Joint motor asymmetric stiffness adjustment method and system based on virtual reference model

Abstract

The application discloses a kind of joint motor asymmetric stiffness adjustment method and system based on virtual reference model, and the method contains four major core steps of constructing virtual reference model, real-time state deviation calculation, disturbance observation and asymmetric stiffness adjustment.By running virtual model without friction term in microprocessor, actual running state is compared with theoretical state, and equivalent external disturbance torque is solved by combining friction compensation;Based on the comparison result of disturbance value and dynamic safety threshold, the proportional gain and integral gain of position loop controller are adjusted in real time using asymmetric strategy.The application does not need additional torque sensor, and is low in cost, fast in response, can effectively distinguish friction and external impact force, instant flexible retreat when collision is detected, and smooth recovery after collision disappears, and safety of human-computer interaction and overall rigidity of mechanical structure are considered.

Description

Technical Field

[0001] This invention relates to motor drive control and automation control technology, specifically to a control method that can realize flexible interaction and anti-collision protection of articulated motors by relying on microprocessor algorithms, and more specifically to a method and system for adjusting the asymmetric stiffness of articulated motors based on a virtual reference model. Background Technology

[0002] Human-robot interaction safety is a core indicator for collaborative robots, but existing flexible control solutions struggle to simultaneously achieve low cost, high rigidity, and high precision. Traditional hardware solutions require integrating expensive torque sensors at the joints, significantly increasing cost and size while reducing the overall rigidity of the robotic arm. Existing sensorless force control solutions based on current loops are limited by the complex nonlinear friction of harmonic reducers and the temperature-dependent viscosity of lubricating grease, making it difficult to accurately distinguish between inherent frictional forces and external collision forces. This results in insufficient sensitivity to weak collisions at low speeds and a high risk of false alarms due to errors in dynamic modeling at high speeds.

[0003] Furthermore, existing variable stiffness control strategies pose serious safety hazards during the stiffness recovery phase after a collision. Most solutions employ linear or symmetrical parametric recovery logic, causing the motor to output a huge restoring torque when attempting to eliminate accumulated position errors, resulting in severe secondary oscillations in the robotic arm. This secondary impact is often more dangerous than the initial collision and can easily cause secondary injuries to personnel.

[0004] In summary, existing robot joint control technologies suffer from several technical challenges, including the high cost of hardware torque sensors, the significant susceptibility of traditional current loop force control to nonlinear friction interference, and the tendency for conventional variable stiffness control to generate secondary oscillations during the recovery phase. Therefore, there is an urgent need to develop an intelligent anti-collision control technology that requires no external sensors, can accurately eliminate nonlinear friction interference, and possesses smooth asymmetric recovery characteristics. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a flexible anti-collision control method, system, and storage medium for articulated motors based on a virtual reference model and asymmetric stiffness adjustment. This method enables highly sensitive collision detection and active safety protection throughout the entire process without the need for expensive sensors.

[0006] According to a first aspect of the present invention, a method for adjusting the asymmetric stiffness of a joint motor based on a virtual reference model is provided. The method operates in a motor servo control device and includes the following steps:

[0007] S1, Constructing a virtual reference model: A virtual reference model is constructed and run in real time in the microprocessor. The virtual reference model is defined as a linear second-order system containing only equivalent rotational inertia and viscous damping coefficient. The torque command issued by the upper control level is used as input. Under the premise of eliminating the interference of the nonlinear friction term and cogging torque term inherent in the actual physical motor, the virtual reference model is iterated in real time and outputs the virtual theoretical state of the motor in the ideal operation without interference. The virtual theoretical state includes at least the virtual theoretical position and the virtual theoretical velocity.

[0008] S2: Status Acquisition and Deviation Observation: Real-time acquisition of the actual motor's operating signals, which include at least the actual phase current and the actual rotor position, and calculation of the motor's actual operating state based on these signals; construction of a status deviation observer to calculate the dynamic deviation between the actual operating state and the virtual theoretical state in real time within the same control cycle;

[0009] S3: Multi-level signal processing and collision force extraction: Perform multi-level signal processing on the dynamic deviation value to calculate the equivalent external disturbance torque;

[0010] S4: Dynamic safety threshold determination: Obtain a dynamic safety threshold that is non-linearly related to the actual speed of the motor, and compare the equivalent external disturbance torque with the dynamic safety threshold in real time;

[0011] S5: Asymmetric stiffness adjustment: Based on the comparison results, an asymmetric strategy is used to adjust the gain parameters of the position loop controller in real time.

[0012] When the equivalent external disturbance torque exceeds the dynamic safety threshold, the gain parameter of the position loop controller is instantly reduced, causing the actual motor to immediately lose its position holding capability and enter a flexible low-impedance mode that adapts to external forces.

