Large component robot smooth grinding and polishing surface quality control method and system

Abstract

The invention discloses a large component robot smooth grinding and polishing surface quality control method and system. The method comprises the steps that a grinding and polishing initial track is generated based on a large component design model or actual measurement point cloud; based on the compliant grinding and polishing removal model, with the target removal amount and the surface roughness as core optimization targets, optimal process parameters are decided through a PB-MOPSO multi-target optimization algorithm; on the basis of the optimal parameters, the contact deformation and the removal profile in smooth grinding and polishing are fused to generate a uniform removal grinding and polishing track; the redundant machining configuration is optimized in combination with joint performance, singularity avoidance and anti-collision indexes; and after the contact force is adaptively controlled through the compliant device to implement grinding and polishing, process parameters of a defect area are secondarily optimized, and the track and the configuration are corrected according to a detection result. According to the method, the grinding and polishing removal uniformity and the surface quality consistency of the large component are remarkably improved, and closed-loop self-adaptive operation and high-quality grinding and polishing are achieved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of large component grinding and polishing processing control, more particularly, relates to a large component robot compliant grinding and polishing surface quality control method and system. BACKGROUND

[0002] Large complex components are widely used in high-end equipment in the fields of aviation, transportation, energy, etc. Such components are generally used as whole machine load-bearing parts, and their surfaces need to go through multiple grinding and polishing processes to meet the shape and position accuracy and surface quality requirements of the final product. With the continuous improvement of product service performance requirements, the special structure of large components and the stringent manufacturing performance requirements have posed severe challenges to existing grinding and polishing equipment and processes.

[0003] For a long time, large component grinding and polishing mainly relies on manual and semi-automatic equipment. Manual grinding and polishing is directly operated by the operator holding the grinding and polishing tool, and the grinding and polishing force, speed and trajectory are adjusted according to experience; semi-automatic equipment grinding and polishing uses fixed mechanical structures to realize the mechanization of part of the grinding and polishing process, but still needs manual participation in process parameter adjustment and quality detection. In recent years, in order to break through the limitations of traditional methods, the industry has gradually tried to use robots as grinding and polishing equipment executors, coupled with mobile platforms to expand the working range of large-size components, and equipped with compliant force control devices to adapt to the changes in the surface topography of the components and maintain stable contact force during grinding and polishing, forming a grinding and polishing system architecture with preliminary automation capability.

[0004] Using robots as grinding and polishing equipment executors, equipped with mobile platforms and compliant force control devices, can make them have the ability to work flexibly in a large scene. However, the change of the two changes the processing characteristics and motion characteristics of the system, and the original abrasive cutting theory is not applicable to describe the compliant grinding and polishing removal behavior. At the same time, the problems of stable control of contact force and deviation of system large-range motion trajectory are highlighted, and it is difficult to systematically realize the active regulation of grinding and polishing surface quality, and the uniformity of large component grinding and polishing removal and the consistency of surface quality cannot be guaranteed. SUMMARY

[0005] In view of the above defects or improvement needs of the prior art, the present application provides a large component robot compliant grinding and polishing surface quality control method. The method starts from the basic principle of compliant grinding and polishing removal, reveals the removal behavior and surface quality formation mechanism, realizes the core target of target removal amount and surface roughness, focuses on the collaborative optimization of multi-target process parameter decision and uniformity removal trajectory planning, synchronously combines compliant contact force control and over-grinding suppression method, redundant processing pose optimization technology, forms a large component robot compliant grinding and polishing surface quality regulation technology with process parameter-trajectory collaborative optimization as the core, accurately outputs process parameters and corresponding trajectories that meet the target removal amount and surface roughness requirements, so as to realize high-quality grinding and polishing of large component robots.

[0006] To achieve the above object, the application provides a large component robot compliant polishing surface quality control method, comprising the following steps:

[0007] S100: generating polishing initial trajectory based on large component design model or measured point cloud;

[0008] S200: based on the compliant polishing removal model, taking the preset target removal amount and surface roughness as the core optimization target, adopting PB-MOPSO multi-objective optimization algorithm to determine the polishing optimal process parameters, which need to meet the target removal amount deviation threshold and surface roughness deviation threshold requirements;

[0009] S300: based on the optimal process parameters, generating uniform removal polishing trajectory by fusing contact deformation and removal profile in compliant polishing, ensuring that the material removal amount matches the target removal amount, the surface texture meets the surface roughness requirements and the uniformity meets the polishing quality requirements during the trajectory execution process;

[0010] S400: combining joint performance, singularity avoidance and anti-collision index to optimize redundant machining posture, generating optimal robot joint trajectory sequence and base position sequence;

[0011] S500: according to the actual polishing program, controlling the polishing contact force through the compliant device, relying on the mobile platform to expand the working range, and implementing global polishing on the large component;

[0012] S600: detecting the surface quality of the polished component, judging whether the removal amount and surface roughness meet the standards according to the detection results, re-optimizing the process parameters and correcting the trajectory for the defective areas that do not meet the standards, and if there are local defects in the detection, carrying out local polishing for the defective areas to realize closed-loop quality control.

