An automatic alignment method and system for an upper bolt device based on a lifting robot

Abstract

The present invention relates to the technical field of bolt automation, and discloses an automatic alignment method and system for an upper bolt device based on a lifting robot arm, including: configuring a vision sensor on the lifting robot arm of the upper bolt device to collect vision information; performing preliminary recognition on the vision information; according to the result of the preliminary recognition, performing vision positioning on a predetermined area for bolt installation, and determining the installation angle of the bolt according to the positioning result; aligning the bolt according to the positioning result and the installation angle; judging the bolt state in real time, and if the alignment requirement is not met, performing a re-alignment operation. The automatic operation reduces manual intervention and improves construction safety and consistency. Overall, the present invention greatly improves the quality and reliability of bolt installation and is applicable to various engineering scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of bolt automation, and specifically to an automatic alignment method and system for upper bolt equipment based on a lifting robot. Background Art

[0002] In modern engineering construction, bolts, as an important support structure, are widely used in fields such as tunnels, mines, and slope reinforcement. However, traditional bolt installation methods usually rely on manual alignment, which has problems such as insufficient accuracy, low efficiency, and poor safety. Especially in complex geological conditions, it is difficult to precisely control the angle and position of bolt installation, often resulting in misalignment, angle errors, and even installation failures. This not only affects the construction quality but also increases the subsequent maintenance cost. In response to the above problems, with the development of artificial intelligence and automation technologies, the automatic alignment technology of bolts combining a lifting robot and visual recognition has become a new solution. Summary of the Invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the technical problem solved by the present invention is that the existing bolt methods cannot achieve precise automation, and have poor accuracy in positioning and installation, etc.

[0005] To solve the above technical problem, the present invention provides the following technical solution: An automatic alignment method for upper bolt equipment based on a lifting robot, comprising:

[0006] Configuring a visual sensor on the lifting robot of the upper bolt equipment to collect visual information;

[0007] Performing preliminary recognition on the visual information;

[0008] According to the result of the preliminary recognition, performing visual positioning on the predetermined area for bolt installation, and determining the installation angle of the bolt according to the positioning result;

[0009] According to the positioning result and the installation angle, aligning the bolt;

[0010] Judging the bolt state in real time, and if the alignment requirement is not met, performing a re-alignment operation;

[0011] The alignment includes, based on the positioning result, determining the position of the bolt according to the installation angle; after determining the position of the bolt, performing the action of anchoring.

[0012] As a preferred solution of the automatic alignment method of the upper anchor rod device based on the lifting robot of the present invention, wherein: the visual information includes, in the site, delimiting the actual area where the anchor rod needs to be installed as the predetermined area, and making markings in the predetermined area; recognizing whether there are markings in the image through the real-time image information collected by the visual sensor;

[0013] When the predetermined area completely enters the image frame, record the acquisition result.

[0014] As a preferred solution of the automatic alignment method of the upper anchor rod device based on the lifting robot of the present invention, wherein: the preliminary recognition includes, in the sampling result, screening the image and recognizing the screened image;

[0015] Among them, the screening process includes analyzing the clarity, screening out the images with clarity reaching the preset conditions and the proportion of the predetermined area in the picture being higher than τ; performing secondary screening on the screened images, calculating the actual distance A between the center of the predetermined area and the center of the picture, and selecting the image with the smallest A; if the result of the secondary screening is multiple, then select the image with the highest clarity among the multiple results of the secondary screening as the final screening result;

[0016] Among them, represents the preset proportional threshold; , a represents the distance between the center of the predetermined area and the center of the picture in the image, represents the actual horizontal width of the predetermined area, represents the horizontal width of the predetermined area in the figure.

[0017] As a preferred solution of the automatic alignment method of the upper anchor rod device based on the lifting robot of the present invention, wherein: the recognition of the screened image includes using the Canny algorithm to extract the edges in the image, using morphological operations to connect the broken edges to form a closed area;

[0018] Performing recognition based on the convolutional neural network architecture;

[0019] Input: the result of the preliminary recognition;

[0020] Output: the coordinate result of visual positioning;

[0021] The convolutional neural network architecture includes a feature extraction layer, a region division layer, and a coordinate positioning layer;

[0022] Extract the features of each closed area through the feature extraction layer, and obtain a feature map as the input to the area division layer; the area division layer identifies the types and distribution densities of features in the feature map and divides the feature areas; the coordinate positioning layer selects areas in the map with divided feature areas and locates at the centers of the selected areas;

[0023] Among them, the feature extraction layer, the area division layer, and the coordinate positioning layer each have independent functions and can be trained separately. Finally, recognition is completed through cooperation;

[0024] After the recognition is completed, map the coordinate results to the positioning positions in the actual site, continuously monitor the alignment behavior of the bolt installation, map the positioning positions in the actual site to the real-time monitoring screen, and adjust until the bolt installation is completed.

[0025] As a preferred scheme of the automatic alignment method for the upper bolt device based on the lifting robot arm of the present invention, wherein: the installation angle includes matching the positioning result in the design drawing, and obtaining the angle between the bolt and the anchoring plane in the drawing as Angle 1;

[0026] Analyze the anchoring geology and adjust Angle 1 to obtain the actual installation angle.

