Lightweight high-efficiency precise motor control system for robot

Abstract

The invention discloses a lightweight high-efficiency precise motor control system for a robot, and relates to the technical field of motor control and robots, and the system comprises an intelligent integrated communication hub which is in communication connection with a sensing and monitoring module, a control and planning module and an analysis and decision support module; the sensing and monitoring module monitors the position of a motor shaft and motor current data through an absolute magnetic encoder and a Hall effect sensor; the control and planning module designs a hierarchical control architecture, integrates PID control, adaptive control and model prediction control, calculates and sends a control instruction, performs robot fault diagnosis, mode recognition and decision optimization through a machine learning algorithm, and insights the long-term trend and abnormal conditions of the robot through a big data analysis platform; and the analysis and decision support module develops a graphical user interface, displays the state of the robot and adjusts parameters. According to the scheme design, accurate control and real-time response of robot control are improved.

Description

Technical Field

[0001] The present invention relates to the field of motor control and robot technology, and in particular to a lightweight, efficient and precise motor control system for a robot. Background Art

[0002] At present, the application of robots has penetrated into many industries and fields, showing strong application potential and value. In the manufacturing industry, robots are widely used in key links such as welding, assembly, spraying, and handling, which improves production efficiency and product quality. In the agricultural field, robots are used for operations such as tillage, sowing, irrigation, and plant protection, effectively alleviating the problem of labor shortage in rural areas. In addition, robots also play an irreplaceable role in the fields of construction, energy, trade and logistics, such as material distribution and steel bar processing in the construction industry, inspection, operation, and maintenance in the energy industry, and automatic guided vehicles and distribution robots in trade and logistics. With the continuous advancement of technology, the application of robots in medical health, elderly care services, education and other fields is also becoming more and more extensive. For example, in the medical field, surgical robots, nursing robots, etc. have gradually been used in clinical diagnosis and treatment, providing strong support for medical staff. In summary, robots have become an indispensable part of modern society, and their application scenarios are constantly expanding, making important contributions to the development and progress of human society.

[0003] With the rapid development of science and technology, robots are increasingly used in all walks of life, but their precision control and real-time response capabilities have become key factors restricting their further development. In order to break through this bottleneck, a lightweight, efficient and precise motor control system for robots is proposed. By integrating advanced perception and monitoring modules, the motor shaft position data and motor current data can be obtained in real time and accurately to ensure the stable operation of the robot. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a lightweight, efficient and precise motor control system for a robot, which solves the problems in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A lightweight, efficient and precise motor control system for a robot, comprising an intelligent integrated communication hub, wherein the intelligent integrated communication hub is communicatively connected with a perception and monitoring module, a control and planning module and an analysis and decision support module; The sensing and monitoring module detects the robot motor shaft position data through an absolute magnetic encoder and monitors the motor current data through a Hall effect sensor; The control and planning module designs a hierarchical control architecture, comprehensively uses PID control, adaptive control and model predictive control, calculates and sends control instructions, performs robot fault diagnosis, pattern recognition and decision optimization through DNN algorithm, autoencoder algorithm and reinforcement learning algorithm, and predicts the long-term trend and abnormal situation of the robot through the big data analysis platform; The analysis and decision support module develops a graphical user interface to display the robot status and adjust parameters, and realizes remote monitoring and maintenance of the robot through the Internet of Things technology.

[0006] Furthermore, in the perception and monitoring module, the process of detecting the position data of the robot motor shaft by an absolute magnetic encoder includes: An absolute magnetic encoder is deployed on the robot motor rotor and directly connected to the motor shaft. The absolute magnetic encoder determines the motor shaft position by sensing the magnetic field changes generated by the magnetic ring fixed on the rotor. When the motor shaft rotates, the sensor element inside the absolute magnetic encoder detects the change in magnetic field strength, converts the change in magnetic field strength into an electrical signal, and provides a coding value for each motor shaft position, so that the current position can be located immediately after restarting after power failure. The high-frequency noise in the electrical signal is removed by a low-pass filter, the reading is adjusted according to the known standard position, the offset and scaling errors are compensated, and the nonlinear response of the encoder output is converted into a linear output by mathematical transformation; Parse the binary output representing the shaft position, calculate the motor speed through the position change rate, further analyze the speed change rate to obtain acceleration information, monitor the sudden changes and abnormal patterns of the absolute encoder electrical signal, and identify sticking and loss of step in time.

