Method and system for evaluating motion stability of a climbing robot facing a large curved surface

Abstract

The application discloses a large-scale curved surface-oriented climbing robot motion stability evaluation method and system, relates to the technical field of robot motion stability evaluation, and comprises the following steps: collecting multi-source motion data in the robot climbing motion process; establishing a robot motion simulation model; performing analysis and fusion of the static position stability margin and the static force stability margin to obtain a static stability index; performing comprehensive analysis to obtain a dynamic stability index; and performing weighted summation on the static stability index and the dynamic stability index to obtain a comprehensive stability evaluation value. The application solves the technical problem that the existing climbing robot stability evaluation does not combine the dynamic distribution of the curvature of a curved surface, the fixed fusion mode of dynamic and static indexes, and the low accuracy of the evaluation result, and achieves the technical effects of realizing adaptive stability evaluation of the curvature of a curved surface and improving the fitting degree and accuracy of the robot motion stability evaluation result under a large-scale variable-curvature curved surface.

Description

Technical Field

[0001] This invention relates to the field of robot motion stability evaluation technology, specifically to a method and system for evaluating the motion stability of climbing robots for large curved surfaces. Background Technology

[0002] Existing large curved surface climbing robots often operate in scenarios where the curvature of the surface continuously changes. Traditional stability evaluation schemes only use fixed weights to integrate static and dynamic stability indicators, without dynamically adjusting the influence ratio of the two types of indicators according to the severity of the surface curvature. Furthermore, they rely solely on force or geometric position for stability analysis, lacking comprehensive evaluation methods that combine simulation models to simultaneously quantify motion deviations and local adsorption instability of the wheels. This results in a large deviation between the stability assessment results and the actual operating state of the robot, making it difficult to accurately reflect the real motion stability risks of the robot under curved surfaces with varying curvature.

[0003] Existing stability evaluation methods for climbing robots do not incorporate dynamic weighting of surface curvature and use a fixed method for fusing dynamic and static indicators, resulting in low accuracy of evaluation results. Summary of the Invention

[0004] This application provides a method and system for evaluating the motion stability of climbing robots for large curved surfaces, which addresses the technical problem of low accuracy in existing stability evaluations of climbing robots due to the lack of dynamic weight allocation based on surface curvature and the fixed method of fusion of dynamic and static indicators.

[0005] In view of the above problems, this application provides a method and system for evaluating the motion stability of climbing robots for large curved surfaces.

[0006] The first aspect of this application provides a method for evaluating the motion stability of a climbing robot for large curved surfaces, the method comprising: Multi-source motion data is collected during the robot's climbing motion; a motion simulation model of the robot facing a surface with variable curvature is established; based on the motion simulation model, static position stability margin and static force stability margin are analyzed and fused to obtain a static stability index; based on the motion simulation model and the multi-source motion data, a comprehensive analysis based on motion deviation and local instability is performed to obtain a dynamic stability index; the degree of curvature of the current surface is judged based on the real-time acquired rate of change of surface curvature radius, the weight ratio of the static stability index and the dynamic stability index is determined, and the static stability index and the dynamic stability index are weighted and summed to obtain a comprehensive stability evaluation value.

[0007] A second aspect of this application provides a motion stability evaluation system for climbing robots targeting large curved surfaces, the system comprising: The system includes a multi-source motion data acquisition module for collecting multi-source motion data during robot climbing; a motion simulation model establishment module for establishing a motion simulation model of the robot facing a variable curvature surface; a static stability index acquisition module for analyzing and fusing static position stability margin and static force stability margin based on the motion simulation model to obtain a static stability index; a comprehensive analysis module for performing a comprehensive analysis based on motion deviation and local instability state based on the motion simulation model and the multi-source motion data to obtain a dynamic stability index; and an evaluation value acquisition module for judging the degree of curvature of the current surface based on the real-time acquired rate of change of surface curvature radius, determining the weight ratio of the static stability index and the dynamic stability index, and performing a weighted summation of the static stability index and the dynamic stability index to obtain a comprehensive stability evaluation value.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: Multi-source motion data is collected during the robot's climbing motion; a motion simulation model of the robot facing a variable curvature surface is established; static position stability margin and static force stability margin are analyzed and fused to obtain static stability indices; a comprehensive analysis based on motion deviation and local instability is performed to obtain dynamic stability indices; the weight ratio of the static stability indices and the dynamic stability indices is determined, and the static stability indices and the dynamic stability indices are weighted and summed to obtain a comprehensive stability evaluation value. This achieves the technical effect of realizing adaptive stability evaluation of curved surfaces, improving the fit and accuracy of robot motion stability evaluation results under large variable curvature surfaces. Attached Figure Description

[0009] 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.

[0010] Figure 1 This is a schematic diagram of the motion stability evaluation method for climbing robots oriented towards large curved surfaces, provided in an embodiment of this application.

[0011] Figure 2 This is a schematic diagram of the structure of a motion stability evaluation system for climbing robots oriented towards large curved surfaces, provided in an embodiment of this application.

[0012] Figure 3 A line graph showing the relationship between the rate of change of surface curvature radius and the weights of dynamic and static stability indices provided in the embodiments of this application.

[0013] Figure labeling: 10 Multi-source motion data acquisition module, 20 Motion simulation model establishment module, 30 Static stability index acquisition module, 40 Comprehensive analysis module, 50 Evaluation value acquisition module. Detailed Implementation

[0014] This application provides a method and system for evaluating the motion stability of climbing robots for large curved surfaces, which addresses the technical problem of low accuracy in existing stability evaluations of climbing robots due to the lack of dynamic weight allocation based on surface curvature and the fixed method of fusion of dynamic and static indicators.

