A method and system for controlling a robotic finger joint

Abstract

The application relates to the technical field of robot control, and discloses a control method and system for a robot finger joint, which comprises the following steps: performing multi-physical field signal acquisition, injecting a high-frequency excitation into a driving voltage, and extracting a current response; using a high-frequency response to analyze a resistance value, combining a heat conduction model to calculate a real-time temperature of a coil, and reconstructing an effective magnetic chain of a permanent magnet according to a temperature coefficient; according to the temperature, lubricating viscous characteristics are mapped, and an adaptive chatter signal with the amplitude and frequency dynamically mapped to a thermal state is generated in combination with a real-time rotating speed; based on the effective magnetic chain, a torque conversion coefficient is corrected, and the chatter signal is superimposed to a current loop to generate a target current instruction to drive a motor. Through the construction of a sensorless thermal magnetic observation mechanism and a thermal viscous coupling compensation strategy, the application effectively solves the problems of thermal parameter drift and friction nonlinearity of a miniature joint caused by space limitation, and guarantees high-precision torque output and low-noise smooth operation in a full-temperature-range working condition.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, specifically to a control method and system for robot finger joints. Background Technology

[0002] As an end effector, the dexterous hand of a robot typically features highly integrated and high-power-density finger joint modules to simulate the manipulative capabilities of a human hand. Due to the requirement for miniaturization, the motor stator windings and permanent magnet rotor are compressed within a limited housing space, resulting in narrow heat dissipation channels, low heat capacity, and a tendency for rapid temperature rise under high-load conditions.

[0003] Existing micro servo drive systems are limited by internal physical space, making it difficult to directly place contact temperature sensors inside the motor windings. Conventional solutions often use thermistors mounted on the drive board for indirect estimation, but due to conduction delays caused by thermal resistance and capacitance between the drive board and the motor coils, external measurement methods cannot monitor the true thermal state of the motor's internal coils in real time. The lack of coil temperature monitoring prevents the control system from sensing the dynamic drift of the motor's electrical parameters.

[0004] In torque control, the remanence of permanent magnet materials exhibits a negative temperature coefficient. As the coil temperature increases, the effective flux linkage amplitude of the permanent magnet undergoes reversible or irreversible decay. Existing control algorithms typically treat the torque-to-current conversion coefficient as a constant, neglecting the effect of thermally induced magnetic decay. This results in the actual output torque being less than the commanded torque during high-temperature, high-load operation, reducing the linearity of joint output and the stability of end-effector operation.

[0005] Furthermore, the transmission friction characteristics of miniature precision reducers are closely related to temperature. The viscosity of the lubricating grease inside the joint changes non-linearly with temperature; at low temperatures, the increased viscosity leads to a significant increase in static friction, while at high temperatures, the decreased viscosity weakens damping. Existing friction compensation schemes mostly use chatter signals with fixed amplitude and frequency, which cannot be adaptively adjusted according to the current viscosity-temperature state. This fixed-parameter compensation method is difficult to cover all temperature range conditions, and is prone to creeping phenomena due to insufficient compensation during low-temperature startup, or high-frequency oscillations and electromagnetic noise in the system due to overcompensation in high-temperature and medium-to-high-speed operating regions. Summary of the Invention

[0006] To address the problems in existing robot micro-joint modules where limited size leads to heat dissipation difficulties and a lack of space for high-precision sensors, resulting in decreased control accuracy, lag response, and increased operating noise due to motor parameter drift and changes in friction characteristics when temperatures change, this invention provides a robot finger joint control method and system that solves the technical challenge of micro servo systems being unable to simultaneously achieve torque control linearity and low-speed smoothness under full temperature range conditions.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] The first aspect of this invention provides a method for controlling the finger joints of a robot, comprising the following steps:

[0009] First, multiphysics signal acquisition and high-frequency excitation injection are performed. The drive stage temperature, motor rotor mechanical angle, and three-phase current of the micro-joint module are acquired in real time, and a high-frequency voltage excitation signal is superimposed on the base voltage vector of the motor drive. The corresponding high-frequency current response is extracted from the feedback loop.

[0010] Secondly, a thermomagnetic dynamic state observer is constructed. The real-time resistance value of the motor winding is analyzed using the extracted high-frequency current response, and the real-time temperature of the motor coil is calculated by combining the collected drive stage temperature. Based on the temperature coefficient of the permanent magnet material and the real-time temperature of the motor coil, the effective flux linkage amplitude of the permanent magnet is reconstructed.

[0011] Next, an adaptive chatter signal based on thermoviscous coupling is generated. Based on the calculated real-time temperature of the motor coils, the current viscous friction coefficient is determined, and combined with the real-time rotational speed calculated from the motor rotor's mechanical angle, an adaptive chatter compensation signal with amplitude and frequency dynamically mapped to the thermal state is generated.

[0012] Finally, temperature-compensated variable impedance control is executed. The basic impedance torque command is calculated based on the desired joint trajectory and the mechanical angle of the motor rotor. The torque-current conversion coefficient is corrected using the reconstructed effective flux linkage amplitude, and the generated adaptive chatter compensation signal is superimposed on the current control loop to generate the final quadrature-axis target current command to drive the motor.

[0013] Furthermore, in the multi-physics signal acquisition and high-frequency excitation injection steps, the process of extracting the high-frequency current response includes: processing the direct-axis current feedback signal using a digital bandpass filter with a center frequency equal to the frequency of the high-frequency voltage excitation signal, filtering out the DC component and high-frequency switching noise, and outputting the direct-axis high-frequency current feedback value; using synchronous detection logic, performing calculations and low-pass filtering on the direct-axis high-frequency current feedback value with a unit high-frequency cosine carrier signal and a unit high-frequency sine signal respectively, to obtain the real component and the imaginary component of the high-frequency current response.

[0014] Furthermore, in the step of constructing the thermomagnetic dynamic state observer, the process of calculating the real-time temperature of the motor coil includes: calculating the resistance-derived temperature using the real-time resistance value; calculating the heat conduction temperature based on the drive board using the drive stage temperature; calculating the real-time rotational speed based on the motor rotor mechanical angle; configuring a dynamic weighting factor that is negatively correlated with the real-time rotational speed; and using the dynamic weighting factor to weight and combine the resistance-derived temperature and the heat conduction temperature, and using a dual-source temperature fusion observation formula to calculate the real-time temperature of the motor coil. This calculation process integrates the advantages of resistance temperature measurement based on electrical characteristics and temperature measurement based on a heat conduction model, balancing the weights of the two temperature measurement methods under different speed conditions through the dynamic weighting factor.

[0015] Furthermore, in the step of constructing the thermomagnetic dynamic state observer, the process of reconstructing the effective flux linkage amplitude of the permanent magnet includes: determining the nominal flux linkage value of the permanent magnet and the flux linkage calibration temperature measured under calibration conditions; calculating the flux linkage attenuation as the real-time temperature of the motor coil changes, based on the remanence temperature coefficient of the permanent magnet material; and calculating the effective flux linkage amplitude at the current temperature using the flux linkage thermal attenuation correction formula. This step is used to compensate for the attenuation of the permanent magnet's magnetic field strength caused by temperature rise, maintaining the accuracy of torque output.