[0013] Further, the PB-MOPSO multi-objective optimization algorithm in step S200 is used to solve the optimization problem with target removal amount and surface roughness as the core:

[0014]

[0015] s.t.a p ≤a p,limit ,

[0016] R a,limitl ≤R a ≤R a,limit2 ,

[0017] F n，limitl ≤F n ≤F n，limit2 ,

[0018] vs,limitl ≤v s ≤v s，limit2 ,

[0019] v w，limitl ≤v w ≤v w，limit2 ,

[0020] R∈{R P60 ,R P80 ,R P120 ,R P180 R P240 ...}

[0021] In the formula, This represents the material removal depth model. This represents a surface roughness model, where x represents the input process parameters of the model. and This represents the optimal weight coefficients obtained by the corresponding model based on the training data, a. p,limit This indicates the maximum material removal depth limit set. F n,limitl ≤F n ≤F n,limit2 Indicates contact force constraint, v s,limitl ≤v s ≤v slimit2 This indicates the contact wheel linear velocity constraint, v w，limitl ≤v w ≤v w，limit2 Represents the robot feed rate constraint, R∈{R P60 ,R P80 R P120 ,R P180 R P240 ...} represents the equivalent radius constraint of the abrasive grain bundle in the abrasive belt, and the subscript R indicates the mesh size of the corresponding abrasive belt.

[0022] Furthermore, the particle position and velocity updates in the PB-MOPSO algorithm are expressed by the following formula:

[0023]

[0024] In the formula, v i,t and x i，t Let p represent the velocity and position of the i-th particle in the t-th iteration, respectively. i,t With g t c1 and c2 represent the position of particle i at the t-th search and the global optimal position, respectively. c1 and c2 are acceleration coefficients, r1 and r2 represent random numbers within the range [0, 1], τ represents the inertia weight, used to balance global search capability and local search capability, and σ represents the global optimal position. γ (u)=sign(u)|u|γ Let γ represent the Powerball function, and let γ represent its power coefficient.

[0025] Furthermore, in step S300, the uniformity removal trajectory planning first employs an equal chord height step-size method that considers contact deformation to calculate trajectory C. v (v i The step size interpolation parameter {Δu} i The step size interpolation parameters are further modified using a bisection method to improve adaptability. Furthermore, the equal residual height row spacing method, which considers uniformity removal, is used to calculate the row spacing interpolation parameters {Δv} for the next trajectory. i To ensure the continuity of execution for a single trajectory, {Δv} is selected here. i The minimum value Δv min As the line spacing offset distance, it is iterated until it approaches v. max The u-direction of the contact point is selected as the x-direction (feed direction), and the normal vector is selected as the z-direction to obtain a machining trajectory sequence containing attitude information.

[0026] Furthermore, the machining configuration optimization described in step S400 is based on the robot machining comprehensive performance index RCPI, and the objective function is expressed as:

[0027] F r (s base θ tool )=max(ω1JCI,ω2SCl,ω3CCI)

[0028] In the formula, ω1, ω2, and ω3 represent weighting coefficients, JCI represents the joint constraint index, SCI represents the singularity constraint index, CCI represents the collision constraint index, and S... base Represents the robot's movement along the external axis, θ base This indicates the rotation angle of the end tool around the z-axis.

[0029] Furthermore, the singularity constraint index (SCI) is calculated using the following formula:

[0030]

[0031] In the formula, J N The homogeneous Jacobian matrix normalized by the feature length, K(J) N ) denotes the condition number of the velocity Jacobian matrix.

[0032] Furthermore, the collision constraint index (CCI) is calculated using the following formula:

[0033]

[0034] In the formula, r i and r jd represents the effective radius of the bounding box of the i-th and j-th links of the robot, respectively. ij and Δ ij These are the shortest distance and the preset safe distance between robot link i and other links or objects j, respectively. ij It needs to be obtained from the linear equation of the cylindrical bounding box of the link in Cartesian space, that is:

[0035]

[0036] In the formula, c i With c j Let λ represent a point on the equation of the connecting rod axis. i With λ j These are the control parameters for the axis equation, where p0 represents the origin position of the robot's base coordinate system. and Let {B} and {n} represent the transformation matrices from the robot's base coordinate system {B} to the joint coordinate systems {n} and {n+1}, respectively.

[0037] Furthermore, the "adaptive control of polishing contact force by compliant device" in step S500 specifically includes: the system expands the working range by means of a mobile platform, and adaptively adjusts the z-axis extension amount by compliant force control device to maintain the stability of contact force during polishing, so as to ensure that process parameters are applied stably to achieve the target removal amount and surface roughness.

[0038] Furthermore, the step S600, "optimizing the secondary process parameters and correcting the trajectory of the defect area based on the detection results," specifically includes: after polishing, detecting the surface quality based on the measurement module, focusing on whether the removal amount of the defect area deviates from the target removal amount and whether the surface roughness exceeds the target range, calculating the additional removal amount required for the defect area, re-optimizing the process parameters and planning the trajectory for the area based on the target removal amount and surface roughness, and performing secondary polishing until the large component as a whole meets the processing quality requirements.

[0039] A surface quality control system for compliant grinding and polishing of large components by robots includes:

[0040] Offline programming module generates initial grinding and polishing trajectories based on large component design models or measured point clouds;

[0041] The process optimization and decision module, based on the compliant grinding and polishing removal model, takes the preset target removal amount and surface roughness as the core optimization objectives, and uses the PB-MOPSO multi-objective optimization algorithm to determine the optimal grinding and polishing process parameters. The optimal process parameters must meet the target removal amount deviation threshold and surface roughness deviation threshold requirements.

[0042] The trajectory optimization module, based on the optimal process parameters, integrates contact deformation and removal profile in compliant polishing to generate a uniform polishing trajectory, ensuring that the amount of material removed matches the target amount of removal, the surface texture meets the surface roughness requirements, and the uniformity meets the polishing quality requirements during trajectory execution.

