Method and system for joint control of multiple servo motors

Abstract

The application discloses a joint control method and system of multiple servo motors, which comprises the following steps: collecting real-time working parameters, load feedback data and target feature information of multiple servo motors, and performing space-time alignment and noise filtering processing, simultaneously extracting dynamic response parameters corresponding to each servo motor, and determining the load coupling coefficients between the servo motors; constructing a collaborative task matrix and generating an initial collaborative control strategy; determining different working process stages corresponding to the multiple servo motors, calculating the corresponding collaborative priority weights, and constructing a phased collaborative control model; obtaining corresponding working state parameters and the current state of the target, dynamically adjusting the collaborative coefficients corresponding to each servo motor, substituting into the phased collaborative control model to generate real-time control instructions corresponding to each servo motor, and performing joint control iterative optimization to generate a closed-loop adaptive joint control mechanism. The application can improve the joint control efficiency of multiple servo motors.

Description

Technical Field

[0001] This invention relates to the field of motor control technology, and more specifically, to a method and system for the joint control of multiple servo motors. Background Technology

[0002] Multi-axis linkage systems (such as CNC machine tools, industrial robots, and precision conveyor lines) place higher demands on the coordination, synchronization, load adaptability, and dynamic response speed of multiple servo motors. In these systems, the operating status of a single servo motor directly affects the overall task execution accuracy. The traditional "independent control + fixed timing" mode can only adjust a single motor according to preset parameters, which cannot adapt to the load coupling relationship between multiple motors (such as load fluctuation transmission and torque distribution imbalance) and dynamic task requirements in real time. This can easily lead to increased synchronization errors between motors, causing the synchronization errors to exceed the allowable range, thereby reducing the control efficiency of multiple servo motors and the processing or operation quality. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a method for the joint control of multiple servo motors, comprising the following steps: Step S1: Collect real-time operating parameters, load feedback data, and target feature information of multiple servo motors in the linkage system, and perform spatiotemporal alignment and noise filtering on the collected real-time operating parameters, load feedback data, and target feature information to generate standardized motor operation feature sets, load feature sets, and target feature sets. Step S2: Extract the dynamic response parameters corresponding to each servo motor based on the motor operation feature set and load feature set, including speed tracking error, torque adjustment delay and position control backlash, and determine the load coupling coefficient between each servo motor; construct a collaborative task matrix by combining the target feature set, and clarify the functional division and timing constraint relationship of each servo motor in the target execution process based on the load coupling coefficient and the collaborative task matrix, and generate the initial collaborative control strategy. Step S3: Based on the initial cooperative control strategy, determine the different working process stages corresponding to multiple servo motors, and calculate the cooperative priority weights corresponding to different engineering stages based on the task requirements corresponding to different working process stages and the motor operation feature set to construct a staged cooperative control model; by real-time monitoring of the working state parameters and the current state of the target corresponding to each servo motor, and combining the staged cooperative control model, dynamically adjust the cooperative coefficients corresponding to each servo motor, including speed cooperative coefficient, torque distribution coefficient and position synchronization coefficient. Step S4: Substitute the adjusted coordination coefficients of each servo motor into the staged collaborative control model to generate real-time control commands for each servo motor, and perform joint control iterative optimization based on the real-time control commands for each servo motor to generate a closed-loop adaptive joint control mechanism.

[0004] Furthermore, the present invention also provides a joint control system for multiple servo motors, including a processor, a memory, and a computer program stored in the memory and executable on the processor, for performing the joint control method for multiple servo motors as described above.

[0005] The beneficial effects of this application are as follows: By collecting real-time operating parameters, load feedback, and target feature information of multiple servo motors, and performing spatiotemporal alignment and noise filtering, the problems of data asynchrony and noise interference leading to collaborative analysis failure in traditional multi-axis control are solved. Spatiotemporal alignment ensures the consistency of data from different motors in both time and spatial logic, avoiding misjudgments of coupling relationships caused by time differences in acquisition. Noise filtering eliminates environmental interference or sensor errors, ensuring the purity of motor operating feature sets, load feature sets, and target feature sets. The standardized dataset provides a reliable foundation for subsequent extraction of dynamic response parameters and analysis of load coupling relationships, laying a data foundation for improving the accuracy of collaborative control and reducing synchronization errors caused by data quality issues. Secondly, by extracting dynamic response parameters, determining the load coupling coefficient, and constructing a collaborative task matrix in conjunction with target features, the problem that traditional "fixed timing" control cannot adapt to the interaction and functional division of multiple motor loads is solved. The combination of speed tracking error and load coupling coefficient makes the initial collaborative control strategy no longer a simple time sequence superposition, but rather clearly defines the functional division of each motor (such as main drive and auxiliary adjustment) and timing constraints (such as the start / stop sequence logic). The collaborative task matrix transforms target requirements into executable motor coordination rules, avoiding increased synchronization errors caused by load imbalance. Compared to traditional methods, this enhances the system's adaptability to dynamic loads. Then, by dividing the work process into stages, calculating coordination priority weights to construct a staged model, and dynamically adjusting coordination coefficients, the problem of traditional control modes being unable to adapt to changes in task stages and real-time state fluctuations is solved. Different work stages (such as acceleration start-up, constant speed operation, and deceleration positioning) have significantly different task requirements. The priority weight settings ensure that resources are allocated to critical stages. Real-time monitoring of motor and target states, and dynamic adjustment of parameters such as speed coordination coefficients, can quickly offset synchronization deviations caused by load fluctuations and external interference. This dynamic adaptation mechanism breaks the rigidity of "fixed parameter" control, allowing the coordination strategy to flexibly adjust with task progress and real-time state, significantly reducing synchronization errors between motors and improving the system's response speed to complex operating conditions. Finally, by substituting coordination coefficients to generate real-time control commands and performing joint control iterative optimization, a closed-loop adaptive mechanism is constructed, solving the problems of traditional control lacking feedback optimization and difficulty in continuously improving coordination accuracy. Real-time control commands translate dynamically adjusted coordination coefficients into specific execution parameters, ensuring strategy implementation. Joint control iterations, by comparing actual execution results with target requirements, reverse-correct the coordination coefficients. This mechanism continuously eliminates accumulated errors, constantly improving the coordination accuracy of multi-motor systems through iteration. Compared to the traditional "open-loop control" mode, the closed-loop adaptive mechanism not only copes with sudden disturbances but also adapts to chronic issues such as system aging and load characteristic changes through long-term optimization, fundamentally improving the control efficiency and operational quality of multi-motor linkage. Attached Figure Description

[0006] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of the joint control method for multiple servo motors in this embodiment; Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of step S1. Detailed Implementation

[0007] The following drawings disclose several embodiments of the present invention. For clarity, many practical details will be described in the following description. However, it should be understood that these practical details are not intended to limit the invention. That is, in some embodiments of the invention, these practical details are not essential. Furthermore, for the sake of simplicity, some conventional structures and components will be shown in the drawings in a simple schematic manner.

[0008] To further understand the invention's content, features, and effects, the following embodiments are provided, and detailed descriptions are given below in conjunction with the accompanying drawings: Reference Figure 1 , Figure 1 This is a flowchart of the joint control method for multiple servo motors in this embodiment. The joint control method for multiple servo motors in this embodiment includes the following steps: Step S1: Collect real-time operating parameters, load feedback data, and target feature information of multiple servo motors in the linkage system, and perform spatiotemporal alignment and noise filtering on the collected real-time operating parameters, load feedback data, and target feature information to generate standardized motor operation feature sets, load feature sets, and target feature sets. In this embodiment of the invention, in a precision CNC machine tool system with three servo motors linked together, real-time working parameters (motor A speed 299.8 r / min, current 3.2 A; motor B speed 300.1 r / min, current 3.1 A; motor C speed 299.7 r / min, current 3.3 A) are collected every 10 ms by the motor's built-in encoder, torque sensor collects load feedback data (motor A torque 4.8 N·m, B torque 5.1 N·m, C torque 4.9 N·m), and vision sensor collects feature information of the target (machined part) (coordinates X 50.02 mm, Y 30.01 mm, Z 10.03 mm). Spatiotemporal alignment is achieved using timestamp synchronization technology (unified to the system clock 16:30:00.000 / 010 / 020). Kalman filtering is used to process the speed data (reducing the speed fluctuation of motor A from ±0.5 r / min to ±0.1 r / min). The 3σ criterion is used to remove outliers in the load data (removing the outlier of 6.2 N·m for motor B). Finally, a standardized feature set is generated: the motor operation feature set includes speed, current, and position; the load feature set includes torque and torque fluctuation; and the target feature set includes coordinates and attitude. The data format is unified as "timestamp-parameter value".

[0009] Step S2: Extract the dynamic response parameters corresponding to each servo motor based on the motor operation feature set and load feature set, including speed tracking error, torque adjustment delay and position control backlash, and determine the load coupling coefficient between each servo motor; construct a collaborative task matrix by combining the target feature set, and clarify the functional division and timing constraint relationship of each servo motor in the target execution process based on the load coupling coefficient and the collaborative task matrix, and generate the initial collaborative control strategy. In this embodiment of the invention, by extracting the motor A speed command of 300 r / min and the actual feedback of 299.8 r / min from the motor operation feature set, the speed tracking error is calculated to be 0.2 r / min; the time difference of 1.2 ms from the torque command being sent to the output reaching 88% (4.224 N·m) is recorded, resulting in a torque adjustment delay of 1.2 ms; the maximum deviation between the position command and feedback is calculated to be 0.02 mm, resulting in a position control backlash of 0.02 mm. Similarly, the dynamic response parameters of motors B and C are obtained. Combining the torque correlation data of the load feature set, the load coupling coefficients of motors A and B are calculated to be -0.6 (the negative sign indicates a reverse load influence), A and C to -0.5, and B and C to -0.7. The part processing task (X-axis feed → Y-axis positioning → Z-axis cutting) is extracted from the target feature set, and a collaborative task matrix is ​​constructed (rows: sub-tasks, columns: motors, elements are adaptation degrees), such as sub-task 1 (X-axis feed) motor A adaptation degree 0.9, B 0.6, C 0.7. Based on the coupling coefficient and matrix, it is determined that motor A dominates the X-axis feed, B dominates the Y-axis positioning, and C dominates the Z-axis cutting. The timing constraint is that B's positioning is triggered within 0.5s after A's feed is completed, and an initial cooperative control strategy is generated.