[0016] Furthermore, in the step of generating the adaptive chatter signal for thermoviscosity coupling, the frequency generation process includes: setting a low-temperature reference frequency, a high-temperature upper limit frequency, and a viscosity transition temperature threshold; constructing a frequency adjustment logic that monotonically increases with the real-time temperature of the motor coil; calculating the chatter injection frequency using an S-shaped function related to the temperature change slope adjustment coefficient; and calculating the chatter injection frequency using a temperature-adaptive frequency mapping formula. The viscosity transition temperature threshold refers to the temperature point corresponding to the maximum rate of change of the lubricating oil's viscosity characteristics.

[0017] Furthermore, in the step of generating the thermoviscous coupling adaptive chatter signal, the amplitude generation process includes: constructing an amplitude generation mechanism simultaneously constrained by real-time rotational speed and real-time motor coil temperature; attenuating the chatter amplitude when the real-time rotational speed exceeds the Stribeck critical speed constant; and reducing the chatter amplitude using a thermoviscous attenuation coefficient when the real-time motor coil temperature rises; and calculating the final chatter injection current amplitude using a thermoviscous dual attenuation amplitude calculation formula. This mechanism ensures sufficient compensation amplitude under low-temperature, high-static-friction conditions, while automatically reducing the amplitude to suppress noise under high-temperature or high-speed conditions.

[0018] Furthermore, in the temperature-compensated variable impedance control step, the specific process of generating the final quadrature-axis target current command includes: dividing the basic impedance torque command by the product of the effective flux linkage amplitude, the number of motor pole pairs, and the power constant constraint coefficient to obtain the fundamental current component after flux linkage correction; decomposing the adaptive flutter compensation signal into the flutter injection current amplitude and a sinusoidal signal generated based on the flutter injection frequency; superimposing the product of the fundamental current component, the flutter injection current amplitude, and the sinusoidal signal, and calculating the quadrature-axis target current command using the temperature-compensated current command synthesis formula.

[0019] Furthermore, in the execution of the temperature-compensated variable impedance control step, the specific process of calculating the basic impedance torque command includes: combining the joint target position command, the joint target velocity command, and the actual feedback position and velocity to calculate the position tracking error and velocity tracking error; constructing a virtual impedance model using a proportional-derivative control algorithm based on preset joint stiffness coefficients and joint damping coefficients; using the virtual impedance model, taking the product of the position tracking error and the joint stiffness coefficient as the elastic torque component, and the product of the velocity tracking error and the joint damping coefficient as the damping torque component, and linearly superimposing the two to generate the basic impedance torque command. The virtual impedance model refers to the model constructed using a proportional-derivative control algorithm based on preset joint stiffness coefficients and joint damping coefficients, which simulates the dynamic characteristics of a spring-damped system by linearly superimposing the elastic torque component and the damping torque component.

[0020] Furthermore, in the step of generating the adaptive chatter signal of thermoviscous coupling, the process of determining the current viscous friction coefficient specifically includes: pre-constructing the viscosity-temperature characteristic curve data of the lubricating medium at different temperatures; performing a lookup operation on the viscosity-temperature characteristic curve data of the lubricating medium based on the real-time temperature of the motor coil to determine the current viscous friction coefficient benchmark; when it is determined to be in a low-temperature state, determining a higher friction compensation requirement based on the viscous friction coefficient benchmark to overcome static friction; when it is determined to be in a high-temperature state, reducing the friction compensation requirement based on the viscous friction coefficient benchmark to prevent overshoot oscillation.

[0021] A second aspect of the present invention provides a control system for a robot finger joint, comprising:

[0022] The multi-physics signal acquisition module is configured to acquire the drive stage temperature, phase current and position signals of the micro-joint module in real time, calculate the mechanical angle of the motor rotor from the position signal, superimpose the high-frequency voltage excitation signal into the basic voltage vector of the motor drive, and extract the corresponding high-frequency current response from the feedback loop.

[0023] The multi-source information fusion observation module is configured to analyze the real-time resistance value of the motor winding using high-frequency current response, calculate the real-time temperature of the motor coil by combining the drive stage temperature and using the dual-source temperature fusion observation formula, and reconstruct the effective flux amplitude of the permanent magnet based on the temperature coefficient of the permanent magnet material and the real-time temperature of the motor coil and using the flux thermal decay correction formula.

[0024] The adaptive chatter generation module is configured to determine the current viscous friction coefficient based on the real-time temperature of the motor coil, and combine it with the real-time speed calculated from the mechanical angle of the motor rotor, and use the temperature adaptive frequency mapping formula and the thermal speed double attenuation amplitude calculation formula to generate an adaptive chatter compensation signal whose amplitude and frequency are dynamically mapped with the thermal state.

[0025] The temperature-compensated variable impedance control module is configured to calculate the basic impedance torque command based on the desired joint trajectory and the mechanical angle of the motor rotor, correct the torque-current conversion coefficient using the effective flux amplitude, and superimpose the adaptive chatter compensation signal into the current control loop. The final cross-axis target current command is generated using the temperature-compensated current command synthesis formula to drive the motor.

[0026] This invention provides a method and system for controlling the finger joints of a robot. It has the following beneficial effects:

[0027] 1. This invention adopts a multi-physics field signal acquisition and a high-frequency response-based thermo-magnetic observation mechanism. By analyzing the high-frequency impedance to obtain the winding resistance and combining it with the heat transfer model of the drive board, it realizes accurate calculation of internal temperature under the condition of no built-in coil temperature sensor, and solves the problem of lack of thermal state monitoring caused by the inability to place contact sensors in micro joint modules due to volume limitations.

[0028] 2. This invention constructs a variable impedance control loop that includes correction for thermal attenuation of magnetic flux, reconstructs the effective magnetic flux amplitude of the permanent magnet in real time based on the observed coil temperature, and dynamically corrects the torque current conversion coefficient accordingly. This effectively compensates for the attenuation of magnetic field strength caused by motor temperature rise, ensuring the linearity of torque output and trajectory tracking accuracy of the joint module under long-term high-load operation.

[0029] 3. This invention establishes an adaptive chatter generation mechanism based on thermoviscosity coupling. The frequency and amplitude of the chatter signal are dynamically adjusted according to the viscosity-temperature characteristic curve of the lubricating medium and the real-time rotation speed. High amplitude compensation is provided to overcome static friction under low temperature and high viscosity conditions. The injected signal is automatically attenuated to suppress electromagnetic noise under high temperature or medium-high speed conditions, thus realizing low-speed stable start-up and operation under all temperature range conditions. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the hardware structure of a robot finger joint micro-module according to an embodiment of the present invention;

[0031] Figure 2 This is a connection block diagram of an electronic control system and its internal logic function modules according to an embodiment of the present invention;

[0032] Figure 3 This is a flowchart illustrating a robot finger joint control method according to an embodiment of the present invention;

[0033] Figure 4 This is a schematic diagram of the hardware architecture of an electronic device according to an embodiment of the present invention.

