A knowledge- and data-driven virtual reconstruction method for the relative pose of hydraulic supports

Abstract

This invention relates to a knowledge- and data-driven virtual reconstruction method for the relative pose of a hydraulic support. It constructs a digital twin model system of a physical prototype of the hydraulic support, and uses a sensor position planning system to guide the placement of sensors in a physical monitoring system. Sensor data from the physical monitoring system is collected by a data acquisition module and transmitted to a mechanism model and data processing module. After being corrected by a parameter preprocessing module, the data is used as known quantities and transmitted to a relative position calculation module for calculation. The calculation results are dynamically mapped in real-time onto a pose inference system. A pose correction module iterates on the results of the pose inference system, constructing a knowledge- and data-driven reconstruction method. Finally, the pose information of the hydraulic support is displayed in a human-computer interaction system, and an accuracy evaluation system assesses the error of the inferred pose information. The human-computer interaction system is used to monitor the status of all systems.

Description

Technical Field

[0001] This invention relates to the field of hydraulic support pose monitoring, and more specifically, to a knowledge- and data-driven method for virtual reconstruction of the relative pose of hydraulic supports. Background Technology

[0002] As a crucial support system in fully mechanized mining faces, the stable and reliable operation of hydraulic support groups is fundamental to achieving safe and efficient coal mining. For hydraulic support groups under actual working conditions, it is essential to ensure both the stable and reliable operation of individual support units and the sequential coordinated operation of the entire group.

[0003] Currently, the position and posture monitoring of a single hydraulic support mainly focuses on the angle monitoring of the main structures such as the hydraulic support base, top beam, and side guard plate. This is achieved by installing various contact sensors such as tilt sensors, stroke sensors, and gyroscopes. For the position and posture monitoring of hydraulic support groups, relevant scholars have proposed using multi-source information fusion technology to fuse information from multiple sensors to achieve position and posture monitoring of hydraulic support groups.

[0004] The patent application CN110007309A, entitled "An Intelligent Monitoring System for the Attitude of a Covered Hydraulic Support and Its Measurement Method", uses a 3-axis tilt sensor to measure the initial state of the intelligent monitoring device, uses an external lidar for vertical scanning, and uses a host computer to process the data to finally obtain the attitude parameters of the hydraulic support.

[0005] The paper "A MEMS-based method for monitoring the attitude of a hydraulic support" (publication number CN113340305A) measures the attitude of the hydraulic support using an inertial measurement unit and an inclination sensor, establishes a solution model for the support height of the hydraulic support, and finally achieves an accurate solution for the pitch angle between the top beam and the base of the hydraulic support.

[0006] The published paper "Relative Pose Detection and Online Data Display Technology of Hydraulic Support" establishes a mathematical model of the relative pose of hydraulic supports based on orientation angle monitoring, and uses a combination of orientation angle monitoring device and pull-wire sensor to solve the relative pose of hydraulic supports.

[0007] The published paper, "A Visual Measurement-Based Method for Hydraulic Support Pose Detection," uses an explosion-proof camera installed on a coal mining machine to acquire images of four LED signboards arranged on a hydraulic support. Visual algorithms are then used to process the images to obtain the pose of the hydraulic support.

[0008] While some progress has been made in the study of the position and orientation of individual hydraulic supports, obtaining accurate and reliable support position information for the monitoring of hydraulic support groups remains a major challenge. Furthermore, the complex working conditions underground prevent the application of many sensors, and sensor installation and information fusion are also significant difficulties. Summary of the Invention

[0009] The technical problem to be solved by this invention is to provide a knowledge- and data-driven virtual reconstruction method for the relative pose of hydraulic supports, thereby obtaining the support posture and relative position of the hydraulic supports, ensuring stable support of the working face roof, providing monitoring for the automated process of real underground hydraulic support groups, and providing effective reference for the adjustment and control of hydraulic supports.

[0010] To address the above technical problems, this invention provides a knowledge- and data-driven method for virtual reconstruction of the relative pose of a hydraulic support, comprising a physical monitoring system, a mechanism model and data processing system, and a virtual pose reconstruction system.

[0011] The physical monitoring system includes sensors and a data acquisition module; the mechanism model and data processing system includes a parameter preprocessing module, a relative position calculation module, and a pose correction module; the virtual pose reconstruction system includes a digital twin model system, a sensor position planning system, a pose inference system, a human-computer interaction system, and an accuracy evaluation system.