[0043] The configuration optimization module combines joint performance, singularity avoidance and collision avoidance indicators to optimize redundant machining configurations and generate the optimal robot joint trajectory sequence and base position sequence.

[0044] The global / local compliant grinding and polishing module is designed for the grinding and polishing needs of large components. It uses the process optimization module and the trajectory and configuration optimization module to generate the final grinding and polishing program, and then performs the grinding and polishing operation through a compliant grinding and polishing device.

[0045] The measurement and testing module detects the surface quality of the components after grinding and polishing. Based on the test results, it determines whether the removal amount and surface roughness meet the standards. For defective areas that do not meet the standards, the process parameters are optimized and the trajectory is corrected to achieve closed-loop quality control.

[0046] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:

[0047] 1. The method of the present invention is based on the theory of robot compliant grinding and polishing, and combines offline programming and online trajectory correction. At the same time, it controls the surface quality of grinding and polishing from multiple dimensions such as offline trajectory generation, grinding and polishing process parameter decision, uniformity removal trajectory and processing configuration optimization. Based on the measurement results, it performs secondary trajectory planning for unqualified or defective areas, forming a closed-loop adaptive operation of the robot compliant grinding and polishing system, which can realize high-quality grinding and polishing of large components by robots.

[0048] 2. The method of the present invention proposes a surface quality control technology for compliant grinding and polishing of large components by robots. It takes into account the comprehensive influence of removal behavior and system processing performance on surface quality, and realizes stable and controllable compliant grinding and polishing removal.

[0049] 3. The method of this invention proposes a process parameter decision method, a uniformity removal trajectory planning method, and a processing configuration optimization algorithm. The coordinated optimization of the three can directly generate the optimal process parameters, joint trajectories, and base position sequences necessary for system operation based on the target quality, avoiding the complex and tedious debugging process and realizing high-quality grinding and polishing of large components.

[0050] 4. The method of the present invention can determine the optimal process parameters based on the target quality, generate a uniform removal trajectory and optimal processing configuration, maintain stable processing performance of the system, and ensure a controllable compliant grinding and polishing removal process. Attached Figure Description

[0051] Figure 1 This is a process flow diagram of the active control of a robot for compliant grinding and polishing of large components according to an embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram illustrating the process parameter decision-making principle under multi-objective constraints in compliant grinding and polishing according to an embodiment of the present invention.

[0053] Figure 3 This is a simplified geometric model of the robot and other obstacles in the processing system of this invention.

[0054] Figure 4 This is a diagram showing the redundant parameters of a robot machining system with an external axis guide rail according to an embodiment of the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0056] like Figure 1 As shown, this embodiment of the invention provides a method for controlling the surface quality of large component robots through compliant grinding and polishing. The orange module represents the theoretical method, which includes the following steps:

[0057] S100: Generates the initial grinding and polishing trajectory based on the design model of a large component or the measured point cloud;

[0058] Before grinding and polishing, it is necessary to plan the initial processing trajectory based on the design model of the large component or the measured point cloud. The purpose is to provide the global initial trajectory for subsequent grinding and polishing, while saving the calculation time of generating online correction trajectory using real-time measurement data.

[0059] After identifying and compensating for system errors, and considering the surface polishing requirements of large components, a process decision module and a trajectory and configuration optimization module are used to generate the final polishing program for smooth polishing operations. While ensuring system operational accuracy, the proactive control of polished surface quality relies primarily on three aspects: polishing process parameter decision-making, uniformity removal trajectory planning, and redundant processing configuration optimization.

[0060] S200: Based on the compliant grinding and polishing removal model, with the preset target removal amount and surface roughness as the core optimization objectives, the PB-MOPSO multi-objective optimization algorithm is used to determine the optimal grinding and polishing process parameters. The optimal process parameters must meet the target removal amount deviation threshold and surface roughness deviation threshold requirements.

[0061] The decision-making of polishing process parameters is mainly based on the robotic compliant polishing removal model to achieve the expected removal amount 'a' of polishing material. p and surface roughness R a Based on the prediction, the PB-MOPSO multi-objective optimization algorithm is used to determine the optimal combination of grinding and polishing process parameters, including grinding and polishing contact force. sanding belt linear velocity Robot feed rate And the equivalent radius R of the abrasive grains of the abrasive belt used. * Here, the optimization target and feasible solution space can be adjusted according to the specific grinding and polishing quality requirements.

[0062] During the grinding and polishing process, the compliant device can adaptively adjust its z-axis extension to maintain stable contact force. First, 108 pre-grinding and polishing experiments were conducted under different process parameters. Three commonly used ceramic alumina abrasive belts with different mesh sizes (equivalent abrasive radius R) were selected for the experiments, with a normal force F... n The linear velocity of the sanding belt, v s Robot feed speed v w The three factors are each set to six levels, i.e., using Three groups of three factors and six levels of orthogonal experiment.

[0063] like Figure 2 As shown, the process parameter decision principle under multi-objective constraints in compliant polishing includes: measuring the material removal depth 'a' within the stable polishing region. p and surface roughness R a Based on the above data, a material removal depth model was trained. and surface roughness model Where x represents the input process parameters of the model, and This represents the optimal weight coefficients obtained by the corresponding model based on the training data. Based on these, a process parameter decision algorithm under multi-objective constraints is used to obtain the optimal process parameters.

