Method for controlling running posture of electric power inspection robot

Through the acquisition and fusion of multi-type environmental data, combined with kinematic model and path planning technology, power inspection robots can accurately control attitudes and paths in complex environments and identify abnormal states, solving the attitude control deviation and evaluation accuracy in the existing technology The problem of insufficient improvements in patrol efficiency and safety.

CN118963360APending Publication Date: 2024-11-15ZHONGNENG XINGSHENG (XIANGHE) ELECTROMECHANICAL EQUIP CO LTD
View PDF 0 Cites 24 Cited by

Patent Information

Application Number
CN202411062574.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The accuracy of attitude control deviation and abnormal state evaluation of power inspection robots in complex environments is insufficient, resulting in inaccurate attitude control and poor fault tolerance.

Method used

Multi-type environmental data acquisition equipment (lidar, depth camera, light sensor, temperature sensor) is used to collect three-dimensional information and fusion, generate three-dimensional scene models, combine kinematic equations to estimate the robot posture, dynamically plan the optimal inspection path, and Adjust posture and paths in real time, identify abnormal states, and implement fault-tolerant control.

Benefits of technology

It improves the attitude control accuracy and abnormal recognition capabilities of the power inspection robot in complex environments, reduces the risk of inspection omissions and errors, and ensures the safe and stable operation of the power system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118963360A_ABST
    Figure CN118963360A_ABST
Patent Text Reader

Abstract

The invention discloses an electric power inspection robot operation attitude control method, and relates to the technical field of data processing, and the method comprises the following steps: collecting the three-dimensional information of an inspection environment, the obstacle position, the distance, and the illumination and temperature environment parameters based on a multi-type environment data collection device; and the collected environment data is subjected to preprocessing and data fusion. The precision of inspection operation is improved by adjusting the movement track and posture of the robot, the advancing speed and posture of the robot are automatically adjusted by sensing the change of the surrounding environment through the integrated sensor, it is ensured that information of a target area is clearly captured, the inspection path and posture are dynamically adjusted according to collected environment data, and the inspection efficiency is improved. According to the method, the labor intensity of manual inspection is reduced, inspection omission and errors caused by human factors are reduced, the robot can stably operate in a complex environment, potential safety hazards are found and responded in time, and therefore safe and stable operation of an electric power system is guaranteed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method for controlling the running posture of an electric power inspection robot. Background Art

[0002] Power inspections usually need to be carried out in dangerous environments such as high-voltage transmission lines and substations. These environments pose a greater safety threat to inspectors. Power inspection robots can replace humans to enter these dangerous areas and reduce the risk of accidents. Moreover, with the continuous development of robotics technology, power inspection robots have already possessed higher intelligent control and autonomous navigation capabilities, enabling power inspection robots to move autonomously in complex environments and conduct inspections according to preset inspection routes.

[0003] In the prior art, power inspection robots rely on a variety of sensors for posture control, but the working environment of power inspection is relatively complex, and there are obstacles and environmental factors that interfere, which causes posture control deviations in the actual inspection of the robot. In addition, there is a lack of effective robot posture deviation and abnormal state assessment standards, resulting in insufficient accuracy and poor fault tolerance in predicting and adjusting the current posture of the robot. Therefore, how to improve the posture control ability and abnormal recognition and adaptability of the power inspection robot in a complex environment is a problem we need to solve. To this end, a control method for the operating posture of a power inspection robot is proposed. Summary of the invention

[0004] The present invention aims to provide a method for controlling the running posture of an electric power inspection robot to solve the problems raised in the above-mentioned background technology.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A method for controlling the running posture of an electric power inspection robot comprises the following steps:

[0007] Step 1: Based on multiple types of environmental data collection equipment, collect the three-dimensional information of the inspection environment, obstacle location, distance, and light and temperature environmental parameters, and pre-process and fuse the collected environmental data;

[0008] Step 2, combining the pre-processed and data-fused environmental data to perform environmental scanning and modeling to generate a three-dimensional scene model;

[0009] Step 3: Based on the structural parameters of the robot, establish its kinematic equation, analyze the relationship between the position of each joint of the robot and the posture of the end effector, and estimate the current posture of the robot based on the fused environmental data and kinematic equation;

[0010] Step 4: Based on the fused environmental data, the path planning algorithm is used to dynamically generate the optimal inspection path. The optimal path is selected and the robot's motion trajectory is planned by combining the environmental map and the posture estimation results.

[0011] Step 5: Calculate the deviation between the current posture of the robot and the target posture in real time, obtain the posture deviation evaluation index, generate corresponding control instructions according to the posture deviation, adjust the robot's movement posture and execute it;

[0012] Step 6: Collect the robot's status data for analysis, obtain the abnormal status evaluation index, identify faults and abnormal phenomena, and classify the abnormality levels according to the fault identification results, introduce a fault-tolerant control strategy, and correct the robot's faults and abnormal phenomena.