[0012] A digital twin model system for a physical prototype of a hydraulic support is constructed. A sensor position planning system guides the placement of sensors in the physical monitoring system. Sensor data from the physical monitoring system is collected by a data acquisition module and transmitted to the mechanism model and data processing module. After correction by a parameter preprocessing module, the data is used as known quantities and transmitted to a relative position calculation module for calculation. The calculation results are dynamically linked and mapped in real-time to a pose estimation system. A pose correction module iterates on the results of the pose estimation system, constructing a knowledge- and data-driven reconstruction method. Finally, the pose information of the hydraulic support is displayed in a human-computer interaction system, and the accuracy evaluation system assesses the error of the estimated pose information. The human-computer interaction system is used to monitor the status of all systems.

[0013] Furthermore, sensors, including a laser rangefinder, an infrared transmitter, and an infrared receiver matrix, are installed on the physical prototype of the hydraulic support according to the sensor position planning system. The infrared receiver matrix consists of several infrared receivers. The laser beam of the laser rangefinder on the reference support is perpendicular to the pose plane, and the infrared transmitter on the movable support is perpendicular to its pose plane. The data acquisition module is used to acquire the distance information measured by the laser rangefinder and to acquire the high-level signal of the infrared transmitter on the infrared receiver matrix.

[0014] Furthermore, in the physical monitoring system, the origin of the infrared receiving matrix coincides with the origin of the coordinate system of the reference support. Two laser ranging sensors are located on both sides of the infrared receiving matrix and are symmetrical about the origin of the infrared receiving matrix. Their laser emission points are on a straight line with the origin of the infrared receiving matrix. The infrared transmitter coincides with the origin of the coordinate system of the movable support. The position of the reflector on the movable support is the same as the position of the corresponding laser ranging sensor on the reference support.

[0015] Furthermore, the sensor position planning system includes determining the optimal position of the laser rangefinder sensor and the size of the infrared receiver matrix. The Raycast class in Unity3D is used to create a ray, with parameters including the ray's origin, detection distance, and range. These parameters are adjusted according to the performance of the actual laser rangefinder sensor to ensure the virtual sensor achieves the same performance. The RaycastHit class is used to record the interaction information between the virtual infrared transmitter and receiver, with parameters including whether to launch a collision, the location of the collision point, and the distance between the launch point and the collision point. These parameters are also adjusted according to the sensor's performance. The attitude of the movable support is calculated analytically. A collaborative linkage program is written in C#, and relevant sensors are configured in the pose planes of the base support and the movable support, simulating the operation of the hydraulic support during pushing. The geometric dimensions of the infrared receiver matrix are determined by the landing point information on the infrared receiver matrix. The monitoring quality of the laser rangefinder sensor is judged. If the monitoring quality is poor, the position of the laser rangefinder sensor is optimized until the monitoring quality is excellent. The position of the laser sensor and the geometric dimensions of the infrared receiver matrix at this point are recorded in an XML file, providing a reference for the installation of sensors in the physical monitoring system.

[0016] Furthermore, the pose estimation system is a system based on a data-driven hydraulic support reconstruction method. The relative position calculation module, as input, interacts bidirectionally with the pose estimation system through a dynamic link library. By continuously adjusting the position of the virtual hydraulic support, the pose of the hydraulic support is finally obtained. Specifically:

[0017] ① Obtain the measurement values ​​of the laser rangefinder sensor on the reference hydraulic support after processing by the parameter preprocessing module. and and the coordinates in the infrared receiver matrix And obtain the pitch and yaw angles of the hydraulic support base.

[0018] ② Its knowledge-driven hydraulic support reconstruction method is as follows: The relative pose calculation module is used for calculation, based on the measurement values ​​of the laser rangefinder sensor on the reference hydraulic support. and Calculate the yaw angle of the movable hydraulic support. and the center distance between the two hydraulic supports ;

[0019] The specific calculation process is as follows:

[0020]

[0021]

[0022]

[0023] in The yaw angle of the movable support. The distance between the two laser rangefinders. , The distance between the laser rangefinder and the corresponding reflector is 0;

[0024] The attitude of the virtual movable hydraulic support is adjusted based on the measured pitch and yaw angles and the calculated yaw angle. The position of the virtual movable hydraulic support is at a distance of [missing information] along the X-axis. The distances in the Y and Z axes are deduced as follows;

[0025] ③ Its data-driven hydraulic support reconstruction method is as follows: based on the coordinates in the virtual infrared receiver matrix Coordinates in the infrared receiver matrix In comparison, the deviations of the virtual movable hydraulic support in the Y-axis and Z-axis directions are obtained, and the position of the virtual movable hydraulic support is adjusted in the virtual space.

