Precise control method for PCB multi-axis linkage motion coordination and error suppression

Abstract

The invention provides a precise control method for PCB multi-axis linkage motion coordination and error suppression, and the method comprises the steps: extracting a track drift feature from a corrected deviation estimation sequence, constructing a dynamic coordination model for the linear driving characteristics of a multi-axis linkage structure, adjusting the internal weight of the model through an optimization algorithm, and carrying out the optimization of the linear driving characteristics of the multi-axis linkage structure. Obtaining an optimized coordination parameter set; according to the optimized coordination parameter set, virtual torque compensation is applied to geometric interference points of multi-axis configuration and layout, the matching degree of a compensation vector and actual displacement is judged through real-time feedback circulation, and a compensated torque balance state is obtained; and deriving a multi-axis synchronization instruction sequence from the compensated moment balance state, and determining a synchronization error minimization scheme through signal path difference calibration.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, and in particular to a precision control method for PCB multi-axis linkage motion coordination and error suppression. BACKGROUND

[0002] The field of precision motion control occupies a core position in high-end manufacturing and automation equipment, and its ability to achieve high-precision, multi-degree-of-freedom coordinated motion directly determines product quality and production efficiency.

[0003] PCB multi-axis linkage often has too strong coupling between the drive system and the guide rail configuration, resulting in response delay and vibration accumulation when executing complex trajectories, making it difficult to meet the stability requirements under dynamic load.

[0004] This coupling further amplifies the transmission path difference of the control signal, causing the inter-axis synchronization error to increase sharply when switching directions at high speed.

[0005] PCB six-axis double-X double-Y full-linearity structure as the key architecture in this field, the core difficulty lies in the conflict between the rigidity requirement of full-linearity drive and the flexibility requirement of multi-axis parallel coordination.

[0006] Full-linearity drive requires each axis to use an independent linear motor to eliminate transmission gaps, but under the double-X double-Y layout, the intersection area of the X-axis and the Y-axis will produce geometric interference, causing the motor thrust vector to be unevenly distributed at the intersection point, and further causing inter-axis torque imbalance.

[0007] This imbalance is amplified when the six-axis is coordinated as a whole, forming a nonlinear deviation between the control command and the actual displacement.

[0008] Specifically, when machining large curved surface parts, the system needs to drive the double-X axis to achieve wide translation, the double-Y axis to complete local fine adjustment, and the remaining two axes to adjust the posture.

[0009] If the torque imbalance is not suppressed, the X-axis thrust overload will cause the Y-axis guide rail to twist at the micron level, and the cumulative effect at the end of the six-axis will be a trajectory drift of tens of microns, directly causing the surface roughness to exceed the standard.

[0010] Therefore, how to eliminate the inter-axis torque imbalance and nonlinear deviation under the six-axis double-X double-Y full-linearity architecture has become a key problem for achieving micron-level precision coordinated motion. SUMMARY

[0011] The present application provides a precision control method for PCB multi-axis linkage motion coordination and error suppression, mainly including: Obtaining real-time inter-axis thrust vector data, collecting displacement signals and torque signals from displacement sensors and torque sensors in a multi-axis intersection area, determining an inter-axis torque imbalance distribution pattern by fusing the multi-source input signals, and obtaining an initial imbalance mapping matrix; According to the initial imbalance mapping matrix, a filtering algorithm is used to iteratively estimate the nonlinear displacement deviation, and the machining trajectory parameters are integrated into the estimation process, and if the deviation exceeds a preset threshold, a compensation mechanism is activated to obtain a corrected deviation estimation sequence. From the corrected deviation estimation sequence, trajectory drift features are extracted, a dynamic coordination model is constructed for the linear drive characteristics of the multi-axis linkage structure, the internal weights of the model are adjusted through an optimization algorithm, and an optimized coordination parameter set is obtained. According to the optimized coordination parameter set, virtual torque compensation is applied to the geometric interference points of the multi-axis configuration and layout, and the matching degree of the compensation vector and the actual displacement is judged through real-time feedback loop to obtain a compensated torque balance state. From the compensated torque balance state, a multi-axis synchronization instruction sequence is derived, and a synchronization error minimization scheme is determined through signal path difference calibration. According to the synchronization error minimization scheme, a precise motion trajectory planning path is generated, an inter-axis imbalance suppression module is embedded in the planning path, an execution instruction set under stable dynamic load is obtained, and is output to the control system interface. A closed-loop monitoring mechanism is used to process the residual components of the trajectory drift to determine the coordinated motion output response.

[0012] The technical scheme provided by the embodiment of the application can include the following beneficial effects: The application discloses a precise control method for motion coordination and error suppression of a PCB multi-axis linkage system, and solves the problem of insufficient motion precision caused by torque imbalance and trajectory drift in a multi-axis intersection area. By fusing high-precision signals of displacement sensors and torque sensors, the application constructs an initial imbalance mapping matrix to accurately describe the inter-axis torque distribution pattern. A filtering algorithm is used to iteratively estimate nonlinear displacement deviation, combined with dynamic adjustment of machining trajectory parameters, to activate a compensation mechanism to generate a corrected deviation sequence. Trajectory drift features are extracted therefrom, a dynamic coordination model is constructed and the weights are optimized to obtain a coordination parameter set, and virtual torque compensation is applied to eliminate the influence of geometric interference. Through real-time feedback loop calibration, synchronization error is generated, precise motion trajectory planning path is embedded with imbalance suppression module, and finally stable execution instruction set is output. A closed-loop monitoring mechanism further processes residual drift to ensure high-precision coordinated motion output. The application significantly improves the motion precision and stability of the multi-axis linkage system under dynamic load, and is suitable for high-precision machining scenes. BRIEF DESCRIPTION OF DRAWINGS

[0013] Fig. 1A flow chart of a PCB multi-axis linkage motion coordination and error suppression precision control method of the present application.

[0014] Fig. 2 A schematic diagram of a PCB multi-axis linkage motion coordination and error suppression precision control method of the present application.