[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0016] Example 1, as Figure 1 As shown, this application provides a method for evaluating the motion stability of climbing robots for large curved surfaces, the method comprising: Step S100: Collect multi-source motion data during the robot's climbing motion.

[0017] Specifically, the robot comprises a front drive housing, a rear drive housing, a central hinge connecting the front and rear drive housings, four independent cantilever arms, and four omnidirectional drive adsorption wheels. During the climbing operation, force sensors installed at the connection points between each independent cantilever arm and the drive adsorption wheel collect real-time data on the normal contact force between each drive adsorption wheel and the large curved surface. Encoders installed at the drive motors of each drive adsorption wheel synchronously collect real-time actual rotational speed data of the corresponding drive adsorption wheel. Angle sensors arranged on the steering mechanisms of each drive adsorption wheel collect real-time actual steering angle data of each wheel. The three types of data are collected synchronously and sequentially, and the normal contact force data, actual rotational speed data, and actual steering angle data are summarized to form the multi-source motion data of this climbing process.

[0018] Step S200: Establish a motion simulation model of the robot facing a variable curvature surface.

[0019] Specifically, the first motion modeling data is collected corresponding to the torsional motion of the robot's front and rear drive bodies around the central hinge. Then, the mechanical transmission relationship corresponding to the oscillation of each independent cantilever around its own rotary joint is collected to construct the second motion modeling data that can characterize the microscopic contact state between each omnidirectional drive adsorption wheel and the large curved surface. Next, the preset setting parameters of the independent speed and steering coordination control of the four omnidirectional drive adsorption wheels are read to generate the third motion modeling data. The first, second, and third motion modeling data are integrated to complete the coupling modeling of the robot's overall motion state. Finally, the robot motion simulation model is generated by adapting to the variable curvature surface working condition.

[0020] Step S300: Based on the motion simulation model, perform analysis and fusion of static position stability margin and static force stability margin to obtain static stability index.

[0021] Specifically, the robot's real-time climbing commands are input into the motion simulation model, which analyzes and outputs four contact points between the two sets of adsorption wheels of the front drive compartment and the two sets of adsorption wheels of the rear drive compartment and the curved surface. Using these four contact points as vertices, a support polygon is constructed. The robot's real-time posture is simultaneously acquired, and the overall center of mass is calculated. The positional deviation of the robot's center of mass projection point on the tangent plane of the curved surface relative to the center of the support polygon is calculated, and the static positional stability margin is quantitatively solved based on this positional deviation. At the same time, the normal contact force distribution between each drive adsorption wheel and the curved surface is output based on the motion simulation model, and the overturning moment and overturning moment of the whole machine are calculated respectively. The static force stability margin is determined by the ratio of the two. Finally, a fixed fusion weight is set, and the obtained static positional stability margin and static force stability margin are weighted and fused to calculate a unified quantitative static stability index.

[0022] Step S400: Based on the motion simulation model and the multi-source motion data, perform a comprehensive analysis based on motion deviation and local instability to obtain dynamic stability indices.

[0023] Specifically, the actual rotational speed and actual steering angle data of the four omnidirectional drive adsorption wheels are extracted from the multi-source motion data. These data are then compared with the corresponding expected rotational speed and expected steering angle output by the motion simulation model. The rotational speed deviation and steering deviation of each drive adsorption wheel are calculated and fused to obtain the single-wheel motion deviation. The maximum value among the four wheel motion deviations is taken as the overall motion deviation of the entire machine. Next, the normal contact force data of each wheel is extracted from the multi-source motion data. The rate of change of normal contact force of each drive adsorption wheel is calculated frame by frame. The rate of change of force is compared with a preset rate of change of force threshold. If the rate of change of force exceeds the threshold, the drive adsorption wheel is determined to be in a local unstable state. The proportion of local instability duration and the magnitude of exceeding the threshold of a single wheel are statistically analyzed. The degree of local instability of each drive adsorption wheel is calculated through a preset mapping rule. Subsequently, a comprehensive evaluation function is built. The overall motion deviation and the degree of local instability of the four wheels are uniformly normalized and then assigned corresponding evaluation weights. This completes the coupled comprehensive calculation of the two types of dynamic risk characteristics and finally outputs a dynamic stability index that can quantify the smoothness of the robot's dynamic operation.

[0024] Step S500: Determine the degree of curvature of the current surface based on the real-time obtained rate of change of the surface curvature radius, determine the weight ratio of the static stability index and the dynamic stability index, and perform a weighted summation of the static stability index and the dynamic stability index to obtain a comprehensive stability evaluation value.

[0025] Specifically, multiple curvature change thresholds are predefined, and a curvature-weight mapping relationship is constructed offline based on historical measured robot stability samples. This mapping relationship stores the basic weight allocation ratios of static and dynamic stability indices under different curvature change degrees. The curvature radius change rate of large curved surfaces is collected in real time, and this value is matched and retrieved with the thresholds at each level in the curvature-weight mapping relationship to output the static and dynamic index weights that are adapted to the current curvature severity of the curved surface. The static and dynamic stability indices obtained above are multiplied by their respective matched weights and then added to complete the weighted summation operation, and finally output a comprehensive stability evaluation value that can uniformly characterize the climbing stability of the robot as a whole.

[0026] In one possible implementation, step S100 further includes: Step S110: The robot includes a front drive compartment, a rear drive compartment, a central hinge connecting the front and rear drive compartments, four independent cantilever arms, and four omnidirectional drive adsorption wheels.

