Integrated joint module with torque sensing and energy self-recovery functions and design method

Abstract

The invention relates to the technical field of robot actuators, and particularly discloses an integrated joint module with torque sensing and energy self-recovery functions and a design method. The design method comprises the steps that a physical framework of the joint module is constructed; a nonlinear elastic torsion model and a dynamic friction model of the harmonic reducer are constructed, and torque is calculated and derived; calculating a physical sampling torque; fusing the derived torque and the physical sampling torque by adopting an adaptive Kalman filtering mode to obtain a final torque signal; an energy feedback model in a motor braking mode is constructed, and when the joint module decelerates or is dragged by external force, the instantaneous flow of recyclable electric energy is calculated in real time based on the energy feedback model; a super capacitor and a DC-DC converter are introduced for energy recovery; and when the joint module is started or enters an acceleration stage, the energy stored by the super capacitor is released to the direct current bus. Nonlinear transmission interference caused by a harmonic reducer can be effectively eliminated, and in-situ recovery and peak power assistance of braking kinetic energy can be achieved.

Description

Technical Field

[0001] This invention relates to the field of robot actuator technology, and more specifically to an integrated joint module and design method with torque sensing and energy self-recovery. Background Technology

[0002] With the rapid popularization of industrial automation and collaborative robots, integrated joint modules, as their core power source, directly determine the robot's force control accuracy and operating time. Current integrated joint modules achieve a compact physical structure and convenient communication interfaces by highly integrating permanent magnet synchronous motors, reducers, drive boards, and encoders. In the field of force-controlled robots, real-time acquisition of torque information at the joint output is a prerequisite for achieving compliant control and collision detection; simultaneously, during the frequent start-stop processes of mobile robots, effective management and feedback of kinetic energy are key to improving endurance.

[0003] However, existing joint modules exhibit two core drawbacks in practical applications: First, due to size limitations, most modules struggle to install high-precision physical torque sensors. Torque calculated solely from current is subject to significant errors due to nonlinear elasticity, hysteresis, and complex frictional effects within the reducer, leading to reduced sensitivity. Second, the braking energy generated by the joint during deceleration or under gravity is often dissipated as heat through the braking resistor in the drive circuit. This not only results in severe energy waste but also causes heat buildup within the module, shortening the lifespan of electronic components and severely limiting the robot's operating time in confined spaces.

[0004] In integrated joint applications, due to the thin and long bus cables, the line loss caused by long-distance transmission is as high as 20% or more, and the bus voltage rise caused by the feedback current can easily cause overvoltage damage to the drive circuit. There is a lack of a closed-loop mechanism to achieve efficient energy recovery and circulation in the module.

[0005] Therefore, how to solve the problems of insufficient accuracy of torque sensing caused by nonlinear interference from the reducer in the existing technology, and the large loss of braking energy during long-distance transmission, which easily causes bus voltage instability, are problems that urgently need to be solved by those skilled in the art. Summary of the Invention

[0006] In view of the above problems, the present invention proposes an integrated joint module and design method with torque sensing and energy self-recovery, so as to overcome the above problems or at least partially solve the above problems.

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

[0008] In a first aspect, the present invention provides a design method for an integrated joint module with torque sensing and energy self-recovery, comprising the following steps: S1. Construct the physical architecture of the integrated joint module, which includes a permanent magnet synchronous motor, a harmonic reducer, a motor-end encoder, an output-end encoder, and a full-bridge strain gauge array. S2. Construct the nonlinear elastic torsional model and dynamic friction model of the harmonic reducer, and calculate and derive the torque; S3. Acquire the voltage signal output by the full-bridge strain gauge array, and calculate the physical sampling torque considering temperature changes; S4. The derived torque and the physical sampled torque are fused using an adaptive Kalman filter to obtain the final torque signal; S5. Construct an energy feedback model under motor braking mode. When the integrated joint module decelerates or is dragged by external force, it enters the energy recovery mode. The instantaneous flow of recoverable electrical energy is calculated in real time based on the energy feedback model. S6. Introduce supercapacitors and DC-DC converters for energy recovery; S7. When the integrated joint module starts up or enters the acceleration phase, it enters the power assist mode and releases the energy stored in the supercapacitor to the DC bus.