[0013] A further improvement of the technical solution of the present invention is that in step 1, the collection and fusion process of environmental data is as follows:

[0014] Step 101, deploy multiple types of data collection equipment, including laser radar, depth camera, light sensor and temperature sensor, to collect environmental data, wherein the laser radar is used to collect three-dimensional point cloud data to obtain the fine three-dimensional structure of the environment, the depth camera is used to obtain the obstacle position and distance information through stereo matching, and the light sensor and temperature sensor are used to measure the light intensity and temperature environmental parameters respectively; the laser radar emits a laser beam at a high frequency and receives the reflected light, calculates the distance according to the time difference, and generates three-dimensional point cloud data, the depth camera calculates the depth information by taking multi-view images using the parallax principle, and the light and temperature sensors directly read the environmental parameters;

[0015] Step 102, preprocessing the collected environmental data to ensure the accuracy and availability of the data, the preprocessing includes but is not limited to: data filtering and data calibration, wherein data filtering is used to remove noise and abnormal values ​​that may be generated by the sensor, and data calibration is used to calibrate the data collected by different sensors to ensure the consistency of the data in a unified coordinate system;

[0016] Step 103, using a centralized fusion method to centrally fuse the three-dimensional point cloud data, obstacle location information, and illumination and temperature environment parameters;

[0017] Step 104, extracting key features from the multi-source environmental data, wherein the key features are respectively geometric features, surface features and spatial position features of the three-dimensional point cloud data, obstacle boundary, type and distance information features of the obstacle position and distance information, illumination level and illumination change features of the illumination intensity, and ambient temperature and temperature gradient features of the temperature environment parameters;

[0018] Step 105, spatially align the three-dimensional point cloud data of the lidar with the obstacle position information of the depth camera to ensure that they are in the same coordinate system, and map the light and temperature data to the three-dimensional environment based on the spatial position information, assign corresponding light intensity and temperature values ​​to each point in the three-dimensional point cloud data, and identify specific objects in the environment by analyzing the geometric and surface features of the three-dimensional point cloud data in combination with deep learning technology, and associate them with the obstacle position information, light and temperature data to form a unified environment description.

[0019] A further improvement of the technical solution of the present invention is that in step 2, the generation process of the three-dimensional scene model is:

[0020] Step 201, extract the feature data collected by the laser radar and the depth camera. The laser radar measures the distance and angle by emitting a laser beam and receiving the reflected beam to generate point cloud data. The depth camera takes multi-view images and calculates the depth information using the parallax principle. The expression for measuring the distance is:

[0021]

[0022] Where d is the measured distance, c is the speed of light, and t is the round-trip time of the light beam;

[0023] The expression for parallax calculation is:

[0024]

[0025] Where Z is the depth, f is the camera focal length, B is the camera baseline, and s is the disparity;

[0026] Step 202, filtering and denoising the point cloud data to improve data quality, and aligning the point cloud data collected at different times and locations to the same coordinate system. The Gaussian filter expression of the point cloud data is:

[0027]

[0028] Among them, P is the filtered point, p is the original point, and σ is the standard deviation of the Gaussian distribution;

[0029] Step 203, fusing the point cloud data of the laser radar and the depth data of the visual sensor to generate a three-dimensional scene model.

[0030] A further improvement of the technical solution of the present invention is that in step 3, the estimation process of the current posture of the robot is:

[0031] Step 301, define the structural parameters of the robot, including the base, connecting rods and joints. The connecting rods are the limbs of the robot. Each connecting rod is defined by its length and direction. The direction is represented based on the rotation matrix. The expression of the rotation matrix is:

[0032] R(θ)=I+sin(θ)S+(1-Cos(θ))S 2 ;

[0033] Where R(θ) is the rotation matrix, q i is the joint angle, I is the unit matrix, and S is the skew-symmetric matrix consisting of the joint axis vectors;

[0034] Step 302, for each joint, define its joint variables, establish a forward kinematics equation, and calculate the position and direction of the end effector. The expression of the forward kinematics equation is:

[0035] T=T0·T1(q1)·…·T m (q m );

[0036]

[0037] p i (q i )=l i ·R i-1 ·…·R0;

[0038] Where T is the homogeneous transformation matrix of the end effector, q i is the joint variable, T i (q i ) is the transformation matrix generated by the movement of joint i, R i (q i ) is the rotation matrix generated by the rotation of joint i, p i (q i ) is the translation vector generated by the rotation of joint i, l i is the length of connecting rod i, R i-1 ,…,R0 is the rotation matrix of all previous links from the base to link i;

[0039] Step 303, using the forward kinematics equation in combination with the joint variables, estimates the current posture of the robot.

[0040] A further improvement of the technical solution of the present invention is that in step 4, the planning process of the robot motion trajectory is:

[0041] Step 401, based on the fused environmental data and combined with the three-dimensional scene model, the environmental map is updated. The environmental map includes obstacles, walkable areas, and inspection target points, reflecting the dynamic changes of the environment. Obstacle detection is achieved by using data detection of laser radar and depth camera, and marked on the map. Walkable area identification determines the area where the robot can walk safely by analyzing the terrain and lighting conditions, and markings the inspection target points on the map according to the task requirements;

[0042] Step 402, based on the estimated current posture of the robot, the specific position and orientation of the robot in the environment is determined in combination with the environment map, and SLAM (Simultaneous Localization and Mapping) technology is used to simultaneously locate and build a map during the robot's movement, and the current position of the robot is estimated using sensor data and SLAM algorithm;

[0043] Step 403, in the environment map, according to the inspection target and environmental constraints, use the A* path planning algorithm to search for a path from the starting point to the target point on the environment map, and dynamically adjust the path to avoid new obstacles or adverse environmental conditions;

[0044] Step 404, smoothing line optimization is performed on the generated path, and the path is adjusted in combination with the posture estimation result of the robot so that the robot maintains a stable and safe posture during the movement, and the speed and acceleration curves are calculated according to the adjusted path and the dynamic characteristics of the robot, and then the selected path is converted into a detailed motion trajectory, and smooth movement is achieved according to the planned motion trajectory.