[0027] Specifically, the robot includes a front drive chamber, a rear drive chamber, a central hinge connecting the front and rear drive chambers, four independent cantilever arms, and four omnidirectional drive adsorption wheels. The specific structure and adaptive motion control mechanism are as follows: The front and rear drive chambers serve as the two core load-bearing units of the entire platform, connected by a central hinge. The hinge has a virtual central rotation axis that runs through the kinematic center of the front and rear chambers and is perpendicular to the climbing wall. The front and rear chambers can rotate relative to each other around this axis, forming a first-level adaptive mechanism that can compensate for the macroscopic curvature changes caused by the robot's longitudinal crossing of the tower. When the robot climbs the conical tower, the hinge rotation allows the two sets of adsorption wheels to adaptively adapt to the different diameters of the tower's curved surface, ensuring that the platform body and the tower's generatrix direction remain coordinated. Each independent cantilever arm is hinged to the chamber at one end via a rotary joint, with the rotary joint axis parallel to the robot's forward direction. The other end of the cantilever arm is rigidly fitted with an omnidirectional drive adsorption wheel, allowing the cantilever arm to swing downwards in a plane perpendicular to the forward direction, forming a second-level adaptive mechanism. By absorbing local micro-unevennesses or slight curvature changes such as weld seams and pits on the wall surface, the end adsorption wheels autonomously adhere to and vertically press against the local wall surface under the action of springs and gravity, maximizing the magnetic adsorption force and wall grip. Each omnidirectional drive adsorption wheel integrates an independent rotation drive module and a yaw steering module. The speed and steering angle of the four wheels can be individually adjusted, forming a third-level core motion control mechanism. The central controller calculates and coordinates the steering angle of the four wheels in real time, enabling the robot to achieve slip-free omnidirectional motion such as straight-line movement, arbitrary curvature turning, diagonal movement, and stationary rotation. This avoids the lateral slippage caused by traditional steering structures on vertical walls and significantly reduces the risk of dynamic instability. The entire robot relies on the coordinated work of a three-level mechanism: a central hinge, independent cantilever, and omnidirectional drive adsorption wheels. The central hinge adapts to the curvature of large-scale curved surfaces to provide basic stable support, the independent cantilever swings to optimize the local adhesion of a single wheel, and the independent drive steering of all wheels achieves precise vector motion control. The combination of these three mechanisms enables the robot to complete climbing operations stably and with high mobility on complex, continuously variable curvature large curved surfaces.

[0028] In one possible implementation, step S100 further includes: Step S120: Real-time acquisition of normal contact force data between each driving adsorption wheel and the curved surface using force sensors installed at the connection points between each independent cantilever and the driving adsorption wheel.

[0029] Step S130: The actual rotational speed data of each driving adsorption wheel is collected in real time by the encoder installed at the drive motor of each driving adsorption wheel.

[0030] Step S140: The actual steering angle data of each driving adsorption wheel is collected in real time by angle sensors installed at the steering mechanism of each driving adsorption wheel.

[0031] Step S150: The multi-source motion data is composed of normal contact force data, actual rotation speed data, and actual steering angle data.

[0032] Specifically, force sensors are installed at the assembly and connection positions of the omnidirectional drive adsorption wheels at the ends of each independent cantilever. During the robot's climbing operation along the large variable curvature surface, each drive adsorption wheel is subjected to the combined effects of the surface topography, cantilever swing, and central hinge torsion, resulting in different magnitudes of normal clamping force perpendicular to the wall. The force sensors at each point continuously and synchronously collect the raw data of the real-time changing normal contact force between the corresponding adsorption wheel and the surface. After unified timing alignment and noise preprocessing, the data is output as the core basic data for subsequent judgment of local instability and calculation of static force stability margin.

[0033] Encoders are installed on the output shafts of the rotary drive motors that are matched to the four omnidirectional drive adsorption wheels. During the robot's climbing of the variable curvature surface, the encoders rotate synchronously with the drive motors and continuously collect the real-time rotation pulse signals of the motor shafts. The real-time actual rotation speed of each drive adsorption wheel is obtained by counting the pulses. The rotation speed data of multiple wheels is synchronized and stored in sequence. The collected actual rotation speed data of the four wheels is used to compare with the expected rotation speed of the simulation model and to calculate the motion deviation of each adsorption wheel.

[0034] Angle sensors are installed on the yaw steering mechanism of each of the four omnidirectional drive adsorption wheels. During the robot's climbing operation on a large variable curvature surface, each angle sensor synchronously follows the steering mechanism to deflect, senses and collects the actual steering angle value corresponding to the current deflection state of each drive adsorption wheel in real time, and synchronously completes the time-series acquisition of angle data from multiple rounds and outputs the raw angle data.

[0035] The collected data on the normal contact force of each driving adsorption wheel, the actual rotation speed of each driving adsorption wheel collected by the encoder, and the actual steering angle of each driving adsorption wheel collected by the angle sensor are time-aligned and processed to unify the data format. After removing abnormal noise data generated during the acquisition process, the three types of synchronously acquired working condition data are integrated and packaged to form complete multi-source motion data that can be used for simulation comparison and stability margin analysis.

[0036] In one possible implementation, step S200 further includes: Step S210: Collect the first motion modeling data of the robot's front drive compartment and rear drive compartment torsion around the central hinge.

[0037] Step S220: Collect the mechanical transmission relationship of each independent cantilever swinging around the rotary joint, and establish the second motion modeling data characterizing the microscopic contact state between each driving adsorption wheel and the curved surface.