[0009] Furthermore, in S1, the full-bridge strain gauge array is integrated into the stress concentration area of ​​the rigid output disk of the harmonic reducer through vacuum sputtering.

[0010] Furthermore, in S2, the nonlinear elastic torsional model considers the elastic deformation of the flexure of the harmonic reducer, establishing a nonlinear relationship between the output torque and the angular displacements at the motor and output ends. The expression for the nonlinear elastic torsional model is:

[0011] in, The output torque of the harmonic reducer The reduction ratio, For linear stiffness, It is the third-order hardening coefficient. This refers to the angular displacement of the motor end. This represents the angular displacement at the output end.

[0012] Furthermore, the dynamic friction model first solves for the microscopic state variables of the hairline deformation of the contact surface inside the harmonic reducer through numerical integration, and then calculates the dynamic friction compensation value, the specific expression of which is:

[0013]

[0014]

[0015]

[0016] in, The microscopic state variable is the deformation of the hair-like structure on the internal contact surface of the harmonic reducer. Let z be the rate of change; This refers to the rotational speed of the permanent magnet synchronous motor. This refers to the microscopic stiffness coefficient; A continuous function describing the Stribek effect; The frictional torque is the Coulomb friction torque. This is the static friction torque; This is the Stribeck critical characteristic velocity; The micro damping coefficient; The macroscopic coefficient of viscous friction; This is the dynamic friction compensation value; This is the final derived torque.

[0017] Furthermore, in S3, the formula for calculating the physical sampling torque is:

[0018] in, This is the physical sampling torque after temperature compensation; This is the voltage signal output by the full-bridge strain gauge array; This is the sensitivity constant; Temperature coefficient; The real-time temperature is collected by the embedded thermistor in the integrated joint module; For reference temperature; It is a function of zero-point offset as a function of temperature.

[0019] Furthermore, in S4, the process of fusing the derivation torque and the physical sampling torque using an adaptive Kalman filter includes: The derived torque calculated using S2 is used as the state prediction value, and the prior estimate value is generated through the state transition equation. The prior estimate is corrected using the physical sampling torque obtained from S3 as the observed value; The Kalman filter optimally fuses the derived torque and the physical sampled torque by adaptively adjusting the weights between the predicted and observed values, and outputs the final torque signal.

[0020] Furthermore, in S5, the expression for the energy feedback model is:

[0021] in, The instantaneous flow rate of recyclable electrical energy; The electrical angular velocity of the permanent magnet synchronous motor. It is a permanent magnet flux chain. For stator resistance, This refers to the d-axis current of the permanent magnet synchronous motor. This refers to the q-axis current of the permanent magnet synchronous motor. This refers to the iron loss power of the permanent magnet synchronous motor.

[0022] Furthermore, S6 includes: S61. A local energy management unit consisting of a supercapacitor and a bidirectional DC-DC converter is integrated on the integrated joint module driver board. The local energy management unit is connected in parallel on both sides of the DC bus inside the integrated joint module. S62. Monitor the DC bus voltage rise slope via the controller. When it is predicted that the DC bus voltage will exceed the safety threshold in the next control cycle At that time, calculate the optimal duty cycle of the bidirectional DC-DC converter:

[0023] in, The optimal duty cycle for a bidirectional DC-DC converter. The DC bus voltage inside the integrated joint module; This represents the real-time voltage of the supercapacitor array. For energy storage inductance in bidirectional DC-DC converters; This is the upper limit of the inductor current ripple. The switching cycle of the bidirectional DC-DC converter; S63. Based on the optimal duty cycle, control the instantaneous flow rate of supercapacitors absorbing recyclable electrical energy.

