Method for processing two-side measurement data of underground roadway millimeter wave radar and measurement device

Abstract

The invention provides an underground roadway millimeter wave radar two-side measurement data processing method and a measurement device, and relates to the technical field of roadway radar detection, and the method comprises the steps: carrying out the denoising of detection data based on a Gaussian mixture filtering method, taking the denoised data as the width of a roadway, and judging whether the dangerous deformation occurs or not; poisson optimization processing is carried out on the point cloud data based on local topology constraints to obtain a final roadway model, resampling is carried out based on the final roadway model, and the roadway height is calculated; angle change data of the measuring device is acquired through an accelerometer and a gyroscope loaded on the measuring device, and effective data acquisition control is performed based on the angle change data. According to the invention, Gaussian mixture filtering is adopted to carry out smoothing processing on millimeter wave radar ranging signals, and mutation noise interference is effectively suppressed. Roadway height data are optimized through Poisson optimization of local topology constraints, real surface information is extracted, and the robustness of the system in a noise environment is enhanced.

Description

Technical Field

[0001] This application relates to the field of tunnel radar detection technology, and more specifically, to a method and measuring device for processing measurement data of the two sides of an underground tunnel using millimeter-wave radar. Background Technology

[0002] As a critical structure in mining and tunnel engineering, the stability of underground roadways directly affects production safety and personnel safety. Due to complex geological conditions, changes in mining pressure, and external disturbances, roadway sides are prone to deformation, convergence, and even collapse. Traditional monitoring methods, such as manual measurement, laser scanning, or video monitoring, have limitations such as low accuracy, poor real-time performance, and susceptibility to environmental interference, especially in harsh underground environments with dust, humidity, and vibration, where their reliability and adaptability are insufficient. In recent years, millimeter-wave radar technology has been increasingly applied to underground monitoring due to its strong penetration, anti-interference capabilities, and high-precision ranging characteristics. However, existing millimeter-wave radar systems still suffer from problems in data processing, such as noise sensitivity, sudden signal interference, and low reconstruction accuracy in complex environments. They lack a complete data processing and equipment protection mechanism, making it difficult to achieve real-time, accurate, and robust monitoring of roadway deformation and height. To address the dangers caused by underground roadway deformation and provide reliable technical support for underground safety monitoring and disaster early warning, current patents mainly focus on the design and compensation methods of roadway detection devices, such as: (1) Chinese utility model patent with publication number CN220667646U discloses a device for measuring the displacement of surrounding rock in a roadway, including a support and a rotating shaft. The support is fixed relative to the roadway body, and an anchor claw winch and an indicator winch are installed on the rotating shaft. The anchor claw winch is connected to an anchor claw rope with a tension sensor, and its end is provided with an anchor claw that can be fixed to the rock mass in the roadway sidewall. The indicator winch is connected to an indicator rope, and the force and displacement of the anchor claw rope are simulated by an indicator tension sensor. This device transmits the displacement and stress state of the deep rock mass to the roadway interior, which is easy to measure, through the mechanical equivalent conversion principle, thereby monitoring the deformation of the surrounding rock.

[0003] (2) Chinese invention patent with publication number CN114442099A discloses a method and device for detecting obstacles in underground mines based on multi-line radar. This invention acquires point cloud data of multi-line radar and single-line data in the horizontal direction, and uses the position points of fixed-angle radar lines on both sides of the moving object's direction of travel on the sidewall of the tunnel to determine whether the angle difference is less than a threshold, so as to dynamically identify straight or curved areas.

[0004] However, existing roadway safety monitoring mainly relies on data processing and fixed installation structures, which have limited detection capabilities. They are unable to dynamically compensate for vehicle vibrations in real time and effectively, or dynamically suppress and compensate for complex vibration environments underground. They cannot adequately meet the requirements of modern mine safety monitoring for high precision, high reliability, and strong anti-interference capabilities. Summary of the Invention

[0005] To solve the above problems, the technical solution adopted in this application is a method for processing measurement data of the two sides of underground roadways using millimeter-wave radar, characterized by comprising: Data acquisition and processing: The millimeter-wave radar mounted on the measuring device detects the triangular angle reflection of the measured device to obtain detection data. The detection data is then denoised using a Gaussian mixture filtering method. The denoised data is used as the tunnel width to determine whether dangerous deformation has occurred. The tunnel height is scanned by the millimeter-wave radar to obtain tunnel surface point cloud data. The point cloud data is then processed using Poisson optimization based on local topological constraints to obtain the final tunnel model. Based on the final tunnel model, the tunnel height is resampled and calculated. Effective data acquisition and control: The angular change data of the measuring device is collected by the accelerometer and gyroscope mounted on the measuring device, and effective data acquisition and control is performed based on the angular change data.