[0010] Step S3: Based on the initial cooperative control strategy, determine the different working process stages corresponding to multiple servo motors, and calculate the cooperative priority weights corresponding to different engineering stages based on the task requirements corresponding to different working process stages and the motor operation feature set to construct a staged cooperative control model; by real-time monitoring of the working state parameters and the current state of the target corresponding to each servo motor, and combining the staged cooperative control model, dynamically adjust the cooperative coefficients corresponding to each servo motor, including speed cooperative coefficient, torque distribution coefficient and position synchronization coefficient. In this embodiment of the invention, the machining task is divided into three work process stages based on an initial cooperative control strategy: Stage 1 (X-axis feed, 0-2s), Stage 2 (Y-axis positioning, 2-3s), and Stage 3 (Z-axis cutting, 3-5s). The task requirements for Stage 2 are position accuracy ±0.01mm and load fluctuation ±0.2N·m. Combining the Stage 2 parameters of the motor operation feature set (motor B position control backlash 0.012mm, torque adjustment delay 1.1ms), the cooperative priority weights are calculated using the analytic hierarchy process: position control 0.6, torque control 0.2, and speed control 0.2, thus constructing a staged cooperative control model. During real-time monitoring of the operating parameters of motor B in phase 2 (speed 200.2 r / min, torque 5.0 N·m, position Y 30.015 mm), the current position deviation of the target is 0.015 mm, which exceeds the accuracy threshold. Based on the model, the coordination coefficients are adjusted as follows: the speed coordination coefficient is increased from 0.98 to 0.99 (reducing speed deviation), the torque distribution coefficient is adjusted from 0.4 to 0.42 (stabilizing torque), and the position synchronization coefficient is increased from 0.95 to 0.97 (reducing position deviation).

[0011] Step S4: Substitute the adjusted coordination coefficients of each servo motor into the staged collaborative control model to generate real-time control commands for each servo motor, and perform joint control iterative optimization based on the real-time control commands for each servo motor to generate a closed-loop adaptive joint control mechanism.

[0012] In this embodiment of the invention, by substituting the adjusted coordination coefficients into the staged collaborative control model, real-time control commands are generated: motor B speed command 200 r / min (200.5 r / min before correction), torque command 5.1 N·m (5.0 N·m before correction), and position command Y 30.005 mm (Y 30.01 mm before correction); motors A and C synchronously generate adaptation commands. Based on the command execution joint control, after the first iteration, the position deviation of motor B is reduced to 0.008 mm, but still needs optimization; in the second iteration, the position synchronization coefficient is adjusted to 0.98, the command is updated to Y 30.002 mm, and the deviation is reduced to 0.003 mm. After five iterations, the parameters at each stage meet the requirements, forming a closed-loop adaptive joint control mechanism: real-time monitoring → coefficient adjustment → command generation → execution feedback → iterative optimization, ensuring that the motor coordination accuracy is stable within ±0.01 mm.

[0013] Furthermore, as an embodiment of the present invention, reference is made to... Figure 2 As shown, Figure 1 A detailed flowchart of step S1 is shown below. In this embodiment, step S1 includes the following steps: Step S11: Deploy a distributed sensor network to collect real-time operating parameters of multiple servo motors in the linkage system, including obtaining position feedback and speed data through the motor's built-in encoder, collecting winding current and calculating torque output through a current sensor, and recording the command response time of each servo motor through a high-speed data acquisition module. In this embodiment of the invention, a distributed sensor network is deployed in a robotic arm system with three servo motors. Each motor has a built-in encoder that outputs 1024 pulse signals every 10ms. Position feedback is calculated by counting the pulses (e.g., the position of the first motor at 100ms is 10240 pulses, corresponding to 36 degrees). The rotational speed is obtained by dividing the difference between two adjacent positions by the time interval (e.g., a position difference of 102 pulses within 10-20ms corresponds to 360 r / min). A current sensor is connected in series in the motor winding circuit, with a sampling frequency of 1kHz, to collect current values ​​(e.g., the current of the first motor at 50ms is 3.2A), which is calculated to be 2.24 N·m using the torque formula (torque = 0.7 × current). A high-speed data acquisition module is connected to the controller command output terminal and the motor feedback terminal, recording the time difference between the issuance of the command and the receipt of the feedback signal (e.g., the second motor returns a confirmation signal 1.2ms after receiving the "forward rotation" command, the command response time is 1.2ms), and synchronously storing the parameters of the three motors to form a real-time working parameter sequence.

[0014] Step S12: Use a load monitoring device to collect load feedback data of multiple servo motors in the linkage system, including obtaining load torque fluctuations through a torque sensor, collecting deformation data of the transmission mechanism through strain gauges, and calculating system energy loss and conversion efficiency through a power analyzer. In this embodiment of the invention, a load monitoring device is installed in a conveyor belt system driven by three servo motors. A torque sensor is installed on the drive shaft between the motor and the conveyor belt roller, outputting a torque value every 50ms (e.g., the load torque of the first motor is 5 N·m at 100ms and 5.2 N·m at 150ms, with a calculated fluctuation of 0.2 N·m). Strain gauges are attached to the stress points of the conveyor belt support, and the resistance change is converted into deformation data (e.g., the maximum deformation of 0.12mm corresponds to a tensile threshold of 800N). A power analyzer is connected in parallel between the motor power supply circuit and the conveyor belt load end, measuring the input electrical power (e.g., total input 1.2kW) and output mechanical power (e.g., total output 0.96kW) every 1s, calculating the energy loss of 0.24kW and the conversion efficiency of 0.96 / 1.2=80%, continuously recording to form a load feedback dataset.

[0015] Step S13: Collect target feature information of multiple servo motors in the linkage system through vision sensors or laser trackers, including the target's three-dimensional motion trajectory, attitude changes and motion accuracy requirements, and obtain the target's dynamic response threshold and fault tolerance range in combination with the task planning system. In this embodiment of the invention, in a precision assembly system with multiple servo motors working together, a vision sensor is mounted above the worktable, capturing the movement of the assembled parts at a frame rate of 30fps. The three-dimensional motion trajectory is calculated through image feature matching (e.g., the part moves from 0mm to 50mm on the X-axis, 0mm to 30mm on the Y-axis, and 0mm to 10mm on the Z-axis, taking 2 seconds). A laser tracker emits a laser beam to the part's reflective cursor, recording coordinates every 10ms and calculating attitude changes (e.g., rotating 5 degrees around the Z-axis), with a motion accuracy requirement of ±0.02mm. The task planning system outputs a dynamic response threshold (e.g., a position deviation exceeding 0.05mm requires correction within 10ms) and a tolerance range (e.g., a single deviation ≤0.1mm does not affect assembly quality) based on the assembly process, integrating this with the trajectory and attitude data to form target feature information.

[0016] Step S14: Perform Kalman filtering on the real-time operating parameters to eliminate measurement noise; perform outlier detection and interpolation on the load feedback data to ensure data continuity; perform coordinate transformation and standardization on the target feature information to unify the data dimensions. In this embodiment of the invention, Kalman filtering is applied to the real-time operating parameters of the three servo motors. Taking the speed data of the first motor as an example, the process noise variance is set to 0.01 and the measurement noise variance to 0.02. The fluctuation of the original speed data is reduced from ±5 r / min to ±1 r / min through the prediction equation (predicted value at time k = filtered value at time k-1 + speed change rate × time) and the update equation (filtered value = predicted value + Kalman gain × (measured value - predicted value)). For the load feedback data, the 3σ criterion is used to detect outliers (e.g., a torque of 7 N·m at a certain moment exceeding the mean of 5 N·m + 3 × standard deviation 0.5 N·m, indicating an anomaly). The data is then interpolated using the mean of the previous five data points (4.9 N·m). The target feature information is subjected to coordinate transformation. The pixel coordinates of the visual sensor (e.g., (320, 240)) are converted to mechanical coordinates ((25 mm, 18 mm)) through a calibration matrix, standardized to the 0-1 interval (25 mm / 50 mm = 0.5), and the data dimension is unified.

[0017] Step S15: Map the processed real-time operating parameters, load feedback data, and target feature information to a unified spatiotemporal coordinate system using timestamp synchronization technology to generate a standardized feature dataset containing time-series labels, including motor operation feature set, load feature set, and target feature set.

[0018] In this embodiment of the invention, timestamp synchronization technology is used to add time stamps accurate to 1ms (e.g., 20231001120000000 to 20231001120001000) to the processed real-time working parameters, load feedback data, and target feature information. A three-dimensional spatiotemporal coordinate system is established with the robotic arm base as the origin (X-axis along the conveyor belt direction, Y-axis perpendicular to the conveyor belt, Z-axis vertically upward, and time axis in milliseconds). The position (36 degrees), corresponding load torque (5 N·m), and part position (25 mm, 18 mm, 5 mm) of the first motor at 100 ms are mapped to the coordinate system (time coordinate 100, spatial coordinate associated with motor and part positions). A standardized feature dataset is generated, where the motor operation feature set includes position, speed, torque, and response time; the load feature set includes torque fluctuation, deformation, energy consumption, and efficiency; and the target feature set includes trajectory, attitude, accuracy, and threshold. All data have time-series labels and uniform dimensions.

