Intelligent optimization method for control parameters of servo motor system of mobile robot

Abstract

The invention discloses a control parameter intelligent optimization method for a mobile robot servo motor system, and relates to the technical field of robots, and the method comprises the steps: collecting servo motor information and a robot motion response signal in real time; distinguishing steady-state, transition operation and extreme disturbance working conditions through working condition evolution identification; establishing a mechanism mapping model between a control parameter and a dynamic response index, constructing a multi-target parameter optimization space, and dynamically adjusting a target weight according to a working condition identification result; and carrying out stabilization processing and convergence correction on an online optimization result, outputting a target control parameter combination and loading the target control parameter combination to a motor control ring. The technical problems that in the prior art, servo control parameters under the dynamic complex working condition cannot be adjusted in a self-adaptive mode, and multiple performance indexes are difficult to balance in real time are solved, and comprehensive optimization of the response speed, overshoot suppression, steady-state precision and energy consumption efficiency is achieved. And the control performance and the environmental adaptability of the servo motor system of the mobile robot are improved.

Description

Technical Field

[0001] This invention relates to the field of robot-related technologies, and in particular to an intelligent optimization method for control parameters of a mobile robot servo motor system. Background Technology

[0002] As the core execution unit for motion control in mobile robots, the performance of the servo motor system directly determines the robot's positioning accuracy, dynamic response speed, operational stability, and energy efficiency. In actual operation, mobile robots face complex and ever-changing environments and task requirements, such as sudden load changes, rapid trajectory switching, variations in ground friction, and external collision disturbances. Traditional servo motor control often employs fixed-parameter PID control or fixed control strategies, which struggle to maintain optimal performance under all operating conditions, easily leading to problems such as response lag, increased overshoot, accumulated steady-state error, or abnormal oscillations. Currently, adjusting servo motor control parameters lacks adaptability to real-time operating conditions and has high computational complexity, making real-time online optimization difficult on embedded platforms for mobile robots. Furthermore, it neglects the need to balance multiple performance indicators in real-world multi-task scenarios, such as the trade-off between response speed and overshoot risk, and tracking accuracy and energy efficiency. This fails to meet the comprehensive motion performance, energy efficiency, and stability requirements of mobile robots, severely impacting their motion accuracy and operational reliability.

[0003] At present, there are technical problems in related technologies, such as the inability to adaptively adjust servo control parameters under dynamic and complex working conditions and the difficulty in balancing multiple performance indicators in real time. Summary of the Invention

[0004] This application provides an intelligent optimization method for control parameters of a mobile robot servo motor system, which solves the technical problems in the prior art of being unable to adaptively adjust servo control parameters under dynamic and complex working conditions and having difficulty balancing multiple performance indicators in real time. It achieves the comprehensive optimization of response speed, overshoot suppression, steady-state accuracy and energy efficiency, thereby improving the control performance and environmental adaptability of the mobile robot servo motor system.

[0005] This application provides an intelligent optimization method for control parameters of a mobile robot servo motor system. The method includes: during the operation of the mobile robot, real-time acquisition of current, voltage, and speed feedback signals of the servo motor, as well as robot motion response information associated with the servo motor; constructing a multi-source operating feature set based on the servo motor feedback signals and robot motion response information; performing condition evolution identification on the multi-source operating feature set, including load change trends, motion state change intensity, and external disturbance sensitivity, to distinguish between steady-state operating conditions, transitional operating conditions, and extreme disturbance operating conditions; constructing a control parameter action mechanism mapping model, which characterizes the causal influence of different control parameter combinations on the system's dynamic response indicators, including response speed, overshoot risk, steady-state error accumulation, and energy consumption change; constructing a multi-objective parameter optimization target space based on the control parameter action mechanism mapping model, and introducing target weights dynamically adjusted according to the operating condition evolution identification results within the multi-objective parameter optimization space, performing online parameter optimization; performing stability suppression and convergence correction on the online parameter optimization results, outputting the target control parameter combination, and loading it into the servo motor control loop.

[0006] In a possible implementation, the intelligent optimization method for control parameters of a mobile robot servo motor system further performs the following processing: based on the working condition evolution identification results, the importance of dynamic response indicators corresponding to different operating conditions is mapped to an initial weight vector; based on the real-time updated load change trend, motion state change intensity, and external disturbance sensitivity, the initial weight vector is continuously updated to form a dynamic weight distribution; the dynamic weight distribution is coupled with each response indicator to construct a weighted multi-objective performance evaluation function; the weighted multi-objective performance evaluation function is used to perform a comprehensive scoring of candidate parameters in the multi-objective parameter optimization space, and online parameter optimization is performed based on the comprehensive scoring results.