[0064] Generally, the desired material removal depth a is achieved through grinding and polishing. p To minimize the number of repeated grinding and polishing cycles and improve grinding and polishing efficiency without damaging the workpiece substrate, objective one can be described as follows:

[0065]

[0066] In the formula, a p，limit This indicates the maximum material removal depth limit set.

[0067] At the same time, it is desirable to minimize the surface roughness value after grinding and polishing within a certain range. Therefore, objective two can be expressed as:

[0068]

[0069] Based on the above optimization objectives, the feasible solution space for process parameters can be further limited according to actual conditions, specifically including:

[0070] 1. Contact force constraint:

[0071] F n,limitl ≤F n ≤F n,limit2 (4)

[0072] 2. Contact wheel linear velocity constraint:

[0073] v s,limitl ≤v s ≤v s,limit2 (5)

[0074] 3. Robot feed speed constraints:

[0075] v w，limitl v w ≤v w，limit2 (6)4. Equivalent radius constraint of abrasive belt abrasive grain bundle:

[0076] R∈{R P60 ,R P80 ,R P120 ,R P180 ,R P240 ,...} (7)

[0077] In the formula, the subscript R indicates the mesh size of the corresponding sand belt.

[0078] Therefore, the multi-objective decision equation for the compliant grinding and polishing process parameters is:

[0079]

[0080] sta p ≤a p，limit ,

[0081] R a，limitl ≤R a ≤R a，limit2 ,

[0082] F n，limitl ≤F n ≤F n,limit2 ,

[0083] v s，limitl ≤v s ≤v s，limit2 ,

[0084] v w，limitl ≤v w ≤vw，limit2 , R∈{R P60 R P80 ,R P120 ,R P180 ,R P240 ,...} (8)

[0085] Solving the above equation is a typical multi-objective optimization problem (MOOP). The MOOP problem of process parameter decision-making can be solved by using a multi-objective evolutionary algorithm to find its Pareto optimal solution set, thereby obtaining the combination of process parameters that minimizes the multi-objective optimization function in equation (8). First, we introduce the multi-objective particle swarm optimization algorithm (MOPSO):

[0086] v i，t+1 =τv i，t +c1r1(p i，t -x i，t )+C2r2(g t -x i，t )

[0087] x i，t+1 =x i，t +v i，t+1

[0088]

[0089] In the formula, v i，t and x i，t+1 Let p represent the velocity and position of the i-th particle in the t-th iteration, respectively, where n is the number of particles. i，t With g t Let c1 and c2 represent the position of particle i and the global optimal position at the t-th search, respectively. c1 and c2 are learning factors that determine the influence of the position information of the current particle i and the global particles on the particle's trajectory. r1 and r2 represent random numbers in the range [0, 1]. τ represents the inertia weight coefficient, which is used to balance the global search capability and the local search capability.

[0090] However, the goal and Conflicts between particles increase the complexity of the optimization process and the Pareto front. To improve the convergence performance of the MOPSO algorithm, its velocity update strategy needs to be improved. Therefore, the particle position and velocity update formulas in equation (9) are improved as follows:

[0091]

[0092] In the formula, σ γ (u)=sign(u)|u| γ The Powerball function is represented by γ, where γ represents its power coefficient. This is achieved by analyzing the difference between the optimal position and the current position of an individual particle in the swarm (p...).i，t -x i，t ) and the difference between the global optimal position and the current position (g) t -x i，t Applying the Powerball function enables nonlinear transformation of particle velocity.

[0093] The multi-objective process parameter optimization process based on PB-MOPSO is shown in Algorithm 1. It mainly includes the initialization of the process parameter particle population and velocity, the initialization of the optimal position of individuals and the storage of the external archive of non-dominated solutions, the iterative calculation of the particle objective function value and the updating of the external archive, and finally outputs the optimal process parameter combination sequence for compliant grinding and polishing.

[0094] Algorithm 1: Multi-objective process parameter optimization algorithm based on PB-MOPSO

[0095] Input: MOPSO inertial weight τ, acceleration constants c1, c2, number of particles n;

[0096] Powerball power coefficient γ;

[0097] Maximum number of iterations Iter max .

[0098]

[0099]

[0100] Output: Optimal combination sequence of process parameters for compliant grinding and polishing

[0101] S300: Based on optimal process parameters, it integrates contact deformation and removal profile in compliant polishing to generate a uniform removal polishing trajectory, ensuring that the amount of material removed matches the target amount of removal, the surface texture meets the surface roughness requirements, and the uniformity meets the polishing quality requirements during the trajectory execution.

[0102] The grinding and polishing removal trajectory is corrected or replanned based on the prediction model and workpiece geometry. The step length and row spacing of the processing trajectory are calculated mainly based on the contact deformation and removal profile in compliant grinding and polishing, generating a grinding and polishing processing trajectory with uniform removal. At the same time, it can effectively reduce the number of unnecessary trajectory points and trajectory lines, and improve grinding and polishing efficiency.

[0103] After determining the optimal grinding and polishing process parameters, a trajectory generation module is used to plan a uniform removal trajectory. For large components, there are many trajectory points. For general industrial robots, their operating storage space is limited and cannot process all trajectory point data at once. It is necessary to adopt a method of dynamically calling trajectory data in different regions to import the trajectory data in batches and ensure the stable operation of the system.