[0026] The deviation is determined using the pose correction module, and the specific calculation process is as follows:

[0027]

[0028]

[0029] in This represents the deviation of the virtual movable hydraulic support on the Y-axis. A positive value indicates that the adjustment direction is opposite to the Y-axis, and vice versa. This represents the deviation of the virtual movable hydraulic support on the Z-axis. A positive value indicates that the adjustment direction is opposite to the Z-axis, while a negative value indicates that the adjustment direction is the same as the Z-axis.

[0030] ④ The knowledge- and data-driven hydraulic support reconstruction method is as follows: based on the measurement values ​​of the virtual laser rangefinder sensor... Compared with the actual measurement value of the laser rangefinder sensor In comparison, this is used to determine whether the hydraulic support has been reconfigured.

[0031] The difference is:

[0032]

[0033] if A value of 0 indicates that the hydraulic support reconfiguration is complete.

[0034] if If the result is not 0, repeat steps ② and ③.

[0035] Furthermore, the accuracy evaluation system is used to evaluate the simulation results of the relative pose simulation system. The positional deviation of the movable support can be divided into relative positional offsets in three directions, which can be combined to represent the offset of the center distance. Therefore, the relative pose error generated by the virtual simulation can be represented by the center distance error and the attitude angle error. By assigning weights to the two errors, the error of the relative pose virtual reconstruction is determined as follows:

[0036]

[0037]

[0038] in, This is the relative error of the center distance. The distance between the center of the movable support and the reference support after the simulation is completed. The center distance is the actual distance measured from the support frame.

[0039]

[0040] These are the relative errors of pitch angle, roll angle, and yaw angle, respectively. These are the pitch angle, roll angle, and yaw angle of the movable support after the simulation. The pitch angle, roll angle, and yaw angle are measured for the actual support structure.

[0041] The reconstruction accuracy is: .

[0042] Furthermore, the digital twin model system includes digital twins of adjacent hydraulic supports, a method for describing pose deviations between adjacent hydraulic supports, and a relative pose reconstruction model;

[0043] The model dimensions of the digital twins of adjacent hydraulic supports were obtained from the assembly drawings of the hydraulic supports selected for testing;

[0044] The method for describing the pose deviation between adjacent hydraulic supports is to construct a local coordinate system for the hydraulic supports and divide the pose deviation between adjacent hydraulic supports into relative position deviation and relative attitude deviation. The relative pose reconstruction model uses a nine-axis sensor to obtain the pitch angle and roll angle, and uses a combination of a laser rangefinder and an infrared beam switch to determine the relative position deviation and yaw angle of the hydraulic supports.

[0045] Compared with the prior art, the present invention has the following advantages:

[0046] 1. Based on the posture deviation description method of hydraulic support, a relative posture reconstruction model of hydraulic support is constructed, which simplifies the reconstruction problem of relative posture of hydraulic support and improves the feasibility of practical application in downhole.

[0047] 2. This invention provides a highly efficient and low-cost method for reconstructing the posture of a hydraulic support. It uses non-contact sensors such as laser rangefinders and infrared transmitters to replace contact sensors, which is more adaptable to the complex working conditions downhole.

[0048] 3. Regarding the question of how to install sensors on hydraulic supports, it is necessary to consider factors such as the support space of the hydraulic supports in actual operation, the installation position of the sensors, whether the sensors are obstructed during operation, and whether they are easily damaged by falling coal. Virtual space is used to simulate these factors to obtain the optimal position of the sensors, and a sensor position planning system is constructed to guide the actual installation of sensors, saving time and money.

[0049] 4. To address the problem of limited downhole sensing equipment and the inability to obtain all parameters for solving the coordinates of monitoring points, a method is proposed to use partial information obtained from laser rangefinders and infrared receiver matrices to perform pose deduction in a Unity3D virtual environment through a data and knowledge hybrid driving approach, thereby obtaining the pose of adjacent hydraulic supports.

[0050] 5. By defining reconstruction accuracy, the effect of virtual reconstruction is comprehensively evaluated, an accuracy evaluation system is constructed, and a theoretical model for obtaining the pose information of the entire hydraulic support group is provided by constructing relative pose, and a reference is provided for pose monitoring and support moving operations in the actual fully mechanized mining process. Attached Figure Description

[0051] The accompanying drawings, which are provided to further illustrate the invention and form part of this application, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention.

[0052] Figure 1 This is a schematic diagram illustrating the establishment of the local coordinate system in this invention;

[0053] Figure 2 This is a schematic diagram of the monitoring plane of the present invention;

[0054] Figure 3 This is a schematic diagram of the pose of the present invention.