[0027] As a preferred scheme of the automatic alignment method for the upper bolt device based on the lifting robot arm of the present invention, wherein: the analysis of the anchoring geology includes obtaining a three-dimensional virtual model of the geological structure, and simulating the penetration process of the bolt in the virtual model; assume that in the same medium, the angle of the bolt will not change during the penetration process, and at the junction of two media, the head of the bolt will shift, resulting in a change in the angle of the bolt;

[0028] Analyze the angle of the bolt end according to the situation of the angle change to obtain the adjustment angle for Angle 1;

[0029] In the three-dimensional virtual model of the geological structure, divide the edges of different media: for any two adjacent media B and C, respectively obtain the exploration signals of B and C, and at the junction of B and C, fit the positions where the exploration signal intensities of B and C are equal to obtain the boundary line of the edge of B and C, and draw all the boundary lines to obtain the division result of the edges of different media;

[0030] In the three-dimensional virtual model of the geological structure, assume that the bolt penetrates at Angle 1 and deflects when reaching any boundary line, and the deflection angle is δ; use mechanical analysis software to analyze the forces on the bolt, so as to obtain the offset angle of the bolt end compared with Angle 1;

[0031] Perform real-time analysis on the offset angle. When the offset angle exceeds the threshold, predict that the anchor rod fails and cancel the automatic alignment of the upper anchor rod device at the current position. When the depth of the anchor rod does not meet the requirements and the offset angle does not exceed the threshold, continue the prediction. When the depth of the anchor rod meets the requirements and the offset angle does not exceed the threshold, determine that the anchor rod can be completed at the current position, and record the final offset angle as the adjustment angle.

[0032] Superimpose the adjustment angle and Angle 1 to obtain the installation angle.

[0033] Among them, is the deflection angle, indicating the angle deflection that occurs when the head of the anchor rod reaches the junction line; use the regression method to fit the relationship between the deflection angle and the parameters ;

[0034] f(...) represents the function for fitting. represents the included angle between the head of the anchor rod and the junction line when the head of the anchor rod reaches the junction line, represents the parameters of the medium where the head of the anchor rod is located before reaching the junction line; represents the parameters of the medium where the head of the anchor rod is located after reaching the junction line; represents the external force applied during the penetration of the anchor rod.

[0035] As a preferred solution of the automatic alignment method of the upper anchor rod device based on the lifting robot arm according to the present invention, wherein: the real-time judgment of the state of the anchor rod includes using a vision sensor to obtain the actual angle between the anchor rod and the anchoring plane. If the actual angle exceeds the preset maximum reference value, it is judged that the alignment requirement is not met; otherwise, it is judged that the alignment requirement is met.

[0036] An automatic alignment system for an upper anchor rod device based on a lifting robot arm adopting any method as described in the present invention, characterized in that:

[0037] A collection unit, configure a vision sensor on the lifting robot arm of the upper anchor rod device to collect vision information;

[0038] An identification unit, perform preliminary identification on the vision information;

[0039] A positioning unit, perform vision positioning on the predetermined area of the anchor rod installation according to the result of the preliminary identification, and determine the installation angle of the anchor rod according to the positioning result;

[0040] An execution unit, align the anchor rod according to the positioning result and the installation angle; perform real-time judgment on the state of the anchor rod. If the alignment requirement is not met, perform a re-alignment operation.

[0041] A computer device, comprising: a memory and a processor; the memory stores a computer program, wherein: when the processor executes the computer program, the steps of the method described in any one of the present invention are implemented.

[0042] A computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by a processor, the steps of the method described in any one of the present invention are implemented.

[0043] Advantages of the present invention: The automatic alignment method of the upper bolt device based on the lifting manipulator provided by the present invention realizes the high-precision automatic alignment of the bolt by combining the lifting manipulator, visual recognition and geological model analysis. Its advantages include: improving the bolt installation accuracy and significantly reducing the installation failure rate caused by angle deviation; improving the construction efficiency through intelligent image recognition and dynamic adjustment; optimizing the installation angle of the bolt by combining with the geological virtual model to adapt to complex geological conditions; automated operation reduces manual intervention and improves construction safety and consistency. Overall, the present invention greatly improves the quality and reliability of bolt installation and is applicable to various engineering scenarios. Description of the Drawings

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0045] Figure 1 It is the overall flowchart of an automatic alignment method of an upper bolt device based on a lifting manipulator provided by the first embodiment of the present invention. Detailed Embodiments

[0046] In order to make the above objects, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0047] Embodiment 1, referring to Figure 1 , which is an embodiment of the present invention, provides an automatic alignment method of an upper bolt device based on a lifting manipulator, including:

[0048] S1: Configure a visual sensor on the lifting manipulator of the upper bolt device to collect visual information.

[0049] The visual information includes, in the site, demarcating the actual area where the anchor bolts need to be installed as the predetermined area and making markings in the predetermined area; identifying whether there are markings in the image by collecting real-time image information through the visual sensor; and recording the collection result when the predetermined area completely enters the image frame.