[0006] Optionally, denoising the probe data based on the Gaussian mixture filtering method includes: A dataset is constructed by collecting real-time distances from the radar probe to each corner reflector. A Gaussian mixture filter model is trained based on the dataset. New data is collected to construct a new dataset. The new dataset is denoised using the Gaussian mixture filter model. Finally, the denoised dataset is smoothed using a Gaussian smoothing filter.

[0007] Optionally, training a Gaussian mixture filter model based on a dataset includes defining the parameter set λ of the Gaussian mixture model, which consists of K Gaussian components. The calculation of λ is shown in the following formula: ; In the formula, The total number of Gaussian components is preset. Let Variance be the variance of the k-th Gaussian component. For the first The mean of the Gaussian components, For the first The weights of the Gaussian components represent the proportion of that component in the overall distribution, and the sum of all weights is 1, i.e.: ; Probability calculations are performed on the obtained data, given data points. The probability density function following this Gaussian mixture model distribution is expressed as: ; The parameter set data is trained using the expectation-maximization algorithm.

[0008] Optionally, denoising the new dataset using a Gaussian mixture filter model includes: During the real-time processing phase, the millimeter-wave radar mounted on the measuring device detects the triangular angles of the measured device to obtain detection data, forming a new dataset. New ranging data points within this new dataset are then processed. Models trained offline Perform noise reduction; Calculate the probability that a data point belongs to each component: ; Offline analysis was used to determine that the Gaussian components represent the effective components of the real data, assuming that this component is the first... One component; define the threshold. If new data points The probability of belonging to a valid component is lower than the threshold. ,Right now If the data point is noisy, then this data point is considered noise. .

[0009] Optionally, Gaussian smoothing filtering is applied to the denoised new dataset, including the denoised data stream. Perform a Gaussian smoothing filter for the m-th data point. Filtered numerical values It can be represented as: .

[0010] Optionally, Poisson optimization processing of point cloud data based on local topological constraints includes: processing the collected point cloud data of the roadway surface. ,have ; Based on the tunnel inspection time ,Will Divided into overlapping local data blocks Used for local analysis; for Each point in , find it The nearest neighbor set Calculate the neighborhood centroid : ; In the formula, Representative point Any point in the neighborhood; The roadway surface indicator function was found using the Poisson reconstruction method and global continuous regularization. Where the energy function is minimized This ensures that the reconstruction results remain smooth and continuous along the tunnel axis, and the indicator function is determined. ; Minimize energy function : ; In the formula, To indicate the gradient of the function, For the target vector field, For regularization parameters; From the solution Extracting isosurfaces As the final tunnel model.

[0011] Optionally, effective data acquisition control includes: Accelerometers measure the acceleration components of each axis, and then... , , The acceleration of the shaft is used to calculate the tilt angle of the detection device. ; In the formula, The tilt angle of the detection device is measured and calculated by the accelerometer. Let x be the acceleration component along the x-axis. The acceleration component along the y-axis is... The acceleration component along the z-axis; The tilt angle of the detection device measured by the gyroscope is obtained by integrating the angular velocity measured by the gyroscope. ; In the formula, The tilt angle of the detection device as measured by the gyroscope at the current moment; This is the initial angle; Angular velocity measured by a gyroscope; Ruodang or If the value continuously exceeds the set threshold for a certain period of time, it is determined to be a strong interference condition, and the power supply will be cut off immediately through the relay to stop the machine.

[0012] Optionally, effective data acquisition control also includes when or When the value continuously exceeds the set threshold, or For data whose values ​​continuously exceed the set threshold within a certain time period, the angle inverse intrinsic parameter verification and the linkage verification of the distance between the two walls, yaw angle and pitch angle are performed sequentially based on the angle change data. When the data within this time period passes the verification, the data whose angle changes exceed the threshold within the time period are calibrated based on the angle change data, and the calibrated data is added to the new dataset; otherwise, the data within this time period is removed. Angular reflection internal parameter verification includes using the fixed distance between angular reflections and inferring the measured distance between angular reflections from the measured distance between the radar and the angular reflections to verify the reliability of the ranging data; The verification of the linkage between the distance between the two walls, yaw angle, and pitch angle includes calculating the measured width deviation and the theoretical angle deviation based on the yaw angle, pitch angle, and tunnel width data; calculating the error rate between the measured width deviation and the theoretical angle deviation; and verifying whether the tunnel width deviation is caused by the angle offset. The data calibrated based on angle change data within a time period exceeding the threshold includes both detection data and tunnel surface point cloud data, which are then applied to the calculation of tunnel width and tunnel height.