[0024] Furthermore, in S7, the Boost mode is used to release the energy stored in the supercapacitor to the DC bus, and the external power supply current... for:

[0025] in, This represents the total load power of the current joint module; This refers to the output power of the supercapacitor. For the conversion efficiency of the bidirectional DC-DC converter, This refers to the DC bus voltage inside the integrated joint module.

[0026] Furthermore, it also includes: S8. Establish thermal balance constraint equations, monitor the temperature of power transistors and supercapacitors within the joint module in real time, and when the local temperature exceeds a preset threshold, limit the permanent magnet synchronous motor. Maximum rate of change of shaft current To reduce losses, the thermal balance constraint equation is expressed as follows:

[0027]

[0028] in, This is the preset maximum allowable operating temperature for the joint module; This is the current real-time temperature; This is the thermal limitation derating function. This is the gain coefficient.

[0029] Secondly, the present invention provides an integrated joint module with torque sensing and energy self-recovery, which is designed using the design method described above.

[0030] As can be seen from the above technical solution, compared with the prior art, the present invention has the following beneficial effects: This invention effectively eliminates nonlinear transmission interference caused by harmonic reducers by deeply integrating a physical strain gauge array and a nonlinear elastic torsion model within the joint module and introducing an adaptive extended Kalman filter algorithm, achieving high-bandwidth, high-precision closed-loop torque sensing. Simultaneously, this invention embeds a local energy management unit based on a supercapacitor and a bidirectional DC-DC converter within the joint module, enabling on-site recovery of braking kinetic energy and peak power assistance.

[0031] This invention, through its integrated sensing-drive-storage architecture, not only effectively reduces transmission losses and internal heat accumulation over long distances but also significantly enhances the system's stability under dynamic motion by smoothing out bus voltage. This invention has broad application prospects in areas such as precision operations in collaborative robots, high-dynamic running in quadruped robots, and long-endurance assistance in industrial exoskeletons, and can significantly improve the overall intelligent interaction level and operational efficiency of the machine. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0033] Figure 1 This is a flowchart illustrating the design method of an integrated joint module with torque sensing and energy self-recovery provided in an embodiment of the present invention. Figure 2 This is a topology and energy flow diagram of the integrated joint module system provided in this embodiment of the invention. Detailed Implementation

[0034] The technical solutions of 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.

[0035] like Figure 1 As shown in the figure, this invention discloses a design method for an integrated joint module with torque sensing and energy self-recovery, including the following steps: S1. Construct the physical architecture of the integrated joint module, which includes a permanent magnet synchronous motor, a harmonic reducer, a motor-end encoder, an output-end encoder, and a full-bridge strain gauge array. S2. Construct the nonlinear elastic torsional model and dynamic friction model of the harmonic reducer, and calculate and derive the torque; S3. Acquire the voltage signal output by the full-bridge strain gauge array, and calculate the physical sampling torque considering temperature changes; S4. The derived torque and the physical sampled torque are fused using an adaptive Kalman filter to obtain the final torque signal; S5. Construct an energy feedback model under motor braking mode. When the integrated joint module decelerates or is dragged by external force, it enters the energy recovery mode. The instantaneous flow of recoverable electrical energy is calculated in real time based on the energy feedback model. S6. Introduce supercapacitors and DC-DC converters for energy recovery; S7. When the integrated joint module starts up or enters the acceleration phase, it enters the power assist mode and releases the energy stored in the supercapacitor to the DC bus.

[0036] The following provides further explanation of each of the above steps.

[0037] S1. Construction of the physical architecture of the integrated joint module and spatial layout of multiple sensors: like Figure 2 As shown, the integrated joint module structurally integrates a permanent magnet synchronous motor (PMSM), a harmonic reducer, dual absolute encoders, and an MCU controller. The dual absolute encoders include a motor-end encoder and a module output encoder. The full-bridge strain gauge array is a three-phase inverter bridge. The high-sensitivity full-bridge strain gauge array is integrated into the stress concentration area of ​​the rigid output disk of the harmonic reducer using a vacuum sputtering process, ensuring the linearity of the sensed signal.