[0045] A further improvement of the technical solution of the present invention is that: the generated path is subjected to smooth line optimization, and its expression is:

[0046]

[0047] Where st is a measure of the rate of change between path points, Δg j is the distance between adjacent path points, Δt j It is the time interval between adjacent path points. It should be noted that the smaller st is, the smoother the path changes and the smoother the robot moves;

[0048] The generation expression of the motion trajectory is:

[0049]

[0050]

[0051] Among them, P y (r) is the position of the robot at time r, expressed as a motion trajectory, which refers to the position P of the robot at time r. y (r) is the starting position P start Add the time integral of the velocity function v(τ) from time 0 to time r, P start is the starting position of the robot's motion, v(τ) is the velocity function at time τ, which represents the velocity of the robot at different time points, and τ is the integral variable, which is used to represent the instantaneous time point in the integration process. It is the integral of the velocity from 0 to r over time, indicating the change in position. v0 is the initial velocity, that is, the velocity value at the starting point of time, the integral operation. a(∈) is the acceleration function at ∈, indicating the change of acceleration over time. d∈ is the differential of a small time interval, which is used for the integral operation.

[0052] A further improvement of the technical solution of the present invention is that in step 5, the process of obtaining the posture deviation evaluation index includes:

[0053] Step 501, analyzing the fused environmental data, determining the target posture of the robot, obtaining the target position and target direction, and estimating the current position and current direction of the robot using sensor data and SLAM technology;

[0054] Step 502, comparing the current posture of the robot with the target posture according to the Euclidean norm, and calculating the deviation between the current position of the robot and the target position;

[0055] Step 503, calculating the deviation between the current direction of the robot and the target direction by calculating the trace of the product of the two rotation matrices;

[0056] Step 504, combining the position deviation and the direction deviation to form a comprehensive posture deviation evaluation index to quantify the posture deviation of the robot;

[0057] Step 505, preset a comprehensive deviation threshold, combined with the posture deviation evaluation index, if it exceeds the comprehensive deviation threshold, trigger the posture adjustment mechanism, generate a control instruction, and send the control instruction to the robot's actuator to adjust the robot's position and direction until the target posture is reached.

[0058] A further improvement of the technical solution of the present invention is that the calculation formula of the position deviation is:

[0059] d pos =||P tr -P cr ||2;

[0060] Among them, d pos is the position deviation, which indicates the distance between the target position and the current position, P tr is the target position, indicating the position the robot should reach, P cr is the current position, indicating the current position of the robot, and ||·||2 represents the Euclidean norm;

[0061] The calculation formula of the directional deviation is:

[0062]

[0063] Among them, d rotis the direction deviation, which is used to measure the difference between the current direction and the target direction, R tr is the target direction, indicating the direction the robot should face, R cr is the current direction, indicating the direction the robot is currently facing, R tr and R cr are all rotation matrices, Tr represents the trace of the matrix, YesR tr The transpose of

[0064] The calculation formula of the posture deviation evaluation index is:

[0065]

[0066] Where D is the attitude deviation evaluation index, d max is the maximum allowable position deviation, d pos is the position deviation, d rot is the directional deviation, v tr and v cr are the target speed and current speed, w tr and W cr are the target angular velocity and the current angular velocity, v max and w max are the maximum permissible deviations of velocity and angular velocity, respectively, and W1, w2, w3, and w4 are weight coefficients used to adjust the importance of different deviations in the evaluation index;

[0067] The generation expression of the control instruction is:

[0068]

[0069] Among them, u is the control command used to adjust the position and direction of the robot, k p is the position control gain, k r is the direction control gain, I is the unit matrix. It should be noted that u depends on the control gain and the position and direction deviations. The control instructions reduce the position deviation and direction deviation until the robot reaches the target posture.

[0070] A further improvement of the technical solution of the present invention is that in step 6, the process of obtaining the abnormal state evaluation index is:

[0071] Step 601, collecting the robot's status data and environmental data according to the installed sensors, including real-time data of temperature, pressure, current, and voltage, respectively, through corresponding sensors;

[0072] Step 602, monitor the robot's status data and environmental data in real time, compare them with the baseline data during normal operation, and identify deviations and anomalies;

[0073] Step 603, using machine learning technology to identify abnormal conditions in the data, and combining the deviation ranges of temperature, pressure, current, and voltage to obtain an abnormal condition evaluation index to determine potential faults and abnormal phenomena;

[0074] Step 604, according to the degree of influence and urgency of the faults and abnormal phenomena on the operation of the robot, the faults and abnormal phenomena are classified into different abnormal levels, namely, a slight abnormal level, a medium abnormal level, and a severe abnormal level, and corresponding evaluation thresholds are set for the faults and abnormal phenomena of different levels in combination with the abnormal state evaluation index;

[0075] Step 605, according to the fault identification result, determine the abnormal level, and deeply analyze the cause and mechanism of the fault, implement the corresponding fault-tolerant control strategy, and correct the fault and abnormal phenomenon of the robot.

[0076] A further improvement of the technical solution of the present invention is that the calculation formula of the abnormal state evaluation index is:

[0077]

[0078] Among them, A is the abnormal state evaluation index, Δp o is the real-time pressure change of the oth pressure sensor, Δp r,o is the reference pressure change of the oth pressure sensor, Δp max,o is the maximum allowable value of the pressure change of the oth pressure sensor, n is the number of pressure sensors involved in the calculation, L is the real-time measurement value of the current, L r is the reference value of the current, L max is the maximum allowable value of current, Y is the real-time measured value of voltage, and Y r is the reference value of voltage, Y max is the maximum allowable value of voltage, W is the real-time measurement value of temperature, W r is the reference value of temperature, W max is the maximum permissible value of temperature;

[0079] A plurality of the abnormality levels correspond to a plurality of the evaluation thresholds, wherein the evaluation thresholds include an upper threshold and a lower threshold;

[0080] The multiple abnormal levels and the multiple evaluation thresholds satisfy the following relationship:

[0081] Minor Abnormality Level 0 <A<A q ;

[0082] Medium Abnormal Level A q ≤A y ;

[0083] Severe abnormality level A≥A​y ;

[0084] Among them, A is the abnormal state evaluation index, A q is the lower threshold corresponding to the medium abnormality level and the upper threshold corresponding to the slight abnormality level, A y It is the lower threshold corresponding to the severe abnormality level and the upper threshold corresponding to the moderate abnormality level.

