Magnetic wheel steering driving method and device for wind power tower climbing robot

Abstract

The application discloses a magnetic wheel turning driving method and device for a wind power tower climbing robot, and relates to the technical field of robots. The method comprises the following steps: solving a robot motion target, mapping a multi-degree-of-freedom control to magnetic wheel turning, and obtaining a reference turning scheme; optimizing protection parameters based on a stick slip risk prediction model, and establishing a stick slip protection mechanism; analyzing curved surface interference according to wind power tower basic data, and constructing an interference compensation mechanism; analyzing adsorption resistance and constructing a compensation mechanism, and combining stick slip protection and curved surface interference compensation to cooperatively optimize the reference scheme and obtain an optimized scheme. The technical problems of large magnetic adsorption and variable-curvature wall working conditions, large magnetic wheel turning resistance, easy occurrence of turning stick slip and curved surface interference, low turning control precision of the climbing robot and poor stability are solved, and the technical effect of improving turning driving stability, precision and environmental adaptability by performing stick slip protection, curved surface interference compensation and adsorption resistance compensation on the magnetic wheel turning process is achieved.

Description

Technical Field

[0001] This invention relates to the field of robotics, and more specifically to a magnetic wheel steering drive method and apparatus for a wind turbine tower crawling robot. Background Technology

[0002] With the rapid development of the wind energy industry, the maintenance and upkeep of wind turbine towers—a core component of wind power equipment—has become particularly important. Wind turbine towers are typically hundreds of meters high, supporting the turbine blades and enduring complex aerodynamic loads over long periods. Traditional manual high-altitude operations suffer from high safety risks, low efficiency, high costs, and unstable quality. In recent years, wall-climbing inspection robots have become the most widely used and mature non-destructive testing equipment in the wind power field, primarily used for the inspection and maintenance of high-altitude or vertical components such as wind turbine towers. Their core advantage lies in achieving stable movement on the wall through reliable adsorption. Currently, wall-climbing robots used in wind turbine towers can be categorized by adsorption method into magnetic adsorption and negative pressure adsorption, and by movement method into wheeled, tracked, and legged types. To ensure the robot's safety in strong winds at heights of hundreds of meters, magnetic wheels typically employ permanent magnet arrays or high-power electromagnets to generate extremely strong normal adsorption forces. However, existing permanent magnet adsorption wall-climbing robots generally suffer from drawbacks such as difficulty in controlling the suction force, difficulty in steering, and poor mobility. In omnidirectional wall-climbing robots with four or more wheels, each magnetic wheel needs to be independently controlled to synthesize an arbitrary movement vector in order to achieve precise trajectory tracking with "zero slip" or "minimal slip". However, when the drive wheel needs to turn around its central axis, the huge magnetic attraction force generates a large static friction torque between the wheel and the wall, forming a huge steering resistance torque. Existing technologies usually use high-torque servo motors or rotary servo motors to directly drive the wheel axle to turn through a reducer. However, even if a high-torque servo motor is selected, its rated output torque is still insufficient in the face of the huge steering damping generated by the magnetic wheel. Long-term operation under extreme loads can easily lead to servo motor overheating, gear damage, or motor burnout. At the same time, directly mounting the bulky servo motor on the swingable wheel arm will significantly increase the end moment of inertia. If it is mounted on the main body, motion needs to be transmitted through a complex transmission chain, which has low transmission efficiency, complex structure, and backlash, affecting steering control accuracy. Therefore, it is difficult to meet the high reliability and high precision omnidirectional movement requirements of wall-climbing robots for 100-meter-high wind turbine towers. Summary of the Invention

[0003] This application provides a magnetic chuck steering drive method and device for a wind turbine tower crawling robot, which solves the technical problems of high steering resistance, easy steering jamming and surface interference of magnetic chuck under strong magnetic adsorption and variable curvature wall conditions, resulting in low steering control accuracy and poor stability of the crawling robot.

[0004] A first aspect of this application provides a magnetic wheel steering drive method for a wind turbine tower crawling robot, the method comprising: The motion target of the crawling robot on the wind turbine tower is calculated to determine the motion target vector. Based on the motion target vector, the magnetic wheel steering module of the crawling robot is mapped with multi-degree-of-freedom control to obtain a steering drive baseline scheme. The steering drive baseline scheme is optimized by using a steering jamming risk prediction model to establish a jamming protection mechanism. The steering drive baseline scheme is subjected to surface interference analysis based on the basic data of the wind turbine tower to construct a surface interference compensation mechanism. The adsorption resistance compensation is analyzed based on the steering drive baseline scheme to construct an adsorption resistance compensation mechanism. The steering drive baseline scheme is then optimized by combining the jamming protection mechanism and the surface interference compensation mechanism to obtain an optimized steering drive scheme.

[0005] A second aspect of this application provides a magnetic wheel steering drive device for a wind turbine tower crawling robot, the device comprising: The system comprises the following components: a control mapping component, a control mapping component, and a collaborative optimization component. The control mapping component calculates the motion target vector for the crawling robot on the wind turbine tower and performs multi-degree-of-freedom control mapping on the magnetic wheel steering module of the crawling robot based on the motion target vector to obtain a steering drive baseline scheme. A jamming protection component optimizes the jamming protection parameters of the steering drive baseline scheme using a steering jamming risk prediction model to establish a jamming protection mechanism. A surface interference component performs surface interference analysis on the steering drive baseline scheme based on the wind turbine tower's basic data to construct a surface interference compensation mechanism. A collaborative optimization component performs adsorption resistance compensation analysis on the steering drive baseline scheme to construct an adsorption resistance compensation mechanism. It then combines the jamming protection mechanism and the surface interference compensation mechanism to perform collaborative optimization of the steering drive baseline scheme to obtain an optimized steering drive scheme.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, the motion target of the wind turbine tower crawling robot is analyzed, and a basic steering control scheme for the magnetic chuck is generated based on the target motion direction and state. Then, a steering jamming risk prediction model is used to assess the risk and optimize the parameters of the basic steering control scheme, establishing a steering jamming protection mechanism. Next, considering the curved surface structure characteristics of the wind turbine tower, surface interference analysis is performed on the steering process, and a surface interference compensation mechanism is constructed. Then, a compensation analysis is conducted on the adsorption resistance between the magnetic chuck and the tower body, establishing an adsorption resistance compensation mechanism. Finally, jamming protection, surface interference compensation, and adsorption resistance compensation are synergistically integrated to optimize the basic steering control scheme, obtaining a steering drive optimization scheme suitable for the complex curved surface environment of the wind turbine tower, thereby improving the stability, reliability, and environmental adaptability of the magnetic chuck steering. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic diagram of the magnetic wheel steering drive method for a wind turbine tower crawling robot provided in an embodiment of this application.

[0009] Figure 2 This is a schematic diagram of the training convergence curve of the steering jamming risk prediction model provided in the embodiments of this application.

[0010] Figure 3 This is a schematic diagram of the magnetic wheel steering drive device for a wind turbine tower crawling robot provided in an embodiment of this application.

[0011] Explanation of reference numerals in the attached figures: Control mapping component 11, jamming protection component 12, curved surface interference component 13, collaborative optimization component 14. Detailed Implementation

[0012] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0013] Example 1, as Figure 1 As shown, this application provides a magnetic wheel steering drive method for a wind turbine tower crawling robot, the method comprising: The motion target is calculated for the crawling robot of the wind turbine tower, the motion target vector is determined, and the magnetic wheel steering module of the crawling robot is mapped with multi-degree-of-freedom control based on the motion target vector to obtain the steering drive reference scheme.

[0014] In this embodiment, the wind turbine tower crawling robot employs a four-magnetic-wheel independent steering structure. The four magnetic-wheel steering modules are respectively positioned at the front left, front right, rear left, and rear right of the robot's main body. Each magnetic-wheel steering module consists of a wheel arm, a steering support, a magnetic wheel, and a steering shaft. The system establishes a vehicle coordinate system with the geometric center of the crawling robot's main body as the origin. The X-axis of the vehicle coordinate system points to the crawling direction of the robot along the height of the wind turbine tower, and the Y-axis points to the tangential direction of the wind turbine tower's circumference. The system receives the robot's current target axial crawling velocity, target circumferential correction velocity, and target attitude angular velocity, and combines these three into a motion target vector V = (vx, vy, ω), where vx represents the target velocity of the crawling robot along the height of the wind turbine tower, vy represents the target velocity of the crawling robot along the tangential direction of the wind turbine tower's circumference, and ω represents the target attitude angular velocity of the crawling robot around its own geometric center. Subsequently, considering the installation position and structural parameters of each magnetic wheel steering module on the crawling robot's main body, the overall motion target is decomposed into individual magnetic wheels. The yaw angle, steering actuator drive stroke, required drive torque, and steering speed for each magnetic wheel are calculated. Through this multi-degree-of-freedom control mapping, the overall motion commands of the crawling robot can be converted into specific drive parameters executable by each magnetic wheel steering module, thus forming an initial steering drive baseline scheme. This provides a control basis for subsequent jamming protection, surface interference compensation, and adsorption resistance compensation.