[0015] Fig. 3 Another schematic diagram of a PCB multi-axis linkage motion coordination and error suppression precision control method of the present application. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present application will be clearly and detailedly described below with reference to the drawings in the embodiments of the present application. The described embodiments are only some embodiments of the present application.

[0017] As Figs. 1-3 , the PCB multi-axis linkage motion coordination and error suppression precision control method of the present embodiment can specifically include: In step S101, real-time inter-axis thrust vector data is obtained, displacement signals and moment signals are collected from displacement sensors and moment sensors in a multi-axis intersection area, an inter-axis moment imbalance distribution pattern is determined by fusing the multi-source input signals, and an initial imbalance mapping matrix is obtained.

[0018] The displacement sensor and moment sensor data of the multi-axis intersection area are acquired, the displacement signals and moment signals are collected, and the original signal data set is generated. The original signal data set is preprocessed by a Kalman filtering algorithm to eliminate noise interference and obtain a smooth signal data set. The smooth signal data set is feature extracted by a principal component analysis algorithm to determine the main distribution characteristics of the inter-axis moment, and a feature vector set is generated. If the variance contribution rate of the feature vector set is greater than a preset threshold, the feature vector set is mapped to a low-dimensional space to obtain a dimension-reduced feature matrix. According to the dimension-reduced feature matrix, the deviation values of the inter-axis moments are calculated, and an initial imbalance mapping matrix is generated. The initial imbalance mapping matrix is used to determine the inter-axis moment imbalance distribution pattern and generate a distribution pattern vector. If the deviation value of the distribution pattern vector exceeds a preset range, the initial imbalance mapping matrix is iteratively optimized to obtain an optimized mapping matrix.

[0019] For example, in one possible implementation, for the multi-axis intersection region displacement sensor and torque sensor data acquisition process, an assembly line scenario for an industrial robot arm can be considered, where the multi-axis intersection region refers to the jointed parts of the robot arm that are equipped with displacement sensors to monitor the axial movement distance and torque sensors to detect the magnitude of torsional force. When collecting displacement signals, the sensors sample 100 times per second, recording the displacement values of the X, Y, Z axes, such as the X-axis displacement varying from 0 to 5 millimeters; at the same time, torque signals are collected, recording the torque values of each axis, for example, the Z-axis torque fluctuating between 10 and 15 newton-meters. These signal combinations form the original signal data set, a typical data set may contain thousands of data points, each point including the vector representation of displacement and torque, providing basic data support for subsequent analysis.

[0020] In one possible implementation, the original signal data set is then preprocessed by the Kalman filter algorithm to eliminate noise interference. Kalman filtering is a recursive algorithm that estimates the true state of a system based on a state-space model, the principle is to calculate the expected state of the system through the prediction step, and then correct the prediction error in the update step using the measurement value.

[0021] Specifically, in the example of a robot arm, assuming that there is noise in the original signal due to vibration or electromagnetic interference, first establish a state model, taking displacement and torque as state variables, the prediction equation considers the dynamic characteristics of the system such as acceleration influence, and the measurement equation incorporates sensor readings. Process noise and measurement noise are represented by covariance matrices, for example, set the process noise covariance to 0.1 and the measurement noise to 0.5. Then, the algorithm iteratively calculates the Kalman gain, which is used to fuse the prediction and measurement, and finally obtains a smoothed signal data set, in which the noise is significantly reduced, such as the displacement fluctuation in the original signal from ±2 millimeters to ±0.5 millimeters, which helps to improve data accuracy and achieve more stable robot control effect in business.

[0022] For example, when using principal component analysis algorithm for feature extraction of smoothed signal data set, principal component analysis is a statistical method used to convert high-dimensional data into low-dimensional representation while preserving the main variation information. Its principle is to calculate the eigenvalues and eigenvectors of the data covariance matrix, and select the principal components with high contribution rate. In the robot arm scenario, apply this algorithm to the smoothed signal data set, first standardize the data to have a mean of 0 and a variance of 1, then calculate the covariance matrix, solve the eigenvalues, for example, the first three eigenvalues are 80%, 15%, and 3%, respectively, to determine the main distribution characteristics of the inter-axis torque, such as the X-axis torque is highly correlated with the Y-axis displacement, generate a set of feature vectors, which capture the main patterns of the data.

[0023] In a possible implementation, if the variance contribution rate of the feature vector set is greater than a preset threshold, such as 85%, the feature vector set is mapped to a low-dimensional space. The specific process involves selecting the first k principal components, where k is determined by the cumulative contribution rate, for example, up to 90%, and then projecting the original data onto these principal components through linear transformation to obtain a dimension-reduced feature matrix. In an example, this can reduce the original 10-dimensional data to 3 dimensions, simplifying the analysis complexity of the inter-axis moment of force.

[0024] For example, when calculating the deviation value of each inter-axis moment of force according to the dimension-reduced feature matrix, the deviation value can be measured by Euclidean distance or cosine similarity, for example, the deviation between the X-axis and Y-axis moments of force is 2.5 Nm, and an initial imbalance mapping matrix is generated, which is a matrix representing the imbalance of each axis.

[0025] In a possible implementation, the inter-axis moment of force imbalance distribution pattern is determined by the initial imbalance mapping matrix, and a distribution pattern vector is generated, for example, the vector elements represent the imbalance degree, such as [0.8, 0.3, 1.2] corresponding to each axis.

[0026] For example, if the deviation value of the distribution pattern vector exceeds a preset range, such as greater than 1.0, the initial imbalance mapping matrix is iteratively optimized, and the matrix elements are gradually adjusted using the gradient descent method until the deviation converges, obtaining an optimized mapping matrix, thereby achieving more accurate moment of force balance control in the robot arm business and improving assembly efficiency.