[0007] Furthermore, in the sensing and monitoring module, the process of monitoring the motor current data through the Hall effect sensor includes: Hall effect sensors are installed around the motor controller and directly connected to the power line. When current passes through the conductor, the Hall effect sensor applies a magnetic field perpendicular to the current direction, generating Hall voltage on both sides of the conductor. The Hall voltage is proportional to the current. The current data is indirectly measured by detecting the Hall voltage. For DC environment, the Hall effect sensor provides a linear current and voltage conversion relationship. For AC environment, the Hall effect sensor captures the current waveform changes and analyzes the current effective value and peak characteristics. Preprocessing the raw current signal obtained from the Hall effect sensor, the preprocessing operation includes using an analog low-pass filter to remove high-frequency noise components in the raw current signal, adjusting the amplification factor so that the output signal is within the optimal input range of the subsequent processing circuit, eliminating the bias error so that the output in the no-current state is close to zero, and compensating for the drift caused by temperature changes; Calculate the average power level of the pre-processed current data that reflects the motor load condition, identify the harmonic components in the current waveform, capture current mutations as signs of start, stop and abnormal operating conditions, convert the time domain signal into a frequency domain representation through Fourier transform, assist in discovering potential problem patterns, integrate current data with time, and calculate the energy utilization efficiency of the motor.

[0008] Furthermore, in the control and planning module, a hierarchical control architecture is designed, and the process of maintaining the basic stable operation of the motor through PID control includes: A hierarchical control architecture is constructed, which includes a bottom layer, a middle layer and a top layer, wherein PID is used as a bottom feedback controller to receive motor shaft position data and motor current data from a sensing and monitoring module, perform state estimation on the position data and current data, calculate key parameters for control decisions, and the key parameters include motor shaft position error and speed change rate. A genetic algorithm is used to automatically adjust PID parameters so that the system response has good steady-state accuracy, optimize the control effect through feedforward compensation and nonlinear compensation, implement saturation limitation and anti-integral saturation strategies to ensure safe operation of the drive, continuously monitor the health of the system, regularly evaluate key performance indicators and gradually adjust PID parameters in combination with adaptive technology, and provide basic feedback control, and the key performance indicators include tracking error and energy efficiency.

[0009] Furthermore, in the control and planning module, a hierarchical control architecture is designed, and the process of monitoring and adjusting the dynamic characteristics of the system through adaptive control includes: The adaptive control mechanism of the middle layer calculates the moment of inertia and friction coefficient of the system in real time according to the motor shaft position data and motor current data of the motor using the recursive least squares method. The adaptive controller adjusts the proportional, integral and differential gains of the PID controller in real time based on the estimated values ​​to cope with the influence of load changes and temperature fluctuations. An adaptive algorithm is introduced to perform system identification and model updates regularly, and an uncertainty compensation mechanism is introduced to realize fault detection and isolation. It optimizes the dynamic response for rapidly changing working conditions and smooth transition requirements, and is responsible for coordinating the multi-degree-of-freedom collaborative actions of the robot.

[0010] Furthermore, in the control and planning module, the process of designing a hierarchical control architecture, establishing a kinematic and dynamic model, and realizing the motion planning and trajectory generation of the robot includes: The top layer builds kinematic and dynamic models based on the robot's physical structure and motion characteristics to describe the relative motion relationship and torque balance equation between the robot's joints. It uses historical data and online measurement results to regularly perform system identification and update kinematic and dynamic model parameters to reflect dynamic characteristics. Based on the collected data, the robot behavior is predicted in combination with the pre-established kinematic and dynamic models and constraints, including the maximum torque limit and the upper speed limit, and the influence of load and environmental changes on the robot is comprehensively considered, and the interference model is introduced to improve the prediction accuracy; By solving the optimization problem within a limited time range, the optimal control sequence is determined so that the robot moves along the predetermined trajectory while meeting the performance indicators. Under the principle of optimizing the overall performance, the problem of maximizing speed while minimizing energy consumption and balancing response speed and stability are solved; Develop a quadratic programming algorithm to quickly solve optimization problems in a real-time environment, provide immediate control instructions, and adopt a rolling time domain strategy to only execute the first control action each time and calculate a new optimal sequence in the next sampling period to form a closed-loop control to adapt to changing working conditions.

[0011] Furthermore, in the control and planning module, the process of integrating the results from different levels of control algorithms, forming control instructions and sending them to the drive includes: The control instructions are formed by weighted averaging the results of the PID control algorithm, adaptive control algorithm and MPC algorithm. Before the control instructions are sent, a safety check is performed. The safety check includes saturation limit, anti-integral saturation strategy and hardware limit compliance to prevent overload and dangerous operation. The final control instructions are converted into PWM signals suitable for the drive and transmitted to the motor drive through the intelligent integrated communication hub. The system operation status is continuously monitored, key performance indicators are recorded, and control parameters and weight distribution are adjusted in real time based on the actual execution effect.