[0013] This application also provides a device for measuring the sidewall measurement data of a millimeter-wave radar in an underground roadway, characterized in that it is suitable for performing any of the aforementioned methods for processing the sidewall measurement data of a millimeter-wave radar in an underground roadway, including: The measuring device and the device being measured are respectively fixed to the sidewalls on both sides of the roadway. The measuring device and the device being measured are installed at the same height and are coaxially arranged. The measuring device includes a measuring device housing, which includes a connection end fixed to the sidewall of the roadway and a measuring end facing the device being measured. The measuring end facing the device being measured is equipped with a millimeter-wave radar. The measuring device housing is also equipped with a laser pointer facing the device being measured. The measuring device housing contains a data processor, a relay, and an alarm device. A vibration monitor, including a gyroscope and an accelerometer, is fixedly mounted on the housing of the measuring device. The device under test includes a feedback component consisting of three triangles arranged in reverse triangles. The feedback component is positioned facing the measuring device, and a baffle is provided on the side away from the measuring device. The device under test is fixed to the sidewall of the roadway by anchor bolts.

[0014] Optionally, the side length of the triangular inverted triangle is in the range of 10cm-30cm, and the triangle formed by the installation positions of the three triangular inverted triangles is an equilateral triangle. The connection end is fixed to the roadway sidewall by anchor bolts and fixing rods. A universal joint with a ball joint structure is provided between the anchor bolts and fixing rods. The angle of the universal joint is fixed by locking bolts. The ball joint structure of the universal joint allows the measuring device to be adjusted within the range of pitch angle ±15° and yaw angle ±10°.

[0015] The beneficial effects of the millimeter-wave radar data processing method and measuring device for underground roadways provided in this application are as follows: The underground roadway millimeter-wave radar sidewall measurement device provided in this application achieves high-precision monitoring of roadway sidewall deformation and height through the coordinated operation of a millimeter-wave radar probe, vibration monitor, data processor, and alarm device. The system utilizes Gaussian mixture filtering and Poisson optimization with local topological constraints to denoise and optimize the raw detection data. Combined with real-time distance calculation and deformation judgment mechanisms, it effectively identifies dangerous roadway conditions and provides timely warnings through the alarm device. This significantly improves the reliability and accuracy of the monitoring data.

[0016] The millimeter-wave radar sidewall measurement data processing method provided in this application for underground roadways employs Gaussian mixture filtering to smooth the millimeter-wave radar ranging signal, effectively suppressing abrupt noise interference. Combined with a deformation judgment value and a safety distance comparison mechanism, it achieves intelligent identification and early warning of roadway deformation. Furthermore, it optimizes the roadway height data through Poisson optimization with local topological constraints, extracting true surface information and enhancing the system's robustness in noisy environments. It measures the tilt angle using a fusion of accelerometer and gyroscope measurements and incorporates a power-off protection mechanism to automatically cut off power under strong interference conditions, preventing equipment damage and invalid data acquisition, thus improving the system's adaptability and safety in complex underground environments. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0018] Figure 1 This is a perspective view of the measuring device for both sides provided in the embodiments of this application; Figure 2 This is a side view schematic diagram of the measuring device for both sides provided in an embodiment of this application. Figure 3 This is a cross-sectional schematic diagram of the vibration monitor of the two-sided measuring device provided in the embodiments of this application; Figure 4 This is a cross-sectional schematic diagram of the housing of the two-sided measuring device provided in an embodiment of this application; Figure 5 This is a cross-sectional schematic diagram of the universal joint of the two-sided measuring device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the process for processing measurement data of the two sides of a mine roadway using millimeter-wave radar, as provided in an embodiment of this application.

[0019] Explanation of reference numerals in the attached diagram: 1-Millimeter-wave radar probe; 2-Laser pointer; 3-Vibration monitor; 31-Accelerometer; 32-Gyroscope; 4-Housing; 41-Data processor; 42-Relay; 43-Alarm device; 44-Fixing frame; 5-Anchor bolt; 6-Universal joint; 61-Cylindrical base; 62-Sphere; 63-Limiting ring; 64-Locking bolt; 7-Baffle; 8-Angle reflector; 9-Target center; 10-Laser channel. Detailed Implementation

[0020] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0021] Example 1 like Figures 1-2 As shown, this application provides a millimeter-wave radar measuring device for the two sides of an underground roadway, including a measuring device, a device under test, and a data processing and alarm system; The measuring device includes a housing 4, with a millimeter-wave radar probe 1 mounted on the front end and a laser pointer 2 and a vibration monitor 3 (containing an accelerometer 31 and a gyroscope 32) located on the side. The housing 4 is fixed to one side of the roadway by anchor bolts 5, with universal joints 6 of ball-and-socket structure between the anchor bolts 5, and secured by locking bolts 64. Figure 5 As shown, the universal joint 6 consists of a cylindrical base 61, a ball 62, a limiting ring 63, and a locking bolt 64. The ball 62 is connected to the anchor rod 5 and can move within a certain range. It is fixed by the locking bolt 64, so that the radar probe can be adjusted within the range of pitch angle ±15° and yaw angle ±10° to adapt to the actual installation requirements of the roadway.