[0007] In a possible implementation, the intelligent optimization method for control parameters of a mobile robot servo motor system further performs the following processing: mapping the control parameter action mechanism mapping model into a graph structure to construct a control parameter coupling network; dynamically adjusting the topology and network edge weights of the control parameter coupling network based on the working condition evolution identification results, real-time updated load change trends, motion state change intensity, and external disturbance sensitivity, wherein network nodes represent control parameters, and network edge weights characterize the causal influence intensity between parameters; coupling the dynamic weight distribution with the network edge weights to construct a state-aware causal coupling evaluation channel, which is used to quantify the joint performance and potential risks of each candidate control parameter combination under the current working condition; performing a state-driven causal coupling search within the causal coupling evaluation channel to construct candidate parameter combinations; and using the weighted multi-objective performance evaluation function to comprehensively score the candidate parameter combinations and perform online parameter optimization.

[0008] In a possible implementation, the intelligent optimization method for control parameters of a mobile robot servo motor system further performs the following processing: dynamically determining the search direction and search step size based on the causal influence gradient of node and neighborhood parameters; filtering parameter coupling strength within the parameter coupling network using a preset coupling degree threshold, configuring high-coupling parameter subset identifiers and low-coupling parameter subset identifiers; executing a joint optimization strategy based on the high-coupling parameter subset and updating the parameter combination within the subset, and performing local optimization processing on the low-coupling parameter subset to complete the causal coupling search.

[0009] In a possible implementation, the intelligent optimization method for control parameters of a mobile robot servo motor system further performs the following processing: segmenting the multi-source operating feature set based on a time sliding window, extracting evolutionary features reflecting servo system load changes, speed response dynamics, control output fluctuations, and disturbance injection intensity within each time period; constructing operating feature change relationships based on evolutionary features between adjacent time periods, generating a working condition evolution trajectory to characterize the servo system state transition trend; and using the working condition evolution trajectory to identify steady-state operating conditions, transitional operating conditions, and extreme disturbance operating conditions, constructing a working condition evolution identification result.

[0010] In a possible implementation, the intelligent optimization method for control parameters of a mobile robot servo motor system further performs the following processing: constructing a parameter change safety constraint network based on the operating condition evolution identification results; the parameter change safety constraint network is used to limit the maximum change amplitude, change rate, and parameter coupling change consistency of each control parameter within adjacent control parameter update cycles; mapping the online parameter optimization results to parameter change trajectories; inputting the parameter change trajectories into the parameter change safety constraint network; performing a feasibility assessment; identifying high-risk parameter change segments that may induce system oscillation, overshoot, or servo instability; and performing progressive convergence correction management on the high-risk parameter change segments.

[0011] In a possible implementation, the intelligent optimization method for control parameters of a mobile robot servo motor system further performs the following processing: progressive convergence correction includes segmenting and smoothing the parameter change path, rearranging the order of highly coupled parameter changes, and adaptively decaying the parameter update step size.

[0012] In a possible implementation, the intelligent optimization method for control parameters of a mobile robot servo motor system further performs the following processing: obtaining a short-term predictive response, performing an adaptation evaluation of the target control parameter combination based on the short-term predictive response, and establishing a predictive correction result; and performing servo motor control management based on the predictive correction result.

[0013] This application proposes an intelligent optimization method for control parameters of a mobile robot servo motor system. This method involves real-time acquisition of servo motor information and robot motion response signals to construct a multi-source operating feature set. Through operational condition evolution identification, it distinguishes between steady-state, transient, and extreme disturbance conditions. A mechanistic mapping model between control parameters and dynamic response indicators is established, and a multi-objective parameter optimization space is constructed, with objective weights dynamically adjusted based on operational condition identification results. The online optimization results are stabilized and converged, and the resulting target control parameter combination is output and loaded into the motor control loop. This method solves the technical problems in existing technologies, such as the inability to adaptively adjust servo control parameters under dynamic and complex operating conditions and the difficulty in balancing multiple performance indicators in real time. It achieves comprehensive optimization of response speed, overshoot suppression, steady-state accuracy, and energy efficiency, thereby improving the control performance and environmental adaptability of the mobile robot servo motor system. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0015] Figure 1 This is a flowchart illustrating an intelligent optimization method for control parameters of a mobile robot servo motor system, provided in an embodiment of this application.

[0016] Figure 2 This is a flowchart illustrating the online parameter optimization process in a method for intelligent optimization of control parameters of a mobile robot servo motor system provided in this application embodiment. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention in order to achieve the intended purpose, the following detailed description is provided in conjunction with the accompanying drawings and preferred embodiments, based on the specific implementation methods, structures, features and effects of the present invention.

[0018] This application provides an intelligent optimization method for control parameters of a mobile robot servo motor system, such as... Figure 1 As shown, the method includes: Step S100: During the operation of the mobile robot, the current, voltage, and speed feedback signals of the servo motor and the robot motion response information associated with the servo motor are collected in real time, and a multi-source operation feature set is constructed based on the servo motor feedback signals and the robot motion response information.