[0019] Furthermore, step S2 includes the following steps: Step S21: Extract the speed command sequence and actual speed feedback sequence corresponding to each servo motor from the motor operation feature set, and calculate the difference between the speed command value and the actual speed value at each moment to generate instantaneous speed deviation data; Based on the instantaneous speed deviation data, use a sliding window to calculate the statistical feature value corresponding to the deviation sequence, including the deviation mean, standard deviation and peak value, and use the statistical feature value as the basic parameter of speed deviation; In this embodiment of the invention, in an automated sorting system with three servo motors linked together, the speed command sequence of motor A (one group every 10ms: 300r / min at 100ms, 310r / min at 110ms, and 320r / min at 120ms) and the actual speed feedback sequence (299.5r / min, 309.8r / min, and 319.7r / min) are extracted from the motor operation feature set, and the difference at each moment (0.5r / min, 0.2r / min, and 0.3r / min) is calculated to generate instantaneous speed deviation data. A 50ms sliding window (containing 5 data points) is used to calculate the statistical characteristic values: the deviation values ​​within the 120ms window are 0.5, 0.2, 0.3, 0.4, and 0.3, the mean deviation = (0.5 + 0.2 + 0.3 + 0.4 + 0.3) / 5 = 0.34r / min, and the standard deviation = √[((0.5 - 0.34)] 2 +(0.2-0.34) 2 +(0.3-0.34) 2 +(0.4-0.34) 2 +(0.3-0.34) 2 ) / 5]≈0.114r / min, peak value 0.5r / min, these three values ​​are used as the basic parameters of the speed deviation of motor A; similarly, the basic parameters of the speed deviation of motor B (mean 0.28r / min, standard deviation 0.09r / min, peak value 0.4r / min) and motor C (mean 0.36r / min, standard deviation 0.12r / min, peak value 0.6r / min) are calculated.

[0020] Step S22: Correct the speed deviation basic parameters by combining the load torque fluctuation data in the load feature set, and calculate the speed deviation correction coefficient under different load conditions by establishing a load torque and speed deviation correlation model; perform weighted calculation on the speed deviation basic parameters and speed deviation correction coefficient to generate the speed tracking error characterizing the motor speed control. In this embodiment of the invention, by extracting the load torque fluctuation data corresponding to motor A from the load feature set (5 N·m at 100ms, 5.2 N·m at 110ms, and 5.1 N·m at 120ms), a correlation model between load torque and speed deviation is established: Speed ​​deviation correction coefficient = 1 + 0.02 × (actual load torque - rated load torque) (rated load torque 5 N·m). At 100ms, the correction coefficient = 1 + 0.02 × (5 - 5) = 1, at 110ms = 1 + 0.02 × (5.2 - 5) = 1.004, at 120ms = 1 + 0.02 × (5.1 - 5) = 1.002, and the average value of 1.002 is taken as the speed deviation correction coefficient of motor A. Set the weight of the basic parameter of speed deviation to 0.7 and the weight of the correction coefficient to 0.3. Calculate the speed tracking error = (mean × 0.7 + standard deviation × 0.2 + peak value × 0.1) × correction coefficient = (0.34 × 0.7 + 0.114 × 0.2 + 0.5 × 0.1) × 1.002 ≈ (0.238 + 0.0228 + 0.05) × 1.002 ≈ 0.3108 × 1.002 ≈ 0.311 r / min; similarly calculate the speed tracking error of motor B (0.275 r / min) and motor C (0.365 r / min).

[0021] Step S23: Obtain the torque command sending time and the response time when the torque output reaches 85%-90% of the command value from the motor operation feature set, and calculate the time interval between the two times as the initial torque response delay; based on the transmission mechanism strain data in the load feature set, analyze the effect of transmission stiffness on the torque transmission delay, and calculate the transmission delay compensation amount; superimpose the initial torque response delay and the transmission delay compensation amount to generate the torque adjustment delay that reflects the timeliness of torque control. In this embodiment of the invention, the torque command sending time of motor A (the "5 N·m" command is issued at the 100ms mark) and the response time when the torque output reaches 88% (4.4 N·m) of the command value (at the 100.0012s mark) are obtained from the motor operation feature set. The initial torque response delay is calculated as 100.0012s - 100s = 0.0012s (1.2ms). The transmission mechanism strain data (strain value of 0.001mm / mm at the 100ms mark) is extracted from the load feature set. The transmission delay compensation is calculated as strain value × 0.0005s (preset strain and delay correlation coefficient) = 0.001 × 0.0005 = 0.0000005s ≈ 0s. After superposition, the torque adjustment delay of motor A is 1.2ms + 0s = 1.2ms. Motor B's torque command is sent at 110ms, its response time is 110.0013s, its initial delay is 1.3ms, its strain value is 0.0009mm / mm, its compensation is 0.00045s≈0s, and its torque adjustment delay is 1.3ms; Motor C's initial delay is 1.1ms, its strain value is 0.0011mm / mm, its compensation is 0.00055s≈0s, and its torque adjustment delay is 1.1ms.

[0022] Step S24: Extract the position command trajectory and the actual position feedback trajectory from the motor operation feature set, and calculate the position deviation of the two trajectories at key sampling points to generate an instantaneous position deviation dataset; combine the energy loss data from the load feature set to analyze the influence of frictional resistance on the instantaneous position deviation, and calculate the friction compensation value caused by friction; fuse the instantaneous position deviation dataset and the friction compensation value, and use the interval statistical method to calculate the maximum fluctuation range of the position deviation, and define the maximum fluctuation range as the position control clearance; In this embodiment of the invention, by extracting the position command trajectory of motor A (10mm at 100ms, 15mm at 110ms, 20mm at 120ms) and the actual position feedback trajectory (9.99mm, 15.002mm, 19.998mm), the deviations of key sampling points (0.01mm, -0.002mm, 0.002mm) are calculated to generate an instantaneous position deviation dataset. Combined with the energy loss data (120W at 100ms) from the load feature set, a deviation of 0.012mm is calculated using the friction compensation formula (friction compensation value = 0.0001 × energy loss). The instantaneous position deviation dataset is fused with the friction compensation value (deviation value + compensation value: 0.022mm, 0.01mm, 0.014mm). The maximum fluctuation range is calculated using the interval statistical method: maximum value - minimum value = 0.022mm - 0.01mm = 0.012mm, which is defined as the position control backlash of motor A. The position deviation dataset of motor B (0.008mm, -0.003mm, 0.002mm) has a friction compensation value of 0.011mm, and the maximum fluctuation range after fusion is 0.012mm. The position deviation dataset of motor C (0.012mm, -0.001mm, 0.003mm) has a friction compensation value of 0.013mm, and the maximum fluctuation range after fusion is 0.014mm.

[0023] Step S25: Determine the load coupling coefficient between each servo motor based on the dynamic response parameters corresponding to each servo motor; construct a collaborative task matrix by combining the target feature set, and clarify the functional division and timing constraints of each servo motor in the target execution process based on the load coupling coefficient and the collaborative task matrix, and generate an initial collaborative control strategy.

[0024] In this embodiment of the invention, the speed tracking error (0.311 / 0.275 / 0.365 r / min), torque adjustment delay (1.2 / 1.3 / 1.1 ms), and position control clearance (0.012 / 0.012 / 0.014 mm) of motors A, B, and C are used as dynamic response parameters. Cross-correlation coefficients (such as the speed synchronization fluctuation coefficient of A and B 0.032, torque transmission lag coefficient 0.071, and position coordination change coefficient -0.167) are calculated and combined with the transmission path parameters (meshing clearance 0.02 / 0.018 / 0.022 mm) to generate a load coupling coefficient matrix (diagonal 100, off-diagonal -55.2, -52.8, -57.6). Subtasks (grabbing / moving / placing) and accuracy requirements (±0.02mm) of the sorting task are extracted from the target feature set, and a collaborative task matrix is ​​constructed (subtask 1-A: 0.93, subtask 1-B: 0.71, subtask 1-C: 0.72; subtask 2-A: 0.95, subtask 2-B: 0.73, subtask 2-C: 0.74; subtask 3-A: 0.97, subtask 3-B: 0.75, subtask 3-C: 0.76). Based on the matrix, the functional division of labor is clearly defined (A is the main executor, B / C are auxiliary machines) and the timing constraints are defined (A completes the grabbing, triggering B's auxiliary operation within 0.055s, and C's auxiliary operation within 0.053s), and an initial collaborative control strategy is generated (A's speed is 300-320r / min, torque is 5-5.2N·m, B / C's speed follows A ±5r / min, and its torque accounts for 20% of A's).

[0025] Furthermore, the step S24, which involves calculating the friction compensation value based on energy loss data from the load feature set, includes the following steps: Energy loss time-series data is extracted from the load feature set and combined with speed and torque data from the motor operation feature set. The friction loss component is separated by the energy conservation analysis method to generate a friction energy loss sequence. By calculating the ratio of the friction energy loss sequence to the speed sequence, the friction energy consumption parameter corresponding to the unit speed is obtained. In this embodiment of the invention, energy loss time-series data within a certain period is extracted from the load feature set (recorded once every 10ms, e.g., 120W at 100ms, 125W at 110ms, and 130W at 120ms), combined with the corresponding speed (300r / min, 310r / min, 320r / min) and torque (5N·m, 5.2N·m, 5.3N·m) at the motor operating feature set. Using the energy conservation analysis method, the total energy loss is subtracted from electromagnetic losses (calculated as torque × speed × constant, e.g., 5 × 300 × 0.001 = 1.5W) and wind resistance losses (calculated as the square of the speed × a coefficient, e.g., 300...). 2×0.000001=0.09W, obtaining the friction loss component (120-1.5-0.09=118.41W), and generating a friction energy loss sequence. Calculating the ratio of this sequence to the rotational speed sequence (118.41 / 300≈0.395W·min / r), we obtain the friction energy consumption parameter per unit rotational speed.

[0026] Furthermore, a dynamic model of the friction coefficient is constructed based on the friction energy consumption parameters corresponding to the unit speed. The model parameters are corrected by introducing motor running time and temperature data, and a friction coefficient curve that dynamically changes with the operating conditions is generated. The instantaneous sliding friction force is calculated based on the friction coefficient curve and the current speed value. In this embodiment of the invention, a dynamic model of the friction coefficient is constructed based on the friction energy consumption parameter of 0.395 W·min / r per unit rotational speed. The basic formula of the model is friction coefficient = energy consumption parameter / (rotational speed × contact pressure). The model parameters are corrected by introducing motor running time (currently accumulated running time of 1800s) and temperature data (55℃). The coefficient increases by 0.02 for every 600s increase in running time and by 0.03 for every 10℃ increase in temperature. After correction, the model outputs friction coefficient curves (0.042 at 300r / min, 0.043 at 310r / min, and 0.044 at 320r / min). Based on the current rotational speed of 315r / min, the friction coefficient is found to be 0.0435. The instantaneous sliding friction force is calculated as 43.5N using the formula: instantaneous sliding friction force = friction coefficient × normal force (1000N).