[0055] Figure 4 This is a schematic diagram of the infrared receiving matrix calibration of the present invention;

[0056] Figure 5This is a schematic diagram illustrating the pose calculation principle of the present invention;

[0057] Figure 6 This is a schematic diagram of the system of the present invention.

[0058] In the diagram, 1-base support, 2-movable support, 3-laser rangefinder, 4-infrared transmitter, 5-infrared receiver matrix, 6-nine-axis sensor. Detailed Implementation

[0059] To enable those skilled in the art to better understand the present invention, the present invention will be further described clearly and completely below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0060] A typical embodiment of the present invention provides a knowledge- and data-driven virtual reconstruction method for the relative pose of hydraulic supports, thereby obtaining the support posture and relative position of the hydraulic supports, ensuring stable support of the working face roof, providing monitoring for the automated process of real downhole hydraulic support groups, and providing effective reference for the adjustment and control of hydraulic supports.

[0061] The knowledge- and data-driven virtual reconstruction method for the relative pose of hydraulic supports provided in this embodiment includes: a physical monitoring system, a virtual pose reconstruction system, and a mechanism model and data processing system.

[0062] The physical monitoring system includes sensors and a data acquisition module.

[0063] The aforementioned mechanism model and data processing system belong to a knowledge-driven reconstruction method. The mechanism model and data processing system include a parameter preprocessing module, a relative position calculation module, and a pose correction module.

[0064] The virtual pose reconstruction system includes a digital twin model system, a sensor position planning system, a pose inference system, a human-computer interaction system, and an accuracy evaluation system; the pose inference system belongs to a data-driven reconstruction method.

[0065] This implementation constructs a digital twin model system of a physical prototype of a hydraulic support, and uses a sensor position planning system to guide the placement of sensors in the physical monitoring system. Sensor data from the physical monitoring system is collected by a data acquisition module and transmitted to the mechanism model and data processing module. After being corrected by a parameter preprocessing module, the data is used as known quantities and transmitted to the relative position calculation module for calculation. The calculation results are dynamically mapped onto the pose inference system in real time. The pose correction module iterates on the results of the pose inference system, constructing a knowledge- and data-driven reconstruction method. Finally, the pose information of the hydraulic support is displayed in the human-computer interaction system, and the accuracy evaluation system assesses the error of the inferred pose information of the hydraulic support. The human-computer interaction system is used to monitor the status of all systems.

[0066] The digital twin model of the hydraulic support has the same physical model and properties as the physical hydraulic support. By defining the hydraulic support's pose plane and monitoring points, the relative pose reconstruction of the hydraulic support is transformed into the reconstruction of its pose plane. The sensor position planning system is used to determine the optimal installation position of the sensor, providing a reference for the position installation of the sensor in the physical monitoring system.

[0067] Physical monitoring system

[0068] Non-contact sensors are used, and sensors are installed on the physical prototype of the hydraulic support according to the sensor position planning system. The sensors include a laser rangefinder 3, an infrared transmitter 4, and an infrared receiving matrix 5. The infrared receiving matrix 5 is composed of several infrared receivers, and its external dimensions are determined by the sensor position planning system to ensure that the laser beam of the laser rangefinder 3 on the reference support 1 is perpendicular to the pose plane, and that the infrared transmitter 4 on the movable support 2 is perpendicular to its pose plane. The data acquisition module is used to collect the distance information measured by the laser rangefinder 3 and the high-level signal of the infrared transmitter 4 on the infrared receiving matrix.

[0069] The specific requirements for sensor configuration are as follows: the origin of the infrared receiver matrix coincides with the origin of the coordinate system of the reference bracket 1; the two laser rangefinders 3 are located on both sides of the infrared receiver matrix and are symmetrical about the origin of the infrared receiver matrix 5; their laser emission points are on a straight line with the origin of the infrared receiver matrix; the infrared transmitter 4 coincides with the origin of the coordinate system of the movable bracket 2; and the position of the reflector on the movable bracket 2 is the same as the position of the corresponding laser rangefinder 3 on the reference bracket 1.

[0070] Virtual Pose Reconstruction System

[0071] In the digital twin model system, the model dimensions of adjacent hydraulic support digital twins are obtained from the hydraulic support assembly drawing selected for testing. The hydraulic support model is completed using 3D modeling software and imported into Unity 3D. Parent-child relationships are configured between the components of the hydraulic support, and rigid body components and collision body components are assigned to complete the basic configuration of the virtual hydraulic support.