[0038] Step S230: Collect the coordinated control setting parameters of the independent rotation speed and steering of the four omnidirectional drive adsorption wheels, and establish the third motion modeling data.

[0039] Step S240: Model the overall motion state of the robot using the first motion modeling data, the second motion modeling data, and the third motion modeling data to obtain the motion simulation model.

[0040] Specifically, relying on the angle sensing component installed at the central hinge position, the relative torsional angles of the front drive compartment and the rear drive compartment are collected in real time. Combined with the overall machine's travel posture and the relative displacement changes of the front and rear compartments under different surface curvature conditions, the spatial pose, force distribution, and motion constraint parameters of the compartments corresponding to different torsional angles are recorded simultaneously. All measured torsional-related parameters are summarized and organized to form the first motion modeling data that can characterize the relative torsional motion law of the front and rear compartments around the hinge.

[0041] The cantilever tilt angle, cantilever spring force, and force variation law of the cantilever transmitted to the end omnidirectional driving adsorption wheel during the downward swing of each independent cantilever around the rotating joint were collected in sequence. The complete mechanical transmission correspondence was sorted out. Combined with the contact and pressing state of each adsorption wheel with the curved surface and the normal contact force fluctuation under different wall welds, pits, and small curvature abrupt changes, the cantilever pendulum dynamic parameters were bound to the micro-contact characteristics of the adsorption wheel one by one. The second motion modeling data that can completely describe the micro-contact state of each driving adsorption wheel with the curved surface was obtained by summarizing and organizing.

[0042] The robot reads the independent speed and steering angle commands of each drive adsorption wheel issued by the robot's central controller, as well as the cooperative constraint relationship, synchronous adjustment logic and linkage correction parameters between the four wheel speeds and steering in different motion modes such as straight walking, turning, diagonal movement and stationary rotation. All control setting parameters are collected in a complete manner to form the third motion modeling data for simulating and reproducing the robot's omnidirectional motion process.

[0043] A machine learning modeling scheme employing a multi-layer temporal LSTM neural network combined with an attention mechanism is adopted. The overall process is divided into three parts: dataset construction, model training, and model inference. First, a unified multi-channel temporal dataset is constructed by fusing the first, second, and third motion modeling data. The dataset is then divided into training and validation sets, and normalized and preprocessed using a temporal sliding window. Next, a temporal attention LSTM network with an encoder-decoder structure is built. The encoder uses a two-layer LSTM to extract three types of motion features: hinge torsion, cantilever contact, and wheel drive. A multi-head attention module is embedded to automatically assign feature weights to large-curvature hinge motion, microscopic cantilever contact, and omnidirectional wheel cooperative motion. The encoder outputs the robot's overall posture, contact forces of each wheel, and predicted motion state sequences of the wheel assembly. During the training phase, the mean square error loss function (MSE) is used, along with the Adam optimizer to iteratively update the network weights. The prediction error is continuously verified using validation set samples until the loss converges to a preset accuracy threshold, completing model training. In actual use, the current surface curvature and robot motion control commands are input to the network in real time. The trained attention LSTM model can quickly infer and output all kinematic and dynamic state variables of the entire machine, replacing traditional numerical iterative solutions. It directly generates a robot motion simulation model adapted to variable curvature surface conditions, which is used for the rapid calculation of subsequent static and dynamic stability indices.

[0044] In one possible implementation, step S300 further includes: Step S310: Input the actual motion commands during the robot's climbing motion into the motion simulation model, output the contact points between the two driving adsorption wheels of the front drive box and the curved surface and the two driving adsorption wheels of the rear drive box and the curved surface as four vertices to construct a support polygon, and calculate the static position stability margin with the support polygon as a reference.

[0045] Step S320: Obtain the normal contact force distribution between each driving adsorption wheel and the curved surface through the motion simulation model, calculate the ratio of anti-overturning moment to overturning moment, and determine the static force stability margin.

[0046] Step S330: The static position stability margin and the static force stability margin are weighted and fused to obtain the static stability index.

[0047] Specifically, the robot's current complete climbing motion command is input into the motion simulation model. The model combines the real-time surface curvature, central hinge torsion, and swing posture of each cantilever to obtain the four spatial contact coordinates of the two front wheels and two rear wheels in contact with the wall. The four contact points are projected onto the tangent plane of the current position surface, and the projection points are connected sequentially according to the arrangement of the front and rear wheels to form a two-dimensional quadrilateral support polygon. Simultaneously, the mass and spatial coordinates of all structural components of the robot are calculated to obtain the three-dimensional coordinates of the robot's center of mass. The center of mass is then vertically projected onto the same tangent plane to obtain the center of mass projection point. The vertical distances from the projection point to the four sides of the support polygon are calculated sequentially, and the vertical distance with the smallest value is selected as the critical stability distance. The ratio of this critical distance to the reference value of half the side length of the support polygon is normalized to obtain the dimensionless static position stability margin that can intuitively represent the safety margin of the overturning position.

[0048] Input real-time motion commands into the motion simulation model to calculate the normal contact forces of each of the four wheels. , , , The overturning boundary axis is determined by combining the support polygon, and the normal contact force of each support adsorption wheel is multiplied by the vertical force arm from the corresponding contact point to the overturning axis. The total anti-overturning moment is obtained by summing them up. Based on the total weight of the machine and the eccentricity from the center of mass to the overturning axis Calculate the total overturning moment ; Using formula S 力 = / By performing ratio calculations, the dimensionless value S is obtained. 力 This value is the static force stability margin. The larger the value, the higher the robot's anti-tipping force margin.