[0027] Furthermore, the instantaneous sliding friction force is substituted into the dynamic model of the transmission mechanism to calculate the elastic deformation displacement of the transmission mechanism caused by the friction force, and the basic position deviation caused by the friction force is generated by combining the mechanism stiffness parameters. In this embodiment of the invention, the instantaneous sliding friction force of 43.5 N is substituted into the dynamic model of the transmission mechanism. The transmission mechanism in the model consists of a gear set and a transmission shaft. The gear meshing stiffness is 2000 N / mm, the shaft stiffness is 1500 N / mm, and the total stiffness, calculated in parallel, is 3500 N / mm. According to Hooke's Law, the elastic deformation displacement = 43.5 N / 3500 N / mm ≈ 0.0124 mm. Combining the mechanism stiffness parameters, the foundation position deviation is calculated by coupling the deformation displacement with the transmission ratio (1:5): 0.0124 mm × 5 = 0.062 mm, meaning the foundation position deviation caused by friction is 0.062 mm.

[0028] Furthermore, the temporal correlation between the instantaneous position deviation data and the basic position deviation was analyzed, and the friction influence coefficient was fitted using the least squares method. This friction influence coefficient reflects the contribution weight of friction force to the actual position deviation. In this embodiment of the invention, the instantaneous position deviation dataset (recorded every 10ms, such as 0.07mm at 100ms, 0.075mm at 110ms, and 0.08mm at 120ms) is extracted and compared with the corresponding baseline position deviations (0.06mm, 0.062mm, and 0.065mm). The temporal correlation between the two is analyzed. Using the baseline position deviation as the independent variable x and the instantaneous position deviation as the dependent variable y, the least squares method is used to fit the straight line y=kx+b, yielding k=1.2 (b=0.008). This k value is the friction influence coefficient, reflecting that for every 0.1mm increase in baseline position deviation, the actual position deviation increases by 0.12mm, with a weighting of 1.2.

[0029] Furthermore, the basic position deviation is multiplied by the friction influence coefficient, and then the friction amplification factor is corrected by combining the load fluctuation in the load characteristic set to generate a friction compensation value.

[0030] In this embodiment of the invention, the initial compensation value of 0.062 mm is obtained by multiplying the basic position deviation of 0.062 mm by the friction influence coefficient of 1.2. Load fluctuation data (current fluctuation ± 5%) is extracted from the load feature set, and the amplification factor table for load fluctuation on friction is consulted (5% fluctuation corresponds to an amplification factor of 1.05). The initial compensation value is multiplied by the amplification factor 0.0744 × 1.05 ≈ 0.0781 mm to generate the final friction compensation value of 0.0781 mm, which is used for subsequent correction of motor control commands.

[0031] Furthermore, the determination of the load coupling coefficient between each servo motor based on the dynamic response parameters corresponding to each servo motor in step S25 includes the following steps: The speed tracking error sequence, torque adjustment delay sequence, and position control backlash sequence are extracted from the dynamic response parameters of each servo motor. A multi-parameter correlation matrix is ​​constructed by aligning the timestamps. The matrix elements represent the combination of dynamic response parameters of different servo motors at the same time. In this embodiment of the invention, data is extracted from the dynamic response parameters of each motor in an automated assembly system with three servo motors: motor A speed tracking error sequence (1 group every 10ms, 0.5r / min at 100ms, 0.4r / min at 110ms, and 0.6r / min at 120ms), torque adjustment delay sequence (1.2ms at 100ms, 1.1ms at 110ms, and 1.3ms at 120ms), and position control gap sequence (100ms). 0.02mm, 110ms 0.018mm, 120ms 0.022mm; Motor B corresponding sequence (error 0.4 / 0.5 / 0.3r / min, delay 1.3 / 1.2 / 1.4ms, gap 0.019 / 0.021 / 0.017mm); Motor C corresponding sequence (error 0.6 / 0.4 / 0.5r / min, delay 1.1 / 1.3 / 1.2ms, gap 0.021 / 0.019 / 0.02mm). By aligning with timestamps (using 100ms, 110ms, and 120ms as unified time nodes), a 3×3×3 multi-parameter correlation matrix is ​​constructed (the dimensions are time × motor × parameter). For example, at 100ms, the matrix elements are (Motor A: 0.5, 1.2, 0.02; Motor B: 0.4, 1.3, 0.019; Motor C: 0.6, 1.1, 0.021). Each element represents the dynamic response parameter combination of different motors at the same time.

[0032] Furthermore, the cross-correlation coefficients of dynamic response parameters between any two servo motors are calculated, including the synchronization fluctuation coefficient of speed tracking error, the transmission lag coefficient of torque adjustment delay, and the cooperative change coefficient of position control backlash, to generate a parameter correlation vector. In this embodiment of the invention, the cross-correlation coefficients of the dynamic response parameters of any two motors are calculated: the synchronization fluctuation coefficient of the speed tracking error of motors A and B is calculated using the Pearson correlation formula (covariance 0.002, standard deviation product 0.063) to obtain 0.032; the torque adjustment delay transmission lag coefficient (covariance 0.001, standard deviation product 0.014) to obtain 0.071; and the position control backlash coordination change coefficient (covariance -0.00001, standard deviation product 0.00006) to obtain -0.167, forming a vector [0.032, 0.071, -0.167]. Similarly, the coefficient vectors of motors A and C [0.028, 0.065, -0.152] and the coefficient vectors of motors B and C [0.035, 0.074, -0.171] are calculated, generating a total of 3 sets of parameter correlation vectors, each containing correlation coefficients in 3 dimensions.

[0033] Furthermore, a dynamic response coupling model is constructed based on the parameter correlation vector. The mechanical transmission path parameters of the motor installation position are introduced, and the parameter correlation vector is converted into mechanical coupling strength to generate a preliminary load coupling coefficient matrix. In this embodiment of the invention, a dynamic response coupling model is constructed based on three sets of parameter correlation vectors. The model input is the correlation vector, and the output is the mechanical coupling strength. The mechanical transmission path parameters for the motor installation positions are introduced: the transmission path length between motors A and B is 0.8m with 20 meshing teeth; the path between motors A and C is 0.6m with 18 teeth; and the path between motors B and C is 1.0m with 22 teeth. According to the formula, mechanical coupling strength = correlation vector × (path length coefficient 0.05 + tooth number coefficient 0.03), for example, the strength between motors A and B = 0.032 × (0.8 × 0.05 + 20 × 0.03) + 0.071 × (0.8 × 0.05 + 20 × 0.03) + (-0.167) × (0.8 × 0.05 + 20 × 0.03) = (0.032 + 0.071 - 0.167) × 0.64 = -0.064 × 0.64 = -0.04096. Similarly, the strength between motors A and C is calculated to be -0.0384, and the strength between motors B and C is -0.0435, generating a 3×3 preliminary load coupling coefficient matrix (diagonal lines are 1, off-diagonal lines are the calculated values ​​above).

[0034] Furthermore, by combining the total load distribution data in the load feature set, the preliminary load coupling coefficient matrix is ​​verified, the matching degree between the actual load ratio of each motor and the coupling coefficient is calculated, and a coupling correction factor is generated. In this embodiment of the invention, the total load distribution data is extracted from the load feature set: the total system load is 1000N, the actual load of motor A is 350N (0.35%), motor B is 320N (0.32%), and motor C is 330N (0.33%). The matching degree between the actual load ratio of each motor and the preliminary coupling coefficient is calculated: the matching degree between motor A and B = |0.35×0.32-|-0.04096|| = |0.112-0.04096| = 0.07104; the matching degree between motor A and C = |0.35×0.33-|-0.0384|| = |0.1155-0.0384| = 0.0771; the matching degree between motor B and C = |0.32×0.33-|-0.0435|| = |0.1056-0.0435| = 0.0621. According to the formula Coupling Correction Factor = 1 - Matching Degree × 10 (Amplifying the Influence of Matching Degree), we obtain the factors of motor A and B as 0.2896, A and C as 0.229, and B and C as 0.379, generating a 3×3 correction factor matrix.

[0035] Furthermore, the coupling correction factor is weighted and fused element-wise with the initial load coupling coefficient matrix, and the load coupling coefficient is generated through matrix normalization.

[0036] In this embodiment of the invention, the coupling correction factor is weighted and fused element by element with the initial load coupling coefficient matrix: the fusion value of motor A and B = -0.04096 × 0.2896 ≈ -0.01186; the fusion value of motor A and C = -0.0384 × 0.229 ≈ -0.0088; the fusion value of motor B and C = -0.0435 × 0.379 ≈ -0.01659; the diagonal elements (motor self-coupling) are kept at 1 × 1 = 1. The merged matrix is ​​normalized (each row element is divided by the sum of the absolute values ​​of the elements in that row): Motor A row (1, -0.01186, -0.0088) is divided by 1.02066 to get (0.98, -0.0116, -0.0086); Motor B row (-0.01186, 1, -0.01659) is divided by 1.02845 to get (-0.0115, 0.972, -0.0161); Motor C row (-0.0088, -0.01659, 1) is divided by 1.02539 to get (-0.0086, -0.0162, 0.975), generating the final 3×3 load coupling coefficient matrix.

[0037] Furthermore, the process of constructing a dynamic response coupling model based on parameter correlation vectors, introducing mechanical transmission path parameters of the motor installation position, converting the parameter correlation vectors into mechanical coupling strength, and generating a preliminary load coupling coefficient matrix includes the following steps: Feature weights are assigned to the synchronization fluctuation coefficient, transmission lag coefficient, and coordinated change coefficient in the parameter correlation vector. A parameter weight vector is generated based on the significance of the influence of each parameter on load transmission. The speed synchronization weight, torque transmission weight, and position coordination weight are dynamically adjusted according to the motor linkage task type. In this embodiment of the invention, in a precision welding system with three servo motors linked together, the parameter correlation vector includes the synchronization fluctuation coefficient (0.032), transmission lag coefficient (0.071), and coordination variation coefficient (-0.167) of motors A and B. Since welding tasks require priority to position coordination, followed by torque transmission, and secondary to speed synchronization, the following feature weights are assigned: speed synchronization weight 0.2, torque transmission weight 0.3, and position coordination weight 0.5, generating a parameter weight vector [0.2, 0.3, 0.5]. The same weight allocation is applied to the correlation vectors of motors A and C, and B and C, to ensure consistent weights under the same task type. The weight values ​​are dynamically adjusted according to the task type (e.g., the torque transmission weight is increased to 0.4 during handling tasks).