[0022] The device under test consists of three triangular reflectors (8) arranged in an equilateral triangle. In this embodiment, the side length of each reflector is 10cm. A metal baffle (7) is installed behind each reflector and is fixed to the other side of the tunnel by anchor bolts (5). The vertices of the three reflectors (8) form an equilateral triangle to enhance the radar echo signal. "Corner reflector" refers to a corner reflector.

[0023] like Figure 4 As shown, the housing 4 contains a data processor 41, a relay 42, and an alarm device 43. The data processor 41 is responsible for receiving and processing data from the millimeter-wave radar probe 1, the accelerometer 31, and the gyroscope 32; the relay 42 is used to perform a power-off operation under strong interference conditions; and the alarm device 43 is used to issue an audible and visual alarm when deformation or height abnormalities are detected.

[0024] like Figure 1 and Figure 3As shown, the accelerometer 31 and gyroscope 32 in the vibration monitor 3 are installed at the upper end of the housing 4 to obtain the tilt data of the detection device in real time. The data processor 41 obtains the overall tilt information by collecting the tilt data of the detection device and calculating the tilt angle, and determines whether to send it to the alarm device 43 based on the obtained information.

[0025] When installing the measuring device, the anchor rod 5 needs to be installed on both sides of the roadway. The section of the anchor rod 5 with the ball 62 is inserted into the anchor rod 5 of the cylindrical base 61. The limit ring 63 is fixed by welding or bolts to limit the axial movement of the ball 62. The housing 4 and the anchor rod 5 with the ball 62 are connected by welding. The millimeter-wave radar probe 1 is fixed to the housing 4 by locking buckle.

[0026] The device under test only needs to be fixed to the other side of the roadway by drilling holes in the anchor bolt 5.

[0027] Before operation, a laser pointer 2 is used to align the detection device and the device under test on the same straight line. The laser pointer 2 is mounted above the detection device via a mounting bracket 44, with the axial direction of the laser pointer 2 parallel to the axial direction of the detection device. The ideal position is achieved by fine-tuning the ball 62 of the universal joint 6. Once the ideal position is reached, it is secured by the locking bolts 64 around the cylindrical base 61 to ensure the stability of the detection device.

[0028] When the emitted light from the laser pointer 2 passes through the laser channel 10 of the baffle 7 and forms a light spot on the underground tunnel wall behind the baffle 7, the auxiliary positioning correction based on the laser pointer 2 is completed.

[0029] The distance between the target center 9 and the deepest point inside the three corner reflectors is the same, which is used as an auxiliary determination of the installation position of the corner reflectors. The three corner reflectors need to be installed at equal intervals with the measuring device. However, the initial angle difference between the two walls of the tunnel will affect the installation positioning of the corner reflectors. The target center 9 serves as the distance measurement reference. By adjusting the installation position of the corner reflectors based on the distance between the target center 9 and the measuring device, it can be ensured that the three corner reflectors form an installation plane perpendicular to the axis of the measuring device.

[0030] Example 2 like Figure 6 As shown, this application provides a method for processing measurement data of the two sides of an underground roadway using millimeter-wave radar, including the following steps: S1: The detection data is obtained by detecting the triangular angle of the device under test through the millimeter-wave radar probe in front of the measuring device. The obtained detection data is transmitted to the data processor, which processes the received data and sends the processed data to the alarm device.

[0031] Millimeter-wave radar can provide accurate distance measurement information when the underground environment is relatively stable. However, it is subject to interference under complex tunnel conditions and strong vibration operation conditions, which causes fluctuations in the detected data. By processing the detected data through Gaussian mixture filtering algorithm, errors can be effectively suppressed and measurement accuracy can be improved.

[0032] Calculate the real-time distance from the radar probe to each corner reflector. ， ， Take the maximum value among the three distances. Calculate the deformation judgment value: .

[0033] For ranging signals acquired by millimeter-wave radar ， ， The specific steps for denoising millimeter-wave radar ranging signals using Gaussian mixture filtering to address existing noise are as follows: A dataset is constructed by collecting real-time distances from the radar probe to each corner reflector. A Gaussian mixture filter model is trained based on the dataset. New data is collected to construct a new dataset. The new dataset is denoised using the Gaussian mixture filter model. Finally, the denoised dataset is smoothed using a Gaussian smoothing filter.

[0034] In this embodiment, the original dataset uses continuous millimeter-wave radar detection data that does not show an angle shift exceeding the threshold of 5 seconds. After the Gaussian mixture filter model is trained, the original dataset is updated every 6 hours. After the update, the Gaussian mixture filter model is retrained using the new dataset, and the acquisition frequency is 20-30 times / minute.

[0035] Define the parameter set of the Gaussian mixture model, denoted as λ. λ is a Gaussian mixture model composed of K Gaussian components, and the calculation of λ is shown in the following formula: ; In the formula, The total number of Gaussian components is preset. Let Variance be the variance of the k-th Gaussian component. For the first The mean of the Gaussian components, For the first The weights (or mixing coefficients) of each Gaussian component represent the proportion of that component in the overall distribution. The sum of all weights must be 1, i.e.: ; Probability calculations are performed on the obtained data, given data points. The probability density function following this Gaussian mixture model distribution is expressed as: ; in, It is the first The probability density function of the Gaussian components is calculated using the following formula: ; The model is trained using the expectation-maximization (EM) model, and the parameter set data is trained using the EEM algorithm.