[0085] Due to the adoption of the above technical solution, the present invention has the following technical advances compared with the prior art:

[0086] 1. The present invention provides a control method for the running posture of an electric power inspection robot. The accuracy of the inspection operation is improved by adjusting the motion trajectory and posture of the robot. The robot senses changes in the surrounding environment through integrated sensors, thereby automatically adjusting the robot's travel speed and posture to ensure clear capture of information in the target area. The inspection path and posture are dynamically adjusted based on the collected environmental data, which not only reduces the labor intensity of manual inspections, but also reduces inspection omissions and errors caused by human factors, so that the robot can operate stably in a complex environment, promptly discover and respond to potential safety hazards, thereby ensuring the safe and stable operation of the power system.

[0087] 2. The present invention provides a control method for the running posture of an electric power inspection robot. Through path planning and motion trajectory optimization technology, the inspection efficiency of the robot and the rationality of the path are improved. In addition, according to the environmental map and the inspection target, a path planning algorithm is used to dynamically generate the optimal inspection path. Not only obstacles and walkable areas are analyzed, but also the kinematic and dynamic constraints of the robot are analyzed. By optimizing the smoothness of the path, the inspection task is completed with minimal energy consumption and maximum efficiency. In addition, the robot's motion trajectory planning can generate detailed motion instructions according to the path, speed and acceleration curves, ensuring the stability and continuity of the robot during movement.

[0088] 3. The present invention provides a control method for the operating posture of an electric power inspection robot. By learning historical inspection data, potential failure modes can be predicted and identified, so that measures can be taken in advance to avoid the occurrence of failures. This makes the robot more flexible and reliable when performing inspection tasks, improves its operating efficiency and safety in complex environments, greatly improves inspection efficiency, shortens inspection cycles, and provides strong guarantees for the safe and stable operation of power facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0090] Figure 1 is a flow chart of the method of the present invention;

[0091] Figure 2 A flow chart showing the motion trajectory of the robot of the present invention;

[0092] Figure 3 A flowchart for obtaining the posture deviation evaluation index of the present invention;

[0093] Figure 4 This is a flow chart for obtaining the abnormal state evaluation index of the present invention. DETAILED DESCRIPTION

[0094] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0095] Examples, such as Figure 1-Figure 2 As shown, the present invention provides a method for controlling the running posture of an electric power inspection robot, comprising the following steps:

[0096] Step 1, based on multiple types of environmental data acquisition equipment, collect the three-dimensional information, obstacle position, distance, and light and temperature environmental parameters of the inspection environment, and pre-process and fuse the collected environmental data, deploy multiple types of data acquisition equipment, including lidar, depth camera, light sensor and temperature sensor, to collect environmental data, wherein lidar is used to collect three-dimensional point cloud data to obtain the fine three-dimensional structure of the environment, depth camera is used to obtain obstacle position and distance information through stereo matching, and light sensor and temperature sensor are used to measure light intensity and temperature environmental parameters respectively; lidar emits laser beams at a high frequency and receives reflected light, calculates distance based on time difference, and generates three-dimensional point cloud data, depth camera calculates depth information by taking multi-view images using the parallax principle, light and temperature sensors directly read environmental parameters, and pre-processes the collected environmental data to ensure the accuracy and availability of the data, and the pre-processing includes but is not limited to: data filtering and data calibration, wherein data filtering is used to remove noise and outliers that may be generated by the sensor, and data The data calibration is used to calibrate the data collected by different sensors to ensure the consistency of the data in a unified coordinate system. A centralized fusion method is used to centrally fuse the 3D point cloud data, obstacle position information, and illumination and temperature environmental parameters, and extract key features from multi-source environmental data. The key features are the geometric features, surface features, and spatial position features of the 3D point cloud data, the obstacle boundary, type, and distance information features of the obstacle position and distance information, the illumination level and illumination change features of the illumination intensity, and the ambient temperature and temperature gradient features of the temperature environmental parameters. The 3D point cloud data of the lidar is spatially aligned with the obstacle position information of the depth camera to ensure that they are in the same coordinate system. According to the spatial position information, the illumination and temperature data are mapped to the 3D environment, and the corresponding illumination intensity and temperature values ​​are assigned to each point in the 3D point cloud data. By analyzing the geometric and surface features of the 3D point cloud data and combining deep learning technology, specific objects in the environment are identified and associated with the obstacle position information, illumination and temperature data to form a unified environmental description.