[0015] Furthermore, based on the motion target vector, multi-degree-of-freedom control mapping is performed on the magnetic wheel steering module of the crawling robot to obtain a steering drive reference scheme, including: Feature association analysis is performed on the magnetic wheel steering drive log set of the crawling robot to construct a multi-degree-of-freedom control mapping relationship, which includes yaw angle mapping relationship, drive stroke mapping relationship, drive torque mapping relationship, and steering speed mapping relationship. Based on the multi-degree-of-freedom control mapping relationship, a steering control mapping map is constructed. The installation position of the magnetic wheel steering module and the motion target vector are input into the steering control mapping map to obtain a matching steering control set. The matching steering control set is then filtered by trigger frequency to output the steering drive benchmark scheme.

[0016] Preferably, the system first calls the magnetic wheel steering drive log set of the crawling robot. This log set records the historical control data of the four magnetic wheel steering modules in chronological order. Each log entry includes the magnetic wheel number, the magnetic wheel installation position coordinates, the motion target vector, the current yaw angle, the target yaw angle, the current stroke of the servo linear cylinder, the target drive stroke of the servo linear cylinder, the output thrust of the servo linear cylinder, the output torque of the steering rocker arm, the actual steering speed of the magnetic wheel, and the steering completion error. The system first cleans the magnetic wheel steering drive log set, deleting abnormal logs with steering completion errors greater than a preset error threshold, and retaining valid logs with steering completion errors less than the preset error threshold. Then, using the magnetic wheel installation position coordinates and the motion target vector as input features, and the target yaw angle, target drive stroke, output torque, and actual steering speed as output features, feature association analysis is performed on the valid logs. Specifically, the motion target vector is denoted as V, and the installation position coordinates of the i-th magnetic wheel are denoted as r. i =(x i y i ), and according to v i x=vx-ω·y i v i y=vy+ω·x i Calculate the target velocity component at the end of the i-th magnetic wheel, and then according to θ i =atan2(v i y, v i x) Establish a yaw angle mapping relationship between the moving target vector and the yaw angle of the magnetic chuck target. Then, based on the target yaw angle θ in the valid log... i With the target drive stroke s of the servo linear electric cylinder i The data corresponds one-to-one with the target yaw angles. A yaw angle-stroke calibration table is established in ascending order of target yaw angles. A linear interpolation method is used to establish a drive stroke mapping relationship between two adjacent calibration points. Then, the output thrust F of the servo linear electric cylinder is used as the basis for the calibration. i The effective lever arm L of the steering rocker arm and the angle α between the steering rocker arm and the push rod. i According to T i =F i ·L·sinα i Calculate the output torque of the steering rocker arm and the change in the target yaw angle Δθ. i With output torque T i The corresponding data is stored to form a driving torque mapping relationship. Furthermore, the target yaw angle change Δθ is also considered. i and the corresponding turn completion time t in the valid log. i According to u i =Δθ i / t i Calculate the actual steering speed of the magnetic chuck and the change in the target yaw angle Δθ.i With actual steering speed u i The corresponding data is stored to form a steering speed mapping relationship. This results in a multi-degree-of-freedom control mapping relationship consisting of yaw angle mapping, drive stroke mapping, drive torque mapping, and steering speed mapping.

[0017] Next, each valid log sample is converted into a graph node. This graph node includes an input end and an output end. The input end includes the magnetic wheel number and the installation position coordinates r. i The moving target vector V and the current yaw angle θ i 0 The output includes the target yaw angle θ. i Target-driven journeys i Output torque T i and steering speed u i The system establishes independent sub-graphs for the four magnetic wheel steering modules according to their magnetic wheel numbers, and further categorizes them based on the similarity of the target vector V and the target yaw angle θ. i The similarity of values ​​establishes connection edges between adjacent graph nodes, allowing continuous querying of control parameters corresponding to the same magnetic chuck wheel under similar motion targets. During actual control, the installation position coordinates r of the current magnetic chuck wheel steering module are used. i The system inputs the current motion target vector V into the steering control mapping map. First, it determines the corresponding independent sub-map based on the magnetic wheel number. Then, it locks the corresponding magnetic wheel node set based on the installation position coordinates ri. Next, it searches for several historical nodes in this node set that have the smallest distance to the current motion target vector V. The system reads the target yaw angle, target drive stroke, output torque, and steering speed corresponding to several historical nodes to form a matching steering control set. Then, it performs trigger frequency filtering on the matching steering control set. That is, it counts the historical trigger counts within the same target yaw angle interval, the same target drive stroke interval, the same output torque interval, and the same steering speed interval in the matching steering control set. The control parameters with the highest historical trigger count are determined as the preferred control parameters. When two sets of control parameters have the same historical trigger count, the control parameters with smaller steering completion error are selected as the preferred control parameters. The system performs the above filtering process on the four magnetic wheel steering modules respectively, and finally outputs a steering drive benchmark scheme including the target yaw angle of the four magnetic wheels, the target drive stroke of the four servo linear electric cylinders, the target output torque of the four steering rocker arms, and the target steering speed of the four magnetic wheels. By using the above method, the overall motion target of the robot can be stably mapped to the specific execution parameters of each magnetic wheel steering module, thereby improving the matching degree between the steering drive benchmark scheme and the actual motion requirements and reducing the calculation deviation of subsequent compensation control.

[0018] Furthermore, the magnetic wheel steering module includes a wheel arm, a steering support, a magnetic wheel, a steering shaft, a steering rocker arm, and a steering execution unit. One end of the wheel arm is hinged to the main body of the crawling robot. The steering support is rotatably mounted on the other end of the wheel arm via the steering shaft. The magnetic wheel is rotatably mounted on the steering support. The steering rocker arm is connected to the steering shaft. The steering execution unit is drively connected to the steering rocker arm.

[0019] Preferably, the magnetic wheel steering module is installed on one side of the crawling robot's main body. This module includes a wheel arm, a steering support, a magnetic wheel, a steering vertical shaft, a steering rocker arm, and a steering execution unit. The wheel arm serves as the load-bearing connector between the magnetic wheel steering module and the robot's main body. Its inner end is connected to the robot's main body via a hinge shaft, allowing the wheel arm to swing up and down relative to the robot's main body. This ensures that the magnetic wheel remains in contact with the outer wall of the wind turbine tower as the tower's curvature changes. The outer end of the wheel arm is equipped with a bearing seat for mounting the steering vertical shaft. The steering vertical shaft is arranged in a direction nearly perpendicular to the magnetic wheel's contact surface and is rotatably mounted on the outer end of the wheel arm via a bearing. The steering support is fixedly connected to the lower end of the steering vertical shaft. The steering support deflects synchronously with the steering vertical shaft, and the magnetic wheel is rotatably mounted on the steering support via an axle. This allows the magnetic wheel to both roll around its own axle and change its yaw direction under the influence of the steering vertical shaft. The steering rocker arm is fixedly connected to the upper end of the steering vertical shaft, extending radially outward along the steering vertical shaft. The steering actuator is a servo linear cylinder, with the cylinder body movably connected to the cylinder mounting base on the wheel arm, and the push rod end movably connected to the outer end of the steering rocker arm. During operation, a target drive stroke command is sent to the servo linear cylinder, and the push rod extends or retracts according to the target drive stroke. The push rod pushes the steering rocker arm to rotate around the steering vertical shaft, converting the linear thrust into a rotational torque on the steering vertical shaft. The steering vertical shaft further drives the steering support to deflect, and the steering support drives the magnetic chuck to change the yaw angle. When the magnetic chuck needs to turn in the first direction, the servo linear cylinder push rod extends and pushes the steering rocker arm to swing forward; when the magnetic chuck needs to turn in the second direction, the servo linear cylinder push rod retracts and pulls the steering rocker arm to swing in the opposite direction. Because the magnetic chuck is mounted on the steering support, which is installed at the end of the wheel arm via the steering vertical shaft, and the steering actuator is directly connected to the steering rocker arm, the yaw angle of the magnetic chuck can be controlled simply by controlling the push rod stroke of the servo linear electric cylinder. With this structure, the magnetic chuck steering module can independently steer while the wheel arm adapts to the undulations of the wind turbine tower surface. This ensures the magnetic chuck's adhesion to the tower wall while improving the torque output, structural rigidity, and control accuracy of the magnetic chuck steering drive.