[0036] Expected step: Calculate each data point Posterior probability belonging to each Gaussian component k : ; In the formula, This represents the weight of the k-th Gaussian component in the previous iteration. Let $\mathbf{k}$ represent the mean of the $k$-th Gaussian component from the previous iteration. This represents the variance of the k-th Gaussian component in the previous iteration. This represents the weight of the j-th Gaussian component in the previous iteration. Let represent the mean of the j-th Gaussian component in the previous iteration. This represents the variance of the j-th Gaussian component in the previous iteration, where j is the index of the Gaussian component.

[0037] Maximize step: Utilize Update Gaussian mixture model parameters : ; ; ; In the formula, N represents the total number of data points. This represents the weight of the k-th Gaussian component after the update. This represents the mean of the k-th Gaussian component after the update. This represents the variance of the k-th Gaussian component after the update.

[0038] During the real-time processing phase, when new ranging data points are received... At that time, the model was trained offline. To perform noise reduction.

[0039] Calculate the probability that a data point belongs to each component: ; Offline analysis was used to determine that the Gaussian components represent the effective components of the real data, assuming that this component is the first... Each component. Define the threshold. If new data points The probability of belonging to a valid component is lower than the threshold. ,Right now If the data point is not found to be noise, then it is considered to be noise.

[0040] ; In the formula, This indicates the processed data. Indices representing the index of the valid component. This indicates that the effective components are homogeneous.

[0041] For the denoised data stream Perform a Gaussian smoothing filter with a window size of 51-101, where the windows do not overlap. For the m-th data point... Filtering value Represented as: ; In the formula, j represents the offset within the window. This represents the (m+j)th data point within the window. This is half the length of the Gaussian filter window. The weights of the Gaussian smoothing kernel are given by the following formula: ; In the formula, This is the fixed variance of the Gaussian smoothing filter.

[0042] The real-time distance D of the millimeter-wave radar to each corner reflector is obtained from the denoised signal. A D B D C ; Take the maximum value of the three distances ; Calculate the deformation judgment value: ; Will Safety distance from the initial calibration Compare; like ≥ If a dangerous deformation is detected in the tunnel, the alarm device will immediately issue a warning, including sound and light.

[0043] like < The system determines that the roadway is safe and continues monitoring.

[0044] S2: The millimeter-wave radar scans the tunnel height, and the built-in data processing module performs noise reduction on the raw data. For noise, Poisson optimization with local topological constraints is used, and the final result is sent to the alarm device.

[0045] The millimeter-wave radar of the detection device acquires data on the tunnel height, which is then transmitted to a data processor to perform Poisson optimization with local topological constraints. The specific steps are as follows: For the above-collected original point cloud set ,in ; Based on the tunnel inspection time ,Will Divided into overlapping local data blocks Used for local analysis. Each point in , find it The nearest neighbor set Calculate the neighborhood centroid : ; In the formula, Representative point Any point in the neighborhood.

[0046] The normal vector is corrected based on curvature analysis to apply local constraints: Calculate the covariance matrix : ; right Perform eigenvalue decomposition to obtain eigenvalues. and the corresponding feature vector , , ; Set the initial normal vector : Curvature threshold minimum weight Local curvature estimation : ; Determine the corrected normal vector and weight : ; In the formula, It is the average of the normal vectors of all smooth points in its neighborhood.

[0047] Define the vector field function: ; The roadway surface indicator function was found using the Poisson reconstruction method and global continuous regularization. Where the energy function is minimized This ensures that the reconstruction results remain smooth and continuous along the tunnel axis, and the indicator function is determined. .

[0048] Minimize energy function : ; In the formula, To indicate the gradient of the function, For the target vector field, This is the regularization parameter.

[0049] From the solution Extracting isosurfaces As the final tunnel model, along the tunnel axis sampling, Representative point Calculate the vertical coordinates. In this cross section The alleyway is high : .

[0050] S3: The measuring device is equipped with an accelerometer and gyroscope in the vibration monitor to collect the vibration amplitude of this device. The obtained detection data is transmitted to the data processor. The data processor processes the received data to determine whether the vibration angle is a strong interference condition, and sends the processed information to the relay power-off protection device to avoid collecting invalid data.

[0051] The tilt angle of the detection device is measured simultaneously by an accelerometer and a gyroscope and calculated by a data processor. The measurement method is to measure the acceleration components of each axis by the accelerometer and calculate the tilt angle of the detection device by the acceleration of the x, y, and z axes.