[0049] First, perform uniform interval normalization on the static position stability margin and the static force stability margin to eliminate the difference in the two types of margin dimensions, and then preset the position weight coefficient. Force weighting coefficient And satisfy Combining the two sets of fixed weight values ​​determined offline under static conditions of the tower surface, a fusion calculation formula was adopted. The normalized static position stability margin Static force stability margin After multiplying by their respective weights and summing them, a linear weighted fusion of the two types of static stability characteristics is completed, ultimately yielding a static stability index that balances the safety margin of geometric support and the anti-overturning capacity of the wheel surface under stress. In one possible implementation, step S310 further includes: Step S311: Collect the robot's real-time posture and calculate the robot's center of mass.

[0050] Step S312: Calculate the positional deviation of the robot's center of mass projection point on the curved tangent plane relative to the center of the supporting polygon.

[0051] Step S313: Determine the static position stability margin based on the position deviation.

[0052] Specifically, inertial measurement units mounted on the front and rear drive housings are used to collect real-time data on the three-axis attitude angles and spatial displacements of the entire machine. Simultaneously, the torsional angle of the central hinge and the swing angle of each independent cantilever slewing joint are read. The inherent mass and centroid coordinates of each component stored in the controller are retrieved. A global coordinate system is established using the climbing surface. The local centroids of the front drive housing, rear drive housing, four independent cantilever arms, and four omnidirectional drive adsorption wheels are sequentially transformed to spatial coordinates in the global coordinate system using coordinate transformation matrices. The formula for calculating the centroid of a point system is then used. In the formula Let the mass of the i-th component be... To correspond to the global centroid coordinates of the components, a weighted sum of all components is performed and then divided by the total mass of the whole machine to finally calculate the global three-dimensional centroid coordinates of the robot in real time.

[0053] First, the normal vector of the tangent plane at the corresponding position of the current curved surface is calculated based on the contact point coordinates. Then, the three-dimensional centroid coordinates of the entire machine are perpendicularly mapped to the tangent plane using an orthogonal projection algorithm to obtain the coordinates of the two-dimensional centroid projection point. ; Take the average of the coordinates of the four wheel surface contact projection points that constitute the supporting polygon, and calculate the coordinates of the geometric center of the supporting polygon. Then, using the Euclidean distance calculation formula... Solve and The planar distance between them is recorded, and the difference between the horizontal and vertical coordinates is recorded as the two-dimensional position deviation component of the plane. This distance value is the position deviation of the centroid projection relative to the center of the supporting polygon.

[0054] First, traverse the four sides of the supporting polygon, calculate the perpendicular distance from the centroid projection point to each side, and extract the minimum perpendicular safety distance. The equivalent half-width D of the circumscribed polygon is used as the reference for determining the length, and the calculation formula is adopted. Substitute the positional deviation of the centroid projection relative to the center of the polygon After performing the normalization operation, if the result is less than 0, it is directly assigned the value 0, thus obtaining the dimensionless static position stability margin. The larger the value, the higher the safety margin for the overturning position of the whole machine.

[0055] In one possible implementation, step S400 further includes: Step S410: Based on the real-time collected actual rotation speed data and actual steering angle data of each driving adsorption wheel, compare them with the expected rotation speed and expected steering angle output by the motion simulation model, calculate the motion deviation of each driving adsorption wheel, and take the maximum value of the motion deviation among the four driving adsorption wheels as the overall motion deviation.

[0056] Step S420: Based on the real-time collected data of the normal contact force of each driving adsorption wheel, calculate the rate of change of the normal contact force of each driving adsorption wheel, and calculate the degree of local instability of each driving adsorption wheel by statistically analyzing the local instability state.

[0057] Step S430: The overall motion deviation and the degree of local instability of each driving adsorption wheel are comprehensively evaluated to obtain the dynamic stability index.

[0058] Specifically, for each driving adsorption wheel, the actual rotational speed collected by the sensor is read. Actual steering angle Synchronously retrieve the expected rotational speed of the same wheel output from the motion simulation model. Desired steering angle through respectively Calculate the absolute deviation of the rotational speed. The weighted summation formula is used to calculate the absolute steering deviation. The single-wheel motion deviation of the i-th wheel is obtained, where , Based on the preset speed and steering deviation weighting coefficients, the corresponding values ​​for each of the four wheels are calculated sequentially. , , , Then filter by maximum value. The result This refers to the overall motion deviation of the entire machine.

[0059] Obtain the normal contact force of each driving adsorption wheel at the current moment. Contact force at the previous sampling time Using the difference calculation formula Solve for the rate of change of contact force in each round separately, and pre-set the threshold for sudden change of contact force. With critical support force threshold When a certain wheel simultaneously satisfies and When determining that a wheel has experienced local contact instability, count the total number n of wheels that meet the instability criteria among the four wheels, and then use a normalization formula. The degree of local instability is calculated, and the result ranges from 0 to 1. The higher the value, the greater the risk of local support failure and dynamic instability of the wheelset.

[0060] First, the overall motion deviation Degree of local instability Normalization is performed on the interval from 0 to 1 to eliminate dimensional differences. Motion deviation weighting coefficient 'a' and local instability weighting coefficient 'b' are set, and the following conditions are met: A linear weighted comprehensive evaluation formula is adopted. Substituting the two normalized parameters into the fusion calculation, the system comprehensively reflects the error level of the actual motion of the wheelset deviating from the expected value and the risk of failure of the adsorption wheel contact support. The final output is a dynamic stability index with a value between 0 and 1. The higher the value of the indicator, the higher the risk of the robot's dynamic operation becoming unstable.