[0013] When the equivalent external disturbance torque falls below the dynamic safety threshold, it is determined that the external collision has disappeared, and the gain parameter is controlled to rise back to the rated stiffness in a non-linear smooth trajectory, thereby avoiding secondary oscillations in the system during the stiffness recovery process.

[0014] In this way, the phase current, rotor position, and speed signals of the actual motor can be acquired in real time at high frequency, and torque commands from the host computer can be received simultaneously. Using this command as input, the dynamic deviation between the actual motor operating state and the theoretical state output by the virtual reference model can be calculated in real time on an extremely short time scale. This deviation comprehensively reflects the contact forces between the unmodeled disturbances inside the system and the external environment.

[0015] Furthermore, the system can perform multi-level filtering and calculation on dynamic deviation values ​​to eliminate purely external collision forces. The system incorporates Stribeck friction feedforward compensation logic based on a high-precision lookup table method, and combines this with temperature sensor data during motor operation for temperature drift correction, eliminating friction errors caused by changes in grease viscosity. The total disturbance torque is subtracted from the temperature-corrected reference friction torque, and then processed by a low-pass filter to finally output a high signal-to-noise ratio equivalent external disturbance torque observation value.

[0016] Moreover, the dynamic safety threshold can be dynamically generated based on the actual speed of the motor. It maintains a low fixed value in the low-speed range to ensure extremely high human-machine interaction sensitivity, and introduces a dynamic compensation term related to the speed in the high-speed range to offset the modeling errors caused by centrifugal force and Coriolis force and prevent false triggering.

[0017] Moreover, when the disturbance observation value exceeds the dynamic safety threshold, the system can determine that a collision has occurred, and immediately reduce the proportional gain of the position loop controller and clear the integral gain to zero within the current control cycle, so that the motor instantly enters a low-impedance flexible yielding mode to absorb the collision impact.

[0018] Furthermore, it enables a smooth S-shaped recovery after the collision disappears: when the disturbance observation value falls below the safety threshold, the system determines that the collision has disappeared. At this time, the controller does not immediately jump back to the original gain, but runs an S-shaped soft recovery program based on the system's mechanical time constant, allowing the stiffness to recover nonlinearly and smoothly, avoiding secondary oscillations in the system. During this recovery period, if a collision signal is detected again, the system will unconditionally interrupt the recovery process and re-execute instantaneous unloading.

[0019] Preferably, in step S1, the virtual reference model is constructed based on discretized Newtonian dynamic equations:

[0020]

[0021] in, The equivalent moment of inertia obtained by the system identification, Here, k represents the viscous damping coefficient, and k represents a certain moment during motor operation. This is the torque command output by the position loop or velocity loop at time k. Let k be the virtual theoretical angular velocity. To control the cycle.

[0022] Preferably, the multi-level signal processing in step S3 includes:

[0023] The reference friction torque is obtained by looking up a table based on the actual speed of the motor.

[0024] The real-time operating temperature data of the motor is acquired, and the reference friction torque is scaled and corrected based on a preset temperature viscosity correction coefficient to obtain the compensation torque value.

[0025] The equivalent external disturbance torque is extracted by subtracting the compensation torque value from the total disturbance torque, which includes the electromagnetic torque and inertial torque of the motor, and then filtering out high-frequency noise through a low-pass filter.

[0026] Preferably, the dynamic safety threshold mentioned in step S4 is a function that is non-linearly piecewise related to the actual speed of the motor:

[0027] When the actual speed of the motor is lower than or equal to the preset speed boundary, the dynamic safety threshold is a fixed basic threshold.

[0028] When the actual speed of the motor is higher than the preset speed boundary, the dynamic safety threshold increases dynamically as the speed increases, including a dynamic compensation term that is proportional to the square of the actual speed of the motor.

[0029] Preferably, the real-time adjustment of the gain parameter of the position loop controller using an asymmetric strategy in step S5 specifically includes flexible trigger determination:

[0030] When the absolute value of the equivalent external disturbance torque exceeds the dynamic safety threshold, an external collision is determined to have occurred.

[0031] In the current or next control cycle when an external collision is determined to have occurred, the microprocessor forcibly reduces the position loop proportional gain to a preset minimum sustaining value and clears the position loop integral gain to zero in order to perform an instantaneous stiffness unloading operation.

[0032] Preferably, the real-time adjustment of the gain parameters of the position loop controller using an asymmetric strategy in step S5 further includes stiffness recovery determination:

[0033] When the absolute value of the equivalent external disturbance torque is detected to be lower than the dynamic safety threshold, it is determined that the external collision has disappeared;

[0034] After determining that the external collision has disappeared, the recovery timer is started. Based on the mechanical time constant of the system, the proportional gain of the control position loop slowly and nonlinearly recovers to the rated stiffness value along the S-shaped trajectory of the Sigmoid function curve.