[0050] It should be noted that demarcating the predetermined area where the anchor bolts need to be installed and making markings in the site, and collecting and identifying images in real time through the visual sensor can quickly locate the target position for installing the anchor bolts. By ensuring that the predetermined area completely enters the frame and recording the result, the initial positioning of the installation position is realized, providing high-quality initial data support for subsequent precise identification and alignment operations. The purpose of this design is to reduce the errors in the traditional manual positioning process, optimize the installation process of the anchor bolts, and lay a solid foundation for the automated installation in complex construction environments.

[0051] S2: Conduct a preliminary identification of the visual information.

[0052] The preliminary identification includes, in the sampling result, screening the images and identifying the screened images; among them, the screening process includes analyzing the clarity, screening out the images whose clarity reaches the preset conditions and the proportion of the predetermined area in the frame is higher than τ; conducting a secondary screening of the screened images, calculating the actual distance A between the center of the predetermined area and the center of the frame, and selecting the image with the smallest A; if there are multiple results of the secondary screening, select the image with the highest clarity among the multiple results of the secondary screening as the final screening result.

[0053] Among them, represents the preset proportional threshold; , a represents the distance between the center of the predetermined area and the center of the frame in the image, represents the actual horizontal width of the predetermined area, represents the horizontal width of the predetermined area in the figure.

[0054] This preset condition is a preset value. In the design of this embodiment, the clarity is first quantified by existing means; the preset condition here is the mean value of the top 20% of the sample clarity in the training set. If the clarity does not meet the mean value of the top 20%, re-sampling is taken.

[0055] The identification of the screened images includes using the Canny algorithm to extract the edges in the image, using morphological operations to connect the broken edges to form a closed area. Based on the convolutional neural network architecture for identification:

[0056] Input: The result of the preliminary identification.

[0057] Output: The coordinate result of visual positioning.

[0058] It should be noted that the general requirements for anchor bolts are as follows: for a fractured and uneven slope, loose floating stones and rock debris must be removed, voids filled with grouted rubble masonry, and slope surface cracks sealed. After slope trimming, it should be flat and dense. That is to say, the actual position of the anchor bolt is a separately treated position in an area. Then, when installing the anchor bolt, this position needs to be found. According to the requirements of the anchor bolt, different training sets are used. If the entire predetermined area meets the requirements, simplification will be carried out in the second layer of the neural network, thereby reducing the computational amount and directly calculating and positioning the relative position of the predetermined area through the third layer. If a part of the entire predetermined area does not meet the requirements, the area can be identified through the second layer and then a positioning selection can be made. The convolutional neural network architecture includes a feature extraction layer, a region division layer, and a coordinate positioning layer. The alignment includes, based on the positioning result, determining the position of the anchor bolt according to the installation angle; after determining the position of the anchor bolt, performing the action of anchoring.

[0059] The feature extraction layer extracts the features of each closed area to obtain a feature map and inputs it into the region division layer; the region division layer identifies the types and distribution densities of the features in the feature map and divides the feature regions; the coordinate positioning layer selects regions in the map with divided feature regions and locates at the center of the selected regions.

[0060] Feature extraction layer: Extracts the key features (such as edges, textures, shapes, etc.) of each closed area to generate a feature map. Uses convolutional operations to extract edge features:

[0061] ;

[0062] where is the input image, is the convolutional kernel.

[0063] Processes the features using a non-linear activation function (such as ReLU) to improve the network's expressive ability:

[0064] ;

[0065] Uses pooling operations to reduce the size of the feature map, reduce the computational amount, and retain the main features at the same time:

[0066] ;

[0067] Outputs a low-resolution, high-semantic feature map representing the features of each closed area in the predetermined area.

[0068] Region division layer: Based on the results of the feature extraction layer, analyze the types and distribution densities of regional features, and divide the regions that meet and do not meet the requirements of the anchor bolts. For the feature map classify the regions in it to identify which regions meet the anchor bolt requirements. Use convolutional or fully connected layers to output the category C of each region:

[0069] ;

[0070] The filtered region map , including the markings of the regions that meet the anchor bolt installation requirements.

[0071] Coordinate positioning layer, perform specific anchor bolt position positioning in the divided feature regions. Use a coordinate regression model to accurately predict the position of this point: (x,y)=Regressor(Fr).

[0072] Among them, the feature extraction layer, region division layer, and coordinate positioning layer have independent functions and can be trained separately for the three layers. Finally, they complete the recognition through cooperation. After the recognition is completed, map the coordinate results to the positioning positions in the actual site, continuously monitor the alignment behavior of the anchor bolt installation, map the positioning positions in the actual site to the real-time monitoring screen, and adjust until the anchor bolt is completed.

[0073] S3: According to the preliminary recognition results, perform visual positioning on the predetermined region for anchor bolt installation, and determine the installation angle of the anchor bolt according to the positioning results.

[0074] The installation angle includes matching the positioning results in the design drawings, obtaining the angle between the anchor bolt and the anchoring plane in the drawings as Angle 1; analyzing the anchoring geology and adjusting Angle 1 to obtain the actual installation angle.