[0027] High-precision displacement signals of multi-axis intersection area displacement sensors and high-frequency moment of force signals of moment of force sensors are obtained, and the inter-axis moment of force imbalance distribution pattern is determined through signal fusion processing to obtain an initial imbalance mapping matrix.

[0028] High-precision displacement signals of multi-axis intersection area displacement sensors and high-frequency moment of force signals of moment of force sensors are obtained, and a fusion signal dataset is generated through signal fusion. Fast Fourier transform is used to analyze the frequency domain of the fusion signal dataset to obtain frequency domain feature vectors. The frequency domain feature vectors are processed by principal component analysis for dimension reduction to generate a low-dimensional feature matrix. If the variance contribution rate of the low-dimensional feature matrix is greater than a preset threshold, the inter-axis moment of force deviation value is calculated according to the low-dimensional feature matrix to obtain a moment of force deviation vector. The initial imbalance mapping matrix is constructed by the moment of force deviation vector to determine the inter-axis moment of force imbalance distribution pattern. If the deviation value of the initial imbalance mapping matrix exceeds a preset range, the initial imbalance mapping matrix is iteratively optimized using the gradient descent method to obtain an optimized mapping matrix. The final inter-axis moment of force imbalance distribution pattern is generated according to the optimized mapping matrix, and a distribution pattern vector is obtained.

[0029] In an example, in an industrial robot arm assembly line scenario, the joint of the multi-axis intersection region collects data through high-precision displacement sensors and high-frequency torque sensors to determine the inter-axis torque imbalance distribution pattern and generate an initial imbalance mapping matrix. Assuming that the joint region of the robot arm is equipped with displacement sensors, the sampling frequency is 200 times per second, and the displacement values of the X, Y, and Z axes are recorded, for example, the X-axis displacement varies in the range of -3 mm to 7 mm, the Y-axis displacement varies in the range of 0 mm to 4 mm, and the Z-axis displacement varies in the range of -1 mm to 2 mm. At the same time, the torque sensor collects the torque data of each axis at a frequency of 500 times per second, for example, the X-axis torque fluctuates in the range of 8 Nm to 12 Nm, the Y-axis torque fluctuates in the range of 5 Nm to 9 Nm, and the Z-axis torque fluctuates in the range of 15 Nm to 20 Nm. These data form a high-dimensional data set containing tens of thousands of data points, each data point consisting of a displacement vector [X, Y, Z] and a torque vector [T x, T y, T z]. Then, the original data set is preprocessed using an extended Kalman filter algorithm to process nonlinear dynamic noise. The algorithm constructs a nonlinear state transition model, takes displacement and torque as state variables, sets the process noise covariance to 0.05, and the measurement noise covariance to 0.3, and generates a smoothed data set by iteratively updating the state estimate.

[0030] For example, after dimensionality reduction, the data points are reduced from 6 dimensions to 2 dimensions, retaining the main torque change pattern. Subsequently, by calculating the cosine similarity of the torque of each axis in the reduced feature matrix, the inter-axis deviation is quantified, for example, the torque deviation between the X-axis and the Z-axis is 3.2 Nm, and an initial imbalance mapping matrix is generated, represented as a 3x3 matrix, where the diagonal elements are 0 and the non-diagonal elements represent the deviation values such as [0, 3.2, 1.8; 3.2, 0, 2.4; 1.8, 2.4, 0]. Finally, based on this matrix, the least squares method is used to optimize the inter-axis torque distribution to generate a distribution pattern vector, for example, [0.7, 0.4, 1.1], which represents the imbalance degree of each axis and is used for subsequent robot control system adjustment.

[0031] Step S102, according to the initial imbalance mapping matrix, a filtering algorithm is used to iteratively estimate the nonlinear displacement deviation, and the machining trajectory parameters are integrated into the estimation process. If the deviation exceeds the preset threshold, the compensation mechanism is activated, and the corrected deviation estimation sequence is obtained.

[0032] The displacement sensor data and machining trajectory parameters of the multi-axis intersection region are obtained, the displacement signal is preprocessed by a Kalman filtering algorithm, and a smooth displacement sequence is obtained. If the deviation value of the smooth displacement sequence exceeds the preset threshold, the nonlinear deviation is calculated according to the machining trajectory parameters to generate an initial deviation estimation sequence. The least square method is used to iteratively optimize the initial deviation estimation sequence to obtain an optimized deviation sequence. Through matching analysis of the optimized deviation sequence and the machining trajectory parameters, a deviation distribution feature vector is determined. If the deviation degree of the deviation distribution feature vector exceeds the preset range, a compensation mechanism is activated to generate a compensation deviation sequence. According to the correlation between the compensation deviation sequence and the machining trajectory parameters, a final deviation correction sequence is determined. Through the final deviation correction sequence, the deviation control parameters of the multi-axis system are generated.

[0033] For example, in the precision machining scene of an industrial robot arm, based on the initial imbalance mapping matrix, the nonlinear displacement deviation is iteratively estimated by using an extended Kalman filtering algorithm. Assuming that the robot arm is in the machining trajectory of X, Y and Z axes, the initial imbalance mapping matrix records the deviation between the axes, for example, the X-Y axis deviation is 3.2 mm, and the Y-Z axis is 2.8 mm. The extended Kalman filter processes the displacement deviation through a nonlinear state transition model. First, a state vector is constructed, including the displacement deviation and the machining speed, the initial value is [3.2, 2.8, 0.5] mm and [0.1, 0.2, 0.3] mm / s. The state transition equation considers the nonlinear dynamics of the machining trajectory, such as the quadratic curve trajectory equation x(t)=0.5t^2+2t, to predict the deviation at the next time. The measurement model is based on the displacement sensor data, which is sampled 50 times per second, for example, the X-axis deviation measurement value is 3.25 mm, and the noise covariance is set to 0.2. The algorithm linearizes the nonlinear model through the Jacobian matrix, iteratively calculates the state estimation, and obtains the deviation sequence, for example, the X-axis deviation is adjusted from 3.2 mm to 3.15 mm. When the machining trajectory parameters are integrated, the trajectory curvature radius is set to 50 mm, the speed constraint is 0.5 mm / s, and the weighted factor affecting state prediction is 0.8. If the deviation exceeds the preset threshold of 2.0 mm, the compensation mechanism is activated, and the compensation vector is calculated by the inverse kinematics algorithm, for example, the X-axis compensation amount is -0.15 mm. The compensation mechanism generates a corrected deviation estimation sequence by adjusting the servo motor input in real time, for example, the final sequence is [3.05, 2.75, 0.45] mm. Through multiple iterations, the deviation converges within the threshold, forming a stable sequence, which provides data support for the machining trajectory optimization of the robot arm.