[0035] If an external collision is detected again at any point during the stiffness recovery process, the current stiffness recovery process is unconditionally interrupted, and the instantaneous stiffness unloading operation is re-executed.

[0036] Preferably, the total disturbance torque The torque quantification representation of the dynamic deviation between the actual operating state and the virtual theoretical state described in step S2:

[0037] ,

[0038] in The torque constant is For the measured current component, ω act For actual speed,

[0039] Compensation torque value Used to eliminate the effects of temperature drift:

[0040] ,

[0041] in, This represents the reference friction torque obtained using the actual rotational speed of the motor.

[0042] Equivalent external disturbance torque :

[0043] ,

[0044] in, This refers to a low-pass filter that filters out high-frequency noise.

[0045] Secondly, the present invention provides a flexible anti-collision control system for a joint motor, used to implement the above-mentioned method. The system's hardware architecture includes at least:

[0046] Data acquisition module: responsible for real-time acquisition of the motor's three-phase current, rotor absolute position, and speed signals;

[0047] Model computation module: It is embedded in the microprocessor and is used to perform differential iterative computation of the virtual reference model in real time and parallel, and output the virtual theoretical state of the motor.

[0048] Disturbance observation module: used to perform state deviation calculation and friction compensation calculation, and to calculate the equivalent external disturbance torque;

[0049] Stiffness adjustment module: used to generate asymmetric gain adjustment commands based on the comparison results between the equivalent external disturbance torque and the dynamic safety threshold;

[0050] Motor drive module: Used to output modulated voltage vector to drive the motor based on the adjusted control parameters.

[0051] The modules interact with each other via an internal high-speed bus or shared memory mechanism to ensure that the overall response of the system from collision detection to stiffness adjustment is in a state of extremely low latency.

[0052] Thirdly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a microprocessor, implements the aforementioned method for adjusting the asymmetric stiffness of a joint motor. The storage medium also pre-stores a system identification parameter library and a friction model lookup table for specific motor and reducer models, and provides an interface for updating parameters via an external communication bus.

[0053] Fourthly, the present invention provides a computer program product comprising computer program instructions, which, when executed by a processor, cause the processor to execute any of the aforementioned methods for adjusting the asymmetric stiffness of a joint motor based on a virtual reference model.

[0054] Compared with the prior art, the beneficial effects of this invention are as follows:

[0055] (1) No expensive torque sensor is required; flexible control can be achieved using only microprocessor algorithms, which significantly reduces hardware costs and system size while maintaining the overall rigidity of the mechanical structure.

[0056] (2) By combining the virtual reference model with friction feedforward compensation, nonlinear friction interference can be effectively eliminated, the equivalent external disturbance torque with high signal-to-noise ratio can be calculated, the pure external contact force can be accurately separated, the sensitivity of collision detection can be significantly improved, and the problem of misjudgment under high-speed motion can be solved.

[0057] (3) An asymmetric stiffness adjustment strategy is adopted. During a collision, the system is instantly unloaded to become low impedance to protect personnel and equipment. After the collision, the stiffness slowly recovers to prevent system oscillation, thus achieving a perfect balance between safety and stability.

[0058] (4) The system responds quickly, the algorithm based on the disturbance observer has a small computational load, it is easy to run in real time in conventional embedded microprocessors, and it is suitable for large-scale application. Attached Figure Description

[0059] Figure 1 This is a hardware architecture diagram of the overall control system in this invention;

[0060] Figure 2 This is the program logic flowchart of the overall control system in this invention;

[0061] Figure 3 The timing diagram for asymmetric stiffness adjustment in the embodiment is shown;

[0062] Figure 4 The dynamic safety threshold curve in the embodiment is shown. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely some embodiments of this invention, used only to explain the invention, and not intended to limit the invention. The technical features of each embodiment in this invention can be combined accordingly without mutual conflict. It should be noted that unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of components and steps described in these embodiments do not limit the scope of this invention. Those skilled in the art will understand that terms such as "the nth" and "Sn" in the embodiments of this invention are only used to distinguish different steps, devices, or modules, and do not represent any specific technical meaning, nor do they indicate a necessary logical order between them. It should also be understood that in the embodiments of this invention, "multiple" can refer to two or more, and "at least one" can refer to one, two, or more. It should also be understood that any component, data, or structure mentioned in the embodiments of this invention can generally be understood as one or more unless explicitly limited or given contrary guidance in the preceding or following text. Furthermore, the term "and / or" in this invention is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this invention generally indicates that the preceding and following related objects have an "or" relationship. It should also be understood that the description of the various embodiments in this invention emphasizes the differences between the various embodiments; similarities or commonalities can be referred to mutually, and for the sake of brevity, they will not be described in detail. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further discussed in subsequent figures.