[0061] In one possible implementation, step S420 further includes: Step S421: When the rate of change of the normal contact force of any driving adsorption wheel exceeds the preset rate of change of force threshold, it is determined that the corresponding driving adsorption wheel is in a state of local instability.

[0062] Step S422: Calculate the percentage of time that any driving adsorption wheel is in a local unstable state and the degree to which it exceeds the preset force change rate threshold, and call the preset mapping rule to map it to the degree of local instability of each number of driving adsorption wheels.

[0063] Specifically, the change rate of single wheel contact force is calculated based on the difference in wheel normal contact force at adjacent sampling times. The calculated change rate value is compared with the force change rate threshold pre-stored by the system. If the calculated change rate of single wheel contact force is greater than the threshold, the current driving adsorption wheel is directly marked as entering a local unstable state; otherwise, the wheel contact support state is determined to be stable.

[0064] A fixed-duration sliding time window is used to collect continuous normal contact force data of a single wheel. The ratio of the sampling time of the wheel that is judged to be locally unstable within the window to the total window time is used as the instability time proportion. Simultaneously, the difference between the contact force change rate and the force change rate threshold is calculated to obtain the threshold exceedance. The two feature quantities, instability time proportion and threshold exceedance, are input into a pre-configured piecewise linear mapping rule to complete the quantitative scoring of single-wheel instability. Then, the quantitative scores of the four driving adsorption wheels are summarized and normalized and fused to finally obtain the numerical value of local instability that characterizes the risk of failure of the contact support of the entire wheel group.

[0065] In one possible implementation, step S500 further includes: Step S510: Under different curvature change thresholds, the basic weight allocation ratio is set by analyzing historical measured stability samples to obtain the curvature-weight mapping relationship.

[0066] Step S520: Match the static stability index and the dynamic stability index in the curvature-weight mapping relationship based on the real-time obtained surface curvature radius change rate to obtain the weight ratio.

[0067] Specifically, the system divides the surface curvature change threshold range into multiple gradients, collects historical measured sample datasets of static and dynamic stability indicators collected by the robot under different curvature conditions, performs mean statistics on all samples within the same curvature range to balance the influence weights of the two types of indicators, assigns equal basic weights to static stability and dynamic stability using an average distribution method, traverses all curvature change threshold ranges to complete the weight assignment records for each group, constructs a one-to-one curvature-weight mapping relationship table with the surface curvature change as input and the fusion weights of static and dynamic indicators as output, and stores it in the controller.

[0068] The system collects the radius of curvature at the robot's current position on the surface in real time and calculates the rate of change of the radius of curvature using differential methods. This rate of change is then compared sequentially with the threshold values ​​for each curvature interval in a pre-established curvature-weight mapping relationship to locate the interval to which the current rate of curvature change belongs. Finally, the system retrieves the pre-configured static stability index weight coefficients for that interval. Dynamic stability index weighting coefficient And the two sets of coefficients satisfy This outputs a weighted ratio of static and dynamic stability indices that adapts to the current degree of surface curvature.

[0069] To visually illustrate the dynamic adjustment of weights with curvature, a curvature-weight line graph is plotted as follows: Figure 3 As shown in Table 1, for typical operation scenarios of climbing large curved surfaces of wind turbine towers, the matching weights of each curvature interval are calibrated based on a large number of historical field measurement samples, forming a quantitative mapping relationship: Table 1 Mapping Table of Surface Curvature Radius Change Rate Range and Dynamic / Static Stability Index Weights

[0070] From Table 1, Figure 3 The variation pattern of the broken lines can intuitively show that the weight allocation logic fits the actual working conditions of the tower: under a gentle curved surface, the robot is less likely to deviate from its motion, and the overturning of the whole machine is the main danger, so the weight of the static stability index is increased; the more severe the curvature change, the greater the risk of dynamic instability caused by the fluctuation of the normal force of the adsorption wheel and the deviation of motion tracking, so the weight of the dynamic index is increased accordingly; the two broken lines intersect at ΔR=20mm / s, at which point the weights of the static and dynamic indices are equal, corresponding to the balanced working condition of the transition between the straight and conical sections of the tower.

[0071] The following is a complete example calculation based on actual on-site measurement conditions: When the robot climbs the conical section of the tower, the real-time data collection shows that the rate of change of the surface curvature radius ΔR = 32 mm / s. Looking up the table, this value falls within the range of 20 < ΔR ≤ 50, and the weighting is then applied. =0.3、 =0.7; the static stability index is obtained by synchronous real-time calculation. =0.82, dynamic stability index =0.65; Substitute into the weighted fusion calculation formula The comprehensive stability evaluation value is obtained. In this scheme, the closer the comprehensive stability evaluation value is to 1, the higher the risk of instability. 0.701 indicates that the robot currently has a moderate risk of dynamic instability. The controller can actively reduce the climbing speed based on this evaluation value to avoid detachment and overturning failures.

[0072] Example 2, based on the same inventive concept as the method for evaluating the motion stability of climbing robots facing large curved surfaces in the previous examples, such as... Figure 2 As shown, this application provides a motion stability evaluation system for climbing robots targeting large curved surfaces. The system and method embodiments in this application are based on the same inventive concept. The system includes: The multi-source motion data acquisition module 10 is used to acquire multi-source motion data during the robot's climbing motion.

[0073] The motion simulation model building module 20 is used to build a motion simulation model of the robot facing a surface with variable curvature.

[0074] The static stability index acquisition module 30 is used to perform analysis and fusion of static position stability margin and static force stability margin based on the motion simulation model to obtain the static stability index.