[0104] The uniformity removal trajectory planning first uses the constant chord height step size method that considers contact deformation to calculate trajectory C. v (v i The step size interpolation parameter {Δu} i The step size interpolation parameters are further modified using a bisection method to improve adaptability. Furthermore, the equal residual height row spacing method, which considers uniformity removal, is used to calculate the row spacing interpolation parameters {Δv} for the next trajectory. i To ensure the continuity of execution for a single trajectory, {Δv} is selected here. i The minimum value Δv min This serves as the line spacing offset. The process is repeated until it approaches v. max The u-direction of the contact point is selected as the x-direction (feed direction), and the normal vector is selected as the z-direction to obtain a machining trajectory sequence containing attitude information.

[0105] S400: Combines joint performance, singularity avoidance and collision avoidance indicators to optimize redundant machining configurations and generate the optimal robot joint trajectory sequence and base position sequence;

[0106] After obtaining the machining trajectory, the optimal machining configuration sequence for the overall grinding and polishing of the robot is determined based on the comprehensive performance index of the machining configuration, especially the displacement of the robot base, to ensure the accessibility of all machining trajectory points and the continuity of the robot's motion trajectory, while minimizing the impact of poor configuration on the end-processing quality and ensuring stable machining performance of the system.

[0107] After generating the robot's machining trajectory, the motion trajectory is generally solved using inverse kinematics. However, for machining large components, external axes are needed to extend the robot's range of motion. More degrees of freedom (7–9 DoFs) increase the number of machining configurations available for grinding and polishing tasks. Different machining configurations result in variations in robot positioning accuracy and stiffness characteristics, and are prone to problems such as unreachable trajectory points and singularity traps, affecting the final machining quality. Therefore, for machining large components, it is necessary to establish comprehensive performance indicators for machining configurations and utilize the redundancy characteristics of the machining system to optimize the grinding and polishing machining configurations.

[0108] In practice, unreasonable machining configurations resulting from neglecting robot accessibility and maneuverability often lead to problems such as unreachable trajectory points, singularity traps, and collisions. Therefore, it is necessary to establish a dimensionless Robotic Machining Performance Index (RCPI). The constraints of this index need to be constructed from multiple dimensions, specifically including:

[0109] 1. Joint restriction constraints

[0110] The limits of a robot's motion performance are constrained by its own structure and joint motion performance, which can be expressed as:

[0111] θ i，min ≤θ i ≤θ imax ,|ω i |≤ω i，max ,

[0112] |a i |≤a i，max , |s|≤s max , i = 1, ..., n (11)

[0113] In the formula, θ i ω i and a i These represent the robot's corresponding joint angles, angular velocities, and accelerations. s is the robot's displacement relative to the external axis origin, and θ is... i，max (θ i，min ), a i，max s max These are the corresponding limit values.

[0114] In actual robot motion, the spatial reachability should be comprehensively considered to avoid joint angles approaching their limits, thereby ensuring motion performance. Therefore, the limits of joint angles can be:

[0115]

[0116] Map equations (11) and (12) to the range [0,1], and consider the actual importance of each indicator:

[0117]

[0118] In the formula, w 1，j (j = 1, 2, 3, 4) are the normalized weights.

[0119] In practical optimization, priority should be given to index values ​​close to the limit to achieve good operational performance. Therefore, the Joint Constraint Index (JCI) can be further expressed as:

[0120] (14)

[0121] In the formula, JCI∈[0,1], the larger JCI is, the closer the robot is to the joint limit.

[0122] 2. Singular Configuration Constraints

[0123] Certain configurations during robot motion can lead to a loss of motion capability, resulting in kinematic singularities. Furthermore, approaching singular configurations also reduces the robot's rigidity. Therefore, its motion should avoid singular configurations as much as possible. The degree of approach to a singular configuration can be quantified using the condition number of the velocity Jacobian matrix.

[0124] (15)

[0125] In the formula, J N It is a homogeneous Jacobian matrix normalized by the feature length:

[0126]

[0127] In the formula, I 3×3 0 3×3 J and L are the 3×3 identity matrix, the 3×3 zero matrix, and the robot velocity Jacobian matrix, respectively. I The feature length is obtained by normalizing J. The Singularity Constraint Index (SCI) can be obtained by mapping it to the range [0,1).

[0128] (17)

[0129] In the formula, SCI∈[0,1), the larger the value, the closer the robot is to singularity, and its flexibility, system stiffness and control accuracy will decrease accordingly.

[0130] 3. Collision Constraint Optimization Indicators

[0131] like Figure 3 As shown, to simplify the complex structures of the robot and workpiece, the bounding box method is used to determine the effective radius of the object and the shortest distance between them. Considering the geometry of the object in the actual workspace, spherical and cylindrical bounding boxes are used to simplify the robot's main links and end effector, while cuboid bounding boxes are used to simplify obstacles such as the workpiece, control cabinet, and external axes. Therefore, the shortest distance between robot links is the distance between the axes of the two cylindrical bounding boxes, while the shortest distance between the robot and obstacles is the shortest distance between the surfaces of the cylindrical and cuboid bounding boxes. Thus, the collision constraints during the machining process can be expressed as:

[0132]

[0133] In the formula, r i and r j d represents the effective radius of the bounding box of the i-th and j-th links of the robot, respectively. ij and Δ ij These are the shortest distance and the preset safe distance between robot link i and other links or objects j, respectively.

[0134] The Collision Constraint Index (CCI) can be expressed as:

[0135]

[0136] In the formula, d ij It needs to be obtained from the linear equation of the cylindrical bounding box of the link in Cartesian space, that is:

[0137]

[0138] In the formula, c i With c j Let λ represent a point on the equation of the connecting rod axis. i With λ j These are the control parameters for the axis equation. p0 represents the origin position of the robot's base coordinate system. and Let {B} and {n} represent the transformation matrices from the robot's base coordinate system {B} to the joint coordinate systems {n} and {n+1}, respectively.