[0064] This invention can be applied to electronic devices such as terminal devices, computer systems, and servers, and can operate with a wide range of other general-purpose or special-purpose computing system environments or configurations. Examples of well-known terminal devices, computing systems, environments, and / or configurations suitable for use with terminal devices, computer systems, servers, and other electronic devices include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputer systems, mainframe computer systems, and distributed cloud computing environments including any of the above systems, etc. Terminal devices, computer systems, servers, and other electronic devices can be described in the general context of computer system executable instructions (such as program modules) executed by a computer system. Typically, program modules can include routines, programs, object programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. Computer systems / servers can be implemented in distributed cloud computing environments where tasks are performed by remote processing devices linked via a communication network. In distributed cloud computing environments, program modules can reside on local or remote computing system storage media, including storage devices.

[0065] This embodiment provides a flexible collision avoidance control technology for articulated motors based on a virtual reference model and asymmetric stiffness adjustment. This technology aims to achieve active safety control with high-sensitivity collision detection and adaptive asymmetric stiffness adjustment through deep hardware and software collaboration. Here, the virtual reference model refers to an idealized mathematical model running in a microprocessor.

[0066] <Exemplary System>

[0067] This embodiment uses, as follows Figure 1 The illustrated asymmetric stiffness adjustment system for articulated motors based on a virtual reference model is a highly integrated flexible anti-collision control system for articulated motors (also referred to as the system). This system adopts a hierarchical distributed architecture, primarily consisting of a host computer, a drive system, and the articulated motors at the physical level. Each component is tightly coupled to its electrical interface via a high-speed communication bus, forming a closed-loop control circuit.

[0068] The host computer, serving as the trajectory planning layer and human-machine interface of the entire control system, primarily undertakes the tasks of issuing high-level motion commands and calibrating parameters as the next higher control level. The host computer establishes a real-time connection with the drive system via an industrial Ethernet bus (such as EtherCAT or CANopen). It is responsible for sending periodic position commands, velocity commands, or torque feedforward commands to the drive system according to preset kinematic algorithms. During system initialization or maintenance phases, the host computer downloads system identification parameters and friction model lookup table data for specific joint modules to the drive system and can monitor the drive system's operating status and fault alarm information in real time.

[0069] The drive system is the core computing and execution unit of this invention, and it highly integrates signal acquisition, algorithm calculation, and power output functions in hardware. For example... Figure 1 As shown, it contains four key modules: MCU (Microcontroller Unit), ADC (Analog-to-Digital Converter), SRAM (Static Random-Access Memory), and power stage.

[0070] The MCU, serving as the core computing hub, employs a high-performance 32-bit microprocessor (in this embodiment, the microprocessor is not limited to a specific brand or model; any embedded chip or FPGA with a floating-point unit (FPU) that meets real-time computing requirements is applicable). It integrates a floating-point unit for parallel execution of differential iteration of the virtual reference model, state calculation of the disturbance observer, and asymmetric stiffness adjustment strategies. The MCU schedules various peripheral modules via its internal bus to ensure the real-time performance of the control algorithm.

[0071] The ADC, as a high-precision analog signal acquisition interface, typically uses a high sampling frequency to acquire the phase current signal of the motor in real time. This module is usually used in conjunction with an operational amplifier circuit to convert the analog quantity of the acquired current signal into a digital quantity, which is then used by the MCU to perform Clark and Park transformations to calculate the torque current component and the excitation current component.

[0072] The SRAM serves as a high-speed data cache, storing a high-resolution Stribeck friction force map and real-time operating data. Since friction compensation requires fast table lookup speeds, the MCU utilizes the high read / write bandwidth of the SRAM to quickly index and interpolate the reference friction torque based on the current rotational speed, thus solving the control latency problem caused by the slow read speed of traditional Flash memory.

[0073] The power stage, as an energy conversion unit, includes a power inverter (e.g., a three-phase full-bridge inverter circuit composed of IGBTs or SiC MOSFETs) and a gate drive circuit. It receives PWM or SVPWM modulation signals from the MCU, modulates the DC bus voltage into an AC voltage vector that drives the motor, and achieves precise torque control of the motor.

[0074] A joint motor is the physical actuator of a system, comprising a power source, a transmission mechanism, and multi-dimensional sensing elements. For example... Figure 1 As shown, it specifically includes a PMSM (Permanent Magnet Synchronous Motor), an encoder, a sensor, and a reducer.

[0075] The PMSM, as a power source, has high power density and high dynamic response characteristics. Its stator windings receive current from the drive system, and the rotor outputs electromagnetic torque.