[0020] Furthermore, the steering actuation unit includes a servo linear electric cylinder, the cylinder body end of which is connected to the wheel arm via a first movable connector, and the push rod end of which is connected to the steering rocker arm via a second movable connector.

[0021] Optionally, the steering actuator uses a servo linear cylinder. The servo linear cylinder is inclined above the wheel arm along its length. It includes a cylinder body, a lead screw drive mechanism housed within the cylinder body, a push rod connected to the lead screw drive mechanism, and a servo motor that drives the lead screw drive mechanism to rotate. A cylinder mounting base is fixedly mounted on the wheel arm. The first movable connector uses a first ball joint, and the cylinder body end of the servo linear cylinder is connected to the cylinder mounting base via the first ball joint, allowing the cylinder body end to swing at a small angle relative to the wheel arm. The steering rocker arm is fixedly connected to the upper end of the steering shaft and extends radially outward along the steering shaft. The second movable connector uses a second ball joint, and the push rod end of the servo linear cylinder is connected to the outer end of the steering rocker arm via the second ball joint, allowing the push rod end to automatically adjust the connection angle when pushing or pulling the steering rocker arm. During operation, a target drive stroke command is sent to the servo motor, which drives the lead screw drive mechanism within the cylinder body to rotate. The lead screw drive mechanism converts the rotational motion of the servo motor into the linear extension and retraction motion of the push rod. When the push rod extends, it applies a thrust to the steering rocker arm via the second ball joint, causing the rocker arm to swing forward around the steering axis. This causes the steering axis to drive the steering support and magnetic wheel to deflect in the first steering direction. When the push rod retracts, it applies a pull to the steering rocker arm via the second ball joint, causing the rocker arm to swing backward around the steering axis. This causes the steering axis to drive the steering support and magnetic wheel to deflect in the second steering direction. Because the cylinder end is connected to the wheel arm via the first ball joint, and the push rod end is connected to the steering rocker arm via the second ball joint, the servo linear electric cylinder can automatically adjust its posture according to the angle change of the steering rocker arm during the extension and retraction of the push rod, avoiding off-center loading and additional bending moment between the push rod and the steering rocker arm. At the same time, when the wheel arm swings up and down relative to the robot body due to the curvature change of the wind turbine tower wall, the servo linear electric cylinder swings synchronously with the wheel arm, without restricting the swing freedom of the wheel arm. Through the above connection method, the linear push-pull force of the servo linear electric cylinder can be stably converted into the rotational torque of the steering shaft, thereby improving the steering torque output and steering drive reliability of the magnetic chuck under strong magnetic adsorption conditions.

[0022] The steering drive baseline scheme is optimized by using a steering jamming risk prediction model to optimize the jamming protection parameters and establish a jamming protection mechanism.

[0023] In one embodiment, a magnetic wheel steering simulation is established based on a steering drive baseline scheme. The process of the magnetic wheel turning from the current yaw angle to the target yaw angle under strong magnetic attraction resistance is simulated, and simulation data of jamming accidents corresponding to stroke stagnation, speed decrease, and drive current increase are recorded. Subsequently, the contribution of each parameter—the change in target yaw angle, target steering speed, target output torque, and target drive stroke—is evaluated to determine the degree of influence of each parameter on jamming risk. Based on the contribution degree, steering speed limit, starting force boost, and segmented stroke values ​​are configured to form a first jamming protection scheme set. Then, the first jamming protection scheme set is input into a steering jamming risk prediction model. Schemes with lower jamming risk are selected, and the protection parameters are adjusted in the vicinity of these schemes for neighborhood optimization, ultimately obtaining the jamming protection mechanism. This mechanism can reduce the influence of high-risk parameters on the magnetic wheel steering process before the steering action is executed, reducing steering jamming under strong magnetic attraction conditions.

[0024] Furthermore, by optimizing the configuration of the steering drive baseline scheme's jamming protection parameters using a steering jamming risk prediction model, a jamming protection mechanism is established, including: A jamming accident simulation is performed based on the steering drive benchmark scheme to obtain jamming accident simulation data; a parameter-by-parameter contribution evaluation is performed on the steering drive benchmark scheme based on the jamming accident simulation data to obtain a multi-parameter jamming contribution distribution; jamming protection parameters are configured on the steering drive benchmark scheme based on the multi-parameter jamming contribution distribution to obtain a first jamming protection scheme set; neighborhood optimization is performed on the first jamming protection scheme set based on the steering jamming risk prediction model to generate the jamming protection mechanism.

[0025] Preferably, after obtaining the steering drive baseline scheme, a jamming accident simulation is first performed based on the steering drive baseline scheme. This steering drive baseline scheme includes the target yaw angle of the four magnetic chucks, the target drive stroke of the servo linear electric cylinder, the target steering speed, and the target output torque. This jamming accident simulation is conducted in a pre-established wind turbine tower wall steering simulation environment. The wind turbine tower wall steering simulation environment includes the wind turbine tower curvature parameters, the friction coefficient of the steel wall, the normal attraction force of the magnetic chucks, the contact radius between the magnetic chucks and the wall, the lever arm length of the steering rocker arm, the maximum thrust of the servo linear electric cylinder, and the motor current limit. In this simulation environment, the magnetic chucks are regarded as constrained steering wheels attached to the outer wall of the wind turbine tower. The transmission relationship between the servo linear electric cylinder, the steering rocker arm, the steering vertical shaft, and the magnetic chucks is used as the simulation constraint condition to restore the process of the magnetic chucks completing yaw steering by being pushed and pulled by the electric cylinder under strong magnetic attraction. In the specific simulation, taking the i-th magnetic chuck steering module as the object, the current yaw angle, current target drive stroke, current motor current, current steering speed, and current magnetic attraction force of the magnetic chuck are read. The target yaw angle, target drive stroke, target steering speed, and target output torque in the steering drive baseline scheme are used as simulation inputs. According to the preset steering time step, the process of the servo linear cylinder moving from the current stroke to the target drive stroke is calculated in segments. In each time step, the changes in push rod stroke, steering rocker arm swing angle, steering vertical shaft angular velocity, magnetic chuck yaw angle, predicted motor current, and predicted magnetic chuck steering resistance torque are calculated. When the predicted value of motor current exceeds the current safety threshold, the change in push rod stroke is less than the stroke stagnation threshold, the change in yaw angle of magnetic chuck wheel is less than the angle response threshold, and the target output torque is less than the predicted value of steering resistance torque during the simulation, this time step is marked as the jamming accident time point, and the corresponding target yaw angle, target drive stroke, target steering speed, target output torque, predicted value of motor current, predicted value of steering resistance torque, and amount of yaw angle not completed are recorded, thus forming the jamming accident simulation data.

[0026] After obtaining the simulation data of the jamming accident, the contribution of each control parameter in the steering drive baseline scheme is evaluated parameter by parameter. Specifically, the system keeps three parameters—target yaw angle, target drive stroke, target steering speed, and target output torque—unchanged, and only increases or decreases the remaining parameter according to a preset disturbance amplitude. The jamming accident simulation is then re-executed to obtain the change in jamming risk before and after the disturbance. The disturbance calculations are then performed sequentially for the target yaw angle, target drive stroke, target steering speed, and target output torque. The change in jamming risk caused by the disturbance of each parameter is divided by the sum of the changes in jamming risk caused by the disturbances of all parameters to obtain the jamming contribution value corresponding to that parameter. This forms a multi-parameter jamming contribution distribution, which includes the contribution values ​​of the target yaw angle, target drive stroke, target steering speed, and target output torque, representing the degree of influence of each control parameter on steering jamming risk. Subsequently, the control parameter with the highest contribution value is determined as the dominant jamming parameter, and a first jamming protection scheme set is generated around this dominant jamming parameter. When the contribution value of the target steering speed is the highest, multiple sets of steering speed limiting coefficients are generated according to the preset deceleration step size, and the target steering speed in the steering drive reference scheme is reduced accordingly. When the contribution value of the target output torque is the highest, multiple sets of starting force amplification coefficients are generated according to the preset force amplification step size, and the initial push-pull force of the servo linear electric cylinder is increased accordingly. When the contribution value of the target drive stroke is the highest, the target drive stroke is divided into multiple continuous sub-strokes according to the preset segmentation step size, so that the servo linear electric cylinder reaches the target drive stroke segment by segment. When the contribution value of the target yaw angle is the highest, the target yaw angle is divided into multiple continuous sub-angles according to the preset angle step size, so that the magnetic chuck completes the yaw adjustment segment by segment. The system combines each set of control parameters configured above with the overload shutdown current threshold, stroke stagnation judgment threshold, and reverse micro-retraction stroke to form a set of jamming protection schemes, and summarizes multiple sets of jamming protection schemes into the first jamming protection scheme set.