[0139] like Figure 4 As shown, considering that most robotic machining tasks (such as grinding, milling, drilling, welding, etc.) only require 5 degrees of freedom (3 translations and 2 rotations), while a robotic machining system equipped with an external axis with guide rails has 7 degrees of freedom (4 translations and 3 rotations), therefore its movement d along the external axis... base Rotation angle θ of the end tool around the z-axis tool These are redundant degrees of freedom. Therefore, redundant degrees of freedom can be used to optimize the global machining configuration. The RCPI objective function for a single machining trajectory point based on JCI, SCI, and CCI indices can be expressed as:

[0140] F r (s base ,θ tool )=max(ω1JCI,ω2SCl,ω3CCI) (21)

[0141] In the formula, ω1, ω2 and ω3 represent weighting coefficients: choosing a larger ω2 means that more attention is paid to the flexibility of robot operation, that is, SCI has a higher priority than other indicators, which is conducive to achieving higher motion accuracy; while larger ω1 and ω3 pay more attention to JCI and CCI, in order to reduce robot posture changes and obtain more robust passability.

[0142] Before performing global configuration optimization for grinding and polishing, necessary conditions need to be further considered: joint performance must not exceed limits, collisions are not allowed, and all target trajectory points must be reachable. Therefore, the configuration optimization equation for a single trajectory point can be expressed as:

[0143] MinimizeF r (sbase ,θ tool ),

[0144]

[0145] CCI≤1,

[0146] L o ≤H(θ) max , (twenty two)

[0147] In the formula, L o H(θ) represents the distance between the current optimization target point and the robot origin. max The farthest distance r within the robot's reachable range is calculated using equation (12). u，i Let ΔL be the workpiece curvature. i Let ε be the step size. i This represents the chord height error.

[0148] Equation (21) gives the optimization objective for a single trajectory point. To ensure the continuity of the trajectory, the trajectory point with the worst RCPI index in each machining trajectory can be selected to ensure that the optimal configuration of the entire machining trajectory satisfies the constraints. When the inverse kinematic solution of the first trajectory point is determined, the inverse solution of subsequent trajectory points can be determined based on the continuity constraints of the joint space. Therefore, the optimization objective function of the global RCPI can be expressed as:

[0149]

[0150] In the formula, m i m represents the number of inverse solutions corresponding to the first trajectory point. i ≤8,k i F represents the number of trajectory points of the i-th trajectory. r，j Let represent the objective function of the j-th trajectory point.

[0151] In the entire machining area, the maximum objective function value in the machining path should be selected for optimization to obtain the optimal machining configuration for the entire machining area. Therefore, the comprehensive fitness function can be established as follows:

[0152]

[0153] When the workpiece is large or has a complex geometry, the robot's machining posture may not be able to be continuously adjusted, and only one optimal robot configuration [s] may be available. base θ toolIt is impossible to satisfy all the indicators in the entire machining area. Therefore, the machining area can be divided into multiple sub-regions based on the workpiece's geometric features and the workspace, and the optimal machining configuration within each sub-region can be obtained to ensure that the global optimization requirements are met. Although dividing into more sub-regions will make it easier to find a solution that meets the indicators, this will also cause the robot to reposition frequently, introducing more different positional errors δ. Rail Therefore, the robot base should be divided into as few sub-regions as possible to avoid frequent movement while ensuring a feasible solution for the optimization objective.

[0154] Solving the above optimization problem is a complex nonlinear problem with a huge computational burden. Therefore, a single-objective particle swarm optimization algorithm with a compression factor (CFPSO) is chosen to solve for the redundant parameters under the optimal configuration. and The particle swarm position and velocity update process can be represented as:

[0155]

[0156] In the formula, v i，t and x i，t Let p represent the velocity and position of the i-th particle in the t-th iteration, respectively, where n is the number of particles. i，t With g t c1 and c2 represent the position of particle i and the global optimal position at the t-th search, respectively. c1 and c2 are acceleration coefficients, r1 and r2 represent random numbers in the range [0,1], and τ represents the inertia weight, which is used to balance the global search capability and the local search capability. τ represents the compression factor. The following update strategies can generally be used for calculation:

[0157]

[0158] In the formula, t max τ represents the maximum number of iterations. min and τ max These are the minimum and maximum values ​​of τ, respectively.

[0159] The global optimal processing configuration optimization process is shown in Algorithm 2. It mainly includes steps such as initializing the particle swarm, evaluating the fitness value of the processing configuration, updating the individual and global optimal positions, and updating the particle swarm velocity and position. Finally, the optimal processing configuration of each sub-region is obtained.

[0160] Algorithm 2: Global Machining Configuration Optimization Algorithm

[0161] Input: m, n, c1, c2, ω max ,ω min ,g t,pre .

[0162]

[0163]

[0164] Output: Optimal machining configuration

[0165] S500: Based on the actual grinding and polishing procedure, the grinding and polishing contact force is adaptively controlled by the compliant device, and the working range is expanded by relying on the mobile platform to carry out global grinding and polishing of large components.

[0166] Based on the actual polishing procedure, the system expands the working range with the help of a mobile platform and adaptively adjusts the Z-axis extension and retraction through a compliant force control device to maintain stable contact force and perform global polishing on the components. As a core execution link, it lays the foundation for subsequent surface quality inspection, helps determine whether secondary polishing is required, ensures uniformity of component polishing removal and consistency of surface quality, and achieves high-quality polishing.