[0038] Furthermore, the mechanical transmission path parameters of the motor installation location are collected, including the meshing clearance of the transmission gears, the stiffness coefficient of the transmission shaft and the elastic deformation characteristics of the coupling. The physical property matrix of the transmission path is generated through three-dimensional modeling, and the matrix elements characterize the mechanical transmission characteristics of each component in the path. In this embodiment of the invention, the mechanical transmission path parameters of the three motor installation positions are collected: the meshing clearance of the transmission gears between motors A and B is 0.02 mm, the stiffness coefficient of the transmission shaft is 2000 N / mm, and the elastic deformation coefficient of the coupling is 0.001 mm / N; the meshing clearance between motors A and C is 0.018 mm, the stiffness coefficient is 1800 N / mm, and the deformation coefficient is 0.0012 mm / N; the meshing clearance between motors B and C is 0.022 mm, the stiffness coefficient is 2200 N / mm, and the deformation coefficient is 0.0009 mm / N. These parameters are arranged in the order of "meshing clearance - stiffness coefficient - deformation coefficient" through 3D modeling to generate a 3×3 physical attribute matrix. Each row corresponds to a set of path characteristics between motors. For example, the first row [0.02, 2000, 0.001] characterizes the mechanical transmission characteristics between motors A and B.

[0039] Furthermore, a core transformation layer for the dynamic response coupling model is constructed, which performs a tensor product operation on the parameter correlation vector and the physical property matrix to generate an intermediate matrix of transfer characteristics. In this embodiment of the invention, a core transformation layer of the dynamic response coupling model is constructed, inputting a parameter correlation vector and a physical attribute matrix. For motors A and B, the parameter correlation vector is [0.032, 0.071, -0.167], and the corresponding physical attribute matrix is ​​[0.02, 2000, 0.001]. Tensor product operations are performed: 0.032 × 0.02 = 0.00064, 0.032 × 2000 = 64, 0.032 × 0.001 = 0.000032; 0.071 × 0.02 = 0. .00142, 0.071×2000=142, 0.071×0.001=0.000071; -0.167×0.02=-0.00334, -0.167×2000=-334, -0.167×0.001=-0.000167, generating a 3×3 intermediate matrix of transfer characteristics, the matrix elements reflecting the coupling relationship between parameter association and physical properties.

[0040] Furthermore, a mechanical equivalent transformation is performed on the intermediate matrix of transmission characteristics, and the energy transmission loss of different paths is corrected by introducing a transmission efficiency attenuation factor, so as to calculate the mechanical coupling strength value between each servo motor. In this embodiment of the invention, a transmission efficiency attenuation factor is introduced by performing a mechanical equivalent transformation on the intermediate matrix of transmission characteristics (path efficiency of motor A and B: 0.92, A and C: 0.94, B and C: 0.91). Taking motors A and B as examples, multiply each element of the intermediate matrix by the efficiency factor 0.92: 0.00064×0.92≈0.000589, 64×0.92≈58.88, 0.000032×0.92≈0.000029; 0.00142×0.92≈0.001306, 142×0.92≈130.64, 0.000071×0.92≈0.000065; -0.00334×0.92≈-0.00307, ​​-334×0.92≈-307.28, -0.000167×0.92≈-0.000154. Calculate the mean of the sum of the absolute values ​​of each row to get the mechanical coupling strength value (0.000589+58.88+0.000029+0.001306+130.64+0.000065+0.00307+307.28+0.000154)÷9≈496.805÷9≈55.2. Taking the negative value (because the synergistic change coefficient is negative) gives -55.2.

[0041] Furthermore, the mechanical coupling strength values ​​are arranged in order of motor number to form a preliminary load coupling coefficient matrix. The diagonal elements of the matrix represent the self-coupling coefficient of the motor's own load characteristics, and the off-diagonal elements represent the mutual coupling coefficient between different motors, thus completing the mapping transformation from parameter correlation vector to mechanical coupling strength.

[0042] In this embodiment of the invention, the calculated mechanical coupling strength values ​​are arranged in order of motor number: the strength of motors A and B is -55.2, the strength of motors A and C is -52.8 (calculated in the same way), and the strength of motors B and C is -57.6. The diagonal elements of the matrix (self-coupling coefficients) are set to 100 (a baseline value characterizing the load characteristics of the motors themselves), constructing a 3×3 preliminary load coupling coefficient matrix: the first row [100, -55.2, -52.8], the second row [-55.2, 100, -57.6], and the third row [-52.8, -57.6, 100]. The diagonal elements reflect the independence of the load of each motor, while the off-diagonal elements reflect the load transfer correlation between different motors, completing the mapping transformation from the parameter correlation vector to the mechanical coupling strength. Larger matrix element values ​​indicate stronger coupling.

[0043] Furthermore, step S25, which involves constructing a collaborative task matrix by combining the target feature set and clarifying the functional division and timing constraints of each servo motor during the target execution process based on the load coupling coefficient and the collaborative task matrix, and generating an initial collaborative control strategy, includes the following steps: Extract the motion trajectory complexity, accuracy level requirements and dynamic response bandwidth parameters of the target from the target feature set, and decompose the target task into several sub-task units. Each sub-task unit contains motion parameters, execution time and accuracy threshold, and generate sub-task feature vectors. In this embodiment of the invention, in a precision parts assembly task involving three servo motors, the complexity of the motion trajectory of the target (three-dimensional curved trajectory, curvature change rate 0.05mm⁻¹), accuracy level requirements (position error ≤ ±0.02mm), and dynamic response bandwidth (50Hz) are extracted from the target feature set. The assembly task is decomposed into three sub-task units: sub-task 1 is parts gripping (motion parameters: X-axis translation 50mm, execution time 2s, accuracy threshold ±0.03mm), sub-task 2 is attitude adjustment (motion parameters: Z-axis rotation 15°, execution time 1.5s, accuracy threshold ±0.01mm), and sub-task 3 is assembly docking (motion parameters: Z-axis downward movement 20mm, execution time 2.5s, accuracy threshold ±0.005mm). Subtask feature vectors are generated in the format of "complexity-precision-duration": Subtask 1 [0.03,0.03,2], Subtask 2 [0.06,0.01,1.5], Subtask 3 [0.04,0.005,2.5]. The complexity value in the vector is set according to the trajectory curvature, the precision value is directly adopted as the threshold, and the duration is the execution time.

[0044] Furthermore, the matching degree is calculated based on the subtask feature vector and the dynamic response parameters of each servo motor to obtain the motor-subtask adaptation degree matrix. The matrix elements represent the motor's ability to perform the corresponding subtask adaptation score. In this embodiment of the invention, the dynamic response parameters of the three servo motors are extracted as follows: motor A (speed tracking error 0.5 r / min, torque adjustment delay 1.2 ms, position control backlash 0.02 mm), motor B (error 0.4 r / min, delay 1.3 ms, backlash 0.019 mm), and motor C (error 0.6 r / min, delay 1.1 ms, backlash 0.021 mm). The matching degree formula "fit degree = 1 - (|motor error - subtask accuracy| / subtask accuracy + |motor delay - subtask duration × 0.001| / (subtask duration × 0.001)) × 0.5" is used to calculate: The fit degree of motor A to subtask 1 = 1 - (|0.02 - 0.03| / 0.03 + |1.2 - 2 × 0.001| / (2 × 0.001)) × 0.5 ≈ 1 - (0.033 + 599) × 0.5 (Correction: The delay unit is unified to s, the motor delay is 1.2ms = 0.0012s, the subtask duration is 2s, so |0.0012 - 2 × 0.001| / (2 × 0.001) = 0.2, the fit degree = 1 - (0.033 + 0.2) × 0.5 = 0.8835). Similarly, the motor-subtask fit matrix is ​​calculated as follows: Subtask 1 [0.8835, 0.89, 0.87], Subtask 2 [0.92, 0.91, 0.90], Subtask 3 [0.95, 0.94, 0.93]. The matrix elements range from 0 to 1, and the higher the value, the stronger the fit.

[0045] Furthermore, a collaborative task matrix is ​​constructed by combining the motor-subtask adaptability matrix and the load coupling coefficient. The matrix has subtask units in the row dimension and servo motors in the column dimension. The element values ​​are generated by weighted fusion of capability adaptability score and load coupling coefficient, which represents the comprehensive priority of the motor undertaking subtasks. In this embodiment of the invention, by combining the motor-subtask adaptability matrix and the load coupling coefficient matrix (diagonal 100, off-diagonal -55.2, -52.8, -57.6), the adaptability weight is set to 0.6 and the coupling coefficient weight is set to 0.4 (coupling coefficient normalized to 0-1: -55.2→0.448, -52.8→0.472, -57.6→0.424). We obtain: Subtask 1 - Motor A = 0.8835 × 0.6 + 1 × 0.4 = 0.9301; Subtask 1 - Motor B = 0.89 × 0.6 + 0.448 × 0.4 = 0.7132; Subtask 1 - Motor C = 0.87 × 0.6 + 0.472 × 0.4 = 0.7108; Subtask 2 - Motor A = 0.92 × 0.6 + 1 × 0.4 = 0.952; Subtask 2 - Motor B = 0.91 × 0.6 + 0.44 8 × 0.4 = 0.7252; Subtask 2 - Motor C = 0.90 × 0.6 + 0.472 × 0.4 = 0.7288; Subtask 3 - Motor A = 0.95 × 0.6 + 1 × 0.4 = 0.97; Subtask 3 - Motor B = 0.94 × 0.6 + 0.448 × 0.4 = 0.7332; Subtask 3 - Motor C = 0.93 × 0.6 + 0.472 × 0.4 = 0.7368. The element values ​​represent the overall priority.