[0075] The comprehensive analysis module 40 is used to perform a comprehensive analysis based on motion deviation and local instability state according to the motion simulation model and the multi-source motion data to obtain dynamic stability index.

[0076] The evaluation value acquisition module 50 is used to determine the degree of curvature of the current surface based on the real-time acquired rate of change of the surface curvature radius, determine the weight ratio of the static stability index and the dynamic stability index, and perform a weighted summation of the static stability index and the dynamic stability index to obtain a comprehensive stability evaluation value.

[0077] Furthermore, the system is also used to implement the following functions: The robot includes a front drive compartment, a rear drive compartment, a central hinge connecting the front and rear drive compartments, four independent cantilever arms, and four omnidirectional drive adsorption wheels.

[0078] Furthermore, the system is also used to implement the following functions: Force sensors installed at the connection points between each independent cantilever and the driving adsorption wheel collect real-time data on the normal contact force between each driving adsorption wheel and the curved surface; encoders installed at the drive motors of each driving adsorption wheel collect real-time data on the actual rotational speed of each driving adsorption wheel; angle sensors installed at the steering mechanisms of each driving adsorption wheel collect real-time data on the actual steering angle of each driving adsorption wheel; the multi-source motion data is composed of the normal contact force data, the actual rotational speed data, and the actual steering angle data.

[0079] Furthermore, the system is also used to implement the following functions: First motion modeling data is collected on the torsion of the front and rear drive housings of the robot around the central hinge; second motion modeling data is collected on the mechanical transmission relationship of the swing of each independent cantilever around the rotary joint to establish the microscopic contact state between each drive adsorption wheel and the curved surface; third motion modeling data is collected on the coordinated control setting parameters of the independent speed and direction of the four omnidirectional drive adsorption wheels; the overall motion state of the robot is modeled using the first motion modeling data, the second motion modeling data and the third motion modeling data to obtain the motion simulation model.

[0080] Furthermore, the system is also used to implement the following functions: The actual motion commands during the robot's climbing motion are input into the motion simulation model. The contact points between the two driving adsorption wheels of the front drive housing and the curved surface, and the contact points between the two driving adsorption wheels of the rear drive housing and the curved surface are output as four vertices to construct a support polygon. The static position stability margin is calculated with reference to the support polygon. The normal contact force distribution between each driving adsorption wheel and the curved surface is obtained through the motion simulation model, and the ratio of the anti-overturning moment to the overturning moment is calculated to determine the static force stability margin. The static position stability margin and the static force stability margin are weighted and fused to obtain the static stability index.

[0081] Furthermore, the system is also used to implement the following functions: Collect the robot's real-time posture and calculate the robot's center of mass; calculate the positional deviation of the robot's center of mass projection point on the curved tangent plane relative to the center of the supporting polygon; determine the static position stability margin based on the positional deviation.

[0082] Furthermore, the system is also used to implement the following functions: Based on the real-time collected actual rotational speed and actual steering angle data of each driving adsorption wheel, the motion deviation of each driving adsorption wheel is calculated by comparing it with the expected rotational speed and expected steering angle output by the motion simulation model. The maximum motion deviation among the four driving adsorption wheels is taken as the overall motion deviation. Based on the real-time collected normal contact force data of each driving adsorption wheel, the rate of change of the normal contact force of each driving adsorption wheel is calculated. By statistically analyzing local instability states, the degree of local instability of each number of driving adsorption wheels is calculated. The overall motion deviation and the degree of local instability of each driving adsorption wheel are comprehensively evaluated to obtain the dynamic stability index.

[0083] Furthermore, the system is also used to implement the following functions: When the rate of change of the normal contact force of any driving adsorption wheel exceeds the preset rate of change of force threshold, the corresponding driving adsorption wheel is determined to be in a local unstable state. The percentage of time that any driving adsorption wheel is in a local unstable state and the degree to which it exceeds the preset rate of change of force threshold are statistically analyzed, and the preset mapping rules are used to map the local instability degree of each number of driving adsorption wheels.

[0084] Furthermore, the system is also used to implement the following functions: By analyzing historical measured stability samples under different curvature change thresholds, a basic weight allocation ratio is set to obtain the curvature-weight mapping relationship; the curvature radius change rate of the surface is matched in the curvature-weight mapping relationship according to the real-time acquired data to obtain the weight ratio of the static stability index and the dynamic stability index.

[0085] It should be noted that the order of the embodiments described above is for descriptive purposes only and does not represent the superiority or inferiority of the embodiments. Specific embodiments of this specification have been described above. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0086] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0087] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for evaluating the motion stability of climbing robots for large curved surfaces, characterized in that, include: Collect multi-source motion data during the robot's climbing motion; Establish a motion simulation model of the robot facing a surface with variable curvature; Based on the motion simulation model, the static position stability margin and static force stability margin are analyzed and fused to obtain the static stability index. Based on the motion simulation model and the multi-source motion data, a comprehensive analysis based on motion deviation and local instability is performed to obtain dynamic stability indices. The degree of curvature of the current surface is determined based on the real-time acquired rate of change of the surface curvature radius. The weight ratio of the static stability index and the dynamic stability index is determined, and the static stability index and the dynamic stability index are weighted and summed to obtain a comprehensive stability evaluation value.

2. The method for evaluating the motion stability of climbing robots for large curved surfaces as described in claim 1, characterized in that, The robot includes a front drive compartment, a rear drive compartment, a central hinge connecting the front and rear drive compartments, four independent cantilever arms, and four omnidirectional drive adsorption wheels.