[0167] S600: Inspect the surface quality of components after grinding and polishing, and determine whether the removal amount and surface roughness meet the standards based on the test results. For defective areas that do not meet the standards, re-optimize the process parameters and correct the trajectory to achieve closed-loop quality control.

[0168] After polishing, the surface quality can be inspected using the measurement module to identify areas that still have defects or are unqualified. The removal allowance is calculated, and the polishing process parameters for that area are optimized again. The optimal processing trajectory and configuration are calculated, and a second polishing is performed. If local defects are detected, local polishing is carried out on the defective areas until the large component as a whole meets the processing quality requirements.

[0169] This invention also provides a method and system for controlling the surface quality of compliant grinding and polishing of large components using a robot. The system includes an offline programming module, a process optimization and decision-making module, a trajectory and configuration optimization module, a global / local compliant grinding and polishing operation module, and a measurement and detection module. The offline programming module takes the design model and measured point cloud of the large component as input to perform offline programming simulation of the initial trajectory, generating an initial grinding and polishing program. This provides a global initial trajectory for subsequent grinding and polishing, while saving the computation time required to generate online correction trajectories using real-time measurement data. The process optimization and decision-making module, based on material removal mechanism analysis and grinding and polishing process parameter decisions, first conducts multiple sets of pre-grinding and polishing experiments to measure the material removal depth and surface roughness within the stable grinding and polishing area. This trains a material removal depth model and a surface roughness model. Then, using the PB-MOPSO multi-objective optimization algorithm, under constraints such as contact force, contact wheel linear velocity, robot feed speed, and the equivalent radius of the abrasive belt, it calculates and determines the optimal grinding and polishing process parameters. The trajectory and configuration optimization module solves for trajectory parameters by removing them. It plans the grinding and polishing trajectory for uniform removal using the equal chord height step length method considering contact deformation and the equal residual height row spacing method considering uniform removal, reducing the number of unnecessary trajectory points and trajectory lines. Simultaneously, through system machining configuration optimization, it establishes comprehensive robot machining performance indicators and uses the CFPSO algorithm to optimize machining configurations such as robot base displacement and end-effector posture, generating the actual grinding and polishing program. This ensures the accessibility of machining trajectory points and the continuity of robot motion trajectories, reducing the impact of poor configurations on end-effector machining quality and maintaining stable system machining performance. The global / local compliant grinding and polishing module addresses the grinding and polishing needs of large components. It uses the final grinding and polishing program generated by the process optimization module and the trajectory and configuration optimization module, and performs grinding and polishing operations through a compliant grinding and polishing device, performing compliant grinding and polishing operations globally and locally. After global or localized compliant polishing, the measurement and inspection module measures and inspects large components, identifies defect areas, inspects surface quality, and determines whether the large components meet the processing quality requirements. If not, it provides a basis for re-optimizing the polishing process parameters, calculating the optimal processing trajectory and configuration for secondary polishing, thus forming a closed-loop control.

[0170] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A large component robot compliant polishing surface quality control method, characterized by, The method comprises the following steps: S100: generating a grinding and polishing initial trajectory based on a large component design model or a measured point cloud; S200: based on the compliant grinding and polishing removal model, taking the preset target removal amount and surface roughness as the core optimization objectives, adopting a PB-MOPSO multi-objective optimization algorithm to determine the optimal grinding and polishing process parameters, and the optimal process parameters need to meet the target removal amount deviation threshold and the surface roughness deviation threshold requirements; S300: based on the optimal process parameters, fusing the contact deformation and removal profile in the compliant grinding and polishing to generate a uniform removal grinding and polishing trajectory, and ensuring that the material removal amount in the trajectory execution process matches the target removal amount, the surface texture meets the surface roughness requirement, and the uniformity meets the grinding and polishing quality requirement; S400: optimizing the redundant machining posture in combination with the joint performance, singularity avoidance and anti-collision index, generating an optimal robot joint trajectory sequence and base position sequence; S500: according to the actual grinding and polishing program, adaptively controlling the grinding and polishing contact force through the compliant device, relying on the mobile platform to expand the working range, and implementing global grinding and polishing on the large component; S600: detecting the surface quality of the component after grinding and polishing, judging whether the removal amount and the surface roughness meet the requirements according to the detection results, re-optimizing the process parameters and correcting the trajectory for the defect area that does not meet the requirements, and if there is a local defect, carrying out local grinding and polishing for the defect area to realize closed-loop quality control.

2. The large component robot compliant polishing surface quality control method of claim 1, wherein, The PB-MOPSO multi-objective optimization algorithm in step S200 is used to solve the optimization problem with the target removal amount and surface roughness as the core: s.t. a p ≤a p,limit , R a,limitl ≤R a ≤R a，limit2 , F n，limitl ≤F n ≤F n，limit2 , v s,limitl ≤v s ≤v s，limit2 , v w,limitl ≤v w ≤v w,limit2 , R e {R P60 ,R P80 ,R P120 ,R P180 ,R P240 ,...} wherein, represents a material removal depth model, represents a surface roughness model, wherein x represents input process parameters of the model, x = [F n v s v w R T , with represents the optimal weight coefficients of the corresponding model according to the training data, a p,limit represents the set maximum material removal depth limit, F n,limitl ≤ F n ≤ F n,limit2 represents a contact force constraint, v s,limitl ≤ v s ≤ v s,limit2 represents a contact wheel linear velocity constraint, v w，limitl ≤ v w ≤ v w，limit2 represents a robot feed rate constraint, R ∈ {R P60 , R P80 , R P120 , R P180 , R P240 ,...} represents a belt abrasive grain bundle equivalent radius constraint, R subscript represents the mesh number of the corresponding belt.