[0027] After obtaining the first set of jamming protection schemes, each jamming protection scheme in the first set is input into the steering jamming risk prediction model to obtain the corresponding jamming risk value, expected steering completion time, and expected yaw angle error, forming the first steering jamming risk set. The steering jamming risk prediction model can be constructed based on gradient boosting decision trees, support vector machines, BP neural networks, etc. Then, the first set of jamming protection schemes is optimized and subjected to neighborhood mutation based on the first steering jamming risk set. The steering jamming risk prediction model is then used again to calculate the jamming risk value, expected steering completion time, and expected yaw angle error of each neighborhood protection scheme. The system repeatedly executes neighborhood scheme generation, risk prediction, and optimal scheme selection until the difference between the jamming risk values ​​obtained from two consecutive optimizations is less than a preset convergence threshold. Finally, the converged optimal protection scheme is determined as the jamming protection mechanism. This jamming protection mechanism includes a starting boost coefficient, a steering speed limit coefficient, segmented stroke length, reverse micro-retreat stroke, stroke stall judgment threshold, and overload shutdown current threshold. Through the above process, directional protection configuration can be performed on the control parameters most prone to jamming in the steering drive baseline scheme. By optimizing the neighborhood, a jamming protection mechanism that takes into account low jamming risk, low yaw error and short steering time can be obtained, thereby improving the steering safety and execution reliability of the magnetic chuck under strong magnetic adsorption conditions.

[0028] Furthermore, based on the steering jamming risk prediction model, the first jamming protection scheme set is optimized in a neighborhood search to generate the jamming protection mechanism, including: Based on the steering jamming risk prediction model, the first jamming protection scheme set is subjected to steering jamming risk prediction to obtain a first steering jamming risk set; based on the first steering jamming risk set, the first jamming protection scheme set is optimized and screened to obtain a first potential jamming protection mechanism; based on the first potential jamming protection mechanism, the first jamming protection scheme set is subjected to neighborhood mutation to obtain a second jamming protection scheme set; based on the steering jamming risk prediction model, the first potential jamming protection mechanism is iteratively optimized and updated according to the second jamming protection scheme set to obtain the jamming protection mechanism.

[0029] Optionally, after obtaining the first set of jamming protection schemes, each jamming protection scheme in the first set of jamming protection schemes, along with the corresponding target yaw angle, target drive stroke, target steering speed, target output torque, current motor current, and current steering speed, are used as model inputs. The steering jamming risk prediction model predicts the steering jamming risk and outputs the jamming risk value, expected steering completion time, and expected yaw angle error corresponding to that set of jamming protection schemes. The system predicts all schemes in the first set of jamming protection schemes in the above manner to obtain the first set of steering jamming risks, which corresponds one-to-one with the first set of jamming protection schemes. Subsequently, jamming protection schemes with jamming risk values ​​greater than a preset risk threshold are removed, and jamming protection schemes with expected yaw angle errors greater than a preset angle error threshold are also removed. For the remaining jamming protection schemes, they are sorted in ascending order of jamming risk value. When two groups of schemes have the same jamming risk value, the scheme with the shorter expected steering completion time is selected first. When the expected steering completion time is also the same, the scheme with the smaller starting force boost coefficient is selected first to reduce the instantaneous load on the servo linear cylinder. After the above screening, the group of jamming protection schemes with the best ranking is determined as the first potential jamming protection mechanism. After obtaining the first potential jamming protection mechanism, a neighborhood mutation is performed on the first jamming protection scheme set, centered on the protection parameters within the first potential jamming protection mechanism. Specifically, while keeping the parameter types of the first potential jamming protection mechanism unchanged, positive and negative disturbances with preset neighborhood step sizes are applied to the starting boost coefficient, steering speed limiting coefficient, segmented stroke length, reverse micro-retreat stroke, and overload shutdown current threshold. For example, the starting boost coefficient is increased and decreased by one boost step from its current value; the steering speed limiting coefficient is increased and decreased by one speed limiting step from its current value; and the reverse micro-retreat stroke is increased and decreased by one micro-retreat step from its current value. By recombining the disturbed parameters and deleting combinations exceeding the allowable parameter range, a second jamming protection scheme set distributed around the first potential jamming protection mechanism is formed.

[0030] Subsequently, the system again invokes the steering jamming risk prediction model to predict the jamming risk of each scheme in the second jamming protection scheme set, obtaining the second steering jamming risk set. The optimal prediction result in the second steering jamming risk set is then compared with the prediction result corresponding to the first potential jamming protection mechanism. If a scheme in the second jamming protection scheme set has a lower jamming risk value, a predicted yaw angle error that meets a preset angle error threshold, and a predicted steering completion time that does not exceed a preset time threshold, that scheme is updated as a new potential jamming protection mechanism. If no scheme in the second jamming protection scheme set is superior to the first potential jamming protection mechanism, the first potential jamming protection mechanism remains unchanged, and the neighborhood step size is reduced to half of the previous round's neighborhood step size before continuing the optimization process. The system repeats the neighborhood mutation, risk prediction, scheme comparison, and mechanism update until the difference in jamming risk values ​​after two consecutive updates is less than a preset convergence threshold, or the neighborhood step size is less than a preset minimum step size. Finally, the last obtained potential jamming protection mechanism is determined as the jamming protection mechanism. Through the above process, the combination of protection parameters with low jamming risk, low yaw error and low execution load can be gradually approached based on the first jamming protection scheme set, so that the magnetic wheel steering module has a more stable anti-jamming control capability under strong magnetic adsorption conditions.

[0031] For the steering jamming risk prediction model, taking the multi-task BP neural network model as an example, the multi-task BP neural network model includes one input layer, two shared hidden layers, and three output nodes. The input layer receives each set of jamming protection schemes and corresponding steering state parameters from the first set of jamming protection schemes. The system assembles these parameters into a model input vector according to a preset order. Before the input vector enters the model, each input parameter is normalized according to the pre-saved parameter mean and parameter standard deviation to ensure that each input parameter is within a uniform numerical range. The first shared hidden layer of the multi-task BP neural network model is set to 64 neurons, and the second shared hidden layer is set to 32 neurons. Both shared hidden layers use the ReLU activation function to extract the nonlinear correlation features between jamming protection parameters, target steering parameters, and the current operating state. After the second shared hidden layer, three output nodes are connected. The first output node uses the Sigmoid activation function to output a jamming risk value ranging from 0 to 1; the second output node uses a linear output function to output the estimated steering completion time; and the third output node uses a linear output function to output the estimated yaw angle error. Therefore, the steering jamming risk prediction model can predict the jamming risk, completion time, and steering error for each input jamming protection scheme. During model training, training samples are extracted from the magnetic wheel steering drive log and jamming accident simulation data. Each training sample includes a set of executed or simulated jamming protection schemes, the corresponding target yaw angle, target drive stroke, target steering speed, target output torque, pre-execution motor current, and pre-execution steering speed, with the execution result used as the sample label. During training, the training samples are divided into training and validation sets in an 8:2 ratio. The Adam optimization algorithm is used to iteratively update the model parameters, with an initial learning rate of 0.001, a single training batch size of 32, and a maximum number of training rounds of 200. The model loss function consists of three parts: the jamming risk prediction error, the predicted steering completion time prediction error, and the predicted yaw angle error prediction error. The system weights and sums these three errors to obtain the total loss function. When the validation set loss no longer decreases after 20 consecutive training rounds, training is stopped, and the current model parameters are saved as the final parameters for the steering jamming risk prediction model.