[0075] It should be noted that in the solution, before the anchor bolt, it is necessary to preprocess the region where the anchor bolt is required so that the processed region can be consistent with the anchor bolt plane in the drawings. By processing the slope surface, the angle between the actual slope surface and the horizontal position can be presented as a fixed value. However, the Angle 1 obtained from the drawings is an ideal value, and the final anchor bolt result should be within this ideal value range. Therefore, the actual installation angle can be obtained only through adjustment. Among them, it is very simple to obtain the designed angle between the anchor bolt and the slope surface on the drawings, which can be completed by angle measurement. After aligning the slope surface in the drawings, the position of the anchor bolt can be directly locked according to the alignment position. And the actual slope surface should be a flat part after being processed. Then, when the anchor bolt is anchored, the actual angle can be determined through the included angle between the anchor bolt and the actual slope surface.

[0076] When installing an anchor rod, it is necessary to conduct a preliminary geological survey and analyze the geological conditions where the anchor rod will be inserted before proceeding with the installation. The analysis of the geological conditions for anchor rod insertion includes obtaining a three-dimensional virtual model of the geological structure and simulating the insertion process of the anchor rod in the virtual model. It is assumed that in the same medium, the angle of the anchor rod does not change during the insertion process, but at the interface between two media, the head of the anchor rod will shift, resulting in a change in the angle of the anchor rod. Based on the situation of the angle change, an angle analysis is performed on the end of the anchor rod to obtain the adjustment angle for the angle 1, and data information such as deflection, vibration, and displacement of the anchor rod due to force at the interface between two media is recorded.

[0077] Among them, the "angle change" is completed by means of finite element analysis using a mechanical analysis software. The "data information such as deflection, vibration, and displacement of the anchor rod due to force at the interface between two media" is the data that needs to be observed by technicians during the analysis of the adjustment process. By simulating the anchor rod process with a mechanical analysis software, the stress condition of the anchor rod after being impacted by the external medium is analyzed, and time accumulation is carried out according to the stress condition until the anchor rod passes through the interface between media. By recording the position changes of the anchor rod before and after entering the "interface between media", the displacement data can be obtained; by comparing the angle changes of the anchor rod before and after entering the "interface between media", the deflection data can be obtained; by analyzing the stress on the head of the anchor rod and the drilling power during the installation of the anchor rod, the vibration data can be obtained.

[0078] It should be noted that the establishment of a three-dimensional virtual model of the geological structure is a key link in simulating the insertion process of the anchor rod. This model needs to accurately reflect the formation structure, medium properties, and distribution of geological interfaces to facilitate subsequent mechanical analysis and calculation of anchor rod angle adjustment. To build such a model, it is necessary to first obtain three-dimensional geological structure data, then perform modeling and data processing, and finally form a three-dimensional virtual environment for analysis. Three-dimensional geological data can be obtained through various means, mainly including geophysical exploration, borehole survey, ground-penetrating radar scanning, and remote sensing data. The following are common data collection methods:

[0079] (1)Seismic wave detection:

[0080] Principle: Seismic waves are emitted into the ground through a seismic source (such as explosives, vibrators), and a three-dimensional model of the underground geological structure is established by using the velocity difference of seismic waves propagating in different media.

[0081] Seismic reflection method or seismic refraction method is used for geological data collection. The propagation time, amplitude, and frequency of seismic waves are recorded, and the distribution of different underground geological layers is inverted.

[0082] (2)Borehole survey:

[0083] Principle: Drill multiple boreholes in different areas and use sensors to record information such as formation structure, lithological parameters, and groundwater distribution.

[0084] Data acquisition uses the following methods. Electrical logging: Measure the resistivity of rock formations to identify different media.

[0085] Acoustic logging: Measure the propagation speed of sound waves in rock formations to determine density and bedding.

[0086] Density logging: Obtain the density distribution of rock formations to infer the hardness of media.

[0087] (3) Ground Penetrating Radar (GPR):

[0088] Principle: Use high-frequency electromagnetic waves to detect changes in underground media and analyze the reflected signals to obtain shallow geological information.

[0089] Use a GPR antenna to transmit electromagnetic waves, receive the reflected signals from geological layers, and analyze the boundary positions of different media.

[0090] (4) Remote sensing technology:

[0091] Principle: Use technologies such as satellite imagery, unmanned aerial vehicle (UAV) aerial surveys, and Light Detection and Ranging (LiDAR) to infer underground geological information from surface data.

[0092] Spectral analysis: Analyze the surface reflection spectrum to identify different rock formation types. LiDAR scanning: Accurately measure the terrain undulation to infer the underground structure.

[0093] Furthermore, after obtaining geological data, it is necessary to use computer software to construct a three-dimensional virtual geological model. The main processes are as follows:

[0094] Data preprocessing:

[0095] Data cleaning: Remove noise and outliers to ensure the accuracy of seismic waves, logging, GPR, and remote sensing data. Coordinate unification: Convert different data sources (boreholes, seismic waves, radar scans, etc.) into a unified three-dimensional coordinate system (such as the UTM coordinate system).