[0012] Furthermore, in the control and planning module, the process of performing fault diagnosis of the robot by using the DNN algorithm, performing pattern recognition of the robot by using the autoencoder algorithm, and performing decision optimization of the robot by using the reinforcement learning algorithm includes: DNN receives the motor shaft position data and motor current data pre-processed by the perception and monitoring module, uses the historical data set to train DNN online, adjusts the weights through the back-propagation algorithm to minimize the loss function, automatically extracts local features in the time series data through the convolution layer, integrates global information through the fully connected layer, and finally outputs the fault category label. The trained DNN model is deployed, the collected data is received online, and the system status is evaluated in real time. When the model predicts the fault category label, the alarm mechanism is immediately triggered to notify the operator to take corresponding measures; The autoencoder receives the motor shaft position data and motor current data preprocessed by the perception and monitoring module, maps the data to a low-dimensional latent space through the encoder, removes redundant features, and the decoder reconstructs the original input from the low-dimensional representation to generate reconstructed data. By comparing the reconstruction errors, abnormal patterns and new operation types are discovered. Samples with abnormal reconstruction errors are used to indicate new tasks and precursors to faults. Based on the reconstruction errors, the autoencoder classifies different operation modes and establishes a corresponding pattern library. The operation modes that cannot be identified by the clustering algorithm are added to the pattern library as new categories. According to the physical structure and motion characteristics of the robot, the state space, action space and reward function are defined. The state space describes the position, velocity and acceleration of each joint of the robot. The action space covers the control instructions. The reward function reflects the expected speed and energy efficiency of reaching the target position. The reinforcement learning algorithm learns the best strategy through trial and error, selects different actions to maximize the cumulative reward, and uses the DQN algorithm to approximate the Q-value function using a neural network. In each sampling period, DQN selects the optimal action based on the current state, and updates the Q-value estimate based on the actual effect, gradually approaching the optimal strategy. In actual applications, only the first control action currently calculated is executed each time, and a new optimal sequence is calculated in the next sampling period to form a closed-loop control.

[0013] Furthermore, in the control and planning module, the process of predicting the long-term trend and abnormal conditions of the robot through the big data analysis platform includes: The big data analysis platform integrates the motor shaft position data, motor current data, position error, and speed change rate data from the control and planning module, and after cleaning, standardization, and distributed storage operations, uses time series analysis and statistical process control technology to monitor the changing trends of key performance indicators, and combines regression analysis to explore the influencing factors. Through real-time monitoring and alarm mechanisms, it promptly detects and warns of abnormal events, and conducts root cause analysis to guide maintenance decisions.

[0014] Furthermore, in the analysis and decision support module, a graphical user interface is developed to display the robot status and adjust parameters, and the process of realizing remote monitoring and maintenance of the robot through the Internet of Things technology includes: The user interface integrates motor shaft position data and motor current data, and displays key performance indicators through charts and dashboards to assist operators in understanding system operation status. Users can directly adjust the PID gain and MPC optimization target weight of the control algorithm through the interactive interface, and reflect them in the control system. The built-in historical data analysis function allows users to query historical performance records and abnormal events, assisting in troubleshooting and performance optimization; The intelligent integrated communication hub uploads collected data to the cloud server through the MQTT protocol while receiving remote control commands. It authorizes users to access the system status and adjust parameters through mobile applications. It supports multi-device access. The system sets an automatic alarm mechanism triggered by thresholds. When an abnormal situation is detected, it immediately notifies the operator, generates maintenance reports regularly, provides system health assessments and improvement suggestions, and supports scientific decision-making and preventive maintenance.