[0034] Among them, 100 is a micro joint module; 20 is a micro reduction mechanism; 30 is a drive control circuit board; 31 is a power inverter unit; 32 is a temperature sensing unit; 10 is a motor stator; 11 is a motor rotor; 40 is a micro control unit; 50 is a current sampling circuit; 60 is a position sensor; 201 is a multi-physics field signal acquisition module; 202 is a multi-source information fusion observation module; 203 is an adaptive flutter generation module; and 204 is a temperature-compensated variable impedance control module. Detailed Implementation

[0035] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] See attached document Figure 1 This invention provides a control system for a robot finger joint, which operates on a micro-joint module 100. The micro-joint module 100 includes a permanent magnet synchronous motor, a micro-reduction mechanism 20, and a drive control circuit board 30. The permanent magnet synchronous motor includes a stator 10 and a rotor 11. The micro-reduction mechanism 20 is mechanically connected to the output shaft of the rotor 11 to output joint motion torque. The drive control circuit board 30 adopts a compact layout and is physically bonded to the axial end face or side wall of the stator 10 via a thermally conductive medium, forming a direct heat conduction path between the drive control circuit board 30 and the stator 10. This structure makes the thermal equilibrium state of the drive control circuit board 30 strongly correlated with the temperature change of the stator 10. The drive control circuit board 30 integrates a power inverter unit 31 composed of power metal-oxide-semiconductor field-effect transistors, and a temperature sensing unit 32 disposed in the vicinity of the power inverter unit 31. The temperature sensing unit 32 is used to acquire the substrate temperature signal of the drive control circuit board 30 in real time.

[0037] See attached document Figure 2The control system includes a microcontroller unit 40, a current sampling circuit 50, and a position sensor 60. The current sampling circuit 50 is connected in series between the power inverter unit 31 and the windings of the motor stator 10 to collect the motor phase current. The position sensor 60 is coupled to the motor rotor 11 to collect the mechanical angular position of the motor. The microcontroller unit 40 is electrically connected to the temperature sensing unit 32, the current sampling circuit 50, the position sensor 60, and the power inverter unit 31. The microcontroller unit 40 internally contains logical functional modules constructed through program instructions, which, according to the signal flow direction, sequentially include: a multi-physics field signal acquisition module 201, a multi-source information fusion observation module 202, an adaptive flutter generation module 203, and a temperature-compensated variable impedance control module 204.

[0038] The multi-physics signal acquisition module 201 is configured to synchronously receive the substrate temperature output by the temperature sensing unit 32, the phase current output by the current sampling circuit 50, and the position signal output by the position sensor 60, and perform analog-to-digital conversion and filtering on the aforementioned analog signals. The multi-source information fusion observation module 202 is connected to the multi-physics signal acquisition module 201 and is configured to calculate the real-time resistance value of the motor winding, the internal coil temperature, and the real-time flux linkage amplitude of the permanent magnet based on the substrate temperature and current response signals, using a preset thermal resistance model and a high-frequency impedance analysis algorithm. The adaptive chatter generation module 203 is connected to the multi-source information fusion observation module 202 and is configured to map the viscous resistance characteristics of the lubricating medium inside the micro-reduction mechanism 20 according to the calculated coil temperature and real-time rotational speed, and generate a corresponding adaptive frequency and amplitude chatter compensation signal. The temperature-compensated variable impedance control module 204 is connected to the adaptive flutter generation module 203 and the multi-source information fusion observation module 202. It is configured to combine the real-time flux amplitude and flutter compensation signal to calculate the final voltage vector command and drive the power inverter unit 31 through the pulse width modulation signal.

[0039] See attached document Figure 3 This invention provides a method for controlling the finger joints of a robot, comprising the following steps:

[0040] S1 performs multi-physics signal acquisition and high-frequency excitation injection:

[0041] The system collects the drive stage temperature, motor rotor mechanical angle, and three-phase current of the micro-joint module in real time. At the same time, the controller superimposes a micro-amplitude high-frequency voltage signal with a frequency higher than the fundamental wave and lower than the switching frequency into the basic voltage vector of the motor drive, and extracts the corresponding high-frequency current response from the feedback loop.

[0042] S2, Construct a thermomagnetic dynamic state observer:

[0043] The real part of the high-frequency impedance of the motor winding is obtained by analyzing the high-frequency current response extracted from S1. The real-time resistance value is obtained by combining the temperature of the drive stage and the resistance temperature rise model. Then, based on the temperature coefficient of the permanent magnet material and the real-time temperature of the motor coil, the current real-time magnetic flux of the permanent magnet is reconstructed.

[0044] S3 generates an adaptive flutter signal with thermal-viscosity coupling:

[0045] Based on the real-time temperature of the motor coil obtained in S2, the preset temperature-varying Stribeck friction model is queried to determine the current viscous friction coefficient. Combined with the real-time speed calculated from the mechanical angle of the motor rotor, an adaptive chatter compensation signal with amplitude and frequency dynamically mapped with thermal state is generated.

[0046] S4, performs temperature-compensated variable impedance control:

[0047] The basic impedance torque is calculated based on the deviation between the desired joint trajectory and the mechanical angle of the motor rotor. The torque-current conversion coefficient is corrected by the real-time magnetic flux of the motor reconstructed by S2, and the adaptive chatter compensation signal generated by S3 is superimposed on the current control loop to generate the final driving torque command to drive the micro-joint module.

[0048] The technical implementation details of steps S1 to S4 above will be explained in detail below, combining specific mathematical models and control algorithms.

[0049] When performing step S1, the multiphysics signal acquisition module 201 works in conjunction with the microcontroller unit 40 to establish an electrical signal basis for sensing the internal thermal state of the motor through the following sub-steps.

[0050] Step S101: The multi-physics signal acquisition module 201 establishes a voltage vector injection mechanism based on magnetic field orientation control.

[0051] Based on the fundamental direct-axis voltage command and fundamental quadrature-axis voltage command calculated by the microcontroller unit 40 for driving the motor, the multiphysics signal acquisition module 201 superimposes a preset high-frequency voltage excitation signal onto the fundamental direct-axis voltage command. The amplitude of the high-frequency voltage excitation signal is set to a cosine wave form, and its corresponding high-frequency injection voltage amplitude is denoted as... The corresponding high-frequency injection angular frequency is denoted as . To prevent the injected signal from interfering with the robot's normal motion trajectory, a high-frequency angular frequency is injected. The value is set to be greater than the rated operating base frequency of the motor and less than the switching frequency of the power inverter unit 31, while avoiding the mechanical resonance frequency range of the micro-motion joint module. The final composite voltage vector applied to the motor windings contains a constant DC component and a high-frequency cosine component on the direct axis, while maintaining only the fundamental component required for torque control on the quadrature axis, thereby limiting the torque ripple generated by the injected signal to a preset threshold range.

[0052] In step S102, the multi-physics signal acquisition module 201 performs frequency domain separation of the mixed current signal.