[0096] Three-dimensional grid modeling:

[0097] Use inversion algorithms to convert geological data into a three-dimensional spatial structure. Common methods include: Seismic Tomography: Reconstruct a three-dimensional geological model based on the seismic wave velocity distribution. Kriging Interpolation: Use borehole data for spatial interpolation to construct a three-dimensional grid of underground media. Point cloud data modeling: Point cloud data obtained from LiDAR scanning or GPR can be used to construct a detailed surface model.

[0098] Geological boundary extraction, for example: Based on seismic data: Using the characteristics of seismic reflection layers to determine the interface between different media. Based on logging data: Analyzing borehole logging results, identifying lithological changes, and drawing geological cross-sections.

[0099] Three-dimensional visualization: Using three-dimensional modeling software (such as Leapfrog, GOCAD, Surpac) for geological visualization modeling: Generating a three-dimensional geological model to display the spatial distribution of different strata. Defining the properties (elastic modulus, density, Poisson's ratio, etc.) of different media in the model. Combining with mechanical analysis software (such as ANSYS, Abaqus) to simulate the stress of anchor bolts.

[0100] Furthermore, the constructed three-dimensional geological virtual model can be used for anchor bolt construction simulation:

[0101] Define the initial installation angle of the anchor bolt (angle 1) in the model and simulate the drilling process of the anchor bolt.

[0102] Set mechanical parameters: Set the friction force, support force, and resistance of each medium to simulate the stress of the anchor bolt in different geological conditions.

[0103] Simulate the process of the anchor bolt going deeper: In a single medium, the angle remains unchanged. When the anchor bolt reaches the interface between media, calculate the sudden change in stress and predict the deflection angle δ. Record the displacement, vibration, and stress data of the anchor bolt as the basis for construction adjustment.

[0104] Correct the installation angle: Calculate the final angle at the end of the anchor bolt through simulation and adjust the construction angle of the anchor bolt to ensure stability.

[0105] First, in the three-dimensional geological structure virtual model, divide the edges of different media: For any two adjacent media B and C, obtain the exploration signals of B and C respectively. At the junction of B and C, fit the positions with equal exploration signal intensities of B and C to obtain the boundary line of the edge of B and C, and draw all the boundary lines to obtain the division result of the edges of different media.

[0106] In the three-dimensional geological structure virtual model, assume that the anchor bolt goes deeper at angle 1 and deflects when reaching any boundary line, and the deflection angle is ; Use mechanical analysis software to analyze the stress on the anchor bolt to obtain the offset angle of the end of the anchor bolt compared with angle 1. It should be noted that by comparing the angle when the anchor bolt is aligned before anchoring and the angle between the position of the anchor bolt and the anchoring slope after simulation by the mechanical analysis software, and taking the difference between the two, the offset angle can be obtained.

[0107] Use a mechanical analysis software to perform real-time analysis on the offset angle. When the offset angle exceeds the threshold, it is predicted that the anchor rod fails, and the automatic alignment of the upper anchor rod device at the current position is cancelled; when the depth of the anchor rod does not meet the requirements and the offset angle does not exceed the threshold, the prediction continues; when the depth of the anchor rod meets the requirements and the offset angle does not exceed the threshold, it is determined that the anchor can be completed at the current position, and the final offset angle is recorded as the adjustment angle.

[0108] Among them, there are many professional mechanical analysis softwares that can be used to analyze the stress conditions of the rod body, especially for structural mechanics and material mechanics problems. The mechanical analysis software can use the following softwares: ANSYS, Abaqus, SolidWorks Simulation, etc. Through the finite element analysis method, the predicted results are directly output by the software.

[0109] Superimpose the adjustment angle with Angle 1 to obtain the installation angle. Among them, is the deflection angle, indicating the angle deflection that occurs when the anchor rod head reaches the boundary line; use the regression method to fit the relationship between the deflection angle and the parameters ; f(...) represents the function used for fitting, and methods such as linear regression, support vector machine (SVM), or neural network can be used. represents the included angle between the anchor rod head and the boundary line when the anchor rod head reaches the boundary line, represents the parameters of the medium where the anchor rod head is located before reaching the boundary line; represents the parameters of the medium where the anchor rod head is located after reaching the boundary line; represents the external force applied during the penetration of the anchor rod.

[0110] It should be noted that three-dimensional modeling and virtual simulation of the geological conditions are carried out to accurately identify the distribution and properties of the multi-medium junction, so as to provide data support for the dynamic adjustment of the anchor rod angle. During the penetration of the anchor rod, the anchor rod head may shift at the junction of two media. Use the three-dimensional geological virtual model to predict the deflection angle and superimpose it with the initial angle (Angle 1) to calculate the actual installation angle. Through real-time analysis, dynamically adjust the installation angle to adapt to fractured geology or uneven medium conditions, and avoid installation failure or excessive errors. Among them, "dynamically adjusting the installation angle" represents an iterative process, by continuously using the "actual installation angle" as the input and adjustment, so as to output the final installation angle. Ensure that the final installation angle is a most stable solution.

[0111] Through mechanical analysis software (such as ANSYS, Abaqus, etc.), the finite element analysis method is used to calculate the force and deflection angle of the anchor bolt in real time, ensuring that the feasibility of installation can be quickly evaluated under complex geological conditions. When the depth of the anchor bolt does not meet the requirements, continuous predictive analysis is carried out; when the deflection angle exceeds the threshold, the installation operation at the current position is terminated in time, reducing the trial-and-error time and resource waste.