[0046] Furthermore, based on the element value distribution of the collaborative task matrix, a clustering algorithm is used to divide the functional groups and clarify the functional division of each servo motor; by analyzing the temporal dependency relationship of the sub-task units and combining the load coupling coefficient, the action triggering delay threshold between motors is calculated to generate temporal constraint rules. In this embodiment of the invention, based on the distribution of element values ​​of the collaborative task matrix (subtask 1 - motor A 0.9301 is the highest, subtask 2 - motor A 0.952 is the highest, and subtask 3 - motor A 0.97 is the highest), the K-means clustering algorithm (3 clusters) is used to divide the functions into groups: motor A undertakes the main execution of subtasks 1, 2, and 3, motor B assists subtasks 1 and 3, and motor C assists subtask 2, thus clarifying the functional division results. Analyze the timing dependencies of subtasks: Subtask 1 → Subtask 2 → Subtask 3 (grab, adjust posture, and then dock). Calculate the action trigger delay thresholds based on the load coupling coefficient: After motor A completes subtask 1, the delay threshold for sending a trigger signal to motor B is |coupling coefficient| × 0.001 = 0.0552s, and the threshold for sending to motor C is 52.8 × 0.001 = 0.0528s. Generate timing constraint rules: Trigger motor B's auxiliary action within 0.0552s after subtask 1 is completed, and trigger motor C's auxiliary action within 0.0528s to ensure coordinated actions.

[0047] Furthermore, the functional division results and timing constraint rules are substituted into the model predictive control framework. With the accuracy level requirement in the target feature set as the optimization objective and the load coupling coefficient as the constraint condition, the target speed curve, torque distribution ratio and position synchronization reference of each servo motor are obtained and fused to form an initial collaborative control strategy.

[0048] In this embodiment of the invention, by substituting the functional division results (motor A as the primary driver and B / C as secondary drivers) and timing constraint rules into the model predictive control framework, the optimization objectives are set as position errors ≤ ±0.02mm (subtask 1), ≤ ±0.01mm (subtask 2), and ≤ ±0.005mm (subtask 3), with the constraint condition being an absolute value of the load coupling coefficient ≤ 57.6 (to avoid overload). The following are solved through rolling time-domain optimization: the target speed curve of motor A (subtask 1: from 0 to 300 r / min, held for 2s; subtask 2: decreased to 200 r / min, held for 1.5s; subtask 3: decreased to 150 r / min, held for 2.5s); torque distribution ratio (motor A bears 70%, B bears 20%, and C bears 10%); and position synchronization reference (based on the position signal of motor A, the position deviation of B / C must be ≤ 0.01mm). The speed curve, torque ratio, and synchronization reference are integrated to form an initial collaborative control strategy, which includes the real-time control parameters of each motor and collaborative triggering conditions.

[0049] Furthermore, step S3 includes the following steps: Step S31: Analyze the timing logic and task milestones in the preliminary collaborative control strategy, extract the key nodes of the target trajectory, and divide the task intervals between adjacent key nodes into different work process stages. In this embodiment of the invention, the initial collaborative control strategy for part assembly using three servo motors is analyzed as "grab → translation → attitude adjustment → docking". Task milestones are set as follows: grab complete (part detaches from the worktable), translation in place (reaches the top of the assembly area), attitude adjustment qualified (angle deviation ≤ 0.1°), and docking successful (part fit tolerance ≤ 0.01mm). The coordinates of key nodes in the target trajectory are extracted: grab point (X0, Y0, Z0), translation endpoint (X50, Y30, Z20), adjustment endpoint (X50, Y30, Z20, θ15°), and docking endpoint (X50, Y30, Z10, θ15°). The task interval between adjacent key nodes is divided into four work process stages: Stage 1 (grab stage, 0-2s), Stage 2 (translation stage, 2-5s), Stage 3 (attitude adjustment stage, 5-7s), and Stage 4 (docking stage, 7-10s), with each stage defined by a time node and the completion status of the key action.

[0050] Step S32: Extract historical operating parameters corresponding to each working process stage from the motor operating feature set, and construct a stage task requirement vector by combining the stage performance requirements in the target feature set, including stage duration, allowable accuracy deviation and load fluctuation range. In this embodiment of the invention, historical operating parameters for each stage are extracted from the motor operating feature set: Stage 1 (Motor A speed 200 r / min, torque 4 N·m; Motor B speed 180 r / min, torque 3.5 N·m; Motor C speed 190 r / min, torque 3.8 N·m), Stage 2 (A 250 r / min, 5 N·m; B 230 r / min, 4.5 N·m; C 240 r / min, 4.8 N·m), Stage 3 (A 150 r / min, 3 N·m; B 130 r / min, 2.8 N·m; C 140 r / min, 2.9 N·m), Stage 4 (A 100 r / min, 2.5 N·m; B 80 r / min, 2.2 N·m; C 90 r / min, 2.3 N·m). Based on the stage performance requirements of the target feature set, a stage task requirement vector is constructed: Stage 1 [duration 2s, accuracy tolerance ±0.03mm, load fluctuation range 3-4.5N·m], Stage 2 [3s, ±0.02mm, 4-5.5N·m], Stage 3 [2s, ±0.01mm, 2.5-3.5N·m], Stage 4 [3s, ±0.005mm, 2-3N·m]. The elements of the vector are arranged in the order of "duration-accuracy-load".

[0051] Step S33: Calculate the control dimension weights of each stage based on the stage task requirement vector, and transform the stage parameter performance of the motor operation feature set into the collaborative requirement intensity through the analytic hierarchy process, and fuse them to generate the collaborative priority weights corresponding to different work process stages. In this embodiment of the invention, the control dimension weights are calculated based on the stage task requirement vector, with the rule that "the higher the accuracy requirement, the greater the position control weight; the greater the load fluctuation, the greater the torque control weight; and the shorter the duration, the greater the speed control weight." Stage 3 (accuracy ±0.01mm, load fluctuation 1N·m, duration 2s): position control weight 0.5, torque control weight 0.2, speed control weight 0.3. Using the analytic hierarchy process, the motor operating characteristics concentrated in the stage 3 parameters (such as position control gap 0.012mm, torque adjustment delay 1.2ms, speed tracking error 0.3r / min) are quantified using a 1-9 scale: position performance 8 points, torque performance 7 points, speed performance 6 points. The coordinated demand intensity is calculated as (8×0.5+7×0.2+6×0.3)=4+1.4+1.8=7.2. By integrating the control dimension weights and the intensity of collaborative needs, the collaborative priority weights for stage 3 are generated as follows: position 0.5×(8 / 21)×7.2≈1.37, torque 0.2×(7 / 21)×7.2≈0.48, and speed 0.3×(6 / 21)×7.2≈0.617. After standardization, these values ​​are 0.52 for position, 0.18 for torque, and 0.3 for speed, which are used as the collaborative priority weights for stage 3. Similarly, the weights for other stages are calculated.

[0052] Step S34: Embed the collaborative priority weight into the control algorithm framework, and combine the dynamic response parameter thresholds and load coupling coefficients of each stage to construct a staged collaborative control model that includes stage switching conditions, control parameter baseline values, and collaborative rule sets. In this embodiment of the invention, the collaborative priority weights of each stage are embedded into the PID control algorithm framework. The weights for stage 4 (docking stage) are 0.6 for position, 0.2 for speed, and 0.2 for torque. A staged collaborative control model is constructed by combining the dynamic response parameter thresholds for stage 4 (speed tracking error ≤ 0.2 r / min, torque adjustment delay ≤ 1 ms, position control backlash ≤ 0.008 mm) with the load coupling coefficient matrix (AB: -55.2, AC: -52.8, BC: -57.6). The model includes the following stage switching conditions: when the angular deviation of the attitude adjustment in stage 3 is ≤0.05° and lasts for 200ms, the switch to stage 4 is triggered; control parameter baseline values: in stage 4, motor A speed is 100r / min, torque is 2.5N·m, and position control accuracy is ±0.005mm; cooperative rule set: motor A is the master controller, B / C are slave controllers, the position deviation of B / C must follow A and the deviation must be ≤0.003mm, the torque distribution ratio is A:B:C=5:2:3, ensuring that the absolute value of the load coupling coefficient is ≤57.6, and the model construction is completed.

[0053] Step S35: By real-time monitoring of the working status parameters of each servo motor and the current status of the target, and by combining the phased collaborative control model, dynamically adjust the collaborative coefficients of each servo motor, including the speed collaborative coefficient, torque distribution coefficient and position synchronization coefficient.

[0054] In this embodiment of the invention, the working status parameters of the motors in stage 4 are monitored in real time through a distributed sensor network: motor A speed is 99.8 r / min (deviation 0.2 r / min), torque is 2.4 N·m (deviation 0.1 N·m), and position deviation is 0.006 mm; motor B speed is 79.5 r / min (deviation 0.5 r / min), torque is 2.1 N·m (deviation 0.1 N·m), and position deviation is 0.009 mm; the current state coordinate deviation of the target is 0.007 mm, and the attitude angle offset is 0.02°. By combining the phased collaborative control model, the collaboration coefficients are dynamically adjusted: the speed collaboration coefficient is adjusted from the initial 0.97 to 0.98 (due to the B speed deviation exceeding the threshold, the collaboration is improved), so that the B speed deviation from A is reduced to 0.2 r / min; the torque distribution coefficient is adjusted from 5:2:3 to 4.8:2.2:3 (increasing the B torque proportion), so that the B torque deviation is reduced to 0.05 N·m; the position synchronization coefficient is adjusted from 0.96 to 0.99 (strengthening position collaboration), so that the B position deviation is reduced to 0.004 mm. All parameters meet the stage 4 threshold requirements, realizing dynamic adjustment.