[0034] In step S103, trajectory drift features are extracted from the corrected deviation estimation sequence, a dynamic coordination model is constructed for the linear drive characteristics of the multi-axis linkage structure, the internal weights of the model are adjusted by an optimization algorithm, and an optimized coordination parameter set is obtained.

[0035] The trajectory drift vector is obtained, the independent motion components of the multi-axis linkage structure are separated by a motion decomposition algorithm, and the inter-axis motion sequence is obtained. According to the matching analysis of the inter-axis motion sequence and the linear driving parameters, the inter-axis coupling coefficient is calculated, and the motion coordination deviation is determined. If the motion coordination deviation exceeds the preset threshold, the internal weight of the dynamic coordination model is optimized by the gradient descent algorithm, and the adjusted weight set is obtained. The dynamic coordination model is updated by using the adjusted weight set, and the coordinated motion parameters are generated. Through the correlation analysis of the coordinated motion parameters and the trajectory drift vector, the trajectory correction vector of the multi-axis system is calculated. According to the trajectory correction vector, the driving control signal of the multi-axis linkage structure is generated, and the final coordinated control parameter is determined. Through the final coordinated control parameter, the motion execution instruction of the multi-axis system is updated, and the optimized motion trajectory is obtained.

[0036] For example, when extracting the trajectory drift feature from the corrected deviation estimation sequence, assuming that the industrial robot arm is in a three-axis linkage machining scene, the deviation sequence is [3.05, 2.75, 0.45] mm, the feature is extracted by Fourier transform to analyze the frequency components of the sequence, the frequency of the periodic drift is calculated as 0.1 Hz, and the amplitude is 0.3 mm. The feature vector [0.1, 0.3] is constructed to represent the periodicity and intensity of the drift. Then, a dynamic coordination model is constructed for the linear driving characteristics of the multi-axis linkage structure, and a state space model is used to represent the coupling relationship between the axes. The state vector includes the positions and velocities of the axes, and the initial values are [3.05, 2.75, 0.45, 0.2, 0.15, 0.1] with units of mm and mm / s respectively. The state transition matrix is constructed based on the stiffness coefficient 0.85 and the damping coefficient 0.12 of the linear drive. The optimization algorithm selects the particle swarm optimization algorithm, sets the particle number to 50 and the iteration number to 100, and the objective function is to minimize the coordination error between the axes. The initial weight is [0.5, 0.3, 0.2], and the weight is adjusted by iteration to obtain the optimized coordination parameter set [0.62, 0.28, 0.1], in which the weight of X axis increases to reflect its dominant role. The analysis process first quantifies the drift feature by Fourier transform to ensure that the feature vector accurately captures the periodic fluctuations, and then the dynamic coordination model simulates the multi-axis coupling by the state space method, and the particle swarm optimization algorithm iteratively adjusts the weight based on the sum of squared errors. The error converges from the initial 0.25 mm^2 to 0.08 mm^2. Logically, feature extraction provides input for the model, the dynamic model describes the system behavior, and the optimization algorithm adjusts the parameters to improve coordination, forming a closed-loop optimization process. If the frequency components are insufficient in feature extraction, the sampling rate of the sensor can be increased to 100 times per second to supplement the data and ensure the integrity of the model input.

[0037] By extracting the trajectory drift feature from the deviation estimation sequence, the motion trajectory and error analysis of the multi-axis linkage structure are obtained, and the initial parameters of the dynamic coordination model are obtained.

[0038] The trajectory drift vector is obtained, the independent motion components of the multi-axis linkage structure are separated by a motion decomposition algorithm, and the inter-axis motion sequence is obtained. According to the inter-axis motion sequence, the motion synchronization coefficient between each axis is calculated, and the synchronization deviation distribution is obtained. If the synchronization deviation distribution exceeds the preset threshold, the internal parameters of the dynamic coordination model are optimized by the least square method, and the adjusted parameter set is obtained. The dynamic coordination model is updated using the adjusted parameter set, and the coordinated motion instruction set is generated. Through the matching analysis of the coordinated motion instruction set and the trajectory drift vector, the trajectory adjustment vector of the multi-axis system is calculated. According to the trajectory adjustment vector, the control signal sequence of the multi-axis linkage structure is generated, and the final motion control parameters are determined. The final motion control parameters are used to update the execution instruction sequence of the multi-axis system, and the optimized motion trajectory sequence is obtained.