[0019] Preferably, during the operation of the mobile robot, the real-time phase current or total current value output by the servo motor driver is collected in real time to reflect the actual magnitude and dynamic changes of the motor output torque, and the current feedback signal is determined; the real-time voltage value applied to the motor winding by the servo motor driver is collected in real time to reflect the electrical relationship between the output state of the driver and the motor load, and the voltage feedback signal is determined; the angular velocity or linear velocity of the motor shaft is measured in real time through position sensors such as encoders and rotary transformers to reflect the actual movement speed of the motor, and the speed feedback signal is determined.

[0020] Preferably, the robot motion response information associated with the servo motor is acquired in real time, including robot body motion state feedback, task-level motion command execution deviation, and dynamic response information associated with the servo motor. Specifically, the kinematic state of the robot as a whole or its components is measured using other sensors on the robot body, such as wheel speed encoders for wheeled robots, joint position sensors for legged robots, and IMUs. The real-time deviation between the target position, target velocity, and target acceleration commands generated by the upper controller of the mobile robot and the measured values ​​of the robot body motion state feedback is also measured. The acceleration information of the robot body directly connected to the drive wheels or joints is measured by IMU, and the estimated value of the load-end torque is calculated by current feedback combined with model calculation or directly measured by torque sensor. The actual motion trajectory of the robot end effector or center of mass is also measured.

[0021] Preferably, the servo motor feedback signal and robot motion response information are transformed into a feature vector that comprehensively describes the motion state of the servo motor system. This may include instantaneous values ​​of current, voltage, rotational speed, robot posture / position / velocity, statistical features such as mean, variance, peak value, and effective value within the sliding window, dynamic features such as the rate of change of the signal, frequency domain features, phase difference between current and rotational speed changes, and correlation coefficient, and position error and speed tracking error features between motion commands and measured responses, ultimately obtaining a multi-source operation feature set.

[0022] Step S200: Perform condition evolution identification on the multi-source operating feature set. Condition evolution identification includes load change trend, intensity of motion state change and sensitivity to external disturbances, in order to distinguish steady-state operating conditions, transitional operating conditions and extreme disturbance operating conditions.

[0023] Preferably, the working condition evolution identification refers to extracting and calculating quantitative indicators reflecting the changing trend and system sensitivity from a multi-source operating feature set, including load change trend, intensity of motion state change, and sensitivity to external disturbances. Among them, the load change trend refers to the direction and rate of change of mechanical load such as the robot body driven by the servo motor, the object being transported, and the force on the end of the robotic arm. This is judged by analyzing the slope, continuity, and correlation of the change of motor current within a specific time window. For example, monitoring whether the current continuously rises, falls, or remains constant, and combining this with robot motion commands for comprehensive identification, so as to distinguish different states such as constant load, gradual increase / decrease of load, and sudden change of load.

[0024] Preferably, the intensity of motion state change refers to the degree of drastic change in the kinematic quantities of the robot as a whole or the controlled components. By analyzing the rotational speed feedback signal, the robot's body acceleration, and the set value of the motion command, the higher-order derivative of the motion state or the energy of a specific frequency band is calculated. For example, the spectrum of the speed command or the actual speed is calculated, and the energy of the high-frequency components is analyzed to distinguish motion modes such as uniform speed or micro-motion, acceleration / deceleration, emergency start / stop or rapid trajectory tracking, which respectively represent low, medium and high intensity changes.

[0025] Preferably, external disturbance sensitivity refers to the degree of response of the system to external disturbances such as impacts, wind resistance, collisions, and external forces caused by uneven ground. It is identified by analyzing features such as high-frequency fluctuations in control errors, sudden pulsations in current or voltage, and abnormal jitters in robot posture. For example, if the feedback signal shows unexpected fluctuations without changing the motion command, it may indicate the existence and intensity of external disturbances, so as to assess whether the system is experiencing and to what extent it is affected by uncertain external disturbances.

[0026] Preferably, if the load change trend is gentle or constant, the intensity of motion state change is low, and the sensitivity to external disturbances is low, it is determined to be a steady-state operating condition, such as the robot moving in a straight line at a constant speed, maintaining a stationary position, or performing repetitive and smooth handling tasks. If the load shows a clear changing trend, the intensity of motion state change is moderate to high, and the sensitivity to external disturbances may increase but mainly comes from the system's own state switching, it is determined to be a transitional operating condition, such as the robot starting, braking, turning, curve segments in trajectory tracking, or performing grasping / placing actions. If the sensitivity to external disturbances increases significantly, strong and unexpected signal fluctuations occur, the load changes abruptly, or the motion state becomes uncontrollable or changes drastically, it is determined to be an extreme disturbance operating condition, such as wheel slippage or spinning, collision with obstacles, being subjected to strong crosswinds or external force, or sudden abnormalities in the drive system.

[0027] Step S200 further includes step S210, which involves segmenting and modeling the multi-source operating feature set based on a time sliding window, and extracting evolutionary features reflecting the load changes, speed response dynamics, control output fluctuations, and disturbance injection intensity of the servo system within each time period; step S220, which involves constructing the relationship between operating feature changes based on the evolutionary features between adjacent time periods, and generating an operating condition evolution trajectory to characterize the state transition trend of the servo system; and step S230, which involves using the operating condition evolution trajectory to identify steady-state operating conditions, transitional operating conditions, and extreme disturbance operating conditions, and constructing an operating condition evolution identification result.