[0097] Step 2: Combine the preprocessed and data-fused environmental data to perform environmental scanning and modeling, generate a 3D scene model, extract feature data collected by the LiDAR and the depth camera, and measure the distance and angle by emitting a laser beam and receiving the reflected beam to generate point cloud data. The depth camera takes multi-view images, calculates the depth information using the parallax principle, filters and denoises the point cloud data to improve data quality, and registers the point cloud data collected at different times and locations to the same coordinate system. The point cloud data of the LiDAR and the depth data of the visual sensor are fused to generate a 3D scene model. The expression for measuring distance is:

[0098]

[0099] Where d is the measured distance, c is the speed of light, and t is the round-trip time of the light beam;

[0100] Furthermore, the expression for disparity calculation is:

[0101]

[0102] Where Z is the depth, f is the camera focal length, B is the camera baseline, and s is the disparity;

[0103] Furthermore, the Gaussian filter expression of point cloud data is:

[0104]

[0105] Among them, P is the filtered point, p is the original point, and σ is the standard deviation of the Gaussian distribution;

[0106] Step 3: Based on the structural parameters of the robot, establish its kinematic equations, analyze the relationship between the positions of each joint of the robot and the posture of the end effector, and estimate the current posture of the robot based on the fused environmental data and kinematic equations. Define the structural parameters of the robot, including the base, connecting rods and joints. The connecting rods are the limbs of the robot. Each connecting rod is defined by its length and direction. The direction is represented based on the rotation matrix. The expression of the rotation matrix is:

[0107] R(θ)=I+sin(θ)S+(1-cos(θ))S 2 ;

[0108] Where R(θ) is the rotation matrix, q i is the joint angle, I is the unit matrix, and S is the skew-symmetric matrix consisting of the joint axis vectors;

[0109] For each joint, define its joint variables, establish the forward kinematics equation, and calculate the position and direction of the end effector. The expression of the forward kinematics equation is:

[0110] T=T0·T1(q1)·…·T m (q m );

[0111]

[0112] p i (q i )=l i ·R i-1 ·…·R0;

[0113] Where T is the homogeneous transformation matrix of the end effector, q i is the joint variable, Ti (q i ) is the transformation matrix generated by the movement of joint i, R i (q i ) is the rotation matrix generated by the rotation of joint i, p i (q i ) is the translation vector generated by the rotation of joint i, l i is the length of connecting rod i, R i-1 ,…,R0 is the rotation matrix of all previous links from the base to link i;

[0114] Use the forward kinematics equations, combined with the joint variables, to estimate the robot's current posture;

[0115] Step 4: Based on the fused environmental data, the path planning algorithm is used to dynamically generate the optimal inspection path. Combined with the environmental map and posture estimation results, the optimal path is selected and the robot's motion trajectory is planned. Based on the fused environmental data and combined with the three-dimensional scene model, the environmental map is updated. The environmental map includes obstacles, walkable areas, and inspection target points to reflect the dynamic changes of the environment. Obstacle detection is achieved by using lidar and depth camera data detection and marking them on the map. Walkable area identification determines the area where the robot can walk safely by analyzing the terrain and lighting conditions. Marking inspection target points on the map according to task requirements, and determines the specific position and direction of the robot in the environment based on the estimated current posture of the robot and combined with the environmental map, and Using the synchronous positioning and mapping technology, the robot is positioned and mapped at the same time during its movement. The sensor data and SLAM algorithm are used to estimate the current position of the robot. In the environmental map, according to the inspection objectives and environmental constraints, the A* path planning algorithm is used to search for the path from the starting point to the target point on the environmental map. The path is dynamically adjusted to avoid new obstacles or adverse environmental conditions. The generated path is optimized for smooth lines. Combined with the robot's posture estimation results, the path is adjusted to keep the robot in a stable and safe posture during movement. According to the adjusted path and the dynamic characteristics of the robot, the speed and acceleration curves are calculated, and then the selected path is converted into a detailed motion trajectory. According to the planned motion trajectory, smooth movement is achieved.

[0116] Furthermore, the generated path is optimized by smoothing, and its expression is:

[0117]

[0118] Where st is a measure of the rate of change between path points, Δg j is the distance between adjacent path points, Δt j It is the time interval between adjacent path points. It should be noted that the smaller st is, the smoother the path changes and the smoother the robot moves;

[0119] The expression for generating the motion trajectory is:

[0120]

[0121]

[0122] Among them, P y (r) is the position of the robot at time r, expressed as a motion trajectory, which refers to the position P of the robot at time r. y (r) is the starting position P start Add the time integral of the velocity function v(τ) from time 0 to time r, P start is the starting position of the robot's motion, v(τ) is the velocity function at time T, which indicates the velocity of the robot at different time points, and T is the integral variable, which is used to indicate the instantaneous time point in the integral process. It is the integral of the velocity from 0 to r over time, indicating the change in position. v0 is the initial velocity, that is, the velocity value at the starting point of time, the integral operation. a(∈) is the acceleration function at ∈, indicating the change of acceleration over time. d∈ is the differential of a small time interval, which is used for the integral operation. The velocity v(τ) at any time T is equal to the initial velocity v0 plus the integral of the acceleration function a(∈) over time from the start of time to T, which calculates the total velocity change caused by the change in acceleration since the initial time;

[0123] Step 5: Calculate the deviation between the current posture of the robot and the target posture in real time, obtain the posture deviation evaluation index, generate corresponding control instructions according to the posture deviation, adjust the robot's movement posture and execute it;

[0124] Step 6: Collect the robot's status data for analysis, obtain the abnormal status evaluation index, identify faults and abnormal phenomena, and classify the abnormality levels according to the fault identification results, introduce a fault-tolerant control strategy, and correct the robot's faults and abnormal phenomena.