[0039] For example, in the industrial robot multi-axis linkage processing scene, when extracting the trajectory drift feature from the deviation estimation sequence, assuming that the deviation sequence is [2.8, 1.9, 0.6] mm, the wavelet transform method is used to analyze the sequence, and the high-frequency and low-frequency components are obtained. The main frequency of the periodic drift is 0.15 Hz, and the amplitude is 0.25 mm. The feature vector [0.15, 0.25] is constructed to represent the drift periodicity and amplitude intensity. Then, according to the motion characteristics of the multi-axis linkage structure, a dynamic coordination model is constructed, and a Kalman filter is used to describe the dynamic coupling between axes. The state vector includes position and acceleration, and the initial value is [2.8, 1.9, 0.6, 0.18, 0.12, 0.08] with units of mm and mm / s². The state transition matrix is constructed according to the system stiffness coefficient 0.9 and the damping coefficient 0.15. The optimization process uses a genetic algorithm with a population size of 60 and an iteration number of 120. The objective function is to minimize the sum of squared errors of the trajectory. The initial weight is [0.4, 0.35, 0.25], and the parameter set [0.55, 0.30, 0.15] is obtained after iterative optimization, reflecting the enhanced dominance of the X-axis. The analysis process first extracts the drift feature through wavelet transform to ensure the capture of periodicity and transient fluctuations. Then, the Kalman filter simulates the multi-axis dynamic response, and the genetic algorithm iteratively optimizes the weight. The error converges from the initial 0.22 mm² to 0.07 mm². If the high-frequency component is insufficient in feature extraction, the sensor sampling rate can be increased to 120 times / s to enhance data resolution and ensure complete model input. The entire process forms a closed loop: feature extraction provides accurate input, dynamic model describes system behavior, optimization algorithm adjusts parameters to improve coordination, and the logic is rigorous and mutually related.

[0040] In step S104, according to the optimized coordination parameter set, a virtual torque compensation is applied to the geometric interference points of the multi-axis configuration and layout. The matching degree of the compensation vector and the actual displacement is judged by using a real-time feedback loop, and a compensated torque balance state is obtained.

[0041] identify the geometric interference point position in the multi-axis configuration according to the optimized coordination parameter set, calculate the virtual torque compensation direction and amplitude from the geometric interference point position, generate an initial compensation vector using the virtual torque compensation, obtain actual displacement data through a real-time feedback loop, calculate displacement matching degree according to the actual displacement data and the initial compensation vector, if the displacement matching degree is lower than a preset threshold, adjust the virtual torque amplitude to obtain an updated compensation vector, and recalculate the torque balance state using the updated compensation vector.

[0042] For example, based on the optimized coordination parameter set [0.62, 0.28, 0.1], first, the geometric interference point of the four-axis linkage robot arm is mapped in space, the D-H parameter method is used to construct the coordinate system of each axis link, the closest distance between joint 4 and joint 2 is calculated to be 12.4 mm, the interference point coordinates are determined to be [45.2, 18.7, 30.1] mm, and a virtual force field is constructed with this as the center, and a torque compensation vector is applied. The X-axis compensation torque is 0.62x0.8=0.496 Nm, the Y-axis is 0.28x0.6=0.168 Nm, and the Z-axis is 0.1x0.4=0.04 Nm, forming an initial compensation vector [0.496, 0.168, 0.04] Nm. Subsequently, Analyzing- is started. Real-time feedback loop, the sensor collects actual displacement data at a frequency of 50 Hz, the sequence is [0.51, 0.15, 0.05] mm, and the compensation vector is fused through a Kalman filter. The filter gain matrix is set to diag[0.7, 0.65, 0.6], the updated estimated displacement is [0.49, 0.17, 0.042] mm, and the matching degree is calculated using the Euclidean distance formula √[(0.496-0.49)²+(0.168-0.17)²+(0.04-0.042)²]=0.008 Nm. When the distance is less than the 0.01 Nm threshold, it is determined to be matched, and enters the torque balance state. The analysis process shows that the virtual torque compensation generates a directional force field according to the geometric interference point, the feedback loop corrects the deviation in real time through filtering fusion, the matching degree quantifies the compensation accuracy, and finally the system converges to the balance state within 0.05 seconds, the torque residual error is reduced to 0.003 Nm, and an adaptive compensation chain is formed.

[0043] Step S105, derive a multi-axis synchronization instruction sequence from the compensated torque balance state, and determine a synchronization error minimization scheme through signal path difference calibration.

[0044] Analyzing -sequence is obtained according to the compensation torque balance state. The signal path difference calculation method is used to determine the signal delay parameters of each axis. The calibration delay compensation amount is obtained through the path calibration process. If the calibration delay compensation amount exceeds the preset threshold, the instruction sequence timestamp is adjusted to obtain an updated synchronization sequence. The synchronization error value is obtained from the updated synchronization sequence. The linear regression algorithm is used to determine the correlation between the synchronization error value and the path difference parameter to obtain an error prediction model. The synchronization error minimization scheme is determined through the error prediction model.

[0045] For example, based on the stable output of the torque balance state, the system automatically generates a multi-axis synchronization instruction sequence, and uses a timestamp interpolation algorithm to discretize each axis motion instruction at 2 millisecond intervals. The axis 1 instruction sequence is [1.2, 1.4, 1.6, 1.8] radians, the axis 2 is [0.8, 1.0, 1.2, 1.4] radians, the axis 3 is [2.1, 2.3, 2.5, 2.7] radians, and the axis 4 is [0.5, 0.7, 0.9, 1.1] radians. A cubic spline interpolation function is used to ensure the continuity and smoothness of the instruction sequence. The signal path difference calibration module is then started, and a high-precision clock reference is used to measure the signal transmission delay of each axis controller. The axis 1 delay is 3.2 milliseconds, the axis 2 is 4.1 milliseconds, the axis 3 is 2.8 milliseconds, and the axis 4 is 3.7 milliseconds. A delay compensation matrix [3.2, 4.1, 2.8, 3.7] milliseconds is established, and a delay prediction model y = 0.85x + 2.1 is obtained by least squares fitting, where x is the axis number. The system uses a genetic algorithm to optimize the synchronization error minimization scheme, with a population size of 50, a crossover probability of 0.8, and a mutation probability of 0.02. After 120 generations of evolution, the optimal solution is converged, and the axis timing adjustment parameters are [0.9, 0.7, 1.1, 0.8]. After applying the parameters, the multi-axis synchronization error is reduced from the initial 8.5 milliseconds to 1.2 milliseconds, meeting the requirements of high-precision coordinated motion, and forming a closed-loop synchronization control system.

[0046] Through torque balance state analysis, a multi-axis synchronization instruction sequence is obtained, and an inter-axis coordinated control scheme is obtained.