[0052] ; In the formula, The tilt angle of the detection device is measured and calculated by the accelerometer. Let x be the acceleration component along the x-axis. The acceleration component along the y-axis is... The acceleration component along the z-axis is denoted as .

[0053] The tilt angle of the detection device measured by the gyroscope is obtained by integrating the angular velocity measured by the gyroscope. ; In the formula, The tilt angle of the detection device as measured by the gyroscope at the current moment; This is the initial angle; The angular velocity is measured by the gyroscope.

[0054] Ruodang or θ accel If the value continuously exceeds the set threshold for a certain period of time, it is determined to be a strong interference condition. The power supply is immediately cut off via a relay to protect the equipment and prevent the collection of invalid data. In this embodiment, or θ accel The numerical threshold is 2° and the time is 5s. When actually using the solution provided in this application, the maximum angular displacement of the fixed facilities on the roadway wall caused by a typical vehicle passing through the roadway and the time of a single vehicle passing through are used as the benchmarks for determining the threshold and time.

[0055] Effective data acquisition control also includes when or When the value continuously exceeds the set threshold, or For data whose values ​​continuously exceed the set threshold within a certain time period, the angle inverse intrinsic parameter verification and the linkage verification of the distance between the two walls, yaw angle, and pitch angle are performed sequentially based on the angle change data. When the data within this time period passes the verification, the data whose angle changes exceed the threshold within the time period are calibrated based on the angle change data, and the calibrated data is added to a new dataset; otherwise, the data within this time period is removed.

[0056] Angular inverse distance (D) A D B D C ), distance between the two walls ( ) and angle data (tilt angle measured by accelerometer) tilt angle measured by gyroscope The measurement data are based on the following: angular reflection geometric reference (the distance a0 between the three angular reflections), width reference (W0 represents the average value of data during periods when no angular deviation exceeds the threshold), angle reference (0° since the measuring device and the measured device are coaxially mounted), and included angle reference (based on the installation positions of the three angular reflections being points A, B, and C, respectively). The angle between the line connecting point A and the radar launch position and the line connecting point B and the radar launch position is used as static data. The internal parameter deviation threshold for angle inverse internal parameter verification is set to 5%, and the error rate threshold for the linkage verification of distance between two walls, yaw angle and pitch angle is set to 10%. These are used for angle inverse internal parameter verification and linkage verification of distance between two walls, yaw angle and pitch angle.

[0057] Angular inversion intrinsic parameter verification involves using the fixed spacing between angular inversions and inferring the measured spacing between the angular inversions from the measured distance between the radar and the angular inversions to verify the reliability of the ranging data. Since the spacing between the angular inversions does not change with the deformation of the tunnel, angular inversion intrinsic parameter verification can check whether the radar ranging accuracy has changed.

[0058] First, the measured distance between angle reflections A and B is calculated using the angle reflection distance before noise reduction and the included angle reference. ; In the formula, D A D represents the distance at which the radar reaches the angular reflection. B This indicates the distance the radar reaches angle B. This represents the angle formed by the line connecting the radar to angle reflector A and the line connecting the radar to angle reflector B. The measured distance between angle reflectors A and C is calculated using the same method. And the measured distance between the angle reflections B and C. Then take the average value to get .

[0059] Then, deviation judgment is performed, and the measured spacing is calculated. The deviation from the reference a0 is based on the following formula: ; Since the intrinsic parameter deviation threshold for inverse intrinsic parameter verification is set to 5%, when... ≤5% If the measurement data is considered reliable, then the verification procedures for the distance between the two walls and the yaw angle linkage, and the tunnel height and pitch angle linkage verification will begin. >5% If the measurement data is deemed unreliable, the radar detection will be shut down for maintenance or the data within that time period will be excluded.

[0060] The verification of the linkage between the two wall spacing and the yaw angle includes calculating the measured width deviation and the theoretical yaw angle deviation based on the yaw angle and the tunnel width data, calculating the error rate between the measured width deviation and the theoretical yaw angle deviation, and verifying whether the tunnel width deviation is caused by the angular offset. The impact of angular offset on width was verified by linking the distance between the two walls, yaw angle, and pitch angle. D max As the actual test distance between the two walls, when there is an angular offset in the radar, D max The inaccuracy will be due to beam projection deviation. The cause of the inaccuracy is verified by matching the measured width deviation with the theoretical angle deviation: First, calculate the theoretical width D that can be obtained theoretically due to the angular deviation. 理论 Based on the following formula: ； In the formula, Indicates pitch angle, Indicates the yaw angle.

[0061] Perform a consistency assessment and calculate the error rate between the measured width and the theoretical width. Based on the following formula: ; like If the deviation is ≤10%, confirm whether the roadway width deviation is caused by angular offset. Based on the angular change data, calibrate the detection data and roadway surface point cloud data, and then apply them to the calculation of roadway width and roadway height. like If the deviation is greater than 10%, the ranging inaccuracy is determined to exceed the calibration limit, and the measurement data within that time period is removed.