[0015] The present invention provides a lightweight, efficient and precise motor control system for a robot. It has the following beneficial effects: First, by integrating the intelligent integrated communication hub, efficient and reliable communication between the perception and monitoring module, the control and planning module, and the analysis and decision support module is achieved, thereby ensuring the coordinated work and stable operation of the entire motor control system. Secondly, the perception and monitoring module uses an absolute magnetic encoder and a Hall effect sensor to achieve accurate monitoring of the motor shaft position and motor current data, providing strong data support for the precise control of the robot. This helps to improve the robot's motion accuracy and stability, enabling it to perform well in various application scenarios. Furthermore, the control and planning module adopts a hierarchical control architecture and a variety of advanced control algorithms to achieve efficient and precise control of the motor. At the same time, by introducing machine learning algorithms and big data analysis platforms, the system can also perform fault diagnosis, pattern recognition and decision optimization of the robot, thereby improving the robot's intelligence level and self-healing ability. Finally, the analysis and decision support module provides a graphical user interface, allowing users to intuitively understand the robot's status and adjust parameters. Through the Internet of Things technology, the system also realizes remote monitoring and maintenance of the robot, further improving the robot's availability and reliability. In summary, the invention is of great significance in improving the robot's control accuracy, stability and intelligence level. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a block diagram of a lightweight, efficient and precise motor control system for a robot according to an embodiment of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] like Figure 1As shown, the present invention provides a technical solution: a lightweight, efficient and precise motor control system for a robot, comprising an intelligent integrated communication hub, wherein the intelligent integrated communication hub is communicatively connected with a perception and monitoring module, a control and planning module and an analysis and decision support module; The sensing and monitoring module detects the robot motor shaft position data through an absolute magnetic encoder and monitors the motor current data through a Hall effect sensor. Control and planning module, designing hierarchical control architecture, comprehensively using PID control, adaptive control and model predictive control, calculating and sending control instructions, performing robot fault diagnosis, pattern recognition and decision optimization through DNN algorithm, autoencoder algorithm and reinforcement learning algorithm, and predicting the long-term trend and abnormal situation of the robot through the big data analysis platform; Analysis and decision support module, develop a graphical user interface, display the robot status and adjust parameters, and realize remote monitoring and maintenance of the robot through Internet of Things technology.

[0019] In the perception and monitoring module, the process of detecting the robot motor shaft position data through an absolute magnetic encoder includes: An absolute magnetic encoder is deployed on the robot motor rotor and directly connected to the motor shaft. The absolute magnetic encoder determines the motor shaft position by sensing the magnetic field changes generated by the magnetic ring fixed on the rotor. When the motor shaft rotates, the sensor element inside the absolute magnetic encoder detects the change in magnetic field strength, converts the change in magnetic field strength into an electrical signal, and provides a coding value for each motor shaft position, so that the current position can be located immediately after restarting after power failure. The high-frequency noise in the electrical signal is removed by a low-pass filter, the reading is adjusted according to the known standard position, the offset and scaling errors are compensated, and the nonlinear response of the encoder output is converted into a linear output by mathematical transformation; Parse the binary output representing the shaft position, calculate the motor speed through the position change rate, further analyze the speed change rate to obtain acceleration information, monitor the sudden changes and abnormal patterns of the absolute encoder electrical signal, and identify sticking and loss of step in time.

[0020] In the sensing and monitoring module, the process of monitoring the motor current data through the Hall effect sensor includes: Hall effect sensors are installed around the motor controller and directly connected to the power line. When current passes through the conductor, the Hall effect sensor applies a magnetic field perpendicular to the current direction, generating Hall voltage on both sides of the conductor. The Hall voltage is proportional to the current. The current data is indirectly measured by detecting the Hall voltage. For DC environment, the Hall effect sensor provides a linear current and voltage conversion relationship. For AC environment, the Hall effect sensor captures the current waveform changes and analyzes the current effective value and peak characteristics. Preprocessing the raw current signal obtained from the Hall effect sensor, the preprocessing operation includes using an analog low-pass filter to remove high-frequency noise components in the raw current signal, adjusting the amplification factor so that the output signal is within the optimal input range of the subsequent processing circuit, eliminating the bias error so that the output in the no-current state is close to zero, and compensating for the drift caused by temperature changes; Calculate the average power level of the pre-processed current data that reflects the motor load condition, identify the harmonic components in the current waveform, capture current mutations as signs of start, stop and abnormal operating conditions, convert the time domain signal into a frequency domain representation through Fourier transform, assist in discovering potential problem patterns, integrate current data with time, and calculate the energy utilization efficiency of the motor.

[0021] In the control and planning module, a hierarchical control architecture is designed to maintain the basic stable operation of the motor through PID control. The process includes: A hierarchical control architecture is constructed, which includes the bottom layer, the middle layer and the top layer. Among them, PID is used as the bottom feedback controller to receive the motor shaft position data and the motor current data from the perception and monitoring module, perform state estimation on the position data and the current data, and calculate the key parameters used for control decisions. The key parameters include the motor shaft position error and the speed change rate. A genetic algorithm is used to automatically adjust the PID parameters so that the system response has good steady-state accuracy. The control effect is optimized through feedforward compensation and nonlinear compensation. Saturation limitation and anti-integral saturation strategies are implemented to ensure the safe operation of the drive. The health of the system is continuously monitored. The key performance indicators are regularly evaluated and the PID parameters are gradually adjusted in combination with adaptive technology to provide basic feedback control. The key performance indicators include tracking error and energy efficiency.