[0072] The method for describing the pose deviation between adjacent hydraulic supports constructs a local coordinate system for the hydraulic supports, dividing the pose deviation between adjacent hydraulic supports into relative position deviation and relative attitude deviation. In the local coordinate system, reference support 1 is the left support, and its three-dimensional coordinate origin coincides with the center of the base of reference support 1. Figure 1 As shown, a left-handed coordinate system is used, where The X-axis points to the adjacent support, the Y-axis points to the direction of the moving support, and the Z-axis direction is determined according to the left-hand rule. The right side of the reference support 1 is the movable support 2, which has various positional deviations relative to the reference support 1. The relative positional deviation of the hydraulic support refers to the distance deviation between two adjacent hydraulic supports along the coordinate axis. The relative attitude deviation of the hydraulic support refers to the attitude angle deviation of the tested support relative to the reference support 1. The attitude angle refers to the angle of rotation around the X-axis as the pitch angle, the angle of rotation around the Z-axis as the yaw angle, and the angle of rotation around the Y-axis as the roll angle. The sign of the attitude angle is determined according to the left-hand rule.

[0073] The relative pose reconstruction model utilizes a nine-axis sensor 6 to obtain pitch and roll angles, and a combination of a laser rangefinder sensor 3 and an infrared beam switch to determine the relative position deviation and yaw angle of the hydraulic support; the plane of symmetry of the hydraulic support is used as the monitoring plane, such as... Figure 2 As shown, considering the actual support space of the hydraulic support, the installation of the laser sensor, and ensuring its stable performance, △PQN is selected as the pose plane of the hydraulic support, as follows. Figure 3 As shown, boundary point P is the intersection of the connecting pin between the column and the top beam and the monitoring plane; boundary point Q is the intersection of the mounting pin of the hydraulic support base pushing cylinder and the monitoring plane; boundary point N is the intersection of the pin between the hydraulic support base and the rear connecting rod and the monitoring plane. A laser rangefinder 3 and an infrared receiver matrix are installed in the pose plane of the reference support, and an infrared transmitter 4 and a reflector are installed in the movable support 2.

[0074] The sensor position planning system includes determining the optimal position of the laser rangefinder sensor 3 and the geometric dimensions of the infrared receiver matrix. A ray is created using the Raycast class in Unity3D, with parameters including the ray's origin, detection range, and detection distance. These parameters are adjusted according to the performance of the actual laser rangefinder sensor 3 to ensure the virtual sensor achieves the same performance. The RaycastHit class is used to record the interaction information between the virtual infrared transmitter 4 and the virtual infrared receiver, with parameters including whether to emit a collision, the location of the collision point, and the distance between the emission point and the collision point. These parameters are adjusted according to the sensor's performance. The parameters are corrected; the attitude of the movable support 2 is calculated analytically, a collaborative linkage program is written in C#, and relevant sensors are configured in the pose plane of the reference support 1 and the movable support 2 to simulate the operation of the hydraulic support pushing. The geometric dimensions of the infrared receiving matrix 5 are determined by the landing point information on the infrared receiving matrix 5. The monitoring quality of the laser ranging sensor 3 is judged. If the monitoring quality is poor, the position of the laser ranging sensor 3 is optimized until the monitoring quality is excellent. The position of the laser sensor and the geometric dimensions of the infrared receiving matrix 5 at this time are recorded in an XML file to provide a reference for the installation of sensors in the physical monitoring system.

[0075] The attitude analysis mainly involves the analysis of the four-bar linkage of the hydraulic support, the coordination between the four-bar linkage and the top beam, and the coordination between the four-bar linkage and the column. The method for determining the geometric dimensions of the infrared receiving matrix 5 is as follows: the landing points on the infrared receiving matrix 5 are visualized using Gizmos in Unity3D, where the coordinates of the landing point farthest from the origin of the hydraulic support are... The optimal envelope pattern is determined based on the distribution of other landing points. If the optimal envelope pattern is a circle, then the geometry of the infrared receiving matrix 5 is a circle, with a radius of: If the optimal envelope pattern is rectangular, then the geometry of the infrared receiver matrix 5 is rectangular, with a length of [missing information]. , width is The geometric center is the origin of the hydraulic support. Let M be the physical model dimensions of the infrared receiver; the monitoring quality judgment method is as follows: during the hydraulic support pushing process, the set M is the total number of points where the laser ranging sensor 3 lands on the pose plane of the movable support 2, and the set N is the area of ​​the reflector on the movable support 2. If M... N indicates that the monitoring quality is excellent; otherwise, the monitoring quality is poor. Sensor position optimization is performed under the premise that the monitoring quality is poor. The optimization method is to traverse each position in the monitoring plane, with the initial position being the origin of the hydraulic support. The order is to first traverse the Y direction, then the Z direction, until the monitoring quality is excellent.