[0062] The calibration of the detection data based on angle change data includes correcting the real-time distance from the radar probe to each corner reflector within that time period according to the following formula. , , The corrected D is obtained A修正 D B修正 D C修正 : ; ; ; Correct the roadway surface point cloud data according to the following formula: The raw radar point cloud data is corrected for yaw and pitch angles to obtain the true point cloud data. The yaw angle correction is used to convert the coordinate system S directly measured by the radar under angular offset. r Transform into an intermediate coordinate system S without left or right tilt. m Coordinate system S r The point cloud coordinates are ( ) indicates that the intermediate coordinate system S m The point cloud coordinates are ( ) indicates that, through the rotation matrix R yaw Perform yaw angle correction: ; The point cloud coordinates in the intermediate coordinate system are calculated as follows: ; The point cloud obtained from the oblique section scan is rotated into a point cloud of a vertical cross-section parallel to the tunnel axis. The pitch angle correction is used to adjust the intermediate coordinate system S. m Convert to the real tunnel coordinate system S t The actual tunnel coordinate system S tThe point cloud coordinates are ( ) indicates that, through the rotation matrix R pitch Perform pitch angle correction: ; The point cloud coordinates in the intermediate coordinate system are calculated as follows: ; The corrected point cloud is used to calculate the tunnel height in the final tunnel model.

[0063] The other combinations and connections in this implementation scheme are the same as in Example 1.

[0064] The above methods enable the measuring device to remain stable in complex underground vibration environments, effectively suppressing ranging signal fluctuations and noise interference caused by strong vibrations in millimeter-wave radar. Gaussian mixture filtering is used to smooth and denoise the radar ranging signal, accurately extracting roadway deformation information; Poisson optimization of height data with local topological constraints accurately reconstructs the roadway surface morphology; and accelerometers and gyroscopes in the vibration monitor monitor the equipment attitude in real time, identifying strong interference conditions and triggering power-off protection. This entire method significantly improves the accuracy, anti-interference capability, and reliability of millimeter-wave radar in underground roadway monitoring.

[0065] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for processing measurement data from millimeter-wave radar on both sides of an underground roadway, characterized in that, include: Data acquisition and processing: The millimeter-wave radar mounted on the measuring device detects the triangular angle reflection of the measured device to obtain detection data. The detection data is then denoised using a Gaussian mixture filtering method. The denoised data is used as the tunnel width to determine whether dangerous deformation has occurred. The millimeter-wave radar scans the tunnel height to obtain tunnel surface point cloud data. The point cloud data is then processed using Poisson optimization based on local topological constraints to obtain the final tunnel model. Based on the final tunnel model, the tunnel height is resampled and calculated. Effective data acquisition and control: The angular change data of the measuring device is collected by the accelerometer and gyroscope mounted on the measuring device, and effective data acquisition and control is performed based on the angular change data.

2. The method for processing measurement data of the two sides of an underground roadway using millimeter-wave radar according to claim 1, characterized in that: The noise reduction of the probe data based on the Gaussian mixture filtering method includes: A dataset is constructed by collecting real-time distances from the radar probe to each corner reflector. A Gaussian mixture filter model is trained based on the dataset. New data is collected to construct a new dataset. The new dataset is denoised using the Gaussian mixture filter model. Finally, the denoised dataset is smoothed using a Gaussian smoothing filter.

3. The method for processing measurement data of the two sides of an underground roadway using millimeter-wave radar according to claim 2, characterized in that: The training of the Gaussian mixture filter model based on the dataset includes defining the parameter set λ of the Gaussian mixture model, which consists of K Gaussian components. The calculation of λ is shown in the following formula: ； In the formula, The total number of Gaussian components is preset. Let Variance be the variance of the k-th Gaussian component. For the first The mean of the Gaussian components, For the first The weights of the Gaussian components represent the proportion of that component in the overall distribution, and the sum of all weights is 1, i.e.: ； Probability calculations are performed on the obtained data, given data points. The probability density function following this Gaussian mixture model distribution is expressed as: ； The parameter set data is trained using the expectation-maximization algorithm.