[0038] S2. Construct the nonlinear elastic torsional model and dynamic friction model of the harmonic reducer, and calculate and derive the torque; specifically including: S21. The nonlinear elastic torsional model considers the elastic deformation of the flexure in the harmonic reducer and establishes a nonlinear relationship between the output torque and the angular displacements at the motor and output ends. The expression for the nonlinear elastic torsional model is as follows:

[0039] in, The output torque of the harmonic reducer For the reduction ratio, For linear stiffness, It is the third-order hardening coefficient. This refers to the angular displacement of the motor end. This represents the angular displacement at the output end. This model is used to compensate for dynamic sensing lag caused by mechanical flexibility.

[0040] S22. To eliminate internal losses in the transmission chain, a dynamic friction model based on the LuGre model is established to correct the electromagnetic torque. Specifically, the dynamic friction model first solves for the microscopic state variables of the hairline deformation of the contact surface inside the harmonic reducer through numerical integration, and then calculates the dynamic friction compensation value. The specific expression is:

[0041]

[0042]

[0043] in, The microscopic state variable is the deformation of the hair-like structure on the internal contact surface of the harmonic reducer. Let z be the rate of change; This refers to the rotational speed of the permanent magnet synchronous motor. This refers to the microscopic stiffness coefficient; A continuous function describing the Stribek effect; The frictional torque is the Coulomb friction torque. This is the static friction torque; This is the Stribeck critical characteristic velocity; The micro damping coefficient; The macroscopic coefficient of viscous friction; This is the dynamic friction compensation value.

[0044] By using a dynamic friction model to isolate the interference of Coulomb friction and viscous friction on the perceived torque in real time, the dynamic friction model achieves decoupling of different physical properties through the combination of mathematical terms, specifically: Peeling of the viscous friction term: in the formula The term is proportional to velocity and directly corresponds to the viscous friction torque, which is obtained by real-time velocity acquisition. And combined with the coefficient of viscous friction The viscous drag component that varies with velocity is identified and separated from the total disturbance.

[0045] Stripping the Coulomb friction term: a term in the formula The elastic force generated by the deformation of the bristles, during the steady-state sliding phase, is... When, the term converges to .pass The function describes the static friction in real time. Friction towards Coulomb The transitional Stribeck effect allows for the accurate calculation of the Coulomb friction component at different speeds.

[0046] Microscopic damping term: The term is used to isolate transient damping interference during motion switching.

[0047] The total friction loss calculated in real time above is obtained. Then, compensation calculations are performed on the sensed torque: The calculated Coulomb friction component and viscous friction component are then compared from the torque... Precise elimination of interference is achieved to obtain the net joint output torque after removing interference. This refers to the derived torque, also known as the model torque.

[0048] S3. Digital processing and temperature drift compensation of the voltage signal from the full-bridge strain gauge array, specifically, combined with the temperature data acquired by the embedded thermistor within the joint module. Establish the compensation equation:

[0049] in, This is the physical sampling torque after temperature compensation; This is the voltage signal output by the full-bridge strain gauge array; This is the sensitivity constant; Temperature coefficient; The real-time temperature is collected by the embedded thermistor in the integrated joint module; For reference temperature; The zero-point offset is a function of temperature, and its calculation formula is: .

[0050] in, The temperature of the motor windings or magnets is collected in real time. This is a reference temperature (usually taken as room temperature, 20℃ or 25℃). To be at the reference temperature The reference value for rated magnetic flux is as follows; is the temperature sensitivity coefficient of the permanent magnet material.