[0032] like Figure 2 As shown, the training convergence process of the steering jamming risk prediction model is illustrated. With increasing training rounds, both the total loss of the training set and the total loss of the validation set show a continuous decreasing trend, gradually stabilizing in the later stages. This indicates that the model can effectively learn the mapping relationship between jamming protection parameters, steering target parameters, and jamming risk. Simultaneously, the validation set loss does not show a significant rebound, indicating that the model has good generalization ability and can be used for subsequent jamming risk prediction and protection parameter optimization.

[0033] Based on the foundation data of the wind turbine tower, a surface interference analysis was performed on the steering drive reference scheme to construct a surface interference compensation mechanism.

[0034] In one embodiment, surface interference risk parameters are calculated based on wind turbine tower foundation data and a steering drive baseline scheme. These parameters are then compared with preset interference safety conditions. When any margin is less than the corresponding safety threshold, an abnormal surface interference risk index is determined. Subsequently, the correspondence between the abnormal surface interference risk index and the steering drive baseline scheme is analyzed to determine if the anomaly is caused by an excessively large target yaw angle, excessive target drive stroke, or excessively high target steering speed. Simultaneously, the correspondence between the abnormal surface interference risk index and the wind turbine tower foundation data is analyzed to determine if the anomaly is related to increased tower curvature or weld protrusions. Afterward, surface interference risk suppression parameters are configured based on the above analysis results. When the anomaly is caused by an excessively large target yaw angle, the single yaw angle change is reduced; when the anomaly is caused by an excessively large target drive stroke, the single target drive stroke is reduced; when the anomaly is caused by an excessively high target steering speed, the target steering speed is reduced; and when the anomaly is related to tower curvature or weld protrusions, segmented steering control is triggered in advance. The system combines the corrected target yaw angle, target drive stroke, target steering speed, and segmented steering triggering conditions into a curved surface interference compensation mechanism, thereby reducing the structural interference risk of the magnetic wheel steering module in the curved surface environment of the wind turbine tower.

[0035] Furthermore, based on the wind turbine tower foundation data, a surface interference analysis is performed on the aforementioned steering drive reference scheme to construct a surface interference compensation mechanism, including: Based on the wind turbine tower foundation data and the steering drive benchmark scheme, surface interference risk parameters are calculated; anomaly detection is performed on the surface interference risk parameters according to preset interference safety conditions to determine surface interference risk anomaly indicators; correlation analysis is performed on the surface interference risk anomaly indicators and the steering drive benchmark scheme to establish a first relationship of interference risk anomaly; correlation analysis is performed on the surface interference risk anomaly indicators and the wind turbine tower foundation data to establish a second relationship of interference risk anomaly; surface interference risk suppression parameters are configured based on the first relationship of interference risk anomaly and the second relationship of interference risk anomaly to generate the surface interference compensation mechanism.

[0036] Preferably, after obtaining the steering drive baseline scheme, the system uses wind turbine tower foundation data to perform surface interference analysis on the magnetic wheel steering module. This wind turbine tower foundation data includes the outer radius of the tower wall at the robot's current height, the central axis of the tower wall, the normal direction of the tower wall, the weld protrusion height, and the robot's current attachment posture. The system performs segmented prediction of the steering process for the magnetic wheel steering module based on the steering drive baseline scheme. Then, based on the prediction results, it performs temporal correlation envelope fitting to establish a steering process envelope space. Using this steering process envelope space as a basis, surface interference risk parameters are obtained through surface constraint projection and margin identification. These surface interference risk parameters include the arm swing margin, the steering axis offset margin, and the steering actuator posture margin. Subsequently, anomaly detection is performed on the surface interference risk parameters according to preset interference safety conditions. These preset interference safety conditions are set as follows: the arm swing margin is greater than a first safety threshold, the steering axis offset margin is greater than a second safety threshold, and the steering actuator posture margin is greater than a third safety threshold. When the wheel arm swing margin is less than or equal to the first safety threshold, a wheel arm swing anomaly index is generated; when the steering axis offset margin is less than or equal to the second safety threshold, a steering axis offset anomaly index is generated; when the steering actuator attitude margin is less than or equal to the third safety threshold, a steering actuator attitude anomaly index is generated. The surface interference risk anomaly index consists of the above anomaly indices and records the time of anomaly occurrence, the corresponding target yaw angle, the corresponding target drive stroke, the corresponding target steering speed, and the corresponding margin deficiency.

[0037] After determining the surface interference risk anomaly index, a correlation analysis was performed between the surface interference risk anomaly index and the steering drive benchmark scheme. Specifically, the impact of target yaw angle, target drive stroke, and target steering speed on three types of margin deficiencies was analyzed. When an increase in the target yaw angle leads to a decrease in the boom swing margin or steering axis offset margin, a relationship was established between the target yaw angle and the corresponding margin deficiencies; when an increase in the target drive stroke leads to a decrease in the steering actuator attitude margin, a relationship was established between the target drive stroke and the steering actuator attitude margin deficiencies; when an increase in the target steering speed leads to a decrease in the boom swing margin or steering actuator attitude margin, a relationship was established between the target steering speed and the corresponding margin deficiencies. This yielded the first relationship of interference risk anomalies, used to characterize the degree of influence of control parameters in the steering drive benchmark scheme on surface interference risk. Furthermore, a correlation analysis was performed between the surface interference risk anomaly index and wind turbine tower foundation data, specifically analyzing the impact of the tower outer wall radius, the tower wall normal direction, and the weld protrusion height on the three types of margin deficiencies. When a decrease in the outer radius of the tower wall leads to a reduction in the boom swing margin, a relationship is established between the outer radius of the tower wall and the insufficient boom swing margin. Similarly, when a change in the normal direction of the tower wall leads to a reduction in the steering axis offset margin, a relationship is established between the normal direction of the tower wall and the insufficient steering axis offset margin. Furthermore, when an increase in the weld protrusion height leads to a reduction in the steering actuator's attitude margin, a relationship is established between the weld protrusion height and the insufficient steering actuator's attitude margin. This yields a second relationship for interference risk anomalies, used to characterize the impact of wind turbine tower surface conditions on surface interference risk.

[0038] Subsequently, based on the first and second relationships of interference risk anomalies, surface interference risk suppression parameters are configured. Typically, when the boom swing margin is insufficient, the target steering speed at the corresponding moment is reduced and the number of segmented steering maneuvers is increased, allowing the boom more sufficient time to adapt to the swing on the tower's curved surface. When the steering axis offset margin is insufficient, the single target yaw angle change is reduced, keeping the steering vertical axis offset within a safe range. When the steering actuator attitude margin is insufficient, the single target drive stroke of the servo linear electric cylinder is reduced, and the push rod extension / retraction speed is limited, ensuring the steering actuator completes the push-pull action within the allowable attitude range. The system combines the corrected target steering speed, number of segmented steering maneuvers, single target yaw angle change, single target drive stroke, and push rod extension / retraction speed into a surface interference compensation mechanism. Through this process, surface interference risks can be identified around three parameters: boom swing margin, steering axis offset margin, and steering actuator attitude margin. Targeted compensation is then applied to the steering drive baseline scheme, thereby reducing the interference and jamming risks of the magnetic wheel steering module on the wind turbine tower's curved surface.

[0039] Furthermore, based on the wind turbine tower foundation data and the steering drive reference scheme, surface interference risk parameters are calculated, including: Based on the steering drive benchmark scheme, the magnetic wheel steering module is segmented to predict the steering process, resulting in multiple steering prediction attitude sequences. A temporal correlation envelope fitting is performed on these multiple steering prediction attitude sequences to construct a steering process envelope space. A surface constraint projection is performed on the steering process envelope space based on the wind turbine tower foundation data to obtain a surface constraint envelope space. The magnetic wheel steering module is then subjected to margin identification based on the surface constraint envelope space to obtain surface interference risk parameters, which include wheel arm swing margin, steering axis offset margin, and steering actuator attitude margin.

