate travel distance data in the body coordinate system to ensure a spatial reference system that is consistent with the compensated three-dimensional pose data; S43: Based on the compensated 3D pose data and the mileage data in the body coordinate system, an adaptive Kalman filter is constructed. The two data are fused to generate an optimized state vector. The input parameters include the state vector, the observation matrix, and the process noise matrix. The specific steps are as follows: (1) Define the state vector as follows: ; (Includes position, attitude angle, and velocity information, fused with compensated 3D pose data and converted odometer data) (2) Construction of observation matrix: Inertial navigation pose data and travel mileage and rate are integrated to form a hybrid observation value; (3) The process noise matrix Q is dynamically adjusted, and the adjustment rules are as follows: ; in, This is the vibration intensity coefficient. This is the encoder error rate (the Hamming distance verification error rate of S12).

[0041] S4 also includes: S44: Based on the optimized state vector, perform a multi-cycle prediction-correction loop to generate a fused positioning trajectory and extract the nodes of the fused positioning trajectory. The specific steps are as follows: (1) Time update (prediction step), represented as: ; The state transition matrix F includes the pitch angle. The nonlinear coupling term for z-axis displacement improves prediction accuracy.

[0042] (2) Measurement update (calibration step), represented as: ; The observation matrix H is a 9×9 diagonal matrix, with the diagonal elements determined by weighting the confidence of the compensated pose and the confidence of the odometer data. R is the measurement noise matrix, which is dynamically adjusted according to the drift characteristics of inertial navigation.

[0043] (3) Output the optimal estimated state vector and extract its values. As a node for fusion positioning trajectory; S45: Based on the fused positioning trajectory nodes, trajectory smoothing optimization is performed. Continuous trajectory nodes are fitted using cubic B-spline curves, and kinematic constraints are applied to generate the optimized fused positioning trajectory. The specific steps are as follows: (1) A cubic B-spline curve is used to smooth the fused positioning trajectory in order to reduce abrupt changes and improve trajectory continuity; (2) Apply the following constraints to ensure that the trajectory meets the requirements of coal mine operations: Curvature continuity and maximum curvature limit (≤0.15) ), to prevent excessive trajectory deviation; The angle deviation between the velocity vector and the coal seam strike (≤20°) ensures that the direction of travel conforms to the geological structure; The consistency of the projection of the vibration direction vector and the trajectory tangent direction (cosine similarity ≥ 0.8) ensures that the inertial navigation matches the physical motion direction; (3) Generate optimized fusion positioning trajectories that satisfy kinematic constraints to provide high-precision trajectory data for subsequent navigation calibration.

[0044] S5 includes: S51: Perform spatial registration of the fused positioning trajectory and the 3D geological model to generate a geological-trajectory association data set. The specific steps are as follows: (1) Based on the optimized fusion positioning trajectory of S45, the coordinates of the trajectory nodes are... The conversion to the geographic coordinate system of the geological model is as follows: ; in, For the rotation matrix of the tunnel direction, This represents the current working face advance offset. (2) Based on the transformed geographic coordinates, extract the coal seam characteristic parameters at the corresponding locations in the three-dimensional geological model, including: Coal seam hardness coefficient (0.5-3.0); Crack development index (0-1); Sandwich layer distribution density (kg / m'); (3) Combine trajectory coordinates and coal seam characteristic parameters to generate a geological-trajectory correlation data set for subsequent geological disturbance error calculation; S52: Based on the geological-trajectory correlation data set, construct a geological disturbance error mapping model and generate a three-dimensional error compensation quantity. The specific steps are as follows: (1) Calculate the angle between the vibration direction vector and the coal seam strike. Calculation of vibration direction vector based on S45 (2) The vibration intensity coefficient of S22 is adopted. As a factor influencing geological disturbance; (3) Calculate the cutting resistance of the current trajectory node, expressed as: ; Where k is the cutoff coefficient. The coal seam hardness coefficient, The density of the interlayer of gangue; (4) Calculate the error compensation amount for the trajectory nodes, expressed as: ,in The elastic modulus of the coal seam; (5) Generate the three-dimensional error compensation amount for each trajectory node For future revisions.