[0125] Embodiment 2, as Figure 3-Figure 4As shown, on the basis of Example 1, the present invention further provides a technical solution: preferably, in step 5, the process of obtaining the posture deviation evaluation index includes: analyzing the fused environmental data, determining the target posture of the robot, obtaining the target position and target direction, and estimating the current position and current direction of the robot using sensor data and SLAM technology; comparing the current posture of the robot with the target posture according to the Euclidean norm, calculating the deviation between the current position of the robot and the target position, calculating the deviation between the current direction of the robot and the target direction by calculating the trace of the product of two rotation matrices, combining the position deviation and the direction deviation to form a comprehensive posture deviation evaluation index to quantify the posture deviation of the robot, presetting a comprehensive deviation threshold, combining the posture deviation evaluation index, if it exceeds the comprehensive deviation threshold, triggering the posture adjustment mechanism, generating a control instruction, and sending the control instruction to the robot's actuator to adjust the position and direction of the robot until the target posture is reached;

[0126] Furthermore, the calculation formula of the position deviation is:

[0127] d pos =||P tr -P cr ||2;

[0128] Among them, d pos is the position deviation, which indicates the distance between the target position and the current position, P tr is the target position, indicating the position the robot should reach, P cr is the current position, indicating the current position of the robot, and ||·||2 represents the Euclidean norm. It should be noted that d pos is a non-negative real number, representing the straight-line distance from the current position to the target position, d pos It gradually decreases to zero, indicating that the robot is approaching or reaching the target position;

[0129] The calculation formula for directional deviation is:

[0130]

[0131] Among them, d rot is the direction deviation, which is used to measure the difference between the current direction and the target direction, R tr is the target direction, indicating the direction the robot should face, R cr is the current direction, indicating the direction the robot is currently facing, R tr and R cr are all rotation matrices, Tr represents the trace of the matrix, YesR tr The transpose of d rotis a real number in the interval [0, 1], 0 means the directions are exactly the same, 1 means the directions are completely opposite, d rot Decreases to zero, indicating that the robot is heading towards the target direction;

[0132] Furthermore, the calculation formula of the posture deviation evaluation index is:

[0133]

[0134] Where D is the attitude deviation evaluation index, d max is the maximum allowable position deviation, d pos is the position deviation, d rot is the directional deviation, v tr and v cr are the target speed and current speed, w tr and W cr are the target angular velocity and the current angular velocity, v max and w max are the maximum allowable deviations of velocity and angular velocity, respectively. W1, w2, w3, and w4 are weight coefficients used to adjust the importance of different deviations in the evaluation index. It should be noted that the smaller the value of D, the smaller the posture deviation. When D decreases to zero, it means that the robot's posture is completely consistent with the target posture.

[0135] Furthermore, the generation expression of the control instruction is:

[0136]

[0137] Among them, u is the control command used to adjust the position and direction of the robot, k p is the position control gain, k r is the direction control gain, I is the unit matrix. It should be noted that u depends on the control gain and the position and direction deviations. The control instructions reduce the position deviation and direction deviation until the robot reaches the target posture.

[0138] In step 6, the process of obtaining the abnormal state evaluation index is as follows: collect the state data and environmental data of the robot according to the installed sensors, including real-time data of temperature, pressure, current, and voltage, collect and monitor the state data and environmental data of the robot in real time through the corresponding sensors, compare them with the benchmark data during normal operation, identify deviations and anomalies, use machine learning technology to identify abnormal states in the data, and integrate the deviation ranges of temperature, pressure, current, and voltage to obtain the abnormal state evaluation index, judge potential faults and abnormal phenomena, and classify the faults and abnormal phenomena into different abnormal levels according to the degree and urgency of the impact of the faults and abnormal phenomena on the operation of the robot, namely, slight abnormal level, medium abnormal level, and severe abnormal level, and set corresponding evaluation thresholds for faults and abnormal phenomena of different levels in combination with the abnormal state evaluation index, determine the abnormal level to which it belongs according to the fault identification result, and deeply analyze the cause and mechanism of the fault, implement the corresponding fault-tolerant control strategy, and correct the faults and abnormal phenomena of the robot;

[0139] Furthermore, the calculation formula of the abnormal state evaluation index is:

[0140]

[0141] Among them, A is the abnormal state evaluation index, Δp o is the real-time pressure change of the oth pressure sensor, Δp r,o is the reference pressure change of the oth pressure sensor, Δp max,o is the maximum allowable value of the pressure change of the oth pressure sensor, n is the number of pressure sensors involved in the calculation, L is the real-time measurement value of the current, L r is the reference value of the current, L max is the maximum allowable value of current, Y is the real-time measured value of voltage, and Y r is the reference value of voltage, Y max is the maximum allowable value of voltage, W is the real-time measurement value of temperature, W r is the reference value of temperature, W max is the maximum allowable value of temperature. It should be noted that the value range of A is (0, ∞). When the value of A is close to 0, it means that the robot is in a normal state. When any parameter exceeds its maximum allowable value, the value of A will increase, indicating that the robot is in an abnormal state.

[0142] Furthermore, the multiple abnormality levels correspond to multiple evaluation thresholds, wherein the evaluation threshold includes an upper threshold and a lower threshold;

[0143] Multiple anomaly levels and multiple evaluation thresholds satisfy the following relationship:

[0144] Minor Abnormality Level 0 <A<A q; Continue monitoring and send fault warning information to operators;

[0145] Medium Abnormal Level A q ≤A y ; Isolate the faulty part to prevent the fault from spreading, adjust the robot operating parameters, try to restore the normal state, and send fault warning information to the operator;

[0146] Severe abnormality level A≥A y ; Immediately send fault warning information to the operator, perform emergency braking and stop the machine for inspection;

[0147] Among them, A is the abnormal state evaluation index, A q is the lower threshold corresponding to the medium abnormality level and the upper threshold corresponding to the slight abnormality level, A y It is the lower threshold corresponding to the severe abnormality level and the upper threshold corresponding to the moderate abnormality level.