[0040] Optionally, when predicting the steering process segment by segment for the magnetic wheel steering module based on the steering drive baseline scheme, the target yaw angle is first subtracted from the current yaw angle to obtain the total steering angle difference. Then, the total steering angle difference is divided by a preset angle step size and rounded up to obtain the number of steering segments. Subsequently, the total steering angle difference is divided by the number of steering segments to obtain the angle change per segment. The predicted yaw angle for each steering stage is calculated by accumulating the angle change per segment according to the current yaw angle. At the same time, the target drive stroke is subtracted from the current drive stroke to obtain the total stroke difference. The total stroke difference is then proportionally distributed according to the ratio of the segment number of each steering stage to the number of steering segments to obtain the predicted drive stroke for each steering stage. The system then calculates the corresponding steering arm swing angle based on the preset calibration relationship between the predicted drive stroke and the steering arm length. Specifically, it reads the difference between the predicted drive stroke and the zero-position drive stroke, multiplies this difference by a preset stroke-arm angle conversion coefficient, and adds the zero-position arm angle to obtain the corresponding steering arm swing angle. This steering arm swing angle is used as the steering vertical axis deflection angle. Simultaneously, based on the change in the attachment position of the magnetic chuck to the turret surface at this predicted yaw angle, the system calculates the predicted swing angle of the arm relative to the robot body. Specifically, it calculates the surface attachment height difference between the magnetic chuck contact point and the reference contact point at this predicted yaw angle, divides the surface attachment height difference by the arm length, and obtains the predicted arm swing angle. The system arranges the predicted yaw angle, predicted drive stroke, steering arm swing angle, steering vertical axis deflection angle, and arm swing angle for each steering stage in chronological order to obtain multiple steering prediction attitude sequences. Subsequently, the predicted yaw angle, predicted drive stroke, steering arm swing angle, steering shaft deflection angle, and wheel arm swing angle are read sequentially between two adjacent steering stages. The numerical difference between two adjacent steering stages is calculated, and then divided by the time interval between the two adjacent steering stages to obtain the rate of change of the corresponding parameters. Based on the rate of change of each parameter, the system performs linear point supplementation between two adjacent steering stages, so that the discrete predicted yaw angle, predicted drive stroke, steering arm swing angle, steering shaft deflection angle, and wheel arm swing angle form a continuous change sequence. Next, the maximum wheel arm swing angle, maximum steering shaft deflection angle, maximum predicted drive stroke, and maximum steering arm swing angle in the continuous change sequence are extracted, and all the continuous attitude points after point supplementation are combined to form the steering process envelope space. This steering process envelope space is used to represent the continuous motion boundary formed by wheel arm swing, steering shaft deflection, and extension and retraction of the steering actuator unit during the period when the magnetic chuck moves from the current yaw angle to the target yaw angle.

[0041] After obtaining the steering process envelope space, a surface-constrained projection is performed on the steering process envelope space based on the wind turbine tower foundation data. Specifically, a local surface coordinate system is first established with the current contact point of the magnetic chuck as the origin. The axial direction is along the tower height, the tangential direction is along the tower circumference, and the normal direction is along the tower radius. Then, the predicted yaw angle, predicted drive stroke, steering arm swing angle, steering shaft deflection angle, and wheel arm swing angle in the steering process envelope space are read sequentially over time. Based on the wheel arm length, steering shaft installation position, and servo linear cylinder installation position, the axial, tangential, and normal offsets of each attitude point relative to the magnetic chuck contact point are calculated. For the tangential offset, it is divided by the outer radius of the tower wall corresponding to the current height to obtain the surface angular displacement of the attitude point in the tower circumference direction. Then, combined with the normal direction of the tower wall, the normal offset is corrected to obtain the actual position of the attitude point under surface constraints. The system projects all attitude points onto the local surface coordinate system in the manner described above, and reassembles them in chronological order to obtain the surface constraint envelope space. This surface constraint envelope space is used to represent the continuous motion range of the magnetic chuck steering module under the curvature constraint of the outer wall of the tower.

[0042] Then, the wheel arm swing angle for each steering stage is extracted from the surface constraint envelope space, and the maximum wheel arm swing angle is compared with the allowable wheel arm swing limit. The difference between the allowable wheel arm swing limit and the maximum wheel arm swing angle is determined as the wheel arm swing margin. The system extracts the steering vertical axis deflection angle for each steering stage from the surface constraint envelope space, and calculates the maximum offset of the steering vertical axis relative to the normal direction of the tower wall, combining the wheel arm swing angle of the corresponding stage. The difference between the allowable offset limit of the steering axis and the maximum offset is then determined as the steering axis offset margin. In addition, the predicted drive stroke, steering rocker arm swing angle, and wheel arm swing angle for each steering stage are extracted from the surface constraint envelope space. Based on the predicted drive stroke, it is determined whether the servo linear cylinder is close to the maximum or minimum stroke. Based on the steering rocker arm swing angle, it is determined whether the angle between the push rod and the steering rocker arm is close to the allowable angle limit. Based on the wheel arm swing angle, it is determined whether the installation posture of the servo linear cylinder after swinging with the wheel arm is close to the allowable posture limit. The system determines the minimum value among the aforementioned travel margin, included angle margin, and installation attitude margin as the steering actuator attitude margin. This yields the surface interference risk parameters composed of wheel arm swing margin, steering axis offset margin, and steering actuator attitude margin. Through this process, the predicted yaw angle, predicted drive travel, steering rocker arm swing angle, steering vertical shaft deflection angle, and wheel arm swing angle can be uniformly mapped to three interference risk margins, providing a clear calculation basis for the subsequent surface interference compensation mechanism.

[0043] Based on the aforementioned steering drive benchmark scheme, an adsorption resistance compensation analysis is performed, an adsorption resistance compensation mechanism is constructed, and the steering drive benchmark scheme is synergistically optimized by combining the aforementioned jamming protection mechanism and the aforementioned surface interference compensation mechanism to obtain an optimized steering drive scheme.

[0044] In one embodiment, after obtaining the steering drive benchmark scheme, the system performs adsorption resistance compensation analysis based on the steering drive benchmark scheme. That is, it uses the steering drive benchmark scheme to predict the adsorption state parameter set and contact state parameter set between the magnetic wheel steering module and the wind turbine tower, then identifies multidimensional resistance characteristics based on these two parameter sets, and uses the identified multidimensional resistance characteristics to perform segmented compensation, thus constructing an adsorption resistance compensation mechanism. Subsequently, the system applies the adsorption resistance compensation mechanism, the jamming protection mechanism, and the surface interference compensation mechanism to the steering drive baseline scheme. First, the surface interference compensation mechanism corrects the target yaw angle, target drive stroke, and target steering speed to ensure that the magnetic wheel steering module meets the safety requirements for wheel arm swing margin, steering axis offset margin, and steering actuator attitude margin. Then, the adsorption resistance compensation mechanism increases the target push-pull force and target output torque at the corresponding stage, enabling the magnetic wheel to overcome the steering resistance generated by strong magnetic adsorption. Next, the jamming protection mechanism limits the corrected target steering speed, target push-pull force, and motor current threshold. When the compensated push-pull force exceeds the allowable range, the target steering speed is reduced and the number of segmented steering steps is increased. When the compensated motor current approaches the overload shutdown current threshold, reverse micro-retardation or pause protection is executed. The system combines the target yaw angle, target drive stroke, target steering speed, target output torque, compensated push-pull force, number of segmented steering steps, and protection thresholds after the above-mentioned collaborative optimization into a steering drive optimization scheme. Through the above process, under the premise of ensuring the safety of curved surface interference and jamming protection, the steering resistance caused by strong magnetic adsorption can be actively compensated, thereby improving the stability of magnetic wheel steering drive, steering accuracy and continuous operation reliability.

[0045] Furthermore, based on the aforementioned steering drive benchmark scheme, an adsorption resistance compensation analysis is performed, and an adsorption resistance compensation mechanism is constructed, including: Based on the steering drive benchmark scheme, the adsorption state parameter set and contact state parameter set between the magnetic wheel steering module and the wind turbine tower are predicted; based on the adsorption state parameter set and contact state parameter set, multi-dimensional resistance characteristics are identified, including static friction resistance characteristics, transition resistance characteristics and stable steering resistance characteristics; based on the multi-dimensional resistance characteristics, segmented compensation is performed to obtain the adsorption resistance compensation mechanism.

[0046] Preferably, when performing adsorption resistance compensation analysis based on the steering drive baseline scheme, the target yaw angle, target drive stroke, target steering speed, and target output torque in the steering drive baseline scheme are first read. The process of rotating the magnetic chuck from the current yaw angle to the target yaw angle is then divided into multiple continuous prediction stages according to a preset time step. For each prediction stage, based on the predicted yaw angle and predicted drive stroke of that stage, the normal distance between the center of the magnetic chuck and the local curved surface of the wind turbine tower is calculated. This normal distance is then subtracted from the baseline distance when the magnetic chuck is normally attached, yielding the change in attachment distance for that stage. Subsequently, based on the pre-calibrated correspondence between the change in attachment distance and the normal adsorption force, the normal adsorption force for that stage is obtained. The normal adsorption forces of two adjacent prediction stages are then subtracted to obtain the change in adsorption force. Finally, the normal adsorption force for that stage is multiplied by the wind turbine tower wall friction coefficient and then by the magnetic chuck contact radius to obtain the adsorption resistance torque for that stage, thus forming a set of adsorption state parameters. Simultaneously, the predicted contact position of the magnetic chuck on the curved surface of the wind turbine tower is determined based on the predicted yaw angle of this stage. The contact pressure of this stage is obtained by dividing the normal adsorption force by the contact area between the magnetic chuck and the tower wall. The contact pressure of two adjacent predicted stages is then subtracted to obtain the change in contact pressure. Finally, the difference in yaw angle between two adjacent predicted stages is divided by the time step to obtain the change in steering speed, thus forming a set of contact state parameters.