[0053] The current sampling circuit 50 acquires the three-phase analog current of the motor stator in real time and converts it into direct-axis and quadrature-axis current feedback signals in a synchronous rotating coordinate system. Since the stator windings generate a response current under the action of a high-frequency voltage excitation signal, the direct-axis current feedback signal at this time is a mixed signal containing the fundamental current direct wave, high-frequency response, and switching noise. The multi-physics signal acquisition module 201 utilizes the fact that the center frequency equals the high-frequency injection angular frequency. The digital bandpass filter processes the direct-axis current feedback signal, filtering out the DC component and high-frequency switching noise, and outputting a direct-axis high-frequency current feedback value containing only stator impedance information. The passband width of the digital bandpass filter is configured to effectively extract the target signal at a set frequency and suppress noise interference in adjacent frequency bands.

[0054] In step S103, the multi-physics signal acquisition module 201 uses synchronous detection logic to extract the current component that characterizes the constant resistance feature.

[0055] The system will output the direct-axis high-frequency current feedback value in step S102 and the frequency will be equal to the high-frequency injection angular frequency. The high-frequency cosine carrier signal is multiplied, a process that shifts the effective current information of the same frequency and phase to the zero-frequency DC band. The result of the multiplication is then smoothed using a low-pass filter and multiplied by a demodulation factor of 2 to restore the signal amplitude, thus obtaining the real component of the high-frequency current response, denoted as . Simultaneously, the system will input the direct-axis high-frequency current feedback value at a frequency equal to the high-frequency injection angular frequency. The high-frequency sinusoidal signal is multiplied and low-pass filtered to obtain the imaginary component of the high-frequency current response, denoted as . .

[0056] The real component of the high-frequency current response obtained here It includes information on the real part of the resistance of the motor stator winding at the current temperature, and the imaginary part of the high-frequency current response. This includes inductance information. These two parameters will serve as the core input data for the thermomagnetic state observer in subsequent steps. The specific operational instructions for coordinate transformations, filtering algorithms, and multiplication demodulation involved in the above process can be implemented using conventional digital signal processing techniques by those skilled in the art; the implementation details will not be elaborated here.

[0057] When performing step S2, the multi-source information fusion observation module 202 constructs a data fusion algorithm based on the electrical feature signals extracted in step S1 and the physical sampling signals of the temperature sensing unit 32, and realizes the observation of the internal state of the motor through the following sub-steps.

[0058] Step S201: The multi-source information fusion observation module 202 performs online identification of the stator resistance based on the high-frequency current response. The multi-source information fusion observation module 202 receives the high-frequency injection voltage amplitude determined in step S1. Real component of high-frequency current response after demodulation and the imaginary part of the high-frequency current response Based on the equivalent RL series model of the stator winding of a synchronous motor at high frequencies, the multi-source information fusion observation module 202 uses the complex impedance analytical method to calculate the real part of the high-frequency impedance. Specifically, the system utilizes...

[0059] The real-time resistance observation value at the current moment is calculated by combining the ratio of the injected voltage amplitude to the square of the high-frequency response current magnitude with the real component of the high-frequency current response. The physical formula upon which its calculations are based is:

[0060] ;

[0061] in, Represents the imaginary unit (i.e.) In complex number operations, ), is used to identify the direction of the imaginary axis of the complex plane (usually corresponding to the inductive component). The operator represents the real part extraction operator, which means separating the resistive component from the complex impedance and discarding the inductive component.

[0062] This calculation process is executed within each control cycle, eliminating the influence of the inductive reactance component on the measurement results, thus ensuring the real-time resistance observation values ​​are accurate. It can accurately reflect the change in the real part of the stator impedance caused by the thermal effect of current.

[0063] Step S202: The multi-source information fusion observation module 202 constructs a coil temperature extrapolation model assisted by the driver stage temperature. Addressing the issues of single-resistance temperature measurement being susceptible to high-frequency noise interference and the lack of back-correction in the single thermal model method, the system establishes a temperature calculation mechanism based on weighted fusion. The multi-source information fusion observation module 202 calculates the resistance-based temperature extrapolation using the linear change in resistivity of copper conductors with temperature, and calculates the heat conduction temperature based on the driver board using a thermal circuit effect model, combining the two through a dynamic weight value. The multi-source information fusion observation module 202 uses a dual-source temperature fusion observation formula to calculate the real-time temperature of the motor's internal coils. The dual-source temperature fusion observation formula is:

[0064] ;

[0065] in, This indicates the calculated real-time temperature of the internal coil; This represents the real-time resistance observation value calculated in step S201; This indicates the nominal value of the stator resistance measured at a known reference temperature; The resistance calibration reference temperature when indicating the nominal value of the stator resistance; This represents the temperature coefficient of resistance of the stator winding conductor material; This represents a dynamic weighting factor, with a value set between 0 and 1, and configured to be negatively correlated with motor speed. It is used to increase the weight of resistance temperature measurement at low speeds and increase the weight of thermal model temperature measurement at high speeds. This indicates the temperature of the drive control circuit board substrate collected by the temperature sensing unit 32. This represents the pre-determined equivalent thermal resistance coefficient between the motor coil and the drive control circuit board; This represents the effective value of the motor phase current obtained through the current sampling circuit 50.

[0066] In step S203, the multi-source information fusion observation module 202 performs flux linkage reconstruction based on the thermal decay characteristics of permanent magnets. Considering the physical characteristic that the rotor magnetic field strength of the permanent magnet synchronous motor decreases with increasing temperature, the multi-source information fusion observation module 202 corrects the flux linkage parameters in the control model based on the temperature information obtained in step S202 to ensure the accuracy of torque control. The multi-source information fusion observation module 202 uses the flux linkage thermal decay correction formula to calculate the effective flux linkage amplitude at the current temperature. The flux linkage thermal decay correction formula is as follows:

[0067] ;

[0068] in, This represents the corrected effective flux linkage magnitude; This indicates the nominal flux linkage value of a permanent magnet measured under calibrated conditions. This represents the remanence temperature coefficient of permanent magnet materials; this coefficient is negative. This indicates the real-time temperature of the internal coil; This indicates the flux linkage calibration temperature at which the nominal flux linkage value of the permanent magnet is measured. Through this step, the system transforms the flux linkage parameter, which was originally a constant, into a variable that dynamically adjusts with temperature, enabling the subsequently generated current command to automatically compensate for torque loss caused by thermal decay of the magnetic field.

[0069] During step S3, the adaptive chatter generation module 203 dynamically generates a chatter signal to compensate for nonlinear friction based on the internal thermal state observed in step S2 and the real-time motion state of the motor. This process aims to address the problem that chatter signals with fixed parameters cannot simultaneously achieve both static friction overcoming capability at low temperatures and noise suppression capability at high temperatures.

[0070] In step S301, the adaptive chatter generation module 203 establishes a temperature-velocity correlated tribodynamic analysis benchmark. The system pre-stores the viscosity-temperature characteristic curve data of the lubricating oil. Based on the real-time temperature of the internal coil, the adaptive chatter generation module 203 queries and determines the current viscous friction coefficient benchmark and Coulomb friction torque benchmark. When the motor is at a low temperature, the increased viscosity of the lubricating oil leads to an increase in starting resistance torque, and the system determines that a high-amplitude compensation signal is needed to overcome static friction. When the temperature rises, the shear force of the lubricating oil decreases, and the system determines that the compensation intensity should be reduced to prevent overshoot oscillations. This physical state determination does not directly output control commands but serves as the physical boundary condition for subsequent calculations of chatter frequency and amplitude.