[0055] Furthermore, step S35 includes the following steps: Step S351: Collect the working status parameters of each servo motor in real time through a distributed sensor network, including the effective value of three-phase current, rotor position angle, winding temperature and actual output speed. Extract the current status parameters of the target from the target feature set, including three-dimensional coordinate deviation, attitude angle offset and motion speed vector, and construct a status monitoring dataset. In this embodiment of the invention, in a precision machining system with three servo motors working in tandem, a distributed sensor network collects operating status parameters every 5ms: Motor A's three-phase current RMS value is 3.2A / 3.1A / 3.3A, rotor position angle is 36.2°, winding temperature is 55°C, and actual output speed is 299.8r / min; Motor B's current is 3.0A / 2.9A / 3.1A, position angle is 72.5°, temperature is 54°C, and speed is 300.1r / min; Motor C's current is 3.3A / 3.2A / 3.4A, position angle is 108.3°, temperature is 56°C, and speed is 299.7r / min. Extract the current state parameters of the target (processed part) from the target feature set: three-dimensional coordinate deviation X-axis 0.02mm / Y-axis -0.01mm / Z-axis 0.015mm, attitude angle offset -0.1°, motion velocity vector X-axis 5mm / s / Y-axis 3mm / s / Z-axis 2mm / s. Integrate the motor parameters and target parameters according to the timestamp (e.g., 16:30:00.000 / 005 / 010) to construct a state monitoring dataset.

[0056] Step S352: Compare the working state parameters in the state monitoring dataset with the parameter benchmark values ​​of the corresponding working process stage in the staged collaborative control model, calculate the parameter deviation rate, and at the same time compare the current state parameters of the target with the preset target trajectory parameters to generate the target state deviation vector. In this embodiment of the invention, the current working process in the staged collaborative control model is the "finishing program segment," with corresponding baseline parameter values: motor current 3.0±0.2A, rotor position angle error ≤±0.5°, winding temperature ≤60℃, and speed 300±0.3r / min; target trajectory parameters: coordinate deviation ≤±0.01mm, attitude angle offset ≤±0.05°, and velocity vector error ≤±0.5mm / s. The parameter deviation rates are calculated as follows: motor A current deviation rate (3.2-3.0) / 3.0≈6.67%, position angle deviation rate (36.2-36) / 36≈0.56%, and speed deviation rate (299.8-300) / 300≈-0.07%; the corresponding deviation rates for motors B and C are calculated similarly. The target state deviation vector is generated by comparing the target state parameters: [0.02-0.01,-0.01-0,0.015-0.01,-0.1-(-0.05),5-5,3-3,2-2], which is [0.01,-0.01,0.005,-0.05,0,0,0]. Each element of the vector corresponds to the deviation value of the coordinates, attitude, and velocity.

[0057] Step S353: Construct a collaborative adjustment trigger index based on the parameter deviation rate and the target state deviation vector. This index is generated by fusing the deviation amplitude, duration and trend. When the index exceeds the threshold set by the staged collaborative control model, the collaborative coefficient adjustment mechanism is activated. In this embodiment of the invention, a collaborative adjustment trigger index is constructed, and the formula is "index = (deviation amplitude × 0.4 + duration × 0.3 + change trend × 0.3)". The speed deviation amplitude of motor A is 0.2 r / min, the duration is 100 ms (5 sampling periods), and the trend is -0.02 r / min / period. The calculated value is (0.2×0.4+0.1×0.3+(-0.02)×0.3)=0.08+0.03-0.006=0.104; the target coordinate deviation amplitude is 0.02 mm, the duration is 100 ms, and the trend is 0.002 mm / period. The calculated value is (0.02×0.4+0.1×0.3+0.002×0.3)=0.008+0.03+0.0006=0.0386; the comprehensive index is 0.104×0.6+0.0386×0.4≈0.0624+0.0154=0.0778. The phased model sets the threshold for this phase to 0.05. If 0.0778 > 0.05, the synergy coefficient adjustment mechanism is activated.

[0058] Step S354: Based on the coordination priority weight of the current working process stage, assign adjustment sensitivity weights to the speed coordination coefficient, torque distribution coefficient, and position synchronization coefficient respectively. The adjustment sensitivity of the speed coordination coefficient is related to the speed deviation rate and speed control weight, the adjustment sensitivity of the torque distribution coefficient is related to the current imbalance and torque control weight, and the adjustment sensitivity of the position synchronization coefficient is related to the position deviation and position control weight. In this embodiment of the invention, the priority weights for the current finishing program segment are: position control 0.5, speed control 0.3, and torque control 0.2. Adjustment sensitivity weights are assigned as follows: Speed ​​coordination coefficient adjustment sensitivity = absolute value of speed deviation rate × speed control weight = 0.07% × 0.3 = 0.021%; Torque distribution coefficient adjustment sensitivity = current imbalance (current difference between motors A and B 0.2A / 3.0A ≈ 6.67%) × torque control weight = 6.67% × 0.2 ≈ 1.334%; Position synchronization coefficient adjustment sensitivity = position deviation (0.02mm) × position control weight = 0.02 × 0.5 = 0.01mm. The three sensitivity weights are standardized to the 0-1 range, resulting in 0.021 / 1.334 ≈ 0.0157, 1.334 / 1.334 = 1, and 0.01 / 0.01 = 1, thus clarifying the adjustment priority of each coefficient.

[0059] Step S355: Using the model predictive control algorithm, the adjustment sensitivity weight, parameter deviation rate and target state deviation vector are input into the coordination coefficient optimization module of the staged coordinated control model to solve for the optimal adjustment amount of the coordination coefficient. After load coupling constraint verification, the speed coordination coefficient, torque distribution coefficient and position synchronization coefficient are updated.

[0060] In this embodiment of the invention, a model predictive control algorithm is employed, with the goal of minimizing parameter deviation over the next 50ms (10 sampling periods). The sensitivity weights (0.0157, 1, 1), parameter deviation rates (6.67%, 0.56%, -0.07%), and the target state deviation vector [0.01, -0.01, 0.005, -0.05, 0, 0, 0] are input into the coordination coefficient optimization module. The optimal adjustment amounts for the coordination coefficients are obtained: the speed coordination coefficient is adjusted from 0.98 to 0.99 (+0.01), the torque distribution coefficient is adjusted from 0.3 / 0.3 / 0.4 to 0.28 / 0.32 / 0.4 (A -0.02, B +0.02), and the position synchronization coefficient is adjusted from 0.95 to 0.97 (+0.02). The load coupling constraint was checked (absolute value of coupling coefficient ≤ 57.6). After adjustment, the load coupling coefficients between motors were -55.1, -52.9, and -57.5, all of which met the constraint. The coefficients were then updated to the staged collaborative control model.

[0061] Furthermore, the present invention also provides a joint control system for multiple servo motors, including a processor, a memory, and a computer program stored in the memory and executable on the processor, for performing the joint control method for multiple servo motors as described above.

[0062] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A method for joint control of multiple servo motors, characterized in that, Includes the following steps: Step S1: Collect real-time operating parameters, load feedback data, and target feature information of multiple servo motors in the linkage system, and perform spatiotemporal alignment and noise filtering on the collected real-time operating parameters, load feedback data, and target feature information to generate standardized motor operation feature sets, load feature sets, and target feature sets. Step S2: Extract the dynamic response parameters corresponding to each servo motor based on the motor operation feature set and load feature set, including speed tracking error, torque adjustment delay and position control backlash, and determine the load coupling coefficient between each servo motor; construct a collaborative task matrix by combining the target feature set, and clarify the functional division and timing constraint relationship of each servo motor in the target execution process based on the load coupling coefficient and the collaborative task matrix, and generate the initial collaborative control strategy. Step S3: Based on the initial cooperative control strategy, determine the different working process stages corresponding to multiple servo motors, and calculate the cooperative priority weights corresponding to different engineering stages based on the task requirements corresponding to different working process stages and the motor operation feature set to construct a staged cooperative control model; by real-time monitoring of the working state parameters and the current state of the target corresponding to each servo motor, and combining the staged cooperative control model, dynamically adjust the cooperative coefficients corresponding to each servo motor, including speed cooperative coefficient, torque distribution coefficient and position synchronization coefficient. Step S4: Substitute the adjusted coordination coefficients of each servo motor into the staged collaborative control model to generate real-time control commands for each servo motor, and perform joint control iterative optimization based on the real-time control commands for each servo motor to generate a closed-loop adaptive joint control mechanism.

2. The method for joint control of multiple servo motors according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Deploy a distributed sensor network to collect real-time operating parameters of multiple servo motors in the linkage system, including obtaining position feedback and speed data through the motor's built-in encoder, collecting winding current and calculating torque output through a current sensor, and recording the command response time of each servo motor through a high-speed data acquisition module. Step S12: Use a load monitoring device to collect load feedback data of multiple servo motors in the linkage system, including obtaining load torque fluctuations through a torque sensor, collecting deformation data of the transmission mechanism through strain gauges, and calculating system energy loss and conversion efficiency through a power analyzer. Step S13: Collect target feature information of multiple servo motors in the linkage system through vision sensors or laser trackers, including the target's three-dimensional motion trajectory, attitude changes and motion accuracy requirements, and obtain the target's dynamic response threshold and fault tolerance range in combination with the task planning system. Step S14: Perform Kalman filtering on the real-time operating parameters to eliminate measurement noise; perform outlier detection and interpolation on the load feedback data to ensure data continuity; perform coordinate transformation and standardization on the target feature information to unify the data dimensions. Step S15: Map the processed real-time operating parameters, load feedback data, and target feature information to a unified spatiotemporal coordinate system using timestamp synchronization technology to generate a standardized feature dataset containing time-series labels, including motor operation feature set, load feature set, and target feature set.

3. The method for joint control of multiple servo motors according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Extract the speed command sequence and actual speed feedback sequence corresponding to each servo motor from the motor operation feature set, and calculate the difference between the speed command value and the actual speed value at each moment to generate instantaneous speed deviation data; Based on the instantaneous speed deviation data, use a sliding window to calculate the statistical feature value corresponding to the deviation sequence, including the deviation mean, standard deviation and peak value, and use the statistical feature value as the basic parameter of speed deviation; Step S22: Correct the speed deviation basic parameters by combining the load torque fluctuation data in the load feature set, and calculate the speed deviation correction coefficient under different load conditions by establishing a load torque and speed deviation correlation model; perform weighted calculation on the speed deviation basic parameters and speed deviation correction coefficient to generate the speed tracking error characterizing the motor speed control. Step S23: Obtain the torque command sending time and the response time when the torque output reaches 85%-90% of the command value from the motor operation feature set, and calculate the time interval between the two times as the initial torque response delay; based on the transmission mechanism strain data in the load feature set, analyze the effect of transmission stiffness on the torque transmission delay, and calculate the transmission delay compensation amount; superimpose the initial torque response delay and the transmission delay compensation amount to generate the torque adjustment delay that reflects the timeliness of torque control. Step S24: Extract the position command trajectory and the actual position feedback trajectory from the motor operation feature set, and calculate the position deviation of the two trajectories at key sampling points to generate an instantaneous position deviation dataset; combine the energy loss data from the load feature set to analyze the influence of frictional resistance on the instantaneous position deviation, and calculate the friction compensation value caused by friction; fuse the instantaneous position deviation dataset and the friction compensation value, and use the interval statistical method to calculate the maximum fluctuation range of the position deviation, and define the maximum fluctuation range as the position control clearance; Step S25: Determine the load coupling coefficient between each servo motor based on the dynamic response parameters corresponding to each servo motor; construct a collaborative task matrix by combining the target feature set, and clarify the functional division and timing constraints of each servo motor in the target execution process based on the load coupling coefficient and the collaborative task matrix, and generate an initial collaborative control strategy.