[0051] S4. Torque fusion based on adaptive extended Kalman filter: The derived torque calculated in S2 is then fused. The physical sampling torque obtained from S3 As the observation input, the derived torque is calculated using S2. Using the state prediction baseline, prior estimates are generated through the state transition equation; subsequently, the physical sampling torque (also known as sensor torque) acquired by S3 is used. The prior estimates are corrected based on the observed values. Ultimately, the Kalman filter adaptively adjusts the weights between the predicted and observed values, thus adjusting the model torque. (Has a physical mechanism but suffers from modeling errors) and sensor torque (The sample is real but contains sampling noise) to perform optimal fusion and output a high-precision, low-latency fusion estimation torque T.

[0052] The specific process includes: ① Define state variables Where T represents the estimated actual fusion torque of the joint: This represents the rate of change of torque.

[0053] ② Construct the state transition equation:

[0054] This formula describes the transmission and accumulation of system uncertainties over time. It represents the process at the current moment, based on the state and physical model of the previous moment. To predict the current state, uncertainty will be propagated; simultaneously, due to the derived torque calculated by the physical model (i.e., the nonlinear elastic torsion model and the dynamic friction model)... There are simplified and unmodeled disturbances. This will introduce new uncertainties into the system.

[0055] in, The prior estimate is the covariance matrix, representing the predicted value at the current time. Uncertainty; This is the state transition matrix; The posterior estimated covariance matrix of the previous time step represents the error accuracy of the optimal fusion result at the previous time step. The process noise covariance matrix represents the uncertainty of the physical model itself.

[0056] It is to calculate the Kalman gain. The core variable is the Kalman gain, which determines the weighting of the model value and the sensor value.

[0057]

[0058] Where R is the sensor noise covariance and H is the observation matrix.

[0059] if Smaller size indicates a physical model Very accurate, at this time It will become smaller, and the fusion result will be more biased towards the model's prediction. If The larger value indicates that the physical model is unreliable at this point. It will become larger, and the fusion result will be more biased towards the physical sample values. .

[0060] The initial value is obtained based on the motor's rated parameters and offline identification. This invention employs an adaptive operator based on residuals, making the residuals... When a sudden change in joint acceleration is detected or When the absolute value exceeds the set threshold, it indicates that the physical model... Unable to react promptly to severe impacts. In this case, using a scaling factor... Increase : Where Qinit represents the initial process noise covariance matrix, which is usually preset based on the rated parameters of the motor, the nominal stiffness of the reducer, and the friction parameters obtained offline. The scaling factor λ is not a fixed constant and is highly correlated with the nonlinear torque variation trend in step S2; when While maintaining a small range and steady speed, reduce This allows the filter to make greater use of the physical model to suppress random noise from the sensor.

[0061] ③The final fusion formula is:

[0062] in, It is calculated by step S2 It is generated by the drive and includes the dynamic mechanism of the joint (elasticity, friction compensation, etc.). It is the deviation between the actual observed value and the predicted value; As an adaptive adjustment weight, it determines the magnitude of the correction in real time. The fused torque It has been preserved Its rapid dynamic response, and also possesses High absolute precision.

[0063] set up Kalman gain vector The first element in the matrix (gain with respect to the stress moment dimension), and the observation matrix H=[1,0]. Then the relationship between the final fused torque T and each variable can be expressed as:

[0064] in, K 1 represents a weighting factor that dynamically adjusts according to system operating conditions, calculated in real-time by the adaptive Kalman gain. When the system detects that the model prediction residual is too large, it increases the weighting factor accordingly. Q k Automatically increase K 1. Adjust system weights towards physical sampling torque. The tilt ensures real-time dynamic response; while under stable operating conditions, it reduces... K 1. Make the weights contribute to the model torque. The model is tilted to utilize its analytical properties to suppress sensor noise.