[0071] In step S302, the adaptive flutter generation module 203 performs temperature-adaptive mapping of the flutter frequency. To ensure smooth startup and suppress audible noise of the micro-joint module under different temperature conditions, the system is designed with a frequency adjustment logic that monotonically increases with temperature. At low temperatures, the system outputs a lower fundamental frequency, utilizing the larger displacement characteristic of lower frequency signals to overcome the static friction torque range; as the temperature rises, the system gradually increases the flutter frequency, utilizing the low-pass filtering characteristics of the mechanical structure to attenuate noise generated by high-frequency vibrations. The adaptive flutter generation module 203 calculates the final flutter injection frequency using a temperature-adaptive frequency mapping formula, which is:

[0072]

[0073] in, This represents the calculated flutter injection frequency; This indicates the preset low-temperature reference frequency, which is set in a low-frequency range that can effectively overcome static friction; This indicates the preset upper limit frequency for high temperature, which is set in the high-frequency range far from the mechanical resonance point of the system; This indicates the real-time temperature of the internal coil output from step S202; This indicates the preset viscosity transition temperature threshold, which corresponds to the temperature point where the viscosity characteristics of the lubricating oil change the most. This represents the temperature slope adjustment coefficient, used to adjust the sensitivity of the frequency to temperature changes; The base of the natural logarithm is given. This formula utilizes the nonlinear characteristics of the sigmoid function to avoid abrupt changes during frequency switching.

[0074] In step S303, the adaptive chatter generation module 203 generates a chatter amplitude with thermally-rated dual-variable attenuation characteristics. To eliminate critical torque during steady-state operation while ensuring static friction compensation, the system constructs an amplitude generation mechanism simultaneously constrained by the motor's mechanical angular velocity and coil temperature. Logically, when the motor speed exceeds the critical speed of the Stribeck curve, dynamic friction becomes dominant, requiring attenuation of the chatter amplitude. Simultaneously, when the temperature rises, causing a natural decrease in static friction, the chatter amplitude also needs to be reduced to avoid electromagnetic noise caused by overcompensation. The adaptive chatter generation module 203 calculates the final chatter injection current amplitude using the thermally-rated dual-attenuation amplitude calculation formula, which is:

[0075] ;

[0076] in, This represents the calculated amplitude of the flutter injection current; This represents the maximum static friction compensation amplitude measured at the reference temperature and zero speed. This represents the absolute value of the motor's mechanical angular velocity calculated using a position sensor. This represents the preset Stribeck critical velocity constant, which characterizes the velocity inflection point where frictional force transitions from static friction to kinetic friction. This represents the thermoviscosity decay coefficient, which is used to linearize the proportion of decrease in static friction force demand due to temperature increase. This indicates the real-time temperature of the internal coil; This indicates the reference temperature for resistor calibration.

[0077] The system will eventually calculate the flutter injection frequency. With the amplitude of the flutter injection current The commands are combined to generate a high-frequency sinusoidal chatter command, which will be superimposed in the subsequent current loop control. Through this mechanism, the robot joint can obtain sufficient vibration compensation force when starting at low temperatures, while the command amplitude automatically decays when running at high temperatures or high speeds, thereby achieving low-noise and smooth control across the entire temperature range.

[0078] During step S4, the temperature-compensated variable impedance control module 204, as the output stage of the control architecture, is responsible for converting the torque demand generated by the motion control layer into a low-level current drive command adapted to the current thermal and frictional states of the motor. This module corrects and compensates for the motor output torque through the following steps.

[0079] Step S401: The temperature-compensated variable impedance control module 204 calculates the basic impedance torque based on the joint position error. The system receives the target joint position command and the target joint velocity command, and calculates the position tracking error and velocity tracking error by combining the actual position and actual velocity fed back by the encoder. The temperature-compensated variable impedance control module 204 constructs a virtual impedance model based on preset joint stiffness coefficient and joint damping coefficient using a proportional-derivative (PD) control algorithm. Specifically, the temperature-compensated variable impedance control module 204 linearly superimposes the elastic torque component obtained by multiplying the position tracking error by the joint stiffness coefficient with the damping torque component obtained by multiplying the velocity tracking error by the joint damping coefficient, thereby generating the basic impedance torque command, denoted as... This instruction characterizes the theoretical torque required for the joint module to maintain motion trajectory tracking under ideal linear conditions.

[0080] In step S402, the temperature-compensated variable impedance control module 204 performs current command synthesis that integrates flux linkage correction and chatter superposition. To accurately convert theoretical torque into physical current, the temperature-compensated variable impedance control module 204 introduces a flux linkage correction mechanism based on thermal observation and a chatter injection mechanism based on friction conditions. Since the system has confirmed in step S203 that the permanent magnet flux linkage decays with increasing temperature, if a constant torque coefficient is used for control, the actual output torque of the motor will be lower than the commanded value. Therefore, the system dynamically adjusts the torque-to-current conversion ratio by introducing a real-time corrected effective flux linkage amplitude. Simultaneously, to compensate for nonlinear friction without affecting the accuracy of motion trajectory tracking, the system superimposes the adaptive chatter signal generated in step S303 onto the torque-to-current axis.

[0081] The temperature-compensated variable impedance control module 204 uses a temperature-compensated current command synthesis formula to calculate the final quadrature-axis target current command input to the current loop. The temperature-compensated current command synthesis formula is as follows:

[0082] ;

[0083] in, This represents the calculated quadrature-axis target current command, which serves as the control target value for the current loop in the vector control system. This indicates the basic impedance torque command generated in step S401; Indicates the number of pole pairs of the motor; This represents the effective flux linkage amplitude output in step S203. This parameter is located in the denominator term, which enables the system to automatically increase the fundamental current component to compensate for torque loss when the flux linkage decreases due to temperature rise. This represents the amplitude of the flutter injection current output in step S303; This represents the flutter injection frequency output in step S302; Represents the system's real-time time; numerical value This refers to the constant amplitude coordinate transformation coefficient introduced during the transformation from a three-phase stationary coordinate system to a two-phase rotating coordinate system.

[0084] Through the above calculations, the system achieves adaptive fusion of the control closed loop: when the motor temperature rises, causing a decrease in torque output capability, the first term of the formula compensates for thermal degradation by increasing the current amplitude; when the motor is in a low-temperature, high-viscosity state, the second term overcomes static friction by superimposing a high-amplitude chatter signal; when the motor is running at high speed or high temperature, the second term automatically decays to reduce electromagnetic noise. The final synthesized quadrature-axis target current command... Combined with a preset direct-axis current command, the input is fed into the space vector modulation algorithm to drive the inverter, thereby achieving high-precision torque control across the entire temperature range.

[0085] See attached document Figure 4 This invention provides a non-volatile computer-readable storage medium, specifically implemented in forms including but not limited to flash memory, electrically erasable programmable read-only memory (EEPROM), or ferroelectric memory (FRAM). The computer-readable storage medium stores computer programs or instructions configured as machine-executable code. When the computer program or instructions are read and executed by a processor in an electronic device, the electronic device as a whole executes the control method of steps S1 to S4 and all their sub-steps in the aforementioned embodiments. Specifically, the storage medium stores program code corresponding to the multi-source information fusion observation module 202, the adaptive flutter generation module 203, and the temperature-compensated variable impedance control module 204.