[0022] In the control and planning module, the process of designing a hierarchical control architecture and monitoring and adjusting the system dynamic characteristics through adaptive control includes: The adaptive control mechanism of the middle layer calculates the moment of inertia and friction coefficient of the system in real time according to the motor shaft position data and motor current data of the motor using the recursive least squares method. The adaptive controller adjusts the proportional, integral and differential gains of the PID controller in real time based on the estimated values ​​to cope with the influence of load changes and temperature fluctuations. An adaptive algorithm is introduced to perform system identification and model updates regularly, and an uncertainty compensation mechanism is introduced to realize fault detection and isolation. It optimizes the dynamic response for rapidly changing working conditions and smooth transition requirements, and is responsible for coordinating the multi-degree-of-freedom collaborative actions of the robot.

[0023] In the control and planning module, the process of designing a hierarchical control architecture, establishing kinematic and dynamic models, and implementing robot motion planning and trajectory generation includes: The top layer builds kinematic and dynamic models based on the robot's physical structure and motion characteristics to describe the relative motion relationship and torque balance equation between the robot's joints. It uses historical data and online measurement results to regularly perform system identification and update kinematic and dynamic model parameters to reflect dynamic characteristics. Based on the collected data, the robot behavior is predicted by combining the pre-established kinematic and dynamic models and constraints. The constraints include the maximum torque limit and the upper speed limit. The influence of load and environmental changes on the robot is comprehensively considered, and the interference model is introduced to improve the prediction accuracy. By solving the optimization problem within a limited time range, the optimal control sequence is determined so that the robot moves along the predetermined trajectory while meeting the performance indicators. Under the principle of optimizing the overall performance, the problem of maximizing speed while minimizing energy consumption and balancing response speed and stability are solved; Develop a quadratic programming algorithm to quickly solve optimization problems in a real-time environment, provide immediate control instructions, and adopt a rolling time domain strategy to only execute the first control action each time and calculate a new optimal sequence in the next sampling period to form a closed-loop control to adapt to changing working conditions.

[0024] In the control and planning module, the process of integrating the results from different levels of control algorithms, forming control instructions and sending them to the drive includes: The control instructions are formed by weighted averaging the results of the PID control algorithm, adaptive control algorithm and MPC algorithm. Before the control instructions are sent, a safety check is performed. The safety check includes saturation limit, anti-integral saturation strategy and hardware limit compliance to prevent overload and dangerous operation. The final control instructions are converted into PWM signals suitable for the drive and transmitted to the motor drive through the intelligent integrated communication hub. The system operation status is continuously monitored, key performance indicators are recorded, and control parameters and weight distribution are adjusted in real time based on the actual execution effect.

[0025] In the control and planning module, the process of robot fault diagnosis through DNN algorithm, robot pattern recognition through autoencoder algorithm and robot decision optimization through reinforcement learning algorithm includes: DNN receives the motor shaft position data and motor current data pre-processed by the perception and monitoring module, uses the historical data set to train DNN online, adjusts the weights through the back-propagation algorithm to minimize the loss function, automatically extracts local features in the time series data through the convolution layer, integrates global information through the fully connected layer, and finally outputs the fault category label. The trained DNN model is deployed, the collected data is received online, and the system status is evaluated in real time. When the model predicts the fault category label, the alarm mechanism is immediately triggered to notify the operator to take corresponding measures; The autoencoder receives the motor shaft position data and motor current data preprocessed by the perception and monitoring module, maps the data to a low-dimensional latent space through the encoder, removes redundant features, and the decoder reconstructs the original input from the low-dimensional representation to generate reconstructed data. By comparing the reconstruction errors, abnormal patterns and new operation types are discovered. Samples with abnormal reconstruction errors are used to indicate new tasks and precursors to faults. Based on the reconstruction errors, the autoencoder classifies different operation modes and establishes a corresponding pattern library. The operation modes that cannot be identified by the clustering algorithm are added to the pattern library as new categories. According to the physical structure and motion characteristics of the robot, the state space, action space and reward function are defined. The state space describes the position, velocity and acceleration of each joint of the robot. The action space covers the control instructions. The reward function reflects the expected speed and energy efficiency to reach the target position. The reinforcement learning algorithm learns the best strategy through trial and error and selects different actions to maximize the cumulative reward. The DQN algorithm uses a neural network to approximate the Q-value function. In each sampling period, DQN selects the optimal action according to the current state and updates the Q-value estimate according to the actual effect, gradually approaching the optimal strategy. In actual applications, only the first control action currently calculated is executed each time, and a new optimal sequence is calculated in the next sampling period to form a closed-loop control.