[0028] Preferably, a sliding window of fixed time length is set to segment the multi-source operating feature set for modeling. That is, within each time window, the multi-source operating feature set within the window is modeled and analyzed to extract the evolution characteristics reflecting the load changes, speed response dynamics, control output fluctuations, and disturbance injection intensity of the servo system within each time period. Specifically, the linear fitting slope, root mean square (RMS) value, and variance of the current signal within the window are calculated. The sign and magnitude of the slope reflect the load increase or decrease trend, the RMS value reflects the average load level, and the variance reflects the severity of load fluctuations to reflect the load changes of the servo system. The tracking performance of the speed feedback signal relative to the speed command within the window is analyzed, and the tracking error and the smoothness of speed changes are calculated. The response delay is analyzed through cross-correlation analysis to reflect the dynamic characteristics of the speed response. The dynamic characteristics of the controller's output command are analyzed, and the fluctuation frequency and fluctuation amplitude are calculated through spectrum analysis to reflect the control output fluctuation characteristics. Unexpected disturbances are identified and quantified, and the observed or estimated values ​​of the total disturbance are calculated through an extended state observer (ESO) or model-based residuals to reflect the disturbance injection intensity characteristics.

[0029] Preferably, the relationship between changes in operating characteristics is constructed based on evolutionary features between adjacent time periods. That is, the evolutionary features extracted from the k-th window and the (k-1)-th window are compared, and the rate or amount of change of key features between adjacent windows is calculated, such as the difference in the slope of load change and the increment of the disturbance estimate. Then, the sequence of feature change relationships is arranged in chronological order to form the operating condition evolution trajectory that describes the dynamic characteristics of the system over time, reflecting the system behavior pattern. For example, the curve of the change of load trend slope over time and the curve of the change of disturbance estimate intensity over time.

[0030] Preferably, the operating condition evolution trajectory is used for state identification and classification to construct the operating condition evolution identification result. Specifically, when the operating condition evolution trajectory shows that the absolute value of the load change slope, speed tracking error, and disturbance intensity and their rate of change are all continuously lower than the preset steady-state threshold, it is determined to be a steady-state operating condition; when the operating condition evolution trajectory shows that the intensity characteristics of motion state change are significant, or the load change trend characteristics show a clear and smooth trend, and the disturbance injection intensity characteristics do not exceed the disturbance threshold, it is determined to be a transitional operating condition; when the operating condition evolution trajectory shows that the disturbance injection intensity characteristics suddenly and sharply rise and exceed the disturbance threshold, or the control output fluctuation characteristics and speed response dynamic characteristics show abnormal, non-command-like violent oscillations, it is determined to be an extreme disturbance operating condition.

[0031] Step S300: Construct a control parameter action mechanism mapping model. The control parameter action mechanism mapping model is used to characterize the causal relationship between different combinations of control parameters on the dynamic response index of the system. The dynamic response index includes response speed, overshoot risk, steady-state error accumulation, and energy consumption change.

[0032] Preferably, a causal relationship learning model is trained using historical operating data or active experimental data as a control parameter action mechanism mapping model. This model is used to characterize the causal influence of different control parameter combinations on the dynamic response index of the system. It can predict the expected change direction and magnitude of various performance indicators after applying a specific control parameter combination. Here, the control parameter combination refers to the combination of parameters in the servo motor driver or controller that can be adjusted to change the system behavior, such as the gain of the PID controller, feedforward compensation parameters such as speed / acceleration, filtering parameters of low-pass filter and notch filter, switching gain and adaptive law parameters in sliding mode control, fuzzy rule weights, etc. The dynamic response indicators of a system include response speed, overshoot risk, steady-state error accumulation, and energy consumption variation. Response speed refers to how quickly the system follows changes in commands; quantifiable indicators may include the rise time of a step response, the tracking delay of a setpoint change, or bandwidth. Overshoot risk refers to the tendency and degree to which the system response exceeds the target value; quantifiable indicators may include the maximum percentage overshoot of a step response or the ratio of the peak value to the steady-state value. Steady-state error accumulation refers to the system's ability to eliminate persistent errors; quantifiable indicators may include the integral of steady-state position error or speed tracking error. Energy consumption variation refers to the energy consumed by the system to achieve specific dynamic performance; quantifiable indicators may include the root mean square value or integral value of motor current or power over one motion cycle.

[0033] Step S400: Construct a multi-objective parameter optimization target space based on the control parameter action mechanism mapping model, and introduce target weights that are dynamically adjusted according to the identification results of the working condition evolution in the multi-objective parameter optimization space, and perform online parameter optimization.