[0045] S5 also includes: S53: Based on the 3D error compensation amount, perform particle filtering correction to generate geologically corrected trajectory nodes. The specific steps are as follows: (1) Initialize 1000 particles, each carrying its position state. and weight ; (2) Define the state transition model as follows: ,in, Mileage data taken from the body coordinate system of S42; (3) The particle weights are calculated based on the residual error of S35, and are expressed as follows: ; (4) Perform system resampling, select high-weight particles and redistribute them; (5) Output the optimal estimated coordinates to form geologically corrected trajectory nodes; S54: Based on the geologically corrected trajectory nodes, perform physical constraint verification of the corrected trajectory. The specific steps are as follows: (1) Verify the matching degree between cutting resistance and vibration intensity, expressed as: ,in, For matching coefficients, This is the tolerance threshold; (2) Check whether the Z-axis coordinates of the trajectory nodes conform to the elevation of the top and bottom of the coal seam, and the deviation shall not exceed 5cm; (3) If any verification fails, execute the following correction strategy: (3.1) Re-extract the optimized fused positioning trajectory nodes of S45; (3.2) Adjust the error mapping model parameters of S52 and optimize the compensation calculation; (3.3) Iteratively execute the particle filter correction of S53 until the physical constraints are met.

[0046] S55: Based on the verified trajectory, generate a calibrated navigation coordinate data packet and transmit it to the coal mining machine navigation control system. The specific steps are as follows: (1) Construct the final output data package, including: final three-dimensional coordinates, corresponding coal seam characteristic labels and correction process characteristic parameters; (2) Data is transmitted to the coal mining machine navigation control system through the underground industrial ring network to provide high-precision navigation reference.

[0047] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0048] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for calibrating positioning and navigation errors of a coal mining machine, characterized in that, Includes the following steps: S1: Real-time acquisition of three-dimensional pose data output by the inertial navigation module of the coal mining machine, travel distance data output by the encoder, and multi-axis vibration spectrum data output by the vibration sensor; S2: Extract features from the vibration spectrum data to generate a set of vibration feature parameters corresponding to the current working state. The feature parameters include the main vibration frequency band, vibration intensity coefficient, and vibration direction vector. S3: Input the set of vibration characteristic parameters into the dynamic compensation model to generate a real-time compensation matrix for three-dimensional pose data. The dynamic compensation model is trained based on historical working condition data and the compensation matrix is ​​applied to the three-dimensional pose data collected in S1 to generate the compensated three-dimensional pose. S4: The compensated three-dimensional pose is spatiotemporally fused with the travel mileage data collected in S1 to generate a fused positioning trajectory; S5: Based on the three-dimensional geological model of the working face, the spatial coordinates in the fused positioning trajectory are corrected for geological disturbance, and the calibrated navigation coordinate data is output.

2. The method for calibrating positioning and navigation errors of a coal mining machine according to claim 1, characterized in that, S1 includes: S11: The original pose signal is acquired through the MEMS gyroscope and accelerometer group built into the inertial navigation module. The original pose signal is subjected to sliding window noise reduction processing to generate preprocessed three-dimensional pose data. The three-dimensional pose data includes pitch angle, roll angle, yaw angle and X / Y / Z axial displacement. S12: Set up a pulse signal redundancy verification channel at the encoder output end, synchronously collect A / B phase quadrature pulse signals and Z phase zero position signals, perform integrity verification on the pulse sequence based on the Hamming distance algorithm, and output the verified travel distance data, which includes travel speed, cumulative travel distance and motion direction indicator; S13: Collect time-domain signals of X / Y / Z axial vibration acceleration through a triaxial vibration sensor, and generate multi-axis vibration spectrum data after performing Fourier transform on the time-domain signals; S14: Establish a multi-source data time synchronization mechanism. Using the inertial navigation module clock as a reference, perform timestamp alignment on the preprocessed three-dimensional pose data, the verified travel mileage data, and the multi-axis vibration spectrum data to generate a time-synchronized multi-source data packet. S15: Perform a joint judgment on data validity and trigger data resampling when an anomaly is detected.