[0026] In the control and planning module, the process of predicting the long-term trends and abnormal situations of the robot through the big data analysis platform includes: The big data analysis platform integrates the motor shaft position data, motor current data, position error, and speed change rate data from the control and planning module, and after cleaning, standardization, and distributed storage operations, uses time series analysis and statistical process control technology to monitor the changing trends of key performance indicators, and combines regression analysis to explore the influencing factors. Through real-time monitoring and alarm mechanisms, it promptly detects and warns of abnormal events, and conducts root cause analysis to guide maintenance decisions.

[0027] In the analysis and decision support module, the process of developing a graphical user interface to display the robot status and adjust parameters, and realizing remote monitoring and maintenance of the robot through the Internet of Things technology includes: The user interface integrates motor shaft position data and motor current data, and displays key performance indicators through charts and dashboards to assist operators in understanding system operation status. Users can directly adjust the PID gain and MPC optimization target weight of the control algorithm through the interactive interface, and reflect them in the control system. The built-in historical data analysis function allows users to query historical performance records and abnormal events, assisting in troubleshooting and performance optimization; The intelligent integrated communication hub uploads collected data to the cloud server through the MQTT protocol while receiving remote control commands. It authorizes users to access the system status and adjust parameters through mobile applications. It supports multi-device access. The system sets an automatic alarm mechanism triggered by thresholds. When an abnormal situation is detected, it immediately notifies the operator, generates maintenance reports regularly, provides system health assessments and improvement suggestions, and supports scientific decision-making and preventive maintenance.

[0028] First, start the system and ensure that the intelligent integrated communication hub works properly. As the core of the system, the hub is responsible for data transmission and command issuance between modules. Then, the perception and monitoring module starts working, accurately detecting the position data of the motor shaft through the absolute magnetic encoder, while the Hall effect sensor monitors the motor current data in real time to ensure the accuracy and real-time nature of the data. Then, the control and planning module receives the data transmitted by the perception and monitoring module, and according to the preset hierarchical control architecture, uses PID control, adaptive control and model predictive control algorithms to calculate the optimal control instructions and send them to the driver for execution. At the same time, the module also uses machine learning algorithms to perform fault diagnosis, pattern recognition and decision optimization on the robot to improve the intelligence level of the robot. The big data analysis platform is responsible for analyzing the long-term operation data of the robot to discover potential trends and abnormal situations. Finally, the analysis and decision support module displays the current status of the robot through a graphical user interface and allows users to adjust relevant parameters as needed. In addition, through the Internet of Things technology, the module can also realize remote monitoring and maintenance of the robot to ensure the stable operation and timely maintenance of the robot. During the entire use process, the modules work together to ensure the lightweight, high efficiency and precision of the motor control system, thereby meeting the needs of the robot in various application scenarios.

[0029] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions. The sentence "includes an element defined by ... does not exclude the existence of other identical elements in the process, method, article or device including the element".

[0030] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A lightweight, efficient and precise motor control system for a robot, characterized in that: It includes an intelligent integrated communication hub, which is communicatively connected with a perception and monitoring module, a control and planning module, and an analysis and decision support module; The sensing and monitoring module detects the robot motor shaft position data through an absolute magnetic encoder and monitors the motor current data through a Hall effect sensor; The control and planning module designs a hierarchical control architecture, comprehensively uses PID control, adaptive control and model predictive control, calculates and sends control instructions, performs robot fault diagnosis, pattern recognition and decision optimization through DNN algorithm, autoencoder algorithm and reinforcement learning algorithm, and predicts the long-term trend and abnormal situation of the robot through the big data analysis platform; The analysis and decision support module develops a graphical user interface to display the robot status and adjust parameters, and realizes remote monitoring and maintenance of the robot through the Internet of Things technology.

2. A lightweight, efficient and precise motor control system for a robot according to claim 1, characterized in that: In the sensing and monitoring module, the process of detecting the robot motor shaft position data by an absolute magnetic encoder includes: An absolute magnetic encoder is deployed on the robot motor rotor and directly connected to the motor shaft. The absolute magnetic encoder determines the motor shaft position by sensing the magnetic field changes generated by the magnetic ring fixed on the rotor. When the motor shaft rotates, the sensor element inside the absolute magnetic encoder detects the change in magnetic field strength, converts the change in magnetic field strength into an electrical signal, and provides a coding value for each motor shaft position, so that the current position can be located immediately after restarting after power failure. The high-frequency noise in the electrical signal is removed by a low-pass filter, the reading is adjusted according to the known standard position, the offset and scaling errors are compensated, and the nonlinear response of the encoder output is converted into a linear output by mathematical transformation; Parse the binary output representing the shaft position, calculate the motor speed through the position change rate, further analyze the speed change rate to obtain acceleration information, monitor the sudden changes and abnormal patterns of the absolute encoder electrical signal, and identify sticking and loss of step.