4. The method for processing measurement data of the two sides of an underground roadway using millimeter-wave radar according to claim 2, characterized in that: The method of using a Gaussian mixture filter model to denoise the new dataset includes: During the real-time processing phase, the millimeter-wave radar mounted on the measuring device detects the triangular angles of the measured device to obtain detection data, forming a new dataset. New ranging data points within this new dataset are then processed. Models trained offline Perform noise reduction; Calculate the probability that a data point belongs to each component: ； Offline analysis was used to determine that the Gaussian components represent the effective components of the real data, assuming that this component is the first... One component; define the threshold. If new data points The probability of belonging to a valid component is lower than the threshold. ,Right now If so, this data point is considered noise; 。 5. The method for processing measurement data of the two sides of an underground roadway using millimeter-wave radar according to claim 4, characterized in that: The Gaussian smoothing filter applied to the denoised new dataset includes processing the denoised data stream... Perform a Gaussian smoothing filter for the m-th data point. Filtered values It can be represented as: 。 6. The method for processing measurement data of the two sides of an underground roadway using millimeter-wave radar according to claim 1, characterized in that: The Poisson optimization processing of point cloud data based on local topological constraints includes: processing the collected point cloud data of the roadway surface. ,have ; Based on the tunnel inspection time ,Will Divided into overlapping local data blocks Used for local analysis; for Each point in , find it The nearest neighbor set Calculate the neighborhood centroid : ； In the formula, Representative point Any point in the neighborhood; The roadway surface indicator function was found using the Poisson reconstruction method and global continuous regularization. Where the energy function is minimized This ensures that the reconstruction results remain smooth and continuous along the tunnel axis, and the indicator function is determined. ; Minimize energy function : ； In the formula, To indicate the gradient of the function, For the target vector field, For regularization parameters; From the solution Extracting isosurfaces As the final tunnel model.

7. The method for processing measurement data of the two sides of an underground roadway using millimeter-wave radar according to claim 1, characterized in that: The effective data acquisition control includes: Accelerometers measure the acceleration components of each axis, and then... , , The acceleration of the shaft is used to calculate the tilt angle of the detection device. ； In the formula, The tilt angle of the detection device is measured and calculated by the accelerometer. Let x be the acceleration component along the x-axis. The acceleration component along the y-axis is... The acceleration component along the z-axis; The tilt angle of the detection device measured by the gyroscope is obtained by integrating the angular velocity measured by the gyroscope. ； In the formula, The tilt angle of the detection device as measured by the gyroscope at the current moment; This is the initial angle; Angular velocity measured by a gyroscope; Ruodang or If the value continuously exceeds the set threshold for a certain period of time, it is determined to be a strong interference condition, and the power supply will be cut off immediately through the relay to stop the machine.

8. The method for processing measurement data of the two sides of an underground roadway using millimeter-wave radar according to claim 7, characterized in that: The effective data acquisition control also includes when or When the value continuously exceeds the set threshold, or For data whose values ​​continuously exceed a set threshold within a certain time period, the angle change data is used to sequentially perform angle inverse intrinsic parameter verification and two-wall distance-yaw angle-pitch angle linkage verification. When the data within this time period passes the verification, the data whose angle changes exceed the threshold within a certain time period is calibrated based on the angle change data, and the calibrated data is added to a new dataset; otherwise, the data within this time period is removed. The verification of the angle reflection intrinsic parameters includes using the fixed distance between the angle reflections to infer the measured distance between the angle reflections from the measured distance between the radar and the angle reflections, thereby verifying the reliability of the ranging data. The verification of the linkage between the two wall spacing, yaw angle, and pitch angle includes calculating the measured width deviation and the theoretical angle deviation based on the yaw angle, pitch angle, and tunnel width data; calculating the error rate between the measured width deviation and the theoretical angle deviation; and verifying whether the tunnel width deviation is caused by the angle offset. The calibration of angle change data within a time period exceeding a threshold includes calibrating the detection data and the tunnel surface point cloud data based on angle change data, which are then applied to the calculation of tunnel width and tunnel height.

9. A device for measuring data from millimeter-wave radar on both sides of an underground roadway, characterized in that: The method for processing measurement data of the two sides of a mine roadway using millimeter-wave radar as described in claims 1-8 includes: A measuring device and a measuring device are respectively fixed to the sidewalls on both sides of the roadway. The measuring device and the measuring device are installed at the same height and are coaxially arranged. The measuring device includes a measuring device housing, which includes a connection end fixed to the sidewall of the roadway and a measuring end facing the device being measured. The measuring end facing the device being measured is equipped with a millimeter-wave radar. The measuring device housing is also equipped with a laser pointer facing the device being measured. The measuring device housing is equipped with a data processor, a relay and an alarm device. A vibration monitor is fixedly installed on the housing of the measuring device, and the vibration monitor includes a gyroscope and an accelerometer; The device under test includes a feedback component consisting of three triangles arranged in reverse triangles. The feedback component is positioned facing the measuring device, and a baffle is provided on the side away from the measuring device. The device under test is fixed to the sidewall of the tunnel by anchor bolts.

10. The underground roadway millimeter-wave radar measurement device for measuring data from both sides according to claim 9, characterized in that: The side length of the triangular inverted triangle is between 10cm and 30cm, and the triangle formed by the installation positions of the three triangular inverted triangles is an equilateral triangle. The connecting end is fixed to the roadway sidewall by anchor rods and fixing rods. A universal joint with a ball joint structure is provided between the anchor rods and fixing rods. The angle of the universal joint is fixed by locking bolts. The ball joint structure of the universal joint allows the measuring device to be adjusted within the range of pitch angle ±15° and yaw angle ±10°.

Images