4. The method for joint control of multiple servo motors according to claim 3, characterized in that, Step S24, which involves calculating the friction compensation value based on energy loss data from the load feature set, includes the following steps: Energy loss time-series data is extracted from the load feature set and combined with speed and torque data from the motor operation feature set. The friction loss component is separated by the energy conservation analysis method to generate a friction energy loss sequence. By calculating the ratio of the friction energy loss sequence to the speed sequence, the friction energy consumption parameter corresponding to the unit speed is obtained. A dynamic model of the friction coefficient is constructed based on the friction energy consumption parameters corresponding to unit speed. The model parameters are corrected by introducing motor running time and temperature data, and a friction coefficient curve that dynamically changes with operating conditions is generated. The instantaneous sliding friction force is calculated based on the friction coefficient curve and the current speed value. Substitute the instantaneous sliding friction force into the dynamic model of the transmission mechanism, calculate the elastic deformation displacement of the transmission mechanism caused by the friction force, and generate the basic position deviation caused by the friction force by combining the mechanism stiffness parameters. The temporal correlation between the instantaneous position deviation data and the basic position deviation was analyzed, and the friction influence coefficient was fitted using the least squares method. This friction influence coefficient reflects the contribution weight of friction force to the actual position deviation. The friction compensation value is generated by multiplying the basic position deviation with the friction influence coefficient and then correcting the friction amplification factor by combining the load fluctuation in the load characteristic set.

5. The method for joint control of multiple servo motors according to claim 3, characterized in that, The step S25, which involves determining the load coupling coefficient between servo motors based on the dynamic response parameters of each servo motor, includes the following steps: The speed tracking error sequence, torque adjustment delay sequence, and position control backlash sequence are extracted from the dynamic response parameters of each servo motor. A multi-parameter correlation matrix is ​​constructed by aligning the timestamps, and the matrix elements represent the combination of dynamic response parameters of different servo motors at the same time. Calculate the cross-correlation coefficients of dynamic response parameters between any two servo motors, including the synchronization fluctuation coefficient of speed tracking error, the transmission lag coefficient of torque adjustment delay, and the cooperative change coefficient of position control backlash, and generate a parameter correlation vector. A dynamic response coupling model is constructed based on the parameter correlation vector. The mechanical transmission path parameters of the motor installation position are introduced, and the parameter correlation vector is converted into mechanical coupling strength to generate a preliminary load coupling coefficient matrix. By combining the total load distribution data in the load feature set, the preliminary load coupling coefficient matrix is ​​verified, the matching degree between the actual load ratio of each motor and the coupling coefficient is calculated, and a coupling correction factor is generated. The coupling correction factor is weighted and fused element-wise with the initial load coupling coefficient matrix, and the load coupling coefficient is generated by matrix normalization.

6. The method for joint control of multiple servo motors according to claim 5, characterized in that, The process of constructing a dynamic response coupling model based on parameter correlation vectors, introducing mechanical transmission path parameters of the motor installation location, converting the parameter correlation vectors into mechanical coupling strength, and generating a preliminary load coupling coefficient matrix includes the following steps: Feature weights are assigned to the synchronization fluctuation coefficient, transmission lag coefficient, and coordinated change coefficient in the parameter correlation vector. A parameter weight vector is generated based on the significance of the influence of each parameter on load transmission. The speed synchronization weight, torque transmission weight, and position coordination weight are dynamically adjusted according to the motor linkage task type. The mechanical transmission path parameters of the motor installation location are collected, including the meshing clearance of the transmission gears, the stiffness coefficient of the transmission shaft and the elastic deformation characteristics of the coupling. The physical property matrix of the transmission path is generated through three-dimensional modeling, and the matrix elements characterize the mechanical transmission characteristics of each component in the path. The core transformation layer of the dynamic response coupling model is constructed by performing a tensor product operation between the parameter correlation vector and the physical property matrix to generate an intermediate matrix of transfer characteristics. The intermediate matrix of transmission characteristics is mechanically equivalently transformed, and the energy transmission loss of different paths is corrected by introducing a transmission efficiency attenuation factor. The mechanical coupling strength value between each servo motor is then calculated. The mechanical coupling strength values ​​are arranged in order of motor number to form a preliminary load coupling coefficient matrix. The diagonal elements of the matrix represent the self-coupling coefficient of the motor's own load characteristics, and the off-diagonal elements represent the mutual coupling coefficient between different motors, thus completing the mapping transformation from parameter correlation vector to mechanical coupling strength.

7. The method for joint control of multiple servo motors according to claim 3, characterized in that, Step S25, which involves constructing a collaborative task matrix by combining the target feature set and clarifying the functional division and timing constraints of each servo motor during the target execution process based on the load coupling coefficient and the collaborative task matrix, to generate an initial collaborative control strategy, includes the following steps: Extract the motion trajectory complexity, accuracy level requirements and dynamic response bandwidth parameters of the target from the target feature set, and decompose the target task into several sub-task units. Each sub-task unit contains motion parameters, execution time and accuracy threshold, and generate sub-task feature vectors. The matching degree is calculated based on the feature vector of the subtask and the dynamic response parameters of each servo motor to obtain the motor-subtask adaptation degree matrix. The matrix elements represent the motor's ability to perform the corresponding subtask adaptation score. A collaborative task matrix is ​​constructed by combining the motor-subtask adaptability matrix and the load coupling coefficient. The matrix has subtask units in the row dimension and servo motors in the column dimension. The element values ​​are generated by weighted fusion of capability adaptability score and load coupling coefficient, which represents the comprehensive priority of the motor undertaking the subtask. Based on the element value distribution of the collaborative task matrix, a clustering algorithm is used to divide the functional groups and clarify the functional division of each servo motor; by analyzing the temporal dependency relationship of the sub-task units and combining the load coupling coefficient, the action triggering delay threshold between motors is calculated, and temporal constraint rules are generated. The functional division results and timing constraint rules are substituted into the model predictive control framework. The accuracy level requirement in the target feature set is used as the optimization objective, and the load coupling coefficient is used as the constraint condition. The target speed curve, torque distribution ratio and position synchronization reference of each servo motor are obtained by solving the problem and then fused to form an initial collaborative control strategy.

8. The method for joint control of multiple servo motors according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Analyze the timing logic and task milestones in the preliminary collaborative control strategy, extract the key nodes of the target trajectory, and divide the task intervals between adjacent key nodes into different work process stages. Step S32: Extract historical operating parameters corresponding to each working process stage from the motor operating feature set, and construct a stage task requirement vector by combining the stage performance requirements in the target feature set, including stage duration, allowable accuracy deviation and load fluctuation range. Step S33: Calculate the control dimension weights of each stage based on the stage task requirement vector, and transform the stage parameter performance of the motor operation feature set into the collaborative requirement intensity through the analytic hierarchy process, and fuse them to generate the collaborative priority weights corresponding to different work process stages. Step S34: Embed the collaborative priority weight into the control algorithm framework, and combine the dynamic response parameter thresholds and load coupling coefficients of each stage to construct a staged collaborative control model that includes stage switching conditions, control parameter baseline values, and collaborative rule sets. Step S35: By real-time monitoring of the working status parameters of each servo motor and the current status of the target, and by combining the phased collaborative control model, dynamically adjust the collaborative coefficients of each servo motor, including the speed collaborative coefficient, torque distribution coefficient and position synchronization coefficient.

9. The method for joint control of multiple servo motors according to claim 8, characterized in that, Step S35 includes the following steps: Step S351: Collect the working status parameters of each servo motor in real time through a distributed sensor network, including the effective value of three-phase current, rotor position angle, winding temperature and actual output speed. Extract the current status parameters of the target from the target feature set, including three-dimensional coordinate deviation, attitude angle offset and motion speed vector, and construct a status monitoring dataset. Step S352: Compare the working state parameters in the state monitoring dataset with the parameter benchmark values ​​of the corresponding working process stage in the staged collaborative control model, calculate the parameter deviation rate, and at the same time compare the current state parameters of the target with the preset target trajectory parameters to generate the target state deviation vector. Step S353: Construct a collaborative adjustment trigger index based on the parameter deviation rate and the target state deviation vector. This index is generated by fusing the deviation amplitude, duration and trend. When the index exceeds the threshold set by the staged collaborative control model, the collaborative coefficient adjustment mechanism is activated. Step S354: Based on the coordination priority weight of the current working process stage, assign adjustment sensitivity weights to the speed coordination coefficient, torque distribution coefficient, and position synchronization coefficient respectively. The adjustment sensitivity of the speed coordination coefficient is related to the speed deviation rate and speed control weight, the adjustment sensitivity of the torque distribution coefficient is related to the current imbalance and torque control weight, and the adjustment sensitivity of the position synchronization coefficient is related to the position deviation and position control weight. Step S355: Using the model predictive control algorithm, the adjustment sensitivity weight, parameter deviation rate and target state deviation vector are input into the coordination coefficient optimization module of the staged coordinated control model to solve for the optimal adjustment amount of the coordination coefficient. After load coupling constraint verification, the speed coordination coefficient, torque distribution coefficient and position synchronization coefficient are updated.

10. A joint control system for multiple servo motors, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, for performing the joint control method of multiple servo motors as described in any one of claims 1-9.

Images