3. A lightweight, efficient and precise motor control system for a robot according to claim 2, characterized in that: In the sensing and monitoring module, the process of monitoring the motor current data through the Hall effect sensor includes: Install Hall effect sensors around the motor controller and connect them to the power line. When current passes through the conductor, the Hall effect sensor applies a magnetic field perpendicular to the current direction, generating Hall voltage on both sides of the conductor. The Hall voltage is proportional to the current. The current data is indirectly measured by detecting the Hall voltage. For DC environment, the Hall effect sensor provides a linear current-voltage conversion relationship. For AC environment, the Hall effect sensor captures the current waveform changes and analyzes the current effective value and peak characteristics. Preprocessing the raw current signal obtained from the Hall effect sensor, wherein the preprocessing operation includes using an analog low-pass filter to remove high-frequency noise components in the raw current signal, adjusting the amplification factor, eliminating the bias error, making the output close to zero in a no-current state, and compensating for the drift caused by temperature changes; Calculate the average power level of the pre-processed current data that reflects the motor load condition, identify the harmonic components in the current waveform, capture current mutations as signs of start, stop and abnormal operating conditions, convert the time domain signal into a frequency domain representation through Fourier transform, assist in discovering potential problem patterns, integrate current data with time, and calculate the energy utilization efficiency of the motor.

4. A lightweight, efficient and precise motor control system for a robot according to claim 3, characterized in that: In the control and planning module, a hierarchical control architecture is designed to maintain the basic stable operation of the motor through PID control, including: Construct a hierarchical control architecture, which includes a bottom layer, a middle layer, and a top layer, wherein PID is used as a bottom layer feedback controller to receive motor shaft position data and motor current data from a sensing and monitoring module, perform state estimation on the position data and current data, and calculate key parameters for control decision making, wherein the key parameters include motor shaft position error and speed change rate, and use a genetic algorithm to automatically adjust PID parameters so that the system response has good steady-state accuracy; Optimize control effects through feedforward compensation and nonlinear compensation, implement saturation limitation and anti-integral saturation strategy to ensure drive operation; Continuously monitor system health, regularly evaluate key performance indicators (KPIs), and gradually adjust PID parameters in conjunction with adaptive technology to provide basic feedback control. The KPIs include tracking error and energy efficiency.

5. A lightweight, efficient and precise motor control system for a robot according to claim 4, characterized in that: In the control and planning module, the process of designing a hierarchical control architecture and monitoring and adjusting the system dynamic characteristics through adaptive control includes: The adaptive control mechanism in the middle layer uses the recursive least squares method to calculate the system's moment of inertia and friction coefficient in real time based on the motor shaft position data and motor current data of the motor. The adaptive controller adjusts the proportional, integral, and differential gains of the PID controller in real time based on the estimated values ​​to cope with the effects of load changes and temperature fluctuations. Introduce adaptive algorithms, perform system identification and model updates regularly, introduce uncertainty compensation mechanisms, realize fault detection and isolation, optimize dynamic response for rapidly changing working conditions and smooth transition requirements, and coordinate multi-degree-of-freedom collaborative actions of robots.

6. A lightweight, efficient and precise motor control system for a robot according to claim 5, characterized in that: In the control and planning module, the process of designing a hierarchical control architecture, establishing kinematic and dynamic models, and realizing the robot's motion planning and trajectory generation includes: The top layer builds kinematic and dynamic models based on the robot's physical structure and motion characteristics to describe the relative motion relationship and torque balance equation between the robot's joints. It uses historical data and online measurement results to regularly perform system identification and update kinematic and dynamic model parameters to reflect dynamic characteristics. Based on the collected data, the robot behavior is predicted by combining the pre-established kinematic and dynamic models and constraints, including the maximum torque limit and the upper speed limit, and the influence of the load and environmental changes on the robot is comprehensively considered, and an interference model is introduced; By solving the optimization problem within a limited time range, the optimal control sequence is determined so that the robot moves along the predetermined trajectory while meeting the performance indicators. Under the principle of optimizing the overall performance, the problem of maximizing speed while minimizing energy consumption and balancing response speed and stability are solved; Develop a quadratic programming algorithm to solve optimization problems in a real-time environment, provide immediate control instructions, adopt a rolling horizon strategy, execute only the first control action each time, calculate a new optimal sequence in the next sampling period, and form a closed-loop control to adapt to changing working conditions.