[0047] Torque balance state data is obtained, and multi-axis motion parameters are extracted using signal processing methods. An initial synchronization instruction sequence is generated based on the multi-axis motion parameters. A time series analysis method is used to determine the timing deviation of each axis in the initial synchronization instruction sequence. If the timing deviation exceeds the preset threshold, the timestamp of the synchronization instruction sequence is adjusted to obtain an optimized synchronization sequence. Through the optimized synchronization sequence, the coordinated control parameters of inter-axis motion are calculated. A logistic regression algorithm is used to analyze the correlation between the coordinated control parameters and the torque balance state to determine an inter-axis coordinated control scheme. The final multi-axis synchronization control instruction is generated through the inter-axis coordinated control scheme.

[0048] For example, the system first collects real-time torque data of each axis under dynamic load through torque balance state analysis. Assuming that the torque of the four-axis system is [10.5, 12.3, 9.8, 11.2] Newton-meters, a torque balance model is established based on the Newton-Euler equation to calculate the angular acceleration of each axis, and [0.75, 0.82, 0.68, 0.79] rad / s2 is obtained. Then, the system uses Fourier transform to decompose the torque data into frequency components, extracts signals in the main frequency range of 0.1-5 Hz, generates a multi-axis synchronous instruction sequence, and sets the instruction interval to 1.5 milliseconds. The sequence of axis 1 is [2.0, 2.2, 2.4, 2.6] radians, axis 2 is [1.5, 1.7, 1.9, 2.1] radians, axis 3 is [3.0, 3.2, 3.4, 3.6] radians, and axis 4 is [1.0, 1.2, 1.4, 1.6] radians. A quintic polynomial interpolation algorithm is used to ensure the smoothness of the sequence. Subsequently, the system starts the path delay analysis module, measures the signal transmission time of each axis based on a high-precision crystal oscillator clock, which is 2.5 milliseconds for axis 1, 3.3 milliseconds for axis 2, 2.1 milliseconds for axis 3, and 2.9 milliseconds for axis 4. A delay matrix [2.5, 3.3, 2.1, 2.9] milliseconds is constructed, and the delay trend is predicted by Kalman filtering algorithm to obtain the model y = 0.92x + 1.8, where x is the axis number, which is used for dynamic compensation of signal timing. Finally, the system uses a particle swarm optimization algorithm to optimize the coordination control between axes, sets the number of particles to 60, the inertia weight to 0.7, and the learning factor to 1.5. After 100 iterations of convergence, the timing correction coefficients of each axis are [0.85, 0.95, 0.80, 0.90], forming a closed-loop coordination control scheme to ensure the synchronization of each axis movement. The entire process is automatically executed by an embedded controller, and data processing and algorithm optimization are completed by the system to generate a coordinated control instruction sequence.

[0049] In step S106, a precise motion trajectory planning path is generated according to the synchronization error minimization scheme, an inter-axis imbalance suppression module is embedded in the planning path, an execution instruction set under stable dynamic load is obtained, and the instruction set is output to a control system interface. A closed-loop monitoring mechanism is used to process trajectory drift residual components and determine a coordinated motion output response.

[0050] The precise trajectory planning path is obtained, and an initial trajectory sequence is generated by using a path generation algorithm. Through an inter-axis imbalance suppression module, the load deviation parameters of each axis are calculated to obtain a load balance adjustment amount. If the load balance adjustment amount exceeds a preset threshold, the trajectory sequence timestamp is adjusted to generate an updated trajectory sequence. The execution instruction set under dynamic load is extracted from the updated trajectory sequence and output to a control system interface. A closed-loop monitoring mechanism is used to obtain a trajectory drift residual component and determine a drift compensation parameter. The execution instruction set is adjusted by the drift compensation parameter to obtain a coordinated motion output response. According to the coordinated motion output response, a support vector machine algorithm is used to generate a response optimization model to determine a final trajectory execution sequence.

[0051] For example, the system first generates a precise motion path by a trajectory planning algorithm, uses a quintic polynomial interpolation method, sets the target positions of each axis to [2.0, 1.5, 3.0, 0.6] radians for a four-axis system, and generates a trajectory point sequence with a time interval of 3 milliseconds. The trajectory point of axis 1 is [0.0, 0.5, 1.0, 1.5, 2.0] radians, the trajectory point of axis 2 is [0.0, 0.4, 0.8, 1.1, 1.5] radians, the trajectory point of axis 3 is [0.0, 0.8, 1.6, 2.2, 3.0] radians, and the trajectory point of axis 4 is [0.0, 0.2, 0.3, 0.4, 0.6] radians, ensuring that the path is smooth and meets the maximum speed constraint of 2 radians / s. Next, the inter-axis imbalance suppression module is embedded, the load deviation of each axis is calculated through dynamic load analysis, and it is assumed that the loads of axis 1 to axis 4 are [5.0, 4.8, 5.2, 4.9] Newtons. A proportional-integral control algorithm is used with a proportional gain of 0.6 and an integral gain of 0.05 to generate compensation instructions and adjust the output torque of each axis to [0.3, 0.2, 0.4, 0.25] Newton-meters to suppress inter-axis imbalance. Subsequently, the adjusted instruction set is output through the control system interface using the CAN bus protocol with data packets sent at 500 microsecond intervals containing position and torque information. The closed-loop monitoring mechanism uses high-precision encoder feedback with a sampling frequency of 1 kHz to detect trajectory drift and calculate the deviation of each axis to be [0.02, 0.03, 0.01, 0.04] radians. A Kalman filter algorithm is applied with a process noise covariance of 0.001 and a measurement noise covariance of 0.005 to estimate the true trajectory and generate correction instructions. Finally, through the real-time coordinated motion response module, the final output sequence is generated based on the deviation-corrected instruction set, and the velocity of each axis is adjusted to [1.8, 1.4, 2.0, 1.2] radians / s to ensure the stability and precision of the system under dynamic load. The above process is automatically implemented by the algorithm, the logic is rigorous, and the steps are connected to each other to form a complete trajectory planning and control system.