[0148] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.​

Claims

1. A method for controlling the running posture of an electric power inspection robot, characterized in that: The following steps are involved: Step 1: Based on multiple types of environmental data collection equipment, collect the three-dimensional information of the inspection environment, obstacle location, distance, and light and temperature environmental parameters, and pre-process and fuse the collected environmental data; Step 2, combining the pre-processed and data-fused environmental data to perform environmental scanning and modeling to generate a three-dimensional scene model; Step 3: Based on the structural parameters of the robot, establish its kinematic equation, analyze the relationship between the position of each joint of the robot and the posture of the end effector, and estimate the current posture of the robot based on the fused environmental data and kinematic equation; Step 4: Based on the fused environmental data, the path planning algorithm is used to dynamically generate the optimal inspection path. The optimal path is selected and the robot's motion trajectory is planned by combining the environmental map and the posture estimation results. Step 5: Calculate the deviation between the current posture of the robot and the target posture in real time, obtain the posture deviation evaluation index, generate corresponding control instructions according to the posture deviation, adjust the robot's movement posture and execute it; Step 6: Collect the robot's status data for analysis, obtain the abnormal status evaluation index, identify faults and abnormal phenomena, and classify the abnormality levels according to the fault identification results, introduce a fault-tolerant control strategy, and correct the robot's faults and abnormal phenomena.

2. The method for controlling the running posture of a power inspection robot according to claim 1, characterized in that: In step 1, the process of collecting and integrating environmental data is as follows: Step 101, deploy multiple types of data collection equipment, including laser radar, depth camera, light sensor and temperature sensor, to collect environmental data, wherein the laser radar is used to collect three-dimensional point cloud data to obtain the fine three-dimensional structure of the environment, the depth camera is used to obtain the obstacle position and distance information through stereo matching, and the light sensor and temperature sensor are used to measure the light intensity and temperature environment parameters respectively; Step 102, preprocessing the collected environmental data; Step 103, using a centralized fusion method to centrally fuse the three-dimensional point cloud data, obstacle location information, and illumination and temperature environment parameters; Step 104, extracting key features from the multi-source environmental data, wherein the key features are respectively geometric features, surface features and spatial position features of the three-dimensional point cloud data, obstacle boundary, type and distance information features of the obstacle position and distance information, illumination level and illumination change features of the illumination intensity, and ambient temperature and temperature gradient features of the temperature environment parameters; Step 105, spatially align the three-dimensional point cloud data of the lidar with the obstacle position information of the depth camera, and map the light and temperature data to the three-dimensional environment based on the spatial position information, assign corresponding light intensity and temperature values ​​to each point in the three-dimensional point cloud data, and identify specific objects in the environment by analyzing the geometry and surface features of the three-dimensional point cloud data in combination with deep learning technology, and associate them with the obstacle position information, light and temperature data to form a unified environment description.

3. The method for controlling the running posture of a power inspection robot according to claim 2, characterized in that: In step 2, the generation process of the 3D scene model is as follows: Step 201, extracting feature data collected by the laser radar and the depth camera. The laser radar measures the distance and angle by emitting a laser beam and receiving the reflected beam to generate point cloud data. The depth camera takes multi-view images and calculates the depth information using the parallax principle. Step 202, filtering and denoising the point cloud data, and registering the point cloud data collected at different times and locations into the same coordinate system; Step 203, fusing the point cloud data of the laser radar and the depth data of the visual sensor to generate a three-dimensional scene model.

4. The method for controlling the running posture of a power inspection robot according to claim 3 is characterized in that: In step 3, the estimation process of the robot's current posture is: Step 301, defining the structural parameters of the robot, including the base, connecting rods and joints, where the connecting rods are the limbs of the robot, and each connecting rod is defined by its length and direction, where the direction is represented based on a rotation matrix; Step 302, for each joint, define its joint variables, establish forward kinematics equations, and calculate the position and direction of the end effector; Step 303, using the forward kinematics equation in combination with the joint variables, estimates the current posture of the robot.

5. The method for controlling the running posture of a power inspection robot according to claim 4, characterized in that: In step 4, the planning process of the robot motion trajectory is: Step 401, updating the environment map according to the fused environment data and in combination with the three-dimensional scene model, the environment map includes obstacles, walkable areas, inspection target points, and reflects the dynamic changes of the environment; Step 402, based on the estimated current posture of the robot, the specific position and orientation of the robot in the environment is determined in combination with the environment map, and SLAM technology is used to simultaneously locate and construct a map during the robot's movement, and the current position of the robot is estimated using sensor data and SLAM algorithm; Step 403, in the environment map, according to the inspection target and environmental constraints, use the A* path planning algorithm to search for a path from the starting point to the target point on the environment map, and dynamically adjust the path to avoid new obstacles or adverse environmental conditions; Step 404, smoothing line optimization is performed on the generated path, and the path is adjusted in combination with the posture estimation result of the robot so that the robot maintains a stable and safe posture during the movement, and the speed and acceleration curves are calculated according to the adjusted path and the dynamic characteristics of the robot, and then the selected path is converted into a detailed motion trajectory, and smooth movement is achieved according to the planned motion trajectory.