[0034] Preferably, each possible candidate control parameter combination is evaluated using a control parameter action mechanism mapping model. All control parameter combinations are mapped to points in a target multidimensional space, representing the predicted values ​​of various dynamic performance indicators under the current or predicted operating conditions. All candidate points constitute a multi-objective parameter optimization target space. Within this space, target weights are introduced that are dynamically adjusted based on the operating condition evolution identification results. This means that each dynamic performance indicator is assigned a corresponding target weight to represent its current relative importance. Furthermore, the target weights are dynamically adjusted according to the operating condition evolution identification results. Specifically, for a steady-state operating condition, the main objective is... For precision and energy saving, the weighting configuration might prioritize higher steady-state error and energy consumption, moderate overshoot, and lower response speed, with the focus on maintaining a stable, accurate, and efficient state. For transitional operating conditions, the primary objective is to quickly and smoothly complete state transitions, with a weighting configuration that prioritizes higher response speed and overshoot, moderate steady-state error, and lower energy consumption, allowing for some energy consumption in exchange for rapid, overshoot-free tracking. For extreme disturbance conditions, the primary objective is to suppress oscillations and maintain stability, with a weighting configuration that prioritizes extremely high overshoot / oscillation risk, temporarily reducing response speed, and lowering the weights for steady-state error and energy consumption, with the core objective being stability, sacrificing speed and precision to avoid system instability. Then, the target weights are fused with the predicted values ​​of various performance indicators to generate a weighted multi-objective performance evaluation function, and online parameter optimization is performed within the multi-objective parameter optimization space to ultimately determine the control parameter combination that yields the optimal weighted comprehensive score under the current operating condition.

[0035] Furthermore, such as Figure 2 As shown, step S400 further includes step S410, which maps the importance of dynamic response indicators corresponding to different operating conditions to an initial weight vector based on the operating condition evolution identification results; step S420, which continuously updates the initial weight vector based on the real-time updated load change trend, motion state change intensity, and external disturbance sensitivity to form a dynamic weight distribution; step S430, which couples the dynamic weight distribution with each response indicator to construct a weighted multi-objective performance evaluation function; and step S440, which uses the weighted multi-objective performance evaluation function to perform a comprehensive scoring of candidate parameters in the multi-objective parameter optimization space, and performs online parameter optimization based on the comprehensive scoring results.

[0036] Preferably, based on the operating condition evolution identification results, the importance of dynamic response indicators corresponding to different operating conditions is mapped to an initial weight vector. For example, the weight of energy consumption is 0.4, the weight of steady-state error is 0.3, the weight of overshoot risk is 0.2, and the weight of response speed is 0.1 for steady-state operating conditions; the weight of response speed is 0.4, the weight of overshoot risk is 0.3, the weight of steady-state error is 0.2, and the weight of energy consumption is 0.1 for transitional operating conditions; and the weight of overshoot risk is 0.6, the weight of response speed is 0.2, the weight of steady-state error is 0.1, and the weight of energy consumption is 0.1 for extreme disturbance operating conditions.

[0037] Preferably, based on the initial weights, the initial weight vector is continuously updated with real-time updated load change trends, motion state change intensity, and external disturbance sensitivity as inputs, and the fine-tuned weights are output. For example, under transitional operating conditions, if abnormally drastic start-stop changes in motion state intensity are detected, the weight of overshoot risk is automatically increased to suppress possible oscillations; under extreme disturbance conditions, if external disturbance sensitivity shows that the disturbance is rapidly decaying, the weight of overshoot risk is gradually decreased and the weight of response speed is correspondingly increased to recover tracking more quickly; under steady-state conditions, if the load change trend shows that the load is starting to increase slowly, the weights of steady-state error and energy consumption are fine-tuned; through continuous fine-tuning and updating, a dynamic weight distribution vector is output.

[0038] Preferably, the four performance index values ​​predicted by the control parameter action mechanism mapping model are transformed to an approximately comparable numerical range using a normalization function. The dynamic weight distribution vector and the normalized performance index vector are then weighted and summed to construct a weighted multi-objective performance evaluation function, used to find the optimal balance point among four conflicting objectives. The weighted multi-objective performance evaluation function is used to perform a comprehensive scoring of candidate parameters within the multi-objective parameter optimization space. Specifically, for each candidate control parameter combination, the control parameter action mechanism mapping model predicts and outputs its corresponding four performance indices. The predicted performance indices and the current dynamic weight distribution are then substituted into the weighted multi-objective performance evaluation function to calculate the comprehensive score of that control parameter combination. The search direction is then guided based on the comprehensive scores of all candidate control parameter combinations, ultimately converging to determine the control parameter combination with the highest comprehensive score, which is the control parameter combination that best meets the current task requirements and the instantaneous environment.