3. The large component robot compliant polishing surface quality control method of claim 2, wherein, The particle position and velocity update in the PB-MOPSO algorithm adopts the following formula: where v i，t and x i，t denote the velocity and position of the ith particle at the tth iteration, p i，t and g t denote the position of the ith particle at the tth search and the global optimal position, c1and c2are acceleration coefficients, r1and r2represent random numbers in the range [0, 1], τ denotes the inertia weight used to balance the global search ability and the local search ability, σ γ (u) = sign(u) |u| γ denotes the Powerball function, and γ denotes its power coefficient.

4. The large component robot compliant surfacing quality control method of any of claims 1-3, wherein, The uniformity removal trajectory planning in step S300 first uses the equal chord height step size method considering contact deformation to calculate trajectory C. v (v i The step size interpolation parameter {Δu} i The step size interpolation parameters are further modified using a bisection method to improve adaptability. Furthermore, the equal residual height row spacing method, which considers uniformity removal, is used to calculate the row spacing interpolation parameters {Δv} for the next trajectory. i To ensure the continuity of execution for a single trajectory, {Δv} is selected here. i The minimum value Δv min As the line spacing offset distance, it is iterated until it approaches v. max The u-direction of the contact point is selected as the x-direction (feed direction), and the normal vector is selected as the z-direction to obtain a machining trajectory sequence containing attitude information.

5. The large component robotic compliant surfacing quality control method of any of claims 1-4, wherein, The machining posture optimization in step S400 is based on the robot machining comprehensive performance index RCPI, and the objective function is represented as: F r (s base ,θ tool )=max(ω1JCI,ω2SCI,ω3CCI) In the formula, ω1, ω2 and ω3 represent weight coefficients, JCI represents a joint constraint index, SCI represents a singularity constraint index, CCI represents a collision constraint index, S base represents the movement of the robot along the external axis, θ base represents the rotation angle of the end tool around the z axis.

6. The large component robot compliant polishing surface quality control method of claim 5, wherein, The singularity constraint index SCI is calculated by the following formula: where J N is the homogenous Jacobian matrix normalized by the characteristic length, K(J N ) denotes the condition number of the velocity Jacobian matrix.

7. The large component robotic compliant surfacing quality control method of claim 5, wherein, The collision constraint index CCI is calculated by the following formula: where r i and r j are the effective radii of the bounding boxes of the i-th and j-th links of the robot, respectively. ij and Δ ij are the shortest distance between the i-th link of the robot and the j-th link or object, and the preset safety distance, respectively. ij need to be obtained according to the linear equations of the cylindrical bounding boxes of the links in the Cartesian space, i.e.: where c i with c j denotes a point on the link axis equation, λ i with λ j is a control parameter of the axis equation, p0denotes the origin position of the robot base frame, and denote the transformation matrices from the robot base coordinate frame {B} to the {n} and {n+1} joint coordinate frames, respectively.

8. The large component robotic compliant surfacing quality control method of any of claims 1-7, wherein, In step S500, the "adaptive control of grinding and polishing contact force through the compliant device" specifically includes: the system expands the working range by means of the mobile platform, and adjusts the z-axis extension amount through the compliant force control device to maintain the stability of the contact force in the grinding and polishing process, so as to ensure the stable application of the process parameters to achieve the target removal amount and surface roughness.

9. The large component robotic compliant surfacing quality control method of any of claims 1-8, wherein, In step S600, the "secondary process parameter optimization and trajectory correction for the defect area according to the detection results" specifically includes: after completing the grinding and polishing, the surface quality is detected according to the measurement module, the removal amount of the defect area is judged whether it deviates from the target removal amount, and the surface roughness is judged whether it exceeds the target range, the removal amount that needs to be supplemented in the defect area is calculated, the process parameters are re-optimized and the trajectory is planned for the area with the target removal amount and surface roughness as the core, and secondary grinding and polishing is carried out until the overall large component meets the machining quality requirements.

10. A large component robot compliant surface finishing quality control system, characterized by, It comprises: an offline programming module for generating a grinding and polishing initial trajectory based on a large component design model or a measured point cloud; The process optimization and decision module, based on the compliant polishing removal model, takes the preset target removal amount and surface roughness as the core optimization targets, and adopts the PB-MOPSO multi-objective optimization algorithm to decide the optimal polishing process parameters, which need to meet the target removal amount deviation threshold and surface roughness deviation threshold requirements. The trajectory optimization module, based on the optimal process parameters, generates the uniform removal polishing trajectory by fusing the contact deformation and removal profile in the compliant polishing, and ensures that the material removal amount matches the target removal amount, the surface texture meets the surface roughness requirements, and the uniformity meets the polishing quality requirements during the trajectory execution process. The configuration optimization module optimizes the redundant machining configuration by combining the joint performance, singularity avoidance and anti-collision index, and generates the optimal robot joint trajectory sequence and base position sequence. The global / local compliant polishing operation module, for large component polishing requirements, adopts the final polishing program generated by the process optimization module and the trajectory and configuration optimization module, and performs polishing operation through the compliant polishing device. The measurement and detection module detects the surface quality of the polished component, judges whether the removal amount and surface roughness meet the standards according to the detection results, and re-optimizes the process parameters and corrects the trajectory for the defective areas that do not meet the standards, to realize closed-loop quality control.

Images