[0052] The above description is only the preferred embodiment of one or more embodiments of the specification, and is not used to limit one or more embodiments of the specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of the specification should be included in the protection range of one or more embodiments of the specification.

Claims

1. A precision control method of PCB multi-axis linkage motion coordination and error suppression, characterized in that, The method comprises: acquiring real-time inter-axis thrust vector data, collecting displacement signals and torque signals from displacement sensors and torque sensors of a multi-axis intersection area, determining an inter-axis torque imbalance distribution pattern by fusing the multi-source input signals, and obtaining an initial imbalance mapping matrix; According to the initial imbalance mapping matrix, a filtering algorithm is used to iteratively estimate the nonlinear displacement deviation, and the machining trajectory parameters are integrated into the estimation process. If the deviation exceeds the preset threshold, the compensation mechanism is activated, and a corrected deviation estimation sequence is obtained; From the corrected deviation estimation sequence, extract the trajectory drift characteristics, construct a dynamic coordination model according to the linear driving characteristics of the multi-axis linkage structure, adjust the internal weights of the model through an optimization algorithm, and obtain an optimized coordination parameter set; According to the optimized coordination parameter set, a virtual torque compensation is applied to the geometric interference points of the multi-axis configuration and layout, and the matching degree of the compensation vector and the actual displacement is judged by using a real-time feedback loop, and a compensated torque balance state is obtained; From the compensated torque balance state, derive a multi-axis synchronization instruction sequence, and determine a synchronization error minimization scheme through signal path difference calibration; According to the synchronization error minimization scheme, a precise motion trajectory planning path is generated, an inter-axis imbalance suppression module is embedded in the planning path, an execution instruction set under stable dynamic load is obtained, and the execution instruction set is output to a control system interface. A closed-loop monitoring mechanism is used to process the trajectory drift residual component to determine the coordinated motion output response.

2. The method of claim 1, wherein, The method comprises: acquiring real-time inter-axis thrust vector data, collecting displacement signals and torque signals from displacement sensors and torque sensors of a multi-axis intersection area, determining an inter-axis torque imbalance distribution pattern by fusing the multi-source input signals, and obtaining an initial imbalance mapping matrix, comprising: acquiring displacement sensor and torque sensor data of the multi-axis intersection area, collecting displacement signals and torque signals, and generating an original signal data set; Through the Kalman filtering algorithm, the original signal data set is preprocessed to eliminate noise interference, and a smooth signal data set is obtained; Using principal component analysis algorithm, the smooth signal data set is feature extracted to determine the main distribution characteristics of the inter-axis torque, and a feature vector set is generated; If the variance contribution rate of the feature vector set is greater than the preset threshold, the feature vector set is mapped to a low-dimensional space to obtain a reduced dimension feature matrix; According to the reduced dimension feature matrix, the deviation value of each inter-axis torque is calculated to generate an initial imbalance mapping matrix; Through the initial imbalance mapping matrix, the inter-axis torque imbalance distribution pattern is determined, and a distribution pattern vector is generated; 3. The method of claim 2, wherein, If the deviation value of the distribution pattern vector exceeds the preset range, the initial imbalance mapping matrix is iteratively optimized to obtain an optimized mapping matrix. Further comprising: acquiring high-precision displacement signals of the displacement sensor and high-frequency torque signals of the torque sensor in the multi-axis intersection area, determining the inter-axis torque imbalance distribution pattern through signal fusion processing, and obtaining the initial imbalance mapping matrix, specifically comprising: acquiring high-precision displacement signals of the displacement sensor and high-frequency torque signals of the torque sensor in the multi-axis intersection area, and generating a fusion signal data set through signal fusion; The frequency domain characteristic vector is obtained by performing frequency domain analysis on the fusion signal data set by using fast Fourier transform; The low-dimensional feature matrix is generated by performing dimension reduction processing on the frequency domain characteristic vector by principal component analysis; If the variance contribution rate of the low-dimensional feature matrix is greater than a preset threshold, the inter-axis moment deviation value is calculated according to the low-dimensional feature matrix, and a moment deviation vector is obtained; An initial imbalance mapping matrix is constructed by using the moment deviation vector, and the inter-axis moment imbalance distribution pattern is determined; If the deviation value of the initial imbalance mapping matrix exceeds a preset range, the initial imbalance mapping matrix is iteratively optimized by using a gradient descent method, and an optimized mapping matrix is obtained; The final inter-axis moment imbalance distribution pattern is generated according to the optimized mapping matrix, and a distribution pattern vector is obtained.

4. The method of claim 1, wherein, According to the initial imbalance mapping matrix, a filtering algorithm is used to iteratively estimate the nonlinear displacement deviation, and the machining trajectory parameters are integrated into the estimation process. If the deviation exceeds a preset threshold, a compensation mechanism is activated to obtain a corrected deviation estimation sequence, including: Obtain the displacement sensor data and the machining trajectory parameters of the multi-axis intersection region, and pre-process the displacement signal by using a Kalman filtering algorithm to obtain a smoothed displacement sequence; If the deviation value of the smoothed displacement sequence exceeds a preset threshold, the nonlinear deviation is calculated according to the machining trajectory parameters to generate an initial deviation estimation sequence; The initial deviation estimation sequence is iteratively optimized by using a least squares method to obtain an optimized deviation sequence; The deviation distribution characteristic vector is determined by matching analysis of the optimized deviation sequence and the machining trajectory parameters; If the deviation distribution characteristic vector deviates from a preset range, a compensation mechanism is activated to generate a compensation deviation sequence; According to the correlation between the compensation deviation sequence and the machining trajectory parameters, a final deviation correction sequence is determined; The deviation control parameters of the multi-axis system are generated through the final deviation correction sequence.