[0076] An absolute position sensor is installed on the motor shaft end or side; in this embodiment, a 23-bit or higher resolution multi-turn absolute magnetic encoder is specifically used. It feeds back the rotor's absolute position information to the drive system in real time via a high-speed serial interface. The MCU uses this position information to further differentially calculate the rotor speed.

[0077] The sensors mainly include a PT1000 temperature sensor embedded inside the motor windings or an NTC thermistor near the power stage, as well as a small-value current sampling resistor connected in series with the three-phase circuit. This module monitors the operating temperature and phase current in real time and feeds them back to the MCU. Based on this, the MCU queries the "temperature-viscosity" correction curve and performs temperature drift compensation on the friction model to ensure the stability of the observation accuracy under different operating conditions of the cold and hot engine.

[0078] The reducer is typically a harmonic reducer, used to increase output torque. Although the reducer introduces complex nonlinear friction and flexible deformation, this system effectively eliminates its inherent physical nonlinear interference through the virtual reference model and observation algorithm in the aforementioned drive system, achieving accurate perception of external collision forces.

[0079] Thus, the present invention provides an asymmetric stiffness adjustment system for a joint motor based on a virtual reference model for use in the methods described below, the system comprising at least the following logical architecture:

[0080] The data acquisition module is used to acquire the three-phase current, rotor absolute position, and speed signals of the motor in real time.

[0081] The model calculation module, which is embedded in the microprocessor, is used to perform differential iterative calculations of the virtual reference model in real time and parallel, and output the virtual theoretical state of the motor (also referred to as the theoretical state).

[0082] The disturbance observation module is used to perform state deviation calculation and friction compensation calculation to solve for the equivalent external disturbance torque.

[0083] The stiffness adjustment module is used to generate an asymmetric gain adjustment command based on the comparison result between the equivalent external disturbance torque and the dynamic safety threshold.

[0084] The motor drive module is used to output a modulated voltage vector to drive the motor based on the adjusted control parameters.

[0085] <Exemplary Method>

[0086] According to the present invention, a method for asymmetric stiffness adjustment of a joint motor based on a virtual reference model is provided, which can be run in a motor servo control device (e.g., a servo drive system including a microprocessor, power inverter, current acquisition circuit, and high-precision position sensor). As detailed below, it includes four core steps: constructing a virtual reference model, calculating real-time state deviation, observing disturbances, and adjusting asymmetric stiffness. By running a virtual model without friction terms in the microprocessor, the actual operating state is compared with the theoretical state, and the equivalent external disturbance torque is calculated using friction compensation. Based on the comparison result of this disturbance value and the dynamic safety threshold, an asymmetric strategy is used to adjust the proportional gain and integral gain of the position loop controller in real time.

[0087] Based on the aforementioned hardware architecture, the control logic in this embodiment operates with high real-time performance within the microprocessor. To clearly illustrate how this control system achieves flexible collision avoidance, the following section combines... Figure 2 The program flowchart is provided, with explanations interspersed throughout the corresponding steps. Figure 3 Asymmetric stiffness adjustment timing and Figure 4 The dynamic safety threshold curve.

[0088] S1: Parameter Initialization and Virtual Reference Model Construction. After the system powers on, the program initializes and automatically reads the system identification parameters for the current joint module from non-volatile memory, mainly including the equivalent moment of inertia J and viscous damping coefficient B of the joint motor rotation system. Simultaneously, the system loads the pre-stored Stribeck mapping table and temperature-viscosity correction coefficient table into high-speed SRAM. Based on this, a virtual reference model is built and runs in real-time in the microprocessor. This model is defined as a linear second-order system containing only the aforementioned equivalent moment of inertia J and viscous damping coefficient B. The nonlinear Coulomb friction, static friction, and cogging torque terms are removed through discretized Newton's equations of motion, thus constructing the ideal state of the actual physical motor. The model's dynamic equations are as follows:

[0089]

[0090] in, The torque command output by the position loop or velocity loop. Let τ be the virtual theoretical angular velocity, k represent a certain moment during the operation of the motor, and τ be the angular velocity. ref (k) represents the torque command output by the position loop or velocity loop at time k, ω v (k) represents the virtual theoretical angular velocity at time k. To control the cycle, the theoretical ideal position can be obtained by integrating the velocity. :

[0091]

[0092] in, This represents the theoretical ideal position at time k.

[0093] S2: High-frequency status acquisition and deviation observation. Within each control cycle, the microprocessor receives torque and speed commands from the previous motion planning stage. These commands are simultaneously input into the physical path (closed circuit) controlling the actual motor and the virtual reference model. The data acquisition module acquires the actual motor's operating status in real-time at high frequency, obtaining the measured torque-current components, actual rotor position, and actual speed. Subsequently, a state deviation observer is constructed to calculate the dynamic deviation between the actual operating state and the theoretical state output by the virtual reference model in real time.