6. The method for controlling the running posture of a power inspection robot according to claim 5, characterized in that: The smooth line optimization of the generated path is expressed as: Where st is a measure of the rate of change between path points, Δg j is the distance between adjacent path points, Δt j is the time interval between adjacent waypoints; The generation expression of the motion trajectory is: Among them, P y (r) is the position of the robot at time r, expressed as a motion trajectory, P start is the starting position of the robot's motion, v(τ) is the velocity function at time τ, which represents the velocity of the robot at different time points, and τ is the integral variable, which is used to represent the instantaneous time point in the integration process. It is the integral of the velocity from 0 to r over time, indicating the change in position. v0 is the initial velocity, that is, the velocity value at the starting point of time, the integral operation. a(∈) is the acceleration function at ∈, indicating the change of acceleration over time. d∈ is the differential of a small time interval, which is used for the integral operation.

7. A method for controlling the running posture of a power inspection robot according to claim 6, characterized in that: In step 5, the process of obtaining the posture deviation evaluation index includes: Step 501, analyzing the fused environmental data, determining the target posture of the robot, obtaining the target position and target direction, and estimating the current position and current direction of the robot using sensor data and SLAM technology; Step 502, comparing the current posture of the robot with the target posture according to the Euclidean norm, and calculating the deviation between the current position of the robot and the target position; Step 503, calculating the deviation between the current direction of the robot and the target direction by calculating the trace of the product of the two rotation matrices; Step 504, combining the position deviation and the direction deviation to form a comprehensive posture deviation evaluation index to quantify the posture deviation of the robot; Step 505, preset a comprehensive deviation threshold, combined with the posture deviation evaluation index, if it exceeds the comprehensive deviation threshold, trigger the posture adjustment mechanism, generate a control instruction, and send the control instruction to the robot's actuator to adjust the robot's position and direction until the target posture is reached.

8. The method for controlling the running posture of a power inspection robot according to claim 7, characterized in that: The calculation formula of the position deviation is: d pos =||P tr -P cr ||2; Among them, d pos is the position deviation, which indicates the distance between the target position and the current position, P tr is the target position, indicating the position the robot should reach, P cr is the current position, indicating the current position of the robot, ‖·‖2 represents the Euclidean norm; The calculation formula of the directional deviation is: Among them, d rot is the direction deviation, which is used to measure the difference between the current direction and the target direction, R tr is the target direction, indicating the direction the robot should face, R cr is the current direction, indicating the direction the robot is currently facing, R tr and R cr are all rotation matrices, Tr represents the trace of the matrix, YesR tr The transpose of The calculation formula of the posture deviation evaluation index is: Where D is the attitude deviation evaluation index, d max is the maximum allowable position deviation, d pos is the position deviation, d rot is the directional deviation, v tr and v cr are the target speed and current speed, w tr and w cr are the target angular velocity and the current angular velocity, v max and w max are the maximum allowable deviations of velocity and angular velocity respectively, and w1, w2, w3, w4 are weight coefficients.

9. A method for controlling the running posture of a power inspection robot according to claim 8, characterized in that: In step 6, the process of obtaining the abnormal state evaluation index is as follows: Step 601, collecting the robot's status data and environmental data according to the installed sensors, including real-time data of temperature, pressure, current, and voltage, respectively, through corresponding sensors; Step 602, monitor the robot's status data and environmental data in real time, compare them with the baseline data during normal operation, and identify deviations and anomalies; Step 603, using machine learning technology to identify abnormal conditions in the data, and combining the deviation ranges of temperature, pressure, current, and voltage to obtain an abnormal condition evaluation index to determine potential faults and abnormal phenomena; Step 604, according to the degree of influence and urgency of the faults and abnormal phenomena on the operation of the robot, the faults and abnormal phenomena are classified into different abnormal levels, namely, a slight abnormal level, a medium abnormal level, and a severe abnormal level, and corresponding evaluation thresholds are set for the faults and abnormal phenomena of different levels in combination with the abnormal state evaluation index; Step 605, according to the fault identification result, determine the abnormal level, and deeply analyze the cause and mechanism of the fault, implement the corresponding fault-tolerant control strategy, and correct the fault and abnormal phenomenon of the robot.

10. The method for controlling the running posture of a power inspection robot according to claim 9, characterized in that: The calculation formula of the abnormal state evaluation index is: Among them, A is the abnormal state evaluation index, Δp o is the real-time pressure change of the oth pressure sensor, Δp r,o is the reference pressure change of the oth pressure sensor, Δp max,o is the maximum allowable value of the pressure change of the oth pressure sensor, n is the number of pressure sensors involved in the calculation, L is the real-time measurement value of the current, L r is the reference value of the current, L max is the maximum allowable value of current, Y is the real-time measured value of voltage, and Y r is the reference value of voltage, Y max is the maximum allowable value of voltage, W is the real-time measurement value of temperature, W r is the reference value of temperature, W max is the maximum permissible value of temperature; A plurality of the abnormality levels correspond to a plurality of the evaluation thresholds, wherein the evaluation thresholds include an upper threshold and a lower threshold; The multiple abnormal levels and the multiple evaluation thresholds satisfy the following relationship: Minor Abnormality Level 0 <A<A q ; Medium Abnormal Level A q ≤A y ;​ Severe abnormality level A≥A y ; Among them, A is the abnormal state evaluation index, A q is the lower threshold corresponding to the medium abnormality level and the upper threshold corresponding to the slight abnormality level, A y It is the lower threshold corresponding to the severe abnormality level and the upper threshold corresponding to the moderate abnormality level.

Citation Information

Cited By

  • AR-based intelligent integrated production line special circuit control method and system

    CN119229064A

  • AR-based intelligent integrated production line dedicated circuit control method and system

    CN119229064B

  • Attitude control system based on single-rail inspection vehicle

    CN119440079A

  • Attitude Control System Based on Single-Rail Inspection Vehicle

    CN119440079B

  • Inspection robot centralized monitoring system and method

    CN119690079A