[0076] The human-computer interaction system utilizes the powerful interactive capabilities of Unity3D, enabling multiple jobs and systems to work together. Furthermore, Unity3D's built-in UGUI component can visualize data, reserving positions for the reconstructed information display results of the movable support 2, the data panel of the physical monitoring system, the pose correction module's calculation panel, and the reconstruction accuracy display panel. The calculation results of each system are connected to the reserved positions on the panel, providing real-time monitoring and facilitating fault diagnosis.

[0077] Mechanism Model and Data Processing System

[0078] The parameter preprocessing module processes the distance information measured by the laser rangefinder 3 and the landing point information of the infrared transmitter 4. The processed data is then fed into the relative position calculation module for calculation. The results are integrated into the attitude inference system via a dynamic link library to drive the movable support 2 to reconstruct its attitude in virtual space. Simultaneously, key information from the reconstruction results is transmitted to the attitude correction module via the dynamic link library for multiple corrections, achieving a hybrid knowledge and data-driven attitude reconstruction of the hydraulic support. The human-computer interaction system provides interfaces for each information support and displays the working status of each system in real time. The accuracy evaluation system decomposes the combination of the six degrees of freedom attitude deviations of the movable support 2 into deviations in the central moment and angular deviations in rotation around the three coordinate axes, thereby describing the accuracy of the reconstruction.

[0079] More specifically, the parameter preprocessing module filters the distance information from the laser rangefinder 3 and the angle information from the nine-axis sensor 6, calibrates the landing point of the infrared transmitter 4 on the infrared receiver matrix 5, and transmits the processed data as known quantities to the relative position calculation module.

[0080] The landing point calibration of the infrared transmitter 4, taking the infrared receiver matrix 5 as a rectangular shape as an example, is specifically as follows: Figure 4 As shown, its origin is the geometric center of the infrared receiving matrix 5. Each infrared receiving end is marked, and the center of each infrared receiving matrix end is taken as its coordinate point. Utilizing the photoelectric effect, the landing point of the infrared transmitter 4 at the infrared receiving matrix end will cause a change in the photosensitive transistor of the infrared receiver, thereby outputting a high level. The coordinate position of the infrared receiver that outputs a high level is the landing point of the infrared transmitter 4.

[0081] A knowledge- and data-driven virtual pose reconstruction method for hydraulic supports is proposed. The pose inference system is a system based on a data-driven hydraulic support reconstruction method. The relative position calculation module, as input, interacts bidirectionally with the pose inference system via a dynamic link library. By continuously adjusting the position of the virtual hydraulic support, the pose of the hydraulic support is ultimately obtained. Specifically:

[0082] ① Obtain the measurement values ​​of the laser rangefinder 3 on the reference hydraulic support after processing by the parameter preprocessing module. and and the coordinates in the infrared receiver matrix And obtain the pitch and yaw angles of the hydraulic support base.

[0083] ② Its knowledge-driven hydraulic support reconstruction method is as follows: Figure 5 As shown, the relative pose calculation module is used for calculation, based on the measurement values ​​of the laser rangefinder sensor on the reference hydraulic support. and Calculate the yaw angle of the movable hydraulic support. and the center distance between the two hydraulic supports ;

[0084] The specific calculation process is as follows:

[0085]

[0086]

[0087]

[0088] in The yaw angle of the movable support 2. The distance between the two laser rangefinders 3. , The distance between the laser rangefinder 3 and the corresponding reflector is 0, and the initial value of D is 0.

[0089] The attitude of the virtual movable hydraulic support is adjusted based on the measured pitch and yaw angles and the calculated yaw angle. The position of the virtual movable hydraulic support is at a distance of [missing information] along the X-axis. The distances in the Y and Z axes are deduced as follows.

[0090] ③ Its data-driven hydraulic support reconstruction method is as follows: based on the coordinates in the virtual infrared receiver matrix Coordinates in the infrared receiver matrix In comparison, the deviations of the virtual movable hydraulic support in the Y-axis and Z-axis directions are obtained, and the position of the virtual movable hydraulic support is adjusted in the virtual space.

[0091] The deviation is determined using the pose correction module, and the specific calculation process is as follows:

[0092]

[0093]

[0094] in This represents the deviation of the virtual movable hydraulic support on the Y-axis. A positive value indicates that the adjustment direction is opposite to the Y-axis, and vice versa. This represents the deviation of the virtual movable hydraulic support on the Z-axis. A positive value indicates that the adjustment direction is opposite to the Z-axis, while a negative value indicates that the adjustment direction is the same as the Z-axis.