[0094] For example, the actual motor's operating signal can be acquired in real time at a high frequency of not less than 10kHz, the operating signal including at least the actual phase current and the actual rotor position, and the actual operating state of the motor can be calculated accordingly; a state deviation observer is constructed to calculate the dynamic deviation value between the actual operating state and the virtual theoretical state output in step S1 in real time within the same control cycle.

[0095] S3: Multi-level signal processing and collision force extraction: Perform multi-level signal processing on the dynamic deviation value to calculate the equivalent external disturbance torque with a high signal-to-noise ratio.

[0096] In order to obtain the total state deviation (i.e., the "total disturbance torque" mentioned later) To extract the pure external collision force, this embodiment performs multi-level refined signal processing.

[0097] First, calculate the total disturbance torque, which includes the motor's electromagnetic torque and inertial torque, based on the actual parameters of the motor. The total disturbance torque This is the torque quantification representation of the dynamic deviation between the actual operating state and the virtual theoretical state described in step S2:

[0098]

[0099] in The torque constant is This represents the measured current component.

[0100] Secondly, the system uses the actual motor speed as an index to quickly look up the Stribeck mapping table in SRAM to obtain the reference friction torque, and combines this with the real-time temperature collected by the temperature sensor to find the viscosity correction coefficient. Scaling corrections are performed to eliminate the effects of temperature drift, resulting in the compensation torque value. :

[0101]

[0102] in, This represents the reference friction torque obtained using the actual rotational speed of the motor.

[0103] Finally, the total disturbance is subtracted from the corrected frictional force (i.e., the difference between the total disturbance torque and the compensation torque), and a low-pass filter (LPF) with a cutoff frequency set to 1 / 5 to 1 / 10 of the current loop bandwidth is used to filter out high-frequency noise, ultimately outputting an equivalent external disturbance torque with a high signal-to-noise ratio. :

[0104]

[0105] in, This refers to a low-pass filter that filters out high-frequency noise.

[0106] S4: Dynamic Safety Threshold Determination. After calculating the equivalent external disturbance torque, the system needs to determine whether the disturbance originates from a collision. This method employs a dynamic safety threshold function that exhibits a non-linear, piecewise correlation with the actual motor speed. :

[0107] =

[0108] Where K represents the dynamic compensation coefficient. Indicates the preset speed boundary.

[0109] When the actual motor speed is lower than or equal to the preset speed limit At that time, the dynamic safety threshold is a fixed base threshold, that is, a lower fixed threshold is used in the low-speed range. ; In the high-speed range (speed greater than the preset speed limit) Introducing a dynamic compensation term proportional to the square of the rotational speed. .

[0110] Combination Figure 4For low-speed operation (such as below 500 RPM) on horizontal straight sections, a lower base threshold is used. This ensures extremely high sensitivity under precise operation; high-speed operation corresponds to a parabolic segment, and the dynamic safety threshold dynamically increases with the speed. It includes a dynamic compensation term proportional to the square of the actual motor speed to offset false triggering caused by Coriolis force and centrifugal force errors. As long as the observed disturbance torque exceeds the dynamic safety threshold at the corresponding speed, the system determines that an external collision has occurred.

[0111] S5: Asymmetric stiffness adjustment state machine operation. Based on the decision result of S4, the microprocessor uses an asymmetric strategy to adjust the gain parameters of the position loop controller in real time. Figure 3 The position loop gain on the ordinate refers to the position loop proportional gain Kp.

[0112] (1) Flexible triggering determination stage ( Figure 3 (T1 point): When the equivalent external disturbance torque is detected The absolute value exceeds the dynamic security threshold. When an external collision is detected, the microprocessor forcibly reduces the position loop proportional gain Kp to a preset minimum holding value and clears the integral gain to zero, causing the actual motor to immediately lose its position holding capability and enter a flexible low-impedance mode that adapts to external forces. Figure 3 (Mid-T1-T2 stage).

[0113] For example, in the current or next control cycle when an external collision is determined to have occurred, the microprocessor forces the position loop proportional gain to be reduced to a preset minimum sustaining value and clears the position loop integral gain to zero in order to perform an instantaneous stiffness unloading operation.

[0114] (2) Stiffness recovery determination stage Figure 3 (T2-T3 stage): When the equivalent external disturbance torque is detected When the temperature drops below the dynamic safety threshold, the external collision is considered to have disappeared. At this point, the gain is slowly and non-linearly increased back to the rated stiffness value (corresponding to) along the Sigmoid function trajectory. Figure 3 initial state (This is to prevent secondary oscillations.)