[0112] Using regression methods (such as linear regression, SVM or neural networks), the relationship between the deflection angle of the anchor bolt and the medium parameters and external forces is fitted to achieve intelligent prediction of the deflection angle, providing generalization ability for the installation of anchor bolts under different site conditions. Dynamically monitor the installation process of the anchor bolt, and adjust the angle and depth parameters in real time to reduce human intervention and achieve efficient and automated anchor bolt construction.

[0113] Through the setting of the deflection angle threshold and real-time prediction, when the deviation of the anchor bolt exceeds the threshold, the alignment of the anchor bolt is terminated in time to avoid the failure of the anchor bolt installation or construction risk caused by excessive errors. Record the final adjusted angle in real time, provide reference for subsequent construction, and improve the construction safety in complex geological areas.

[0114] S4: Align the anchor bolt according to the positioning result and the installation angle.

[0115] S5: Judge the state of the anchor bolt in real time. If the alignment requirement is not met, perform the re-alignment operation.

[0116] Specifically, a vision sensor is used to obtain the actual angle between the anchor bolt and the anchoring plane. If the actual angle exceeds the preset maximum reference value, it is judged that the alignment requirement is not met; otherwise, it is judged that the alignment requirement is met.

[0117] On the other hand, this embodiment also provides an automatic alignment system for the upper anchor bolt device based on a lifting robot arm, which includes:

[0118] An acquisition unit, which configures a vision sensor on the lifting robot arm of the upper anchor bolt device to acquire vision information.

[0119] An identification unit, which performs preliminary identification on the vision information.

[0120] A positioning unit, which performs visual positioning on the predetermined area of the anchor bolt installation according to the result of the preliminary identification, and determines the installation angle of the anchor bolt according to the positioning result.

[0121] An execution unit, which aligns the anchor bolt according to the positioning result and the installation angle; judges the state of the anchor bolt in real time. If the alignment requirement is not met, perform the re-alignment operation.

[0122] Embodiment 2, an embodiment of the present invention, provides a real-time state diagnosis method for a hydraulic system of an automatic alignment method for an upper bolt device based on a lifting robot arm.

[0123] Configure multiple sensors (displacement sensors, vibration acceleration sensors, acceleration sensors, etc.) on the robot arm and the hydraulic system; collect displacement information and various acceleration information.

[0124] During the alignment installation, transmit the data information to the hydraulic system in real time; realize automatic data detection, automatic processing calculation and storage of parameters (pressure, flow rate, acceleration, vibration, displacement, etc.), which is convenient for adjusting the angle of the bolt robot arm and monitoring the working state of the hydraulic system as well as signal acquisition;

[0125] Regard the information data during the operation of the hydraulic system as a linear process for analysis, explore its data change rules, create a data set, construct a state space model based on machine learning, and train the data to diagnose in real time whether the hydraulic system is abnormal.

[0126] For the state monitoring of the hydraulic system, use the machine learning K-Means method to perform clustering analysis on the data during the operation of the hydraulic system; process the data through Fourier basis function transformation, EM algorithm and Kalman filtering method, and establish a state space model.

[0127] Collect sensor data. When the data acquisition frequency is high and the number of sampling points is large, the obtained data will show certain function characteristics in the data space. The linear data analysis method can perform unevenly spaced data sampling when obtaining observation data. For discrete data samples with different sampling rates, they can be converted into linear data by using interpolation or smoothing methods.

[0128] Functionally express discrete time series data objects based on the linear data analysis method:

[0129] ;

[0130] In the formula: is the function value of the function of the original data sequence at the kth sampling point; is the fitting error corresponding to this data point, usually referring to the disturbance factor, error or other exogenous factors in the observed data; The fitted sequence can be expressed as the function .

[0131] The general steps of linear data analysis are:

[0132] Step 1: Expand the discrete data based on the basis function smoothing method.

[0133] Step 2: Add a penalty term to smooth and flatten the fitting function.

[0134] When using the basis function smoothing method to fit discrete data, in order to avoid overfitting of the function curve, overemphasizing the local fluctuations of the data and being unable to accurately represent the main features of the original data, a roughness penalty factor is introduced into the fitting process, and a penalty term is added to control the curvature of the fitting curve, that is, to control the fitting degree of the basis function to the original discrete data, making its representation more reasonable.

[0135] Step 3: Calculate the coefficient vector using the least squares method.

[0136] The specific process of the K-means clustering algorithm:

[0137] Step 1: Use the sparrow optimization algorithm to select K clustering centroids: .

[0138] Step 2: Calculate the distance between each data point x in the data object and each centroid in the centroid set , and divide the data point x into the category with the closest centroid, as shown in Equation (1-1):

[0139] (1-1);

[0140] In the formula, i = 1, 2,..., N; j = 1, 2,..., K and ; represents the distance between the data point and the centroid, which is calculated using the Euclidean distance formula.

[0141] Step 3: Recalculate the centroid of each cluster, as shown in Equation (1-2):

[0142] (1-2);

[0143] Among them, represents the number of data points in

[0144] Step 4: Repeat Step 2 and Step 3 until the algorithm converges.