3. The method for calibrating positioning and navigation errors of a coal mining machine according to claim 2, characterized in that, S2 includes: S21: Perform wavelet packet decomposition on the multi-axis vibration spectrum data, perform energy calculation of eight-layer tree nodes in the 3-5kHz frequency band, extract continuous frequency bands with energy proportion of each node exceeding 15% as candidate main vibration frequency bands, and select the top three candidate frequency bands according to the frequency band energy integral value to form the main vibration frequency band set. S22: Based on the set of main vibration frequency bands obtained in S21, calculate the vibration energy integral values ​​of the X / Y / Z axes in the selected frequency bands respectively, multiply the integral values ​​of each axis by the center frequency of its corresponding frequency band to generate the axial vibration intensity factor, and sum the triaxial factors after dynamic weight allocation to generate the vibration intensity coefficient. S23: Construct the covariance matrix of the three-axis vibration acceleration, perform eigenvalue decomposition on it, extract the eigenvector corresponding to the largest eigenvalue as the vibration direction vector, and record the cosine values ​​of the angles between the eigenvectors and each axis of the coal mining machine body coordinate system. S24: Normalize the main vibration frequency band set, vibration intensity coefficient and vibration direction vector, synthesize the vibration characteristic parameter set, and add timestamp and working condition type label to generate a standardized characteristic parameter data package; S25: Perform feature parameter validity verification and trigger feature recalculation when a specific anomaly is detected.

4. The method for calibrating positioning and navigation errors of a coal mining machine according to claim 3, characterized in that, S3 includes: S31: Initialize the dynamic compensation model, load the pre-trained basic LSTM network weights and the initial parameters of the working condition adaptation layer. The basic LSTM network is trained based on 12 typical vibration modes of a laboratory standard vibration table, and the input layer dimension matches the number of elements in the vibration feature parameter set. S32: Sort the current set of vibration characteristic parameters according to the time series, input them into the dynamic compensation model for real-time inference, and output the basic compensation matrix and working condition correction factor; S33: Perform matrix fusion operation to synthesize the real-time compensation matrix by combining the basic compensation matrix and the operating condition correction factor.

5. The method for calibrating positioning and navigation errors of a coal mining machine according to claim 4, characterized in that, S3 further includes: S34: Convert the 3D pose data acquired by S1 into homogeneous coordinate form, perform matrix multiplication with the real-time compensation matrix, and generate the compensated 3D pose. S35: Perform online verification of the compensation effect, calculate the residual error of the 3D pose data before and after compensation, and trigger online update of the dynamic compensation model when the error exceeds the limit.

6. The method for calibrating positioning and navigation errors of a coal mining machine according to claim 5, characterized in that, S4 includes: S41: Establish a time synchronization queue for the compensated 3D pose data and mileage data. Based on the time synchronization multi-source data packets generated in S14, a double buffering mechanism is used to align the data streams and generate a synchronized spatiotemporal data group. S42: Based on the generated synchronous spatiotemporal data set, perform spatial coordinate system transformation to transform the travel mileage data from the local coordinate system of the traveling wheel to the coordinate system of the coal mining machine body, and generate travel mileage data in the body coordinate system; S43: Based on the compensated 3D pose data and the travel distance data in the body coordinate system, an adaptive Kalman filter is constructed, and the two data are fused to generate an optimized state vector.

7. The method for calibrating positioning and navigation errors of a coal mining machine according to claim 6, characterized in that, S4 further includes: S44: Based on the optimized state vector, perform a multi-cycle prediction-correction loop to generate a fused positioning trajectory and extract the nodes of the fused positioning trajectory; S45: Based on the fused positioning trajectory nodes, perform trajectory smoothing optimization, and then perform fused positioning trajectory optimization.

8. The method for calibrating positioning and navigation errors of a coal mining machine according to claim 7, characterized in that, S5 includes: S51: Perform spatial registration of the fused positioning trajectory and the 3D geological model to generate a geological-trajectory association data set; S52: Based on the geological-trajectory association data set, construct a geological disturbance error mapping model and generate a three-dimensional error compensation amount.

9. The method for calibrating positioning and navigation errors of a coal mining machine according to claim 8, characterized in that, The S5 also includes: S53: Based on the three-dimensional error compensation amount, perform particle filter correction to generate geologically corrected trajectory nodes; S54: Based on the geologically corrected trajectory nodes, perform physical constraint verification of the corrected trajectory; S55: Based on the verified trajectory, generate a calibrated navigation coordinate data packet and transmit it to the coal mining machine navigation control system.

Images