[0115] For example, after determining that the external collision has disappeared, a recovery timer is started. Based on the mechanical time constant of the system, the proportional gain of the control position loop slowly and nonlinearly recovers to the rated stiffness value along the S-shaped trajectory of the Sigmoid function curve.

[0116] (3) Safety interruption logic: If an external collision is determined to occur again at any time during the recovery process, the current recovery process is unconditionally interrupted and the instantaneous stiffness unloading operation is re-executed.

[0117] This asymmetric adjustment strategy, which involves fast unloading and slow recovery, effectively eliminates the secondary impact and robotic arm oscillation caused by traditional step recovery, and significantly improves the stability of the system.

[0118] Thus, this method establishes and runs a virtual ideal motor model in the microprocessor of the motor controller to remove friction interference. This model is defined as a linear second-order system. Through algorithmic logic, it removes the unavoidable nonlinear Coulomb friction term, static friction term, and cogging torque term caused by magnetic circuit design in the actual physical motor operation, thereby constructing an ideal dynamic reference for the actual motor inside the microprocessor. The operating state of the actual motor is compared with the theoretical state of the model at high frequency to extract the pure external collision force. Subsequently, a safety threshold that dynamically changes with the actual speed of the motor is used for judgment, and an asymmetric strategy of instantaneous unloading and nonlinear slow recovery after the collision is triggered is adopted to adjust the position loop gain in real time.

[0119] In summary, this embodiment achieves its goals by constructing a high-precision virtual reference model, implementing multi-dimensional signal compensation and filtering, and employing a combination of... Figure 3 Time series and Figure 4 The asymmetric stiffness adjustment strategy of the threshold curve has successfully achieved high-performance flexible anti-collision control on a low-cost hardware platform. This solution not only solves industry challenges such as nonlinear friction interference and secondary oscillations, but also provides strong technical support for the safe interaction of collaborative robots with its high integration and high response speed.

[0120] Through the aforementioned hardware and software co-design, this embodiment successfully achieves industrial-grade reliable joint flexible collision avoidance control without adding expensive torque sensors. This invention requires no additional torque sensors, is low-cost, responds quickly, effectively distinguishes between frictional forces and external collision forces, and instantly and flexibly yields upon detecting a collision, smoothly recovering after the collision disappears, thus balancing the safety of human-machine interaction with the overall rigidity of the mechanical structure.

[0121] In addition to the methods and systems described above, embodiments of this disclosure can also be computer program products comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps of the methods according to various embodiments of this disclosure described in the "Exemplary Methods" section of this specification. The computer program product can be written in any combination of one or more programming languages ​​to perform operations of embodiments of this disclosure, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on a user computing device, partially on a user device, as a standalone software package, partially on a user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0122] Furthermore, embodiments of this disclosure can also be a computer-readable storage medium storing a computer program that, when executed by a microprocessor, implements the asymmetric stiffness adjustment method for a joint motor based on a virtual reference model as described above. The computer-readable storage medium may also pre-store a system identification parameter library for a specific motor model and mapping table data for calculating the reference friction torque. The computer-readable storage medium can be any combination of one or more readable media. A readable medium can be a readable signal medium or a readable storage medium. A readable storage medium can be, for example, including but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0123] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for asymmetric stiffness adjustment of a joint motor based on a virtual reference model, characterized in that, The method operates in a motor servo control device and includes the following steps: S1, Constructing a virtual reference model: A virtual reference model is constructed and run in real time in the microprocessor. The virtual reference model is defined as a linear second-order system containing only equivalent rotational inertia and viscous damping coefficient. The torque command issued by the upper control level is used as input. Under the premise of eliminating the interference of the nonlinear friction term and cogging torque term inherent in the actual physical motor, the virtual reference model is iterated in real time and outputs the virtual theoretical state of the motor in the ideal operation without interference. The virtual theoretical state includes at least the virtual theoretical position and the virtual theoretical velocity. S2: Status Acquisition and Deviation Observation: Real-time acquisition of the actual motor's operating signals, which include at least the actual phase current and the actual rotor position, and calculation of the motor's actual operating state based on these signals; construction of a status deviation observer to calculate the dynamic deviation between the actual operating state and the virtual theoretical state in real time within the same control cycle; S3: Multi-level signal processing and collision force extraction: Perform multi-level signal processing on the dynamic deviation value to calculate the equivalent external disturbance torque; S4: Dynamic safety threshold determination: Obtain a dynamic safety threshold that is non-linearly related to the actual speed of the motor, and compare the equivalent external disturbance torque with the dynamic safety threshold in real time; S5: Asymmetric stiffness adjustment: Based on the comparison results, an asymmetric strategy is used to adjust the gain parameters of the position loop controller in real time. When the equivalent external disturbance torque exceeds the dynamic safety threshold, the gain parameter of the position loop controller is instantly reduced, causing the actual motor to immediately lose its position holding capability and enter a flexible low-impedance mode that adapts to external forces. When the equivalent external disturbance torque falls below the dynamic safety threshold, it is determined that the external collision has disappeared, and the gain parameter is controlled to rise back to the rated stiffness in a non-linear smooth trajectory, thereby avoiding secondary oscillations in the system during the stiffness recovery process.