5. The method of claim 1, wherein, The trajectory drift features are extracted from the corrected deviation estimation sequence, and a dynamic coordination model is constructed for the linear driving characteristics of the multi-axis linkage structure. The internal weights of the model are adjusted by using an optimization algorithm to obtain an optimized coordination parameter set, including: Obtain the trajectory drift vector, separate the independent motion components of the multi-axis linkage structure by using a motion decomposition algorithm to obtain an inter-axis motion sequence; According to the matching analysis of the inter-axis motion sequence and the linear driving parameters, the inter-axis coupling coefficient is calculated to determine the motion coordination deviation; If the motion coordination deviation exceeds a preset threshold, the internal weights of the dynamic coordination model are optimized by using a gradient descent algorithm to obtain an adjusted weight set; The dynamic coordination model is updated by using the adjusted weight set to generate the coordinated motion parameters; The trajectory correction vector of the multi-axis system is calculated through the correlation analysis of the coordinated motion parameters and the trajectory drift vector; According to the trajectory correction vector, the driving control signal of the multi-axis linkage structure is generated, and the final coordination control parameter is determined; The motion execution instruction of the multi-axis system is updated by using the final coordination control parameter to obtain an optimized motion trajectory.

6. The method of claim 5, wherein, Further comprising: The trajectory drift features are extracted from the deviation estimation sequence, the motion trajectory and error analysis of the multi-axis linkage structure are analyzed, and the initial parameters of the dynamic coordination model are obtained, including: Obtaining a trajectory drift vector, separating independent motion components of a multi-axis linkage structure through a motion decomposition algorithm to obtain an inter-axis motion sequence; According to the inter-axis motion sequence, calculating the motion synchronization coefficient between each axis to obtain a synchronization deviation distribution; If the synchronization deviation distribution exceeds a preset threshold, optimizing the internal parameters of the dynamic coordination model through the least square method to obtain an adjusted parameter set; Using the adjusted parameter set, updating the dynamic coordination model to generate a coordinated motion instruction set; Through matching analysis of the coordinated motion instruction set and the trajectory drift vector, calculating the trajectory adjustment vector of the multi-axis system; According to the trajectory adjustment vector, generating a control signal sequence of the multi-axis linkage structure to determine the final motion control parameters; Through the final motion control parameters, updating the execution instruction sequence of the multi-axis system to obtain an optimized motion trajectory sequence.

7. The method of claim 1, wherein, According to the optimized coordination parameter set, a virtual torque compensation is applied to the geometric interference points of the multi-axis configuration and layout, and a real-time feedback loop is used to judge the matching degree of the compensation vector and the actual displacement to obtain a compensated torque balance state, including: According to the optimized coordination parameter set, the position of the geometric interference point in the multi-axis configuration is identified, the direction and amplitude of the virtual torque compensation are calculated from the geometric interference point position, the initial compensation vector is generated using the virtual torque compensation, the actual displacement data is obtained through the real-time feedback loop, and the displacement matching degree is calculated according to the actual displacement data and the initial compensation vector. If the displacement matching degree is lower than the preset threshold, the virtual torque amplitude is adjusted to obtain an updated compensation vector, and the torque balance state is recalculated using the updated compensation vector.

8. The method of claim 1, wherein, From the compensated torque balance state, a multi-axis synchronization instruction sequence is derived, and a synchronization error minimization scheme is determined through signal path difference calibration, including: According to the compensated torque balance state, a multi-axis synchronization instruction initial Analyzing-sequence is obtained; Using a signal path difference calculation method, the signal delay parameters of each axis are determined; Through the path calibration process, a calibrated delay compensation amount is obtained; If the calibrated delay compensation amount exceeds the preset threshold, the time stamp of the instruction sequence is adjusted to obtain an updated synchronization sequence; The synchronization error value is obtained from the updated synchronization sequence; Using a linear regression algorithm, the correlation between the synchronization error value and the path difference parameter is determined to obtain an error prediction model; Through the error prediction model, a synchronization error minimization scheme is determined.

9. The method of claim 8, wherein, Further including: Through torque balance state analysis, a multi-axis synchronization instruction sequence is obtained to obtain an inter-axis coordination control scheme, specifically including: Obtaining torque balance state data, and using a signal processing method to extract multi-axis motion parameters; Through the multi-axis motion parameters, an initial synchronization instruction sequence is generated; Using a time series analysis method, the time sequence deviation of each axis in the initial synchronization instruction sequence is determined; If the time sequence deviation exceeds the preset threshold, the time stamp of the synchronization instruction sequence is adjusted to obtain an optimized synchronization sequence; Through the optimized synchronization sequence, the coordination control parameters of inter-axis motion are calculated; Using a logistic regression algorithm, the correlation between the coordination control parameters and the torque balance state is analyzed to determine an inter-axis coordination control scheme; Through the inter-axis coordination control scheme, the final multi-axis synchronization control instruction is generated.

10. The method of claim 1, wherein, The precise motion trajectory planning path is generated according to the synchronization error minimization scheme, an inter-axis imbalance suppression module is embedded in the planning path, an execution instruction set under stable dynamic load is obtained, and output to a control system interface, a closed-loop monitoring mechanism is used to process residual components of trajectory drift, determine a coordinated motion output response, and the coordinated motion output response comprises: A precise trajectory planning path is acquired, and an initial trajectory sequence is generated by using a path generation algorithm; Through an inter-axis imbalance suppression module, load deviation parameters of each axis are calculated, and a load balance adjustment amount is obtained; If the load balance adjustment amount exceeds a preset threshold, the trajectory sequence time stamp is adjusted, and an updated trajectory sequence is generated; An execution instruction set under dynamic load is extracted from the updated trajectory sequence and output to a control system interface; A closed-loop monitoring mechanism is used to obtain residual components of trajectory drift and determine drift compensation parameters; The execution instruction set is adjusted by using the drift compensation parameters, and a coordinated motion output response is obtained; According to the coordinated motion output response, a support vector machine algorithm is used to generate a response optimization model, and a final trajectory execution sequence is determined.

Images