[0047] Subsequently, based on the obtained adsorption state parameter set and contact state parameter set, multidimensional resistance characteristics are identified. When the current steering speed of the magnetic chuck is zero or close to zero, and the target driving stroke has begun to change, this stage is identified as the initial steering stage, and the maximum initial resistance torque calculated from the normal adsorption force, wall friction coefficient, and contact radius at this time is determined as the static friction resistance characteristic. When the magnetic chuck starts rotating from rest and the steering speed gradually increases, the difference in adsorption resistance torque between adjacent moments is divided by the time interval to obtain the resistance change rate, and the resistance change rate and the change in steering speed are determined as the transition resistance characteristic. When the steering speed of the magnetic chuck reaches the target steering speed and remains within the preset fluctuation range, the average value of the adsorption resistance torque at multiple consecutive moments is taken to obtain the stable steering resistance characteristic. Then, segmented compensation is performed based on the static friction resistance characteristics, transitional resistance characteristics, and stable steering resistance characteristics. Specifically, for the initial steering phase, the maximum initial resistance torque corresponding to the static friction resistance characteristics is compared with the target output torque, and the insufficient part is converted into the starting compensation thrust of the servo linear electric cylinder and added to the initial push-pull force. For the transitional steering phase, the compensation thrust is gradually adjusted according to the resistance change rate, so that the push-pull force of the servo linear electric cylinder increases or decreases smoothly with the resistance change. For the stable steering phase, a constant compensation thrust is generated based on the average resistance torque corresponding to the stable steering resistance characteristics, so that the magnetic chuck maintains the target steering speed. Finally, the system combines the starting compensation thrust, transitional compensation thrust, stable compensation thrust, and the corresponding compensation execution sequence into an adsorption resistance compensation mechanism. Through the above process, segmented compensation can be performed for different resistance states of the magnetic chuck from start-up, acceleration to stable steering, reducing the impact of strong magnetic adsorption on steering execution accuracy and steering continuity.

[0048] In summary, the embodiments of this application have at least the following technical effects: First, motion target calculation is performed on the crawling robot of the wind turbine tower to determine the motion target vector. Based on this vector, multi-degree-of-freedom control mapping is applied to the magnetic wheel steering module of the crawling robot to obtain a steering drive baseline scheme. Then, a steering jamming risk prediction model is used to optimize the jamming protection parameters of the baseline scheme, establishing a jamming protection mechanism. Next, surface interference analysis is performed on the baseline scheme based on wind turbine tower data to construct a surface interference compensation mechanism. Finally, adsorption resistance compensation analysis is performed on the baseline scheme to construct an adsorption resistance compensation mechanism. The jamming protection mechanism and the surface interference compensation mechanism are combined to perform collaborative optimization of the baseline scheme, obtaining an optimized steering drive scheme. This solves the technical problems of high steering resistance, easy jamming, and surface interference in the magnetic wheel steering under strong magnetic adsorption and variable curvature wall conditions, leading to low steering control accuracy and poor stability of the crawling robot. It achieves the technical effect of improving steering drive stability, accuracy, and environmental adaptability by implementing jamming protection, surface interference compensation, and adsorption resistance compensation during the magnetic wheel steering process.

[0049] Example 2, based on the same inventive concept as the magnetic wheel steering drive method for the wind turbine tower crawling robot in the foregoing examples, such as... Figure 3 As shown, this application provides a magnetic wheel steering drive device for a wind turbine tower crawling robot, wherein the device includes: Control mapping component 11: Performs motion target calculation on the crawling robot of the wind turbine tower, determines the motion target vector, and performs multi-degree-of-freedom control mapping on the magnetic wheel steering module of the crawling robot based on the motion target vector to obtain a steering drive benchmark scheme; Jamming protection component 12: Optimizes the jamming protection parameters of the steering drive benchmark scheme through a steering jamming risk prediction model to establish a jamming protection mechanism; Curved surface interference component 13: Performs curved surface interference analysis on the steering drive benchmark scheme based on the basic data of the wind turbine tower to construct a curved surface interference compensation mechanism; Collaborative optimization component 14: Performs adsorption resistance compensation analysis on the steering drive benchmark scheme to construct an adsorption resistance compensation mechanism, and performs collaborative optimization of the steering drive benchmark scheme in combination with the jamming protection mechanism and the curved surface interference compensation mechanism to obtain a steering drive optimization scheme.

[0050] Furthermore, the control mapping component 11 is used to perform the following methods: Feature association analysis is performed on the magnetic wheel steering drive log set of the crawling robot to construct a multi-degree-of-freedom control mapping relationship, which includes yaw angle mapping relationship, drive stroke mapping relationship, drive torque mapping relationship, and steering speed mapping relationship. Based on the multi-degree-of-freedom control mapping relationship, a steering control mapping map is constructed. The installation position of the magnetic wheel steering module and the motion target vector are input into the steering control mapping map to obtain a matching steering control set. The matching steering control set is then filtered by trigger frequency to output the steering drive benchmark scheme.

[0051] Furthermore, the control mapping component 11 is used to perform the following methods: The magnetic wheel steering module includes a wheel arm, a steering support, a magnetic wheel, a steering vertical shaft, a steering rocker arm, and a steering execution unit. One end of the wheel arm is hinged to the main body of the crawling robot. The steering support is rotatably mounted on the other end of the wheel arm via the steering vertical shaft. The magnetic wheel is rotatably mounted on the steering support. The steering rocker arm is connected to the steering vertical shaft. The steering execution unit is drivenly connected to the steering rocker arm.

[0052] Furthermore, the control mapping component 11 is used to perform the following methods: The steering actuator includes a servo linear electric cylinder. The cylinder body end of the servo linear electric cylinder is connected to the wheel arm via a first movable connector, and the push rod end of the servo linear electric cylinder is connected to the steering rocker arm via a second movable connector.

[0053] Furthermore, the jamming protection component 12 is used to perform the following method: A jamming accident simulation is performed based on the steering drive benchmark scheme to obtain jamming accident simulation data; a parameter-by-parameter contribution evaluation is performed on the steering drive benchmark scheme based on the jamming accident simulation data to obtain a multi-parameter jamming contribution distribution; jamming protection parameters are configured on the steering drive benchmark scheme based on the multi-parameter jamming contribution distribution to obtain a first jamming protection scheme set; neighborhood optimization is performed on the first jamming protection scheme set based on the steering jamming risk prediction model to generate the jamming protection mechanism.

[0054] Furthermore, the jamming protection component 12 is used to perform the following method: Based on the steering jamming risk prediction model, the first jamming protection scheme set is subjected to steering jamming risk prediction to obtain a first steering jamming risk set; based on the first steering jamming risk set, the first jamming protection scheme set is optimized and screened to obtain a first potential jamming protection mechanism; based on the first potential jamming protection mechanism, the first jamming protection scheme set is subjected to neighborhood mutation to obtain a second jamming protection scheme set; based on the steering jamming risk prediction model, the first potential jamming protection mechanism is iteratively optimized and updated according to the second jamming protection scheme set to obtain the jamming protection mechanism.

[0055] Furthermore, the curved surface interference component 13 is used to perform the following method: Based on the wind turbine tower foundation data and the steering drive benchmark scheme, surface interference risk parameters are calculated; anomaly detection is performed on the surface interference risk parameters according to preset interference safety conditions to determine surface interference risk anomaly indicators; correlation analysis is performed on the surface interference risk anomaly indicators and the steering drive benchmark scheme to establish a first relationship of interference risk anomaly; correlation analysis is performed on the surface interference risk anomaly indicators and the wind turbine tower foundation data to establish a second relationship of interference risk anomaly; surface interference risk suppression parameters are configured based on the first relationship of interference risk anomaly and the second relationship of interference risk anomaly to generate the surface interference compensation mechanism.