[0086] The electronic device includes at least one processor, which serves as the control core and interacts with a computer-readable storage medium via a system bus. To meet the high-frequency control requirements of the micro servo system, the processor is specifically selected from digital signal processors (DSPs), microcontrollers (MCUs), or field-programmable gate arrays (FPGAs). At the hardware interface level, the processor integrates a multi-channel high-precision analog-to-digital converter (ADC) interface. The ADC interface is electrically connected to the current sampling circuit and temperature sensing unit, used to acquire analog voltage signals in real time and convert them into digital quantities for the processor to read. The processor's internal arithmetic logic unit (ALU) is configured to execute the aforementioned dual-source temperature fusion observation formula, flux linkage thermal attenuation correction formula, temperature adaptive frequency mapping formula, and thermal rate dual attenuation amplitude calculation formula. Furthermore, the processor is connected to a pulse width modulation (PWM) signal generation unit, used to generate switching signals to drive the inverter based on the current command output by the temperature-compensated variable impedance control module 204.

[0087] The method and electronic device of the present invention are specifically configured as a highly integrated, one-piece joint module, which is suitable for the end effector of humanoid dexterous hands, exoskeleton robots, or surgical robots. Structurally, the joint module encapsulates a permanent magnet synchronous motor, a precision reducer, a drive control circuit board, a position sensor, and a temperature sensor within a single housing.

[0088] In this application scenario, due to the physical size of the joint module limiting the heat dissipation surface area and the low heat capacity of the internal lubricating and magnetic materials, the internal temperature of the module will change significantly when the motor is under high load conditions of frequent start-stop or long-term stall, which in turn leads to the drift of the grease viscosity characteristics and the attenuation of the permanent magnet flux amplitude.

[0089] The joint module using this embodiment of the invention integrates a processor that executes the above-mentioned control algorithm on the drive control circuit board, enabling real-time monitoring and compensation of the operating environment and the internal state of the motor. When the joint module starts up in a low-temperature environment, the processor calls the adaptive chatter generation module 203 to generate a high-frequency chatter torque to overcome the high viscous static friction of the mechanical transmission chain. When the joint module's temperature rises due to continuous operation, the processor uses the flux thermal attenuation determined by the multi-source information fusion observation module 202 to correct the current command, and simultaneously controls the adaptive chatter generation module 203 to attenuate the chatter amplitude to maintain the linearity of the joint output torque and suppress electromagnetic noise.

[0090] For the reduction transmission mechanism involved in the joint module, those skilled in the art can select a harmonic reducer or a miniature planetary gearbox according to the actual load requirements. The mechanical transmission principle is well-known in the field and will not be elaborated further here. Through the above-mentioned hardware and software collaborative configuration, this electronic device effectively solves the problem of decreased control performance of the miniature servo system due to time-varying parameters under full temperature range conditions.