[0145] Use the CatBoost algorithm for abnormal diagnosis of the operating state of the hydraulic system.

[0146] According to the feature importance ranking of CatBoost, determine the optimal feature subset to judge the severity or generating factors of the fault. The steps are as follows:

[0147] Step 1: Import the original data such as displacement, pressure, and vibration sensors collected from the hydraulic system. Through methods such as random sampling, divide the original experimental data into a training set A and a test set B. Both the training set A and B contain different operating states of the same number of components, and perform maximum-minimum processing on all samples.

[0148] Step 2: On the training set, use the Catboost method to rank the importance of all features and determine the correlation ranking of the features.

[0149] Step 3: Screen the feature variables to ensure the recognition accuracy and determine the feature set.

[0150] Step 4: On the feature subset trained in the previous step, construct a Catboost classification model and evaluate the model's ability to accurately identify the fault type through the test set.

[0151] Step 5: According to the preliminary diagnostic effect of the model, adjust the parameters of the Catboost classification model until the expected effect is achieved.

[0152] Step 6: On the premise of meeting the iteration conditions, store the current optimal model parameters and use these parameters to analyze the test data, and finally obtain an accurate classification and diagnosis result.

[0153] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0154] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.

[0155] More specific examples (nonexhaustive list) of computer-readable media include the following: electrical connections (electronic devices) with one or more wirings, portable computer disk cartridges (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0156] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or combinations thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or combination of the following techniques well known in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits with suitable combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0157] Example 3, an embodiment of the present invention, provides an automatic alignment method for an upper bolt device based on a lifting robot arm. To verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0158] The experimental site selects a fragmented and uneven simulated slope area. The basic conditions of the slope are trimmed to form multiple potential bolt predetermined installation areas. The rock mass conditions of the slope are obtained through survey tools, and a three-dimensional geological structure virtual model is constructed for predicting angle deflection.

[0159] Test equipment and materials:

[0160] Lifting robot arm equipped with a vision sensor (model: G-RobotX2).

[0161] Self-developed three-dimensional geological model simulation platform.

[0162] Mechanical analysis software (ANSYS).

[0163] High-resolution camera (resolution: 1920×1080).

[0164] During the test process, the following steps are carried out in sequence:

[0165] Visual information acquisition and preprocessing: The lifting robotic arm is installed at a predetermined test site to acquire visual information of the set slope area. The visual sensor captures real-time images through a high-resolution camera, extracts the image edges using the Canny algorithm, and connects the broken edges through morphological operations to form multiple closed regions. After screening, the image with high clarity, a large proportion of the predetermined area, and the smallest distance between the center of the image and the center of the predetermined area in the visual information is selected as the basis for subsequent processing.

[0166] Positioning of the predetermined area and determination of the installation angle: The screened images are input into a convolutional neural network for recognition. The feature extraction layer extracts the region edges and texture features, the region division layer marks the eligible regions with higher feature density, and the coordinate positioning layer outputs the coordinates of the center points of the regions that meet the installation conditions. Subsequently, combining the initial installation angle (angle 1) obtained from the design drawings, it is matched with the three-dimensional geological virtual model, the angle deflection of the anchor rod at different depths is simulated, the adjustment angle is calculated, and finally the actual installation angle is determined.

[0167] Simulation of the anchor rod penetration process and data recording: The anchor rod penetrates gradually at the actual installation angle, and a mechanical analysis software is used to monitor its force and deflection angle in real time. The deflection angle is recorded every 1 meter of penetration. When the deflection angle does not exceed the threshold and the depth of the anchor rod reaches the requirement, the installation is determined to be successful, and the final deflection angle is recorded as the adjustment basis; if the deflection angle exceeds the threshold, the installation is terminated and the coordinates of the failure point are recorded.

[0168] Through experimental tests, the present invention can accurately complete the alignment without the need for manual participation in correction.