[0039] Further, step S440 also includes step S441, mapping the control parameter action mechanism mapping model into a graph structure to construct a control parameter coupling network; step S442, dynamically adjusting the topology and network edge weights of the control parameter coupling network based on the operating condition evolution identification results, real-time updated load change trends, motion state change intensity, and external disturbance sensitivity, wherein network nodes represent control parameters, and network edge weights characterize the causal influence intensity between parameters; step S443, coupling the dynamic weight distribution with the network edge weights to construct a state-aware causal coupling evaluation channel, which is used to quantify the joint performance and potential risks of each candidate control parameter combination under the current operating condition; step S444, performing a state-driven causal coupling search within the causal coupling evaluation channel to construct candidate parameter combinations; and step S445, using the weighted multi-objective performance evaluation function to comprehensively score the candidate parameter combinations and perform online parameter optimization.

[0040] Preferably, the control parameter mechanism mapping model is mapped into a graph structure using control parameters as nodes, the coupling relationships between parameters as connecting edges, and the causal influence strength as edge weights, thus constructing a control parameter coupling network. Then, based on the operating condition evolution identification results, real-time updated load change trends, motion state change intensity, and external disturbance sensitivity, the topology and edge weights of the control parameter coupling network are dynamically adjusted. Specifically, under certain operating conditions, some parameter coupling relationships may become irrelevant or crucial, requiring dynamic addition, removal, or activation / freezing of certain edges. For example, under extreme disturbance conditions, the importance of coupling edges directly related to stability is greatly enhanced, while parameter coupling edges related to energy-saving fine adjustment may be temporarily weakened. At the same time, the edge weights are dynamically updated based on real-time characteristics. For instance, when the motion state change intensity is high, the edge weights of couplings between parameters affecting response speed increase; when the external disturbance sensitivity is high, the edge weights of couplings between parameters affecting robustness increase.

[0041] Preferably, dynamic weight distribution is coupled with network edge weights. By propagating performance costs or benefits within the network, a state-aware causal coupling evaluation channel is constructed to quantify the joint performance and potential risks of each candidate control parameter combination under the current operating condition. For example, increasing the proportional gain may directly improve the response speed, but through the coupling edges in the network, it may significantly increase the overshoot risk caused by differential gain mismatch. The causal coupling evaluation channel can comprehensively calculate the direct effects and the indirect effects propagated by the network, thereby evaluating the overall performance and potential risks that the candidate parameter combination can bring under the current specific state. Then, a state-driven causal coupling search is performed within the causal coupling evaluation channel. That is, based on the current operating condition and real-time evolution characteristics, intelligent exploration is performed using the topology and edge weight information of the parameter coupling network to obtain multiple candidate parameter combinations that conform to the current coupling relationship. These are then input into a weighted multi-objective performance evaluation function for comprehensive scoring, and online parameter optimization is performed to finally determine the control parameter combination with the highest comprehensive score. By incorporating a state-aware parameter coupling network, the efficiency, accuracy, and safety of online parameter optimization are improved.

[0042] Furthermore, step S444 also includes: dynamically determining the search direction and search step size based on the causal influence gradient of the node and neighborhood parameters; using a preset coupling degree threshold to filter the parameter coupling strength within the parameter coupling network, configuring high-coupling parameter subset identifiers and low-coupling parameter subset identifiers; executing a joint optimization strategy based on the high-coupling parameter subset and updating the parameter combination within the subset, and performing local optimization processing on the low-coupling parameter subset to complete the causal coupling search.

[0043] Preferably, the causal influence gradient of nodes and their neighborhood parameters is calculated using a gradient propagation algorithm on the network. Specifically, for the current optimization objective, starting from a certain parameter node, the algorithm tracks the network edges to determine the sensitivity direction and magnitude of the impact of its changes on the global objective function. The search then proceeds along the direction of the largest causal influence gradient, dynamically determining the search direction and step size. The step size is dynamically related to the gradient magnitude and the stability requirements of the current operating condition. When the gradient is large and the system is in a stable condition, a larger step size is used for rapid approximation; when the gradient is small or the system is in a sensitive condition such as extreme disturbances, a smaller step size is used for cautious exploration to avoid oscillations. A preset coupling threshold is configured based on historical data to determine whether the coupling relationship between two parameters in the network is significant. All edges of the parameter-coupled network are traversed, and the edge weights are compared with the preset coupling threshold for selection. Nodes with edge weights exceeding the preset coupling threshold are configured as a high-coupling parameter subset and identified, while nodes with sparse connections or edge weights below the preset coupling threshold are configured as a low-coupling parameter subset and identified. Then, a joint optimization strategy is executed based on the highly coupled parameter subset, that is, a fine grid search or gradient optimization under full connectivity is performed, while generating a new combination of all parameters in the subset to update the parameter combination in the subset; the low-coupling parameter subset is subjected to local optimization, that is, a low-dimensional and fast independent optimization is performed in the subset using a small step size local exploration to complete the causal coupling search, and finally construct the candidate parameter combination.

[0044] Step S500: The online parameter optimization results are subjected to stability suppression and convergence correction, and the target control parameter combination is output and loaded into the servo motor control loop.