Claims

1. A method for controlling the finger joints of a robot, characterized in that, Includes the following steps: S1. Perform multi-physics field signal acquisition and high-frequency excitation injection: The system acquires the drive stage temperature of the micro-joint module, the mechanical angle of the motor rotor and the three-phase current in real time, and superimposes the high-frequency voltage excitation signal into the base voltage vector of the motor drive, and extracts the corresponding high-frequency current response from the feedback loop. S2. Construct a thermomagnetic dynamic state observer: Analyze the real-time resistance value of the motor winding using the high-frequency current response extracted in step S1, calculate the real-time temperature of the motor coil using the drive stage temperature collected in step S1, and reconstruct the effective flux linkage amplitude of the permanent magnet based on the temperature coefficient of the permanent magnet material and the real-time temperature of the motor coil. S3. Generate an adaptive chatter signal with thermoviscous coupling: Based on the real-time temperature of the motor coil calculated in step S2, determine the current viscous friction coefficient, and combine it with the real-time speed calculated from the mechanical angle of the motor rotor collected in step S1 to generate an adaptive chatter compensation signal whose amplitude and frequency are dynamically mapped with thermal state. S4. Perform temperature-compensated variable impedance control: Calculate the basic impedance torque command based on the joint desired trajectory and the mechanical angle of the motor rotor collected in step S1, correct the torque current conversion coefficient using the effective flux amplitude reconstructed in step S2, and superimpose the adaptive chatter compensation signal generated in step S3 into the current control loop to generate the final cross-axis target current command to drive the motor. In step S1, the specific process of extracting the corresponding high-frequency current response includes: The direct-axis current feedback signal is processed using a digital bandpass filter whose center frequency is equal to the frequency of the high-frequency voltage excitation signal. The DC component and high-frequency switching noise are filtered out, and the direct-axis high-frequency current feedback value is output. Using synchronous detection logic, the direct-axis high-frequency current feedback value is calculated and low-pass filtered with a unit high-frequency cosine carrier signal and a unit high-frequency sine signal to obtain the real and imaginary components of the high-frequency current response. In step S2, the specific process of calculating the real-time temperature of the motor coil includes: The resistance-induced temperature is calculated using the real-time resistance value, and the heat conduction temperature based on the drive board is calculated using the drive stage temperature. The real-time rotational speed is calculated based on the mechanical angle of the motor rotor collected in step S1, and a dynamic weighting factor that is negatively correlated with the real-time rotational speed is configured. The system uses the dynamic weighting factor to weight and combine the resistance-induced temperature and the heat conduction temperature, and then uses a dual-source temperature fusion observation formula to calculate the real-time temperature of the motor coil. The formula for dual-source temperature fusion observation is: ； in, This indicates the calculated real-time temperature of the internal coil; This represents the real-time resistance observation value calculated in step S201; This indicates the nominal value of the stator resistance measured at a known reference temperature; The resistance calibration reference temperature when indicating the nominal value of the stator resistance; This represents the temperature coefficient of resistance of the stator winding conductor material; This represents a dynamic weighting factor, with a value set between 0 and 1, and configured to be negatively correlated with motor speed. It is used to increase the weight of resistance temperature measurement at low speeds and increase the weight of thermal model temperature measurement at high speeds. This indicates the temperature of the drive control circuit board substrate collected by the temperature sensing unit 32; This represents the pre-determined equivalent thermal resistance coefficient between the motor coil and the drive control circuit board; This represents the effective value of the motor phase current obtained through the current sampling circuit 50; In step S2, the specific process of reconstructing the effective flux linkage amplitude of the permanent magnet includes: Determine the nominal flux linkage value of the permanent magnet and the flux linkage calibration temperature measured under calibration conditions; Based on the remanence temperature coefficient of the permanent magnet material, the flux linkage attenuation as the real-time temperature of the motor coil changes is calculated; The system uses the flux thermal decay correction formula to calculate the effective flux amplitude at the current temperature; The formula for correcting thermal decay of magnetic flux is: ； in, This represents the corrected effective flux linkage magnitude; This indicates the nominal flux linkage value of a permanent magnet measured under calibrated conditions. This represents the remanence temperature coefficient of permanent magnet materials; this coefficient is negative. This indicates the real-time temperature of the internal coil; This indicates the flux linkage calibration temperature used when measuring the nominal flux linkage value of a permanent magnet. In step S3, the process of generating an adaptive flutter compensation signal whose amplitude and frequency are dynamically mapped to thermal state includes: Set the low-temperature reference frequency, the high-temperature upper limit frequency, and the viscosity transition temperature threshold; A frequency regulation logic for the real-time monotonically increasing temperature of the motor coil is constructed, and the chatter injection frequency is calculated using an S-shaped function related to the temperature change slope regulation coefficient. The system uses a temperature-adaptive frequency mapping formula to calculate the flutter injection frequency; The viscosity transition temperature threshold refers to the temperature point corresponding to the largest change rate of the viscosity characteristics of the lubricating oil. The temperature-adaptive frequency mapping formula is: ； in, This represents the calculated flutter injection frequency; This indicates the preset low-temperature reference frequency, which is set in a low-frequency range that can effectively overcome static friction; This indicates the preset upper limit frequency for high temperature, which is set in the high-frequency range far from the mechanical resonance point of the system; This indicates the real-time temperature of the internal coil output from step S202; This indicates the preset viscosity transition temperature threshold, which corresponds to the temperature point where the viscosity characteristics of the lubricating oil change the most. This represents the temperature slope adjustment coefficient, used to adjust the sensitivity of the frequency to temperature changes; is the base of the natural logarithm; In step S3, the generation of an adaptive flutter compensation signal whose amplitude and frequency are dynamically mapped with thermal state includes the following process: Construct an amplitude generation mechanism that is simultaneously constrained by the real-time rotational speed and the real-time temperature of the motor coil; When the real-time rotational speed increases beyond the Stribeck critical speed constant, the chatter amplitude is attenuated; when the real-time temperature of the motor coil increases, the chatter amplitude is reduced by using the thermoviscous attenuation coefficient. The system uses the thermal rate double decay amplitude calculation formula to calculate the final flutter injection current amplitude; The formula for calculating the amplitude of thermal rate double decay is: ； in, This represents the calculated amplitude of the flutter injection current; This represents the maximum static friction compensation amplitude measured at the reference temperature and zero speed. This represents the absolute value of the motor's mechanical angular velocity calculated using a position sensor. This represents the preset Stribeck critical velocity constant, which characterizes the velocity inflection point where frictional force transitions from static friction to kinetic friction. This represents the thermoviscosity decay coefficient, which is used to linearize the proportion of decrease in static friction force demand due to temperature increase. This indicates the real-time temperature of the internal coil; Indicates the resistance calibration reference temperature; In step S4, the specific process of generating the final cross-axis target current command includes: Divide the basic impedance torque command by the product of the effective flux linkage amplitude, the number of motor pole pairs, and the power constant constraint coefficient to obtain the fundamental current component after flux linkage correction. The adaptive flutter compensation signal generated in step S3 is decomposed into the flutter injection current amplitude and a sinusoidal signal generated based on the flutter injection frequency. The fundamental current component is superimposed with the product of the flutter injection current amplitude and the sinusoidal signal, and the system uses the temperature-compensated current command synthesis formula to calculate the cross-axis target current command. The formula for synthesizing temperature compensation current commands is: ； in, This represents the calculated quadrature-axis target current command, which serves as the control target value for the current loop in the vector control system. This indicates the basic impedance torque command generated in step S401; Indicates the number of pole pairs of the motor; This represents the effective flux linkage amplitude output in step S203. This parameter is located in the denominator term, which enables the system to automatically increase the fundamental current component to compensate for torque loss when the flux linkage decreases due to temperature rise. This represents the amplitude of the flutter injection current output in step S303; This represents the flutter injection frequency output in step S302; Represents the system's real-time time; numerical value The constant amplitude coordinate transformation coefficient is introduced during the transformation from a three-phase stationary coordinate system to a two-phase rotating coordinate system; In step S4, the specific process of calculating the foundation impedance torque command includes: By combining the joint target position command, the joint target velocity command, and the actual feedback position and velocity, the position tracking error and velocity tracking error are calculated. Based on the preset joint stiffness coefficient and joint damping coefficient, a virtual impedance model is constructed using a proportional-derivative control algorithm. Using the virtual impedance model, the product of the position tracking error and the joint stiffness coefficient is taken as the elastic torque component, and the product of the velocity tracking error and the joint damping coefficient is taken as the damping torque component. The two are linearly superimposed to generate the basic impedance torque command. The virtual impedance model refers to a model constructed using a proportional-derivative control algorithm based on preset joint stiffness coefficients and joint damping coefficients. It generates a basic impedance torque command by linearly superimposing the elastic torque component and the damping torque component.

2. The method for controlling a robot finger joint according to claim 1, characterized in that, In step S3, the process of determining the current viscous friction coefficient specifically includes: Pre-construct viscosity-temperature characteristic curves of lubricating media at different temperatures; Based on the real-time temperature of the motor coil calculated in step S2, the viscosity-temperature characteristic curve data of the lubricating medium is looked up and calculated to determine the current viscosity-friction coefficient benchmark. When it is determined that the temperature is low, a higher friction compensation requirement is determined based on the viscous friction coefficient benchmark to overcome static friction. When the temperature is determined to be high, the friction compensation requirement is reduced based on the viscous friction coefficient benchmark to prevent overshoot oscillation.