[0095] ④ The knowledge- and data-driven hydraulic support reconstruction method is as follows: based on the measured values ​​of the virtual laser rangefinder sensor 3 The measured value of the actual laser rangefinder sensor 3 In comparison, this is used to determine whether the hydraulic support has been reconfigured.

[0096] The difference is:

[0097]

[0098] if A value of 0 indicates that the hydraulic support reconfiguration is complete.

[0099] if If the result is not 0, repeat steps ② and ③.

[0100] The accuracy evaluation system is used to evaluate the simulation results of the relative pose simulation system. The position deviation of the movable support 2 can be divided into relative position offsets in three directions, which can be combined to represent the offset of the center distance. Therefore, the relative pose error generated by the virtual simulation can be represented by the center distance error and the attitude angle error. By assigning weights to the two errors, the error of the relative pose virtual reconstruction is determined as follows:

[0101]

[0102]

[0103] in, This is the relative error of the center distance. The distance between the center of movable support 2 and reference support 1 after the simulation is completed. The center distance is the actual distance measured from the support frame.

[0104]

[0105] These are the relative errors of pitch angle, roll angle, and yaw angle, respectively. These are the pitch angle, roll angle, and yaw angle of the movable support 2 after the simulation. The pitch angle, roll angle, and yaw angle are measured for the actual support.

[0106] The reconstruction accuracy is: .

[0107] The scope of protection claimed by this invention is not limited to the specific embodiments described above. For those skilled in the art, this invention can have various modifications and alterations. Any modifications, improvements, and equivalent substitutions made within the concept and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A knowledge and data hybrid driven hydraulic support relative pose virtual reconstruction method, characterized in that, The physical monitoring system, the mechanism model and data processing system and the virtual pose reconstruction system are comprised; The physical monitoring system comprises sensors and a data acquisition module; the mechanism model and data processing system comprises a parameter preprocessing module, a relative position solving module and a pose correction module; the virtual pose reconstruction system comprises a digital twin model system, a sensor position planning system, a pose deduction system, a human-computer interaction system and an accuracy evaluation system; The digital twin model system of the physical prototype of the hydraulic support is constructed, and the sensor position planning system is used to guide the arrangement of the sensors in the physical monitoring system; the data of the sensors in the physical monitoring system is collected by the data acquisition module and transmitted to the mechanism model and data processing system, and the data is corrected by the parameter preprocessing module and then transmitted to the relative position solving module as known quantities for solving, and the solving results are mapped in the pose deduction system in real time through dynamic linking; The pose correction module iterates the results of the pose deduction system, and a reconstruction method driven by knowledge and data is constructed; finally, the pose information of the hydraulic support is displayed in the human-computer interaction system, the error of the deduced pose information of the hydraulic support is evaluated in the accuracy evaluation system, and the state of all systems is monitored by the human-computer interaction system; According to the sensor position planning system, sensors are installed on the physical prototype of the hydraulic support, including a laser ranging sensor, an infrared emitter and an infrared receiving matrix; the infrared receiving matrix is composed of a plurality of infrared receivers, the laser beam of the laser ranging sensor on the reference support is perpendicular to the pose plane, and the infrared emitter on the movable support is perpendicular to the pose plane; the data acquisition module is used to collect the distance information measured by the laser ranging sensor and collect the high-level signal of the infrared emitter on the infrared receiving matrix; In the physical monitoring system, the origin of the infrared receiver matrix coincides with the origin of the coordinate system of the reference support, the two laser ranging sensors are on the two sides of the infrared receiving matrix and are symmetrical about the origin of the infrared receiving matrix, the laser emitting point coincides with the origin of the infrared receiving matrix, the infrared emitter coincides with the coordinate origin of the movable support, and the position of the reflector on the movable support is the same as that of the corresponding laser ranging sensor on the reference support; The sensor position planning system includes determination of the optimal position of the laser ranging sensor and determination of the size of the infrared receiver matrix, a ray is created in Raycast class in Unity3D, the parameters of which include the starting point of the ray, the monitoring distance of the ray and the monitoring distance of the ray, the parameters of which are corrected according to the performance of the actual laser ranging sensor to make the virtual sensor achieve the same performance, RaycastHit class is used to record the interaction information of the virtual infrared transmitter and the virtual infrared receiver, the parameters of which include whether the collision is emitted, the position of the collision point and the distance between the emission point and the collision point, and the parameters are corrected according to the performance of the sensor; the analytical method is used to solve the attitude of the movable support, the C# is used to write the cooperative linkage program, the related sensors are configured in the pose plane of the reference support and the movable support, and the work of the hydraulic support when the hydraulic support is pushed is simulated, the landing point information on the infrared receiving matrix is used to determine the geometric size of the infrared receiving matrix, the monitoring quality of the laser ranging sensor is judged, if the monitoring quality is poor, the position of the laser ranging sensor is optimized, until the monitoring quality is good, the position of the laser sensor and the geometric size of the infrared receiving matrix at this time are recorded in the XML file, and the installation of the sensor in the physical monitoring system is provided with reference; The pose deduction system is a system composed of a data-driven hydraulic support reconstruction method, a relative position calculation module is used as input and interacts with the pose deduction system through dynamic link library, the position of the virtual hydraulic support is continuously adjusted to finally obtain the pose of the hydraulic support, which is specifically: ① Obtain the measurement value of the laser ranging sensor on the reference hydraulic support after the parameter preprocessing module is processed d 1 and d 2 and the coordinates in the infrared receiving end matrix y 0 , z 0), and obtain the pitch angle and yaw angle of the hydraulic support base; The knowledge-driven hydraulic support reconstruction method is: solving by using a relative position and posture solving module, calculating the yaw angle γ of the movable hydraulic support and the center distance 3 between the two hydraulic supports according to the measurement values of the laser ranging sensors on the reference hydraulic support d 1 and d 2 d 3; The specific calculation process is as follows: ； where γ is the yaw angle of the active support, l is the distance between the two laser distance sensors, d 1、 d 2 is the distance between the laser distance sensor and the corresponding reflector, D The initial value of is 0; The posture of the virtual mobile hydraulic support is adjusted according to the measured pitch angle and yaw angle and the calculated yaw angle adjustment, and the distance of the position of the virtual mobile hydraulic support in the X-axis direction is d The distances in the Y-axis and Z-axis directions are derived according to the following manner; iii) the data-driven hydraulic support reconstruction method is: according to the coordinates in the virtual infrared receiving end matrix (x, y, z) y 1 , z 1) compared with the coordinates in the infrared receiving end matrix (x, y, z) y 0 , z 0), and then the deviation of the virtual movable hydraulic support in the Y and Z directions is obtained, and the position of the virtual movable hydraulic support in the virtual space is adjusted; The pose correction module is used to determine the deviation, and the specific calculation process is as follows: ； Wherein Δy is the deviation of the virtual movable hydraulic support in Y axis, the value is positive, indicating that the adjustment direction is opposite to Y axis, otherwise, the adjustment direction is the same as Y axis, Δz is the deviation of the virtual movable hydraulic support in Z axis, the value is positive, indicating that the adjustment direction is opposite to Z axis, otherwise, the adjustment direction is the same as Z axis; (4) The hydraulic support reconstruction method driven by the mixed knowledge and data is: comparing the measurement value of the virtual laser ranging sensor D 1 with the measurement value of the actual laser ranging sensor d 1 to determine whether the hydraulic support is reconstructed. The difference is: ； If D 0, indicating that the hydraulic support reconstruction is completed; If D is not 0, repeat steps ②③.