[0045] Step S500 further includes step S510, constructing a parameter change safety constraint network based on the operating condition evolution identification result, wherein the parameter change safety constraint network is used to limit the maximum change amplitude, change rate, and parameter coupling change consistency of each control parameter within adjacent control parameter update cycles; step S520, mapping the online parameter optimization result to a parameter change trajectory, inputting the parameter change trajectory into the parameter change safety constraint network, performing a feasibility assessment, and identifying high-risk parameter change segments that may induce system oscillation, overshoot, or servo instability; step S530, performing progressive convergence correction management on the high-risk parameter change segments.

[0046] Furthermore, step S530 also includes progressive convergence correction, which includes segmenting and smoothing the parameter change path, rearranging the order of highly coupled parameter changes, and adaptively decaying the parameter update step size.

[0047] Preferably, the online parameter optimization results are subjected to stability suppression and convergence correction. Specifically, a parameter change safety constraint network is constructed based on the operating condition evolution identification results. The constraints for steady-state operating conditions are relatively lenient, while the constraints for transitional / extreme disturbance operating conditions are very strict. The parameter change safety constraint network is used to limit the maximum change amplitude, change rate, and parameter coupling change consistency of each control parameter within adjacent control parameter update cycles. The maximum change amplitude is used to constrain the maximum absolute value of a single parameter that can be increased or decreased by no more than 10% between two adjacent updates. The maximum change rate is used to constrain the allowable change of a single parameter per unit time to not exceed 20% of its current value. The parameter coupling change consistency is used to constrain the cooperative rules that a group of strongly coupled parameters must follow when changing. For example, when the proportional gain is increased to improve the response speed, the derivative gain must also be increased accordingly to maintain damping. The integral gain is prohibited from decreasing significantly on its own.

[0048] Preferably, the online parameter optimization results are compared with the currently running parameter values, mapping them to a parameter change trajectory from the current value to the target value. This trajectory is then input into the parameter change safety constraint network for feasibility assessment. Specifically, the parameter change trajectory is compared item by item with the network to evaluate whether the parameter change at each step exceeds the maximum amplitude, whether the change rate exceeds the rate of change, and whether related parameter changes conform to the coupling change consistency rule. This identifies high-risk parameter change segments that may induce system oscillations, overshoot, or servo instability, and marks them as such. Finally, a gradual convergence correction management is implemented for these high-risk segments. This includes techniques such as piecewise smoothing (decomposing aggressive jumps into multiple small-amplitude, continuous change steps), rearranging the order of highly coupled parameter changes (re-arranging the timing of parameter changes for adjustments that violate coupling change consistency), and adaptive state decay (dynamically decreasing the step size of parameter updates based on real-time feedback such as the current error magnitude and oscillation amplitude). Finally, the target control parameter combination is output, which includes the specific values ​​of all parameters that need to be adjusted in the servo motor controller. This value is then loaded into the servo motor control loop via the fieldbus, acting on the position loop, speed loop, and current loop of the servo motor. This ensures long-term stable operation in complex dynamic environments and improves operational reliability.

[0049] Furthermore, step S500 also includes step S540, obtaining a short-term predictive response, performing an adaptation evaluation of the target control parameter combination based on the short-term predictive response, and establishing a predictive correction result; step S550, performing servo motor control management based on the predictive correction result.

[0050] Preferably, before loading the target control parameter combination into the servo motor control loop, a short-term predicted response within a short time window is obtained through simulation in a digital environment. This may include predicted position / velocity tracking error curves, predicted current / torque output curves, predicted overshoot, predicted oscillation trends, etc. The short-term predicted response is then analyzed to evaluate the suitability of the target control parameter combination, including assessing its stability, smoothness, performance compliance, and risk warnings. A prediction correction result is then established, and servo motor control management is executed based on the prediction correction result. If the predicted response performs well and meets all safety and performance expectations, the prediction correction result is approved for loading, and the original target control parameter combination is safely loaded into the servo control loop. If the predicted response shows minor flaws such as slight overshoot or slightly slow convergence, the prediction correction result is a fine-tuning instruction, correcting the target control parameter combination to generate the corrected final parameter combination, which is then loaded into the control loop. If the predicted response shows significant divergence, severe oscillation, or other serious problems, the prediction correction result is rejected, potentially abandoning the current target control parameter combination and activating management contingency plans, such as triggering new online parameter optimization or switching to a safety mode. This enables feedforward adaptive control and improves the control performance and environmental adaptability of the mobile robot's servo motor system.

[0051] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for intelligent optimization of control parameters of a mobile robot servo motor system, characterized in that, The method includes: During the operation of the mobile robot, the current, voltage, and speed feedback signals of the servo motor and the robot motion response information associated with the servo motor are collected in real time. A multi-source operation feature set is constructed based on the servo motor feedback signals and the robot motion response information. The multi-source operating feature set is used to identify the operating condition evolution, which includes load change trend, intensity of motion state change and sensitivity to external disturbances, in order to distinguish steady-state operating conditions, transitional operating conditions and extreme disturbance conditions. A control parameter action mechanism mapping model is constructed. The control parameter action mechanism mapping model is used to characterize the causal relationship between different combinations of control parameters on the dynamic response index of the system. The dynamic response index includes response speed, overshoot risk, steady-state error accumulation and energy consumption change. Based on the control parameter action mechanism mapping model, a multi-objective parameter optimization target space is constructed, and target weights that are dynamically adjusted according to the identification results of the working condition evolution are introduced into the multi-objective parameter optimization space to perform online parameter optimization. The online parameter optimization results are subjected to stability suppression and convergence correction, and the target control parameter combination is output and loaded into the servo motor control loop.