3. The method for evaluating the motion stability of climbing robots for large curved surfaces as described in claim 2, characterized in that, Collect multi-source motion data during the robot's climbing motion, including: Force sensors installed at the connection points between each independent cantilever and the driving adsorption wheel are used to collect real-time data on the normal contact force between each driving adsorption wheel and the curved surface. The actual rotational speed data of each driving adsorption wheel is collected in real time by encoders installed at the drive motors of each driving adsorption wheel. The actual steering angle data of each driving adsorption wheel is collected in real time by angle sensors installed at the steering mechanism of each driving adsorption wheel. The multi-source motion data is composed of normal contact force data, actual rotational speed data, and actual steering angle data.

4. The method for evaluating the motion stability of climbing robots for large curved surfaces as described in claim 2, characterized in that, Establish a motion simulation model for the robot facing a surface with variable curvature, including: Collect the first motion modeling data of the robot's front and rear drive compartments twisting around the central hinge; Collect the mechanical transmission relationship of each independent cantilever swinging around the rotary joint, and establish the second motion modeling data to characterize the microscopic contact state between each driving adsorption wheel and the curved surface; Collect the coordinated control setting parameters of the independent rotation speed and steering of the four omnidirectional drive adsorption wheels to establish the third motion modeling data; The robot's overall motion state is modeled using the first motion modeling data, the second motion modeling data, and the third motion modeling data to obtain the motion simulation model.

5. The method for evaluating the motion stability of climbing robots for large curved surfaces as described in claim 2, characterized in that, Based on the motion simulation model, an analysis and fusion of static position stability margin and static force stability margin is performed to obtain static stability indices, including: The actual motion commands during the robot's climbing motion are input into the motion simulation model. The contact points between the two driving adsorption wheels of the front drive box and the curved surface and the two driving adsorption wheels of the rear drive box and the curved surface are output as four vertices to construct a support polygon. The static position stability margin is calculated with the support polygon as a reference. The normal contact force distribution between each driving adsorption wheel and the curved surface is obtained through the motion simulation model, the ratio of anti-overturning moment to overturning moment is calculated, and the static force stability margin is determined. The static stability index is obtained by weighting and fusing the static position stability margin and the static force stability margin.

6. The method for evaluating the motion stability of climbing robots for large curved surfaces as described in claim 5, characterized in that, Calculating the static position stability margin with reference to the supporting polygon includes: Collect the robot's real-time posture and calculate the robot's center of mass; Calculate the positional deviation of the robot's centroid projection point on the curved tangent plane relative to the center of the supporting polygon; The static position stability margin is determined based on the position deviation.

7. The method for evaluating the motion stability of climbing robots for large curved surfaces as described in claim 3, characterized in that, Based on the motion simulation model and the multi-source motion data, a comprehensive analysis based on motion deviation and local instability is performed to obtain dynamic stability indices, including: Based on the real-time collected actual rotation speed data and actual steering angle data of each driving adsorption wheel, the motion deviation of each driving adsorption wheel is calculated by comparing it with the expected rotation speed and expected steering angle output by the motion simulation model, and the maximum value of the motion deviation among the four driving adsorption wheels is taken as the overall motion deviation. Based on the real-time collected data of the normal contact force of each driving adsorption wheel, the rate of change of the normal contact force of each driving adsorption wheel is calculated. By statistically analyzing the local instability state, the degree of local instability of each driving adsorption wheel is calculated. The dynamic stability index is obtained by comprehensively evaluating the overall motion deviation and the degree of local instability of each driving adsorption wheel.

8. The method for evaluating the motion stability of climbing robots for large curved surfaces as described in claim 7, characterized in that, By statistically analyzing local instability states, the degree of local instability for each number of driving adsorption wheels is calculated, including: When the rate of change of the normal contact force of any driving adsorption wheel exceeds the preset rate of change of force threshold, the corresponding driving adsorption wheel is determined to be in a state of local instability. The percentage of time any driving adsorption wheel is in a state of local instability and the degree to which it exceeds a preset force change rate threshold are statistically analyzed, and then mapped to the degree of local instability of each number of driving adsorption wheels using a preset mapping rule.

9. The method for evaluating the motion stability of climbing robots for large curved surfaces as described in claim 1, characterized in that, The degree of curvature of the current surface is determined based on the real-time acquired rate of change of the surface radius of curvature, and the weight ratio of the static stability index and the dynamic stability index is determined, including: By analyzing historical measured stability samples under different curvature change thresholds, a basic weight allocation ratio is set to obtain the curvature-weight mapping relationship; The static stability index and the dynamic stability index are matched in the curvature-weight mapping relationship based on the real-time obtained surface curvature radius change rate to obtain the weight ratio.

10. A method and system for evaluating the motion stability of climbing robots for large curved surfaces, characterized in that, The system is used to implement the motion stability evaluation method for climbing robots oriented towards large curved surfaces as described in any one of claims 1-9, and the system comprises: The multi-source motion data acquisition module is used to collect multi-source motion data during the robot's climbing motion. The motion simulation model building module is used to build motion simulation models of robots facing surfaces with variable curvature. The static stability index acquisition module is used to perform analysis and fusion of static position stability margin and static force stability margin based on the motion simulation model to obtain the static stability index. The comprehensive analysis module is used to perform a comprehensive analysis based on motion deviation and local instability state according to the motion simulation model and the multi-source motion data to obtain dynamic stability index. The evaluation value acquisition module is used to determine the degree of curvature of the current surface based on the real-time acquired rate of change of the surface curvature radius, determine the weight ratio of the static stability index and the dynamic stability index, and perform a weighted summation of the static stability index and the dynamic stability index to obtain a comprehensive stability evaluation value.

Images