2. The method of claim 1, wherein: The accuracy evaluation system is used to evaluate the deduction result of the relative pose deduction system, the position deviation of the movable support can be combined to represent the center distance deviation in three directions, therefore, the relative pose error generated by virtual deduction can be represented by center distance error and attitude angle error, the two kinds of errors are given weight, and the error of relative pose virtual reconstruction is determined comprehensively: ； wherein, W d is the center distance relative error, D t is the center distance between the active support and the reference support after the derivation is completed, D c is the center distance measured by the actual support; ； W α , W β , W γ pitch, roll, yaw relative errors, respectively, α t , β t , γ t pitch, roll, yaw of the active support after the deduction, respectively, α c , β c , γ c pitch, roll, yaw of the active support after the deduction, respectively, The reconstruction accuracy is: S = 100% - (1 - 0.9999999999999999) = 0.0000000000000001 W .

3. The method of claim 2, wherein: The digital twin model system comprises adjacent hydraulic support digital twins, an adjacent hydraulic support pose deviation description method and a relative pose reconstruction model; The model size of the adjacent hydraulic support digital twin is obtained from the hydraulic support assembly drawing selected by the test; The adjacent hydraulic support pose deviation description method is to divide the pose deviation between adjacent hydraulic supports into relative position deviation and relative attitude deviation by constructing a local coordinate system of the hydraulic support; The relative pose reconstruction model uses a nine-axis sensor to obtain pitch angle and roll angle, and uses a combination of laser ranging sensors and infrared reflection switches to determine the relative position deviation and yaw angle of the hydraulic support.

Images