2. The intelligent optimization method for control parameters of a mobile robot servo motor system as described in claim 1, characterized in that, Within the multi-objective parameter optimization space, target weights are introduced that are dynamically adjusted based on the identification results of the evolving working conditions. Online parameter optimization is then performed, including: Based on the operating condition evolution identification results, the importance of dynamic response indicators corresponding to different operating conditions is mapped to an initial weight vector; Based on the real-time updated load change trend, motion state change intensity, and external disturbance sensitivity, the initial weight vector is continuously updated to form a dynamic weight distribution; The dynamic weight distribution is coupled with each response index to construct a weighted multi-objective performance evaluation function. The weighted multi-objective performance evaluation function is used to perform comprehensive scoring of candidate parameters in the multi-objective parameter optimization space, and online parameter optimization is performed based on the comprehensive scoring results.

3. The intelligent optimization method for control parameters of a mobile robot servo motor system as described in claim 2, characterized in that, Online parameter optimization is performed based on the comprehensive scoring results, including: The control parameter action mechanism mapping model is mapped into a graph structure to construct a control parameter coupling network; Based on the operating condition evolution identification results, the real-time updated load change trend, the intensity of motion state change and the sensitivity to external disturbances, the topology and network edge weights of the control parameter coupling network are dynamically adjusted, where network nodes represent control parameters and network edge weights characterize the causal influence strength between parameters. The dynamic weight distribution is coupled with the network edge weights to construct a state-aware causal coupling evaluation channel, which is used to quantify the joint performance and potential risks of each candidate control parameter combination under the current operating conditions. A state-driven causal coupling search is performed within the causal coupling evaluation channel to construct candidate parameter combinations; The weighted multi-objective performance evaluation function is used to comprehensively score the candidate parameter combinations, and online parameter optimization is performed.

4. The intelligent optimization method for control parameters of a mobile robot servo motor system as described in claim 3, characterized in that, Performing a state-driven causal coupling search within the causal coupling evaluation channel includes: The search direction and search step size are dynamically determined based on the causal influence gradient of node and neighborhood parameters. The parameter coupling strength within the parameter coupling network is filtered using a preset coupling threshold, and high-coupling parameter subset identifiers and low-coupling parameter subset identifiers are configured. A joint optimization strategy is executed based on a subset of parameters with high coupling degree, and the parameter combination within the subset is updated. The subset of parameters with low coupling degree is then locally optimized to complete the causal coupling search.

5. The intelligent optimization method for control parameters of a mobile robot servo motor system as described in claim 1, characterized in that, The process of identifying the evolution of operating conditions for the multi-source operating feature set includes: The multi-source operation feature set is segmented and modeled based on a time sliding window, and the evolutionary features reflecting the load change, speed response dynamics, control output fluctuations and disturbance injection intensity of the servo system are extracted in each time period. Based on evolutionary features, the relationship between changes in operating features is constructed between adjacent time periods to generate an operating condition evolution trajectory that characterizes the state transition trend of the servo system. The operating condition evolution trajectory is used to identify steady-state operating conditions, transitional operating conditions, and extreme disturbance operating conditions, and to construct the operating condition evolution identification results.

6. The intelligent optimization method for control parameters of a mobile robot servo motor system as described in claim 1, characterized in that, The online parameter optimization results are subjected to stability suppression and convergence correction, including: Based on the operating condition evolution identification results, a parameter change safety constraint network is constructed. The parameter change safety constraint network is used to limit the maximum change amplitude, change rate and parameter coupling change consistency of each control parameter within the adjacent control parameter update cycle. The online parameter optimization results are mapped to parameter change trajectories, and the parameter change trajectories are input into the parameter change safety constraint network to perform a feasibility assessment and identify high-risk parameter change sections that may induce system oscillation, overshoot, or servo instability. A gradual convergence correction management is implemented for the high-risk parameter change range.

7. The intelligent optimization method for control parameters of a mobile robot servo motor system as described in claim 6, characterized in that, The incremental convergence correction includes segmenting and smoothing the parameter change path, rearranging the order of highly coupled parameter changes, and adaptively decaying the parameter update step size.

8. The intelligent optimization method for control parameters of a mobile robot servo motor system as described in claim 1, characterized in that, The output target control parameter combination also includes: Obtain short-term predicted response, evaluate the adaptation of target control parameter combinations based on the short-term predicted response, and establish prediction correction results; Servo motor control management is performed based on the predicted correction results.

Images