[0056] Furthermore, the curved surface interference component 13 is used to perform the following method: Based on the steering drive benchmark scheme, the magnetic wheel steering module is segmented to predict the steering process, resulting in multiple predicted steering attitude sequences. A temporal correlation envelope fitting is performed on these multiple predicted steering attitude sequences to construct a steering process envelope space. A surface constraint projection is performed on the steering process envelope space based on the wind turbine tower foundation data to obtain a surface constraint envelope space. The magnetic wheel steering module is then subjected to margin identification based on the surface constraint envelope space to obtain surface interference risk parameters, which include wheel arm swing margin, steering axis offset margin, and steering actuator attitude margin.

[0057] Furthermore, the collaborative optimization component 14 is used to perform the following method: Based on the steering drive benchmark scheme, the adsorption state parameter set and contact state parameter set between the magnetic wheel steering module and the wind turbine tower are predicted; based on the adsorption state parameter set and contact state parameter set, multi-dimensional resistance characteristics are identified, including static friction resistance characteristics, transition resistance characteristics and stable steering resistance characteristics; based on the multi-dimensional resistance characteristics, segmented compensation is performed to obtain the adsorption resistance compensation mechanism.

[0058] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A magnetic wheel steering drive method for a wind turbine tower crawling robot, characterized in that, The method includes: The motion target is calculated for the crawling robot of the wind turbine tower, the motion target vector is determined, and the magnetic wheel steering module of the crawling robot is mapped with multiple degrees of freedom according to the motion target vector to obtain the steering drive reference scheme. The steering drive benchmark scheme is optimized and configured with locking protection parameters using a steering locking risk prediction model to establish a locking protection mechanism. This includes: simulating a locking accident based on the steering drive benchmark scheme to obtain locking accident simulation data; evaluating the parameter contribution of the steering drive benchmark scheme based on the locking accident simulation data to obtain a multi-parameter locking contribution distribution; configuring locking protection parameters for the steering drive benchmark scheme based on the multi-parameter locking contribution distribution to obtain a first locking protection scheme set; and performing neighborhood optimization on the first locking protection scheme set using the steering locking risk prediction model to generate the locking protection mechanism. Based on the wind turbine tower foundation data, a surface interference analysis is performed on the steering drive benchmark scheme to construct a surface interference compensation mechanism. This includes: calculating surface interference risk parameters based on the wind turbine tower foundation data and the steering drive benchmark scheme; performing anomaly detection on the surface interference risk parameters according to preset interference safety conditions to determine surface interference risk anomaly indicators; performing correlation analysis between the surface interference risk anomaly indicators and the steering drive benchmark scheme to establish a first relationship of interference risk anomaly; performing correlation analysis between the surface interference risk anomaly indicators and the wind turbine tower foundation data to establish a second relationship of interference risk anomaly; configuring surface interference risk suppression parameters based on the first and second relationships of interference risk anomaly; and generating the surface interference compensation mechanism. Based on the aforementioned steering drive benchmark scheme, an adsorption resistance compensation analysis is performed, an adsorption resistance compensation mechanism is constructed, and the steering drive benchmark scheme is synergistically optimized by combining the aforementioned jamming protection mechanism and the aforementioned surface interference compensation mechanism to obtain an optimized steering drive scheme.

2. The magnetic wheel steering drive method for a wind turbine tower crawling robot as described in claim 1, characterized in that, Based on the motion target vector, the magnetic wheel steering module of the crawling robot is subjected to multi-degree-of-freedom control mapping to obtain a steering drive reference scheme, including: Based on the magnetic wheel steering drive log set of the crawling robot, feature association analysis is performed to construct a multi-degree-of-freedom control mapping relationship, which includes yaw angle mapping relationship, drive stroke mapping relationship, drive torque mapping relationship and steering speed mapping relationship; Based on the aforementioned multi-degree-of-freedom control mapping relationship, a steering control mapping map is constructed; The installation position of the magnetic wheel steering module and the motion target vector are input into the steering control mapping map to obtain the matching steering control set; The matching steering control set is filtered by trigger frequency, and the steering drive benchmark scheme is output.

3. The magnetic wheel steering drive method for a wind turbine tower crawling robot as described in claim 1, characterized in that, Based on the steering jamming risk prediction model, the first jamming protection scheme set is optimized in a neighborhood search to generate the jamming protection mechanism, including: Based on the steering jamming risk prediction model, the first jamming protection scheme set is subjected to steering jamming risk prediction to obtain the first steering jamming risk set. Based on the first steering jamming risk set, the first jamming protection scheme set is optimized and screened to obtain the first potential jamming protection mechanism; Based on the first potential jamming protection mechanism, the first jamming protection scheme set is subjected to neighborhood mutation to obtain the second jamming protection scheme set. Based on the steering jamming risk prediction model, the first potential jamming protection mechanism is iteratively optimized and updated according to the second jamming protection scheme set to obtain the jamming protection mechanism.

4. The magnetic wheel steering drive method for a wind turbine tower crawling robot as described in claim 1, characterized in that, Based on the wind turbine tower foundation data and the steering drive reference scheme, the surface interference risk parameters are calculated, including: Based on the steering drive baseline scheme, the steering process of the magnetic wheel steering module is segmented and predicted to obtain multiple steering prediction posture sequences. Based on the multiple steering prediction attitude sequences, a temporal correlation envelope fitting is performed to construct the steering process envelope space; Based on the wind turbine tower foundation data, a surface constraint projection is performed on the envelope space of the turning process to obtain the surface constraint envelope space; The magnetic wheel steering module is subjected to margin identification based on the surface constraint envelope space to obtain the surface interference risk parameters, which include wheel arm swing margin, steering axis offset margin, and steering actuator attitude margin.

5. The magnetic wheel steering drive method for a wind turbine tower crawling robot as described in claim 1, characterized in that, Based on the aforementioned steering drive benchmark scheme, an adsorption resistance compensation analysis is performed, and an adsorption resistance compensation mechanism is constructed, including: Based on the steering drive baseline scheme, predict the set of adsorption state parameters and the set of contact state parameters between the magnetic wheel steering module and the wind turbine tower; Based on the adsorption state parameter set and the contact state parameter set, multidimensional resistance characteristics are identified, including static friction resistance characteristics, transition resistance characteristics, and stable turning resistance characteristics. The adsorption resistance compensation mechanism is obtained by performing segmented compensation based on the multidimensional resistance characteristics.

6. The magnetic wheel steering drive method for a wind turbine tower crawling robot as described in claim 1, characterized in that, The magnetic wheel steering module includes a wheel arm, a steering support, a magnetic wheel, a steering shaft, a steering rocker arm, and a steering execution unit. One end of the wheel arm is hinged to the main body of the crawling robot. The steering support is rotatably mounted on the other end of the wheel arm via the steering shaft. The magnetic wheel is rotatably mounted on the steering support. The steering rocker arm is connected to the steering shaft. The steering execution unit is drivenly connected to the steering rocker arm.

7. The magnetic wheel steering drive method for a wind turbine tower crawling robot as described in claim 6, characterized in that, The steering actuator includes a servo linear electric cylinder. The cylinder body of the servo linear electric cylinder is connected to the wheel arm via a first movable connector, and the push rod end of the servo linear electric cylinder is connected to the steering rocker arm via a second movable connector.

8. A magnetic wheel steering drive device for a wind turbine tower crawling robot, characterized in that, The apparatus for implementing the magnetic wheel steering drive method for a wind turbine tower climbing robot according to any one of claims 1-7 comprises: Control mapping component: Performs motion target calculation on the crawling robot of the wind turbine tower, determines the motion target vector, and performs multi-degree-of-freedom control mapping on the magnetic wheel steering module of the crawling robot based on the motion target vector to obtain the steering drive reference scheme; The jamming protection component optimizes the jamming protection parameters of the steering drive benchmark scheme through a steering jamming risk prediction model to establish a jamming protection mechanism. Curved surface interference component: Based on the wind turbine tower foundation data, a curved surface interference analysis is performed on the steering drive reference scheme to construct a curved surface interference compensation mechanism; Collaborative optimization component: Based on the aforementioned steering drive benchmark scheme, an adsorption resistance compensation analysis is performed to construct an adsorption resistance compensation mechanism. Combined with the aforementioned jamming protection mechanism and the aforementioned surface interference compensation mechanism, the collaborative optimization of the steering drive benchmark scheme is executed to obtain a steering drive optimization scheme.

Images