7. A lightweight, efficient and precise motor control system for a robot according to claim 6, characterized in that: In the control and planning module, the process of integrating the results from different levels of control algorithms, forming control instructions and sending them to the drive includes: The control instructions are formed by weighted average comprehensive PID control algorithm, adaptive control algorithm and MPC algorithm results; Before the control command is sent, a safety check is performed, which includes saturation limit, anti-integral saturation strategy and hardware limit compliance, and the final control command is converted into a PWM signal that adapts to the drive; The PWM signal is transmitted to the motor driver through the intelligent integrated communication hub, the system operation status is continuously monitored, key performance indicators are recorded, and control parameters and weight distribution are adjusted in real time based on actual execution results.

8. A lightweight, efficient and precise motor control system for a robot according to claim 7, characterized in that: In the control and planning module, the process of performing fault diagnosis of the robot by using the DNN algorithm, performing pattern recognition of the robot by using the autoencoder algorithm, and performing decision optimization of the robot by using the reinforcement learning algorithm includes: DNN receives the motor shaft position data and motor current data pre-processed by the perception and monitoring module, uses the historical data set to train DNN online, adjusts the weights through the back-propagation algorithm to minimize the loss function, automatically extracts local features in the time series data through the convolution layer, integrates global information through the fully connected layer, and finally outputs the fault category label. The trained DNN model is deployed, the collected data is received online, and the system status is evaluated in real time. When the model predicts the fault category label, the alarm mechanism is immediately triggered to notify the operator to take corresponding measures; The autoencoder receives the motor shaft position data and motor current data preprocessed by the perception and monitoring module, maps the data to a low-dimensional latent space through the encoder, removes redundant features, and the decoder reconstructs the original input from the low-dimensional representation to generate reconstructed data. By comparing the reconstruction errors, abnormal patterns and new operation types are discovered. Samples with abnormal reconstruction errors are used to indicate new tasks and precursors to faults. Based on the reconstruction errors, the autoencoder classifies different operation modes and establishes a corresponding pattern library. The operation modes that cannot be identified by the clustering algorithm are added to the pattern library as new categories. According to the physical structure and motion characteristics of the robot, the state space, action space and reward function are defined. The state space describes the position, velocity and acceleration of each joint of the robot. The action space covers the control instructions. The reward function reflects the expected speed and energy efficiency of reaching the target position. The reinforcement learning algorithm learns the best strategy through trial and error, selects different actions to maximize the cumulative reward, and uses the DQN algorithm to approximate the Q-value function using a neural network. In each sampling period, DQN selects the optimal action based on the current state, and updates the Q-value estimate based on the actual effect, gradually approaching the optimal strategy. In actual applications, only the first control action currently calculated is executed each time, and a new optimal sequence is calculated in the next sampling period to form a closed-loop control.

9. A lightweight, efficient and precise motor control system for a robot according to claim 8, characterized in that: In the control and planning module, the process of predicting the long-term trend and abnormal conditions of the robot through the big data analysis platform includes: The big data analysis platform integrates the motor shaft position data, motor current data, and the position error and speed change rate data from the control and planning modules, and performs cleaning, standardization and distributed storage operations; Time series analysis and statistical process control techniques are used to monitor the changing trends of key performance indicators, and regression analysis is combined to explore the influencing factors. Real-time monitoring and alarm mechanisms are used to timely detect and warn of abnormal events, and root cause analysis is performed to guide maintenance decisions.

10. A lightweight, efficient and precise motor control system for a robot according to claim 9, characterized in that: In the analysis and decision support module, the process of developing a graphical user interface, displaying the robot status and adjusting parameters, and realizing remote monitoring and maintenance of the robot through the Internet of Things technology includes: The user interface integrates motor shaft position data and motor current data, and displays key performance indicators through charts and dashboards to assist operators in understanding system operation status. Users can directly adjust the PID gain and MPC optimization target weight of the control algorithm through the interactive interface, and reflect it in the control system; The intelligent integrated communication hub uploads the collected data to the cloud server through the MQTT protocol, and receives remote control commands at the same time, authorizing users to access the system status and adjust parameters through mobile applications. The system sets up an automatic alarm mechanism triggered by thresholds, notifies operators when abnormal conditions are detected, and generates maintenance reports regularly.

Images