[0169] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for automatically aligning an upper anchor rod device based on a lifting robot, characterized in that: include: A visual sensor is configured on the lifting robot of the upper anchor bolt equipment to collect visual information; Performing preliminary recognition on the visual information; Based on the result of the preliminary identification, visually locate the predetermined area for anchor bolt installation, and determine the installation angle of the anchor bolt based on the positioning result; Align the anchor rods according to the positioning results and installation angles; The anchor bolt status is judged in real time. If the alignment requirements are not met, re-alignment is performed; The alignment includes, based on the positioning result, determining the position of the anchor rod according to the installation angle; after determining the position of the anchor rod, performing an anchoring action; The installation angle includes matching the positioning result in the design drawing, obtaining the angle between the anchor rod and the anchoring plane in the drawing as angle 1; Analyze the anchoring geology and adjust the angle 1 to obtain the actual installation angle; The analysis of anchoring geology includes obtaining a three-dimensional virtual model of geological structure and simulating the penetration process of the anchor rod in the virtual model; assuming that in the same medium, the angle of the anchor rod will not change during the penetration process, and the anchor rod head will deviate at the junction of the two media, so that the angle of the anchor rod will change; Performing angle analysis on the end of the anchor rod according to the angle change to obtain an adjustment angle for the angle 1; In the three-dimensional geological structure virtual model, the edges of different media are divided: for any two connected media B and C, the survey signals of B and C are obtained respectively, and at the connecting part of B and C, the positions where the survey signal intensities of B and C are equal are fitted to obtain the boundary line of the edges of B and C, and all the boundary lines are drawn to obtain the division results of the edges of different media; In the three-dimensional geological structure virtual model, it is assumed that the anchor rod penetrates at an angle of 1 and deflects when reaching any boundary line. The deflection angle is ; The force on the anchor rod is analyzed using mechanical analysis software to obtain the offset angle of the anchor rod end compared to angle 1; The offset angle is analyzed in real time. When the offset angle exceeds a threshold, the anchor bolt is predicted to fail, and the automatic alignment of the upper anchor bolt device at the current position is canceled; when the anchor bolt depth does not meet the requirement and the offset angle does not exceed the threshold, the prediction is continued; when the anchor bolt depth meets the requirement and the offset angle does not exceed the threshold, it is determined that the current position can complete the anchor bolt, and the final offset angle is recorded as the adjustment angle; Superimpose the adjustment angle and angle 1 to obtain the installation angle; in, is the deflection angle, which indicates the angle deflection that occurs when the anchor head reaches the junction line; Use regression method to fit the relationship between deflection angle and parameters ; f(...) represents the function used for fitting; It indicates the angle between the anchor head and the boundary line when it reaches the boundary line. It indicates the parameters of the medium where the anchor head is located before reaching the boundary line; It indicates the parameters of the medium where the anchor head is located after reaching the boundary line; Represents the external force applied during the penetration of the anchor.

2. The automatic alignment method of the upper anchor rod equipment based on the lifting robot according to claim 1 is characterized in that: The visual information includes, in the site, demarcating the actual area where the anchor rod needs to be installed as the predetermined area, and marking it in the predetermined area; Using the real-time image information collected by the visual sensor, identifying whether there is a logo in the image; When the predetermined area completely enters the image frame, the acquisition result is recorded.

3. The automatic alignment method of the upper anchor rod equipment based on the lifting robot according to claim 2 is characterized in that: The preliminary recognition includes screening images in the sampling results and identifying the screened images; The screening process includes: analyzing the clarity to screen out images whose clarity meets the preset conditions and whose proportion of the predetermined area in the picture is higher than τ; performing secondary screening on the screened images to calculate the actual distance A between the center of the predetermined area and the center of the picture, and selecting the image with the smallest A; if there are multiple results of the secondary screening, selecting the image with the highest clarity from the multiple secondary screening results as the final screening result; in, Indicates the preset ratio threshold; , a represents the distance between the center of the predetermined area and the center of the picture in the image, represents the actual horizontal width of the predetermined area, Indicates the horizontal width of the predetermined area in the figure.

4. The automatic alignment method of the upper anchor rod equipment based on the lifting robot according to claim 3 is characterized in that: The identifying of the screened image includes extracting edges in the image using a Canny algorithm, connecting broken edges using a morphological operation, and forming a closed area; Recognition based on convolutional neural network architecture; Input: the result of the preliminary identification; Output: coordinate results of visual positioning; The convolutional neural network architecture includes a feature extraction layer, a region division layer, and a coordinate positioning layer; The feature extraction layer extracts the features of each closed area, obtains a feature map and inputs it into the area division layer; the area division layer identifies the feature types and distribution density in the feature map and divides the feature area; the coordinate positioning layer selects the area in the map of the divided feature area and locates the center of the selected area; Among them, the feature extraction layer, region division layer, and coordinate positioning layer have independent functions, and the three layers can be trained separately, and finally the recognition is completed through collaboration; After the recognition is completed, the coordinate results are mapped to the positioning position of the actual site, the alignment behavior of the anchor installation is continuously monitored, and the positioning position of the actual site is mapped to the real-time monitoring screen, and adjustments are made until the anchor is completed.

5. The automatic alignment method of the upper anchor rod equipment based on the lifting robot according to claim 4 is characterized in that: The real-time determination of the state of the anchor rod includes obtaining the actual angle between the anchor rod and the anchoring plane using a visual sensor, and if the actual angle exceeds a preset maximum reference value, determining that the alignment requirement is not met; Otherwise, it is judged that the alignment requirement is met.

6. An automatic alignment system for upper anchor bolt equipment based on a lifting robot using the method according to any one of claims 1 to 5, characterized in that: The collection unit is equipped with a visual sensor on the lifting robot of the upper anchor equipment to collect visual information; A recognition unit, which performs preliminary recognition on the visual information; A positioning unit, which visually locates a predetermined area for anchor bolt installation according to the result of the preliminary identification, and determines an installation angle of the anchor bolt according to the positioning result; The execution unit aligns the anchor rod according to the positioning result and the installation angle; the anchor rod status is judged in real time, and if the alignment requirements are not met, the re-alignment operation is performed.

7. A computer device comprising: Memory and processor; The memory stores a computer program, wherein the processor implements the steps of any one of the methods of claims 1-5 when executing the computer program.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images