2. The method according to claim 1, characterized in that, In step S1, the virtual reference model is constructed based on discretized Newtonian dynamic equations: ， in, The equivalent moment of inertia obtained by the system identification, Here, k represents the viscous damping coefficient, and k represents a certain moment during motor operation. This is the torque command output by the position loop or velocity loop at time k. Let k be the virtual theoretical angular velocity. To control the cycle.

3. The method according to claim 1, characterized in that, The multi-level signal processing in step S3 includes: The reference friction torque is obtained by looking up a table based on the actual speed of the motor. The real-time operating temperature data of the motor is acquired, and the reference friction torque is scaled and corrected based on a preset temperature viscosity correction coefficient to obtain the compensation torque value. The equivalent external disturbance torque is extracted by subtracting the compensation torque value from the total disturbance torque, which includes the electromagnetic torque and inertial torque of the motor, and then filtering out high-frequency noise through a low-pass filter.

4. The method according to claim 1, characterized in that, The dynamic safety threshold mentioned in step S4 is a function that has a non-linear piecewise correlation with the actual speed of the motor: When the actual speed of the motor is lower than or equal to the preset speed boundary, the dynamic safety threshold is a fixed basic threshold. When the actual speed of the motor is higher than the preset speed boundary, the dynamic safety threshold increases dynamically as the speed increases, including a dynamic compensation term that is proportional to the square of the actual speed of the motor.

5. The method according to claim 1 or 4, characterized in that, The real-time adjustment of the gain parameters of the position loop controller using an asymmetric strategy, as described in step S5, specifically includes flexible trigger determination: When the absolute value of the equivalent external disturbance torque exceeds the dynamic safety threshold, an external collision is determined to have occurred. In the current or next control cycle when an external collision is determined to have occurred, the microprocessor forcibly reduces the position loop proportional gain to a preset minimum sustaining value and clears the position loop integral gain to zero in order to perform an instantaneous stiffness unloading operation.

6. The method according to claim 5, characterized in that, Step S5, which involves real-time adjustment of the gain parameters of the position loop controller using an asymmetric strategy, also includes stiffness recovery determination. When the absolute value of the equivalent external disturbance torque is detected to be lower than the dynamic safety threshold, it is determined that the external collision has disappeared; After determining that the external collision has disappeared, the recovery timer is started. Based on the mechanical time constant of the system, the proportional gain of the control position loop slowly and nonlinearly recovers to the rated stiffness value along the S-shaped trajectory of the Sigmoid function curve. If an external collision is detected again at any point during the stiffness recovery process, the current stiffness recovery process is unconditionally interrupted, and the instantaneous stiffness unloading operation is re-executed.

7. The method according to claim 3, characterized in that, Total disturbance torque The torque quantification representation of the dynamic deviation between the actual operating state and the virtual theoretical state described in step S2: ， in The torque constant is For the measured current component, ω act For actual speed, Compensation torque value Used to eliminate the effects of temperature drift: ， in, This represents the reference friction torque obtained using the actual rotational speed of the motor. Equivalent external disturbance torque : ， in, This refers to a low-pass filter that removes high-frequency noise.

8. An asymmetric stiffness adjustment system for a joint motor based on a virtual reference model, characterized in that, For implementing the method of any one of claims 1 to 7, the system comprises at least the following logical architecture: The data acquisition module is used to acquire the three-phase current, rotor absolute position, and speed signals of the motor in real time. The model calculation module, which is embedded in the microprocessor, is used to perform differential iterative calculations of the virtual reference model in real time and parallel, and output the virtual theoretical state of the motor. The disturbance observation module is used to perform state deviation calculation and friction compensation calculation to solve for the equivalent external disturbance torque. The stiffness adjustment module is used to generate an asymmetric gain adjustment command based on the comparison result between the equivalent external disturbance torque and the dynamic safety threshold. The motor drive module is used to output a modulated voltage vector to drive the motor based on the adjusted control parameters.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a microprocessor, it implements the asymmetric stiffness adjustment method for joint motors based on a virtual reference model as described in any one of claims 1 to 7. The computer-readable storage medium also contains a system identification parameter library for motor models and mapping table data for calculating reference friction torque.

10. A computer program product comprising computer program instructions, characterized in that, When the computer program instructions are executed by the processor, the processor performs the asymmetric stiffness adjustment method for articulated motors based on a virtual reference model as described in any one of claims 1 to 7.

Images