3. A control system for a robot finger joint, characterized in that, A control method for a robot finger joint as described in claim 1 or 2, comprising: The multi-physics signal acquisition module is configured to acquire the drive stage temperature, phase current and position signal of the micro-joint module in real time, calculate the mechanical angle of the motor rotor from the position signal, superimpose a high-frequency voltage excitation signal into the basic voltage vector of the motor drive, and extract the corresponding high-frequency current response from the feedback loop. The multi-source information fusion observation module is configured to analyze the real-time resistance value of the motor winding using the high-frequency current response, calculate the real-time temperature of the motor coil by combining the temperature of the drive stage and using the dual-source temperature fusion observation formula, and reconstruct the effective magnetic flux amplitude of the permanent magnet based on the temperature coefficient of the permanent magnet material and the real-time temperature of the motor coil and using the magnetic flux thermal attenuation correction formula. The adaptive chatter generation module is configured to determine the current viscous friction coefficient based on the real-time temperature of the motor coil, and combine it with the real-time rotational speed calculated from the mechanical angle of the motor rotor, and generate an adaptive chatter compensation signal whose amplitude and frequency are dynamically mapped with the thermal state using the temperature adaptive frequency mapping formula and the thermal rate double attenuation amplitude calculation formula. The temperature-compensated variable impedance control module is configured to calculate the basic impedance torque command based on the desired joint trajectory and the mechanical angle of the motor rotor, correct the torque-current conversion coefficient using the effective flux amplitude, and superimpose the adaptive chatter compensation signal into the current control loop. The final cross-axis target current command is generated using the temperature-compensated current command synthesis formula to drive the motor. In step S1, the specific process of extracting the corresponding high-frequency current response includes: The direct-axis current feedback signal is processed using a digital bandpass filter whose center frequency is equal to the frequency of the high-frequency voltage excitation signal. The DC component and high-frequency switching noise are filtered out, and the direct-axis high-frequency current feedback value is output. Using synchronous detection logic, the direct-axis high-frequency current feedback value is calculated and low-pass filtered with a unit high-frequency cosine carrier signal and a unit high-frequency sine signal to obtain the real and imaginary components of the high-frequency current response. In step S2, the specific process of calculating the real-time temperature of the motor coil includes: The resistance-induced temperature is calculated using the real-time resistance value, and the heat conduction temperature based on the drive board is calculated using the drive stage temperature. The real-time rotational speed is calculated based on the mechanical angle of the motor rotor collected in step S1, and a dynamic weighting factor that is negatively correlated with the real-time rotational speed is configured. The system uses the dynamic weighting factor to weight and combine the resistance-induced temperature and the heat conduction temperature, and then uses a dual-source temperature fusion observation formula to calculate the real-time temperature of the motor coil. The formula for dual-source temperature fusion observation is: ； in, This indicates the calculated real-time temperature of the internal coil; This represents the real-time resistance observation value calculated in step S201; This indicates the nominal value of the stator resistance measured at a known reference temperature; The resistance calibration reference temperature when indicating the nominal value of the stator resistance; This represents the temperature coefficient of resistance of the stator winding conductor material; This represents a dynamic weighting factor, with a value set between 0 and 1, and configured to be negatively correlated with motor speed. It is used to increase the weight of resistance temperature measurement at low speeds and increase the weight of thermal model temperature measurement at high speeds. This indicates the temperature of the drive control circuit board substrate collected by the temperature sensing unit 32; This represents the pre-determined equivalent thermal resistance coefficient between the motor coil and the drive control circuit board; This represents the effective value of the motor phase current obtained through the current sampling circuit 50; In step S2, the specific process of reconstructing the effective flux linkage amplitude of the permanent magnet includes: Determine the nominal flux linkage value of the permanent magnet and the flux linkage calibration temperature measured under calibration conditions; Based on the remanence temperature coefficient of the permanent magnet material, the flux linkage attenuation as the real-time temperature of the motor coil changes is calculated; The system uses the flux thermal decay correction formula to calculate the effective flux amplitude at the current temperature; The formula for correcting thermal decay of magnetic flux is: ； in, This represents the corrected effective flux linkage magnitude; This indicates the nominal flux linkage value of a permanent magnet measured under calibrated conditions. This represents the remanence temperature coefficient of permanent magnet materials; this coefficient is negative. This indicates the real-time temperature of the internal coil; This indicates the flux linkage calibration temperature used when measuring the nominal flux linkage value of a permanent magnet. In step S3, the process of generating an adaptive flutter compensation signal whose amplitude and frequency are dynamically mapped to thermal state includes: Set the low-temperature reference frequency, the high-temperature upper limit frequency, and the viscosity transition temperature threshold; A frequency regulation logic for the real-time monotonically increasing temperature of the motor coil is constructed, and the chatter injection frequency is calculated using an S-shaped function related to the temperature change slope regulation coefficient. The system uses a temperature-adaptive frequency mapping formula to calculate the flutter injection frequency; The viscosity transition temperature threshold refers to the temperature point corresponding to the largest change rate of the viscosity characteristics of the lubricating oil. The temperature-adaptive frequency mapping formula is: ； in, This represents the calculated flutter injection frequency; This indicates the preset low-temperature reference frequency, which is set in a low-frequency range that can effectively overcome static friction; This indicates the preset upper limit frequency for high temperature, which is set in the high-frequency range far from the mechanical resonance point of the system; This indicates the real-time temperature of the internal coil output from step S202; This indicates the preset viscosity transition temperature threshold, which corresponds to the temperature point where the viscosity characteristics of the lubricating oil change the most. This represents the temperature slope adjustment coefficient, used to adjust the sensitivity of the frequency to temperature changes; is the base of the natural logarithm; In step S3, the generation of an adaptive flutter compensation signal whose amplitude and frequency are dynamically mapped with thermal state includes the following process: Construct an amplitude generation mechanism that is simultaneously constrained by the real-time rotational speed and the real-time temperature of the motor coil; When the real-time rotational speed increases beyond the Stribeck critical speed constant, the chatter amplitude is attenuated; when the real-time temperature of the motor coil increases, the chatter amplitude is reduced by using the thermoviscous attenuation coefficient. The system uses the thermal rate double decay amplitude calculation formula to calculate the final flutter injection current amplitude; The formula for calculating the amplitude of thermal rate double decay is: ； in, This represents the calculated amplitude of the flutter injection current; This represents the maximum static friction compensation amplitude measured at the reference temperature and zero speed. This represents the absolute value of the motor's mechanical angular velocity calculated using a position sensor. This represents the preset Stribeck critical velocity constant, which characterizes the velocity inflection point where frictional force transitions from static friction to kinetic friction. This represents the thermoviscosity decay coefficient, which is used to linearize the proportion of decrease in static friction force demand due to temperature increase. This indicates the real-time temperature of the internal coil; Indicates the resistance calibration reference temperature; In step S4, the specific process of generating the final cross-axis target current command includes: Divide the basic impedance torque command by the product of the effective flux linkage amplitude, the number of motor pole pairs, and the power constant constraint coefficient to obtain the fundamental current component after flux linkage correction. The adaptive flutter compensation signal generated in step S3 is decomposed into the flutter injection current amplitude and a sinusoidal signal generated based on the flutter injection frequency. The fundamental current component is superimposed with the product of the flutter injection current amplitude and the sinusoidal signal, and the system uses the temperature-compensated current command synthesis formula to calculate the cross-axis target current command. The formula for synthesizing temperature compensation current commands is: ； in, This represents the calculated quadrature-axis target current command, which serves as the control target value for the current loop in the vector control system. This indicates the basic impedance torque command generated in step S401; Indicates the number of pole pairs of the motor; This represents the effective flux linkage amplitude output in step S203. This parameter is located in the denominator term, which enables the system to automatically increase the fundamental current component to compensate for torque loss when the flux linkage decreases due to temperature rise. This represents the amplitude of the flutter injection current output in step S303; This represents the flutter injection frequency output in step S302; Represents the system's real-time time; numerical value The constant amplitude coordinate transformation coefficient is introduced during the transformation from a three-phase stationary coordinate system to a two-phase rotating coordinate system; In step S4, the specific process of calculating the foundation impedance torque command includes: By combining the joint target position command, the joint target velocity command, and the actual feedback position and velocity, the position tracking error and velocity tracking error are calculated. Based on the preset joint stiffness coefficient and joint damping coefficient, a virtual impedance model is constructed using a proportional-derivative control algorithm. Using the virtual impedance model, the product of the position tracking error and the joint stiffness coefficient is taken as the elastic torque component, and the product of the velocity tracking error and the joint damping coefficient is taken as the damping torque component. The two are linearly superimposed to generate the basic impedance torque command. The virtual impedance model refers to a model constructed using a proportional-derivative control algorithm based on preset joint stiffness coefficients and joint damping coefficients. It generates a basic impedance torque command by linearly superimposing the elastic torque component and the damping torque component.

Images