[0065] S5. Local Energy Feedback Modeling in Motor Braking Mode: When the joint module decelerates in the reverse direction or is dragged by an external force, the motor enters the generation mode and establishes an energy feedback model, the expression of which is:

[0066] in, The instantaneous flow rate of recyclable electrical energy; The electrical angular velocity of the permanent magnet synchronous motor. It is a permanent magnet flux chain. For stator resistance, This refers to the d-axis current of the permanent magnet synchronous motor. This refers to the q-axis current of the permanent magnet synchronous motor. Iron loss power of permanent magnet synchronous motor mainly includes hysteresis loss and eddy current loss caused by alternating magnetic field in the motor core.

[0067] S6. Energy recovery is achieved by introducing supercapacitors and DC-DC converters, specifically including: S61. A local energy management unit consisting of a supercapacitor and a bidirectional DC-DC converter is integrated on the integrated joint module driver board. The local energy management unit is connected in parallel on both sides of the DC bus inside the integrated joint module. S62. Monitor the DC bus voltage rise slope via the controller. When it is predicted that the DC bus voltage will exceed the safety threshold in the next control cycle At that time, calculate the optimal duty cycle of the bidirectional DC-DC converter:

[0068] in, The optimal duty cycle for a bidirectional DC-DC converter. The DC bus voltage inside the integrated joint module; This represents the real-time voltage of the supercapacitor array. For energy storage inductance in bidirectional DC-DC converters; This is the upper limit of the inductor current ripple. The switching cycle of the bidirectional DC-DC converter; S63. Controlling the instantaneous flow rate of supercapacitors absorbing recoverable electrical energy based on the optimal duty cycle, specifically including: During the energy recovery phase, the bidirectional DC-DC converter operates in Buck mode to accurately absorb instantaneous power. First, the target charging current needs to be determined: Optimal duty cycle The relationship with the charging current is as follows: ; in, This is the DC bus voltage. This is the real-time voltage of the supercapacitor. This represents the correction value for the current loop regulator. This duty cycle achieves optimal control of the energy flow balance between the bus and the capacitor.

[0069] The specific charging control logic is as follows: ① Sensing and Commutation: When the motor is detected to be entering the regenerative braking state, i.e. Greater than 0, and bus voltage When the set threshold is exceeded, the recycling branch will be forcibly opened.

[0070] ② Constant power / constant current switching: The controller switches according to... Dynamic adjustment In the initial stages of recycling, to maximize absorption The goal is to reduce the voltage of the supercapacitor as it approaches its upper limit. Switch to constant voltage and current limiting mode.

[0071] ③ Dynamic suppression: through With its rapid regulation, the DC-DC converter behaves as a controlled dynamic load, capable of offsetting voltage ripples caused by rapid mechanical energy feedback, ensuring that the bus voltage remains stable. It remains stable even during periods of intense fluctuation.

[0072] This step converts kinetic energy into electrostatic energy of the supercapacitor in real time by forcibly opening the recovery branch, thus smoothing out voltage surges.

[0073] S7. Secondary Energy Distribution and Peak Current Reduction: During joint module startup or high-acceleration phase, a power-assisted mode is entered. The Boost mode releases the energy stored in the supercapacitor to the internal DC bus, reducing the external DC power supply current to the internal DC bus. for:

[0074] in, This represents the total load power of the current joint module; This refers to the output power of the supercapacitor. For the conversion efficiency of the bidirectional DC-DC converter, This refers to the DC bus voltage inside the integrated joint module.

[0075] This step effectively reduces the instantaneous peak load of the power supply harness and minimizes voltage drop at the module end.

[0076] In a more advantageous embodiment, the method of the present invention further includes: S8. Dynamic adjustment of module performance based on thermal limits: A thermal balance constraint equation is established to monitor the temperature of the power transistors and supercapacitors within the joint module in real time. When the local temperature exceeds a preset threshold, the permanent magnet synchronous motor is restricted. Maximum rate of change of shaft current To reduce losses, the thermal balance constraint equation is expressed as follows:

[0077] in, This is the preset maximum allowable operating temperature for the joint module; This is the current real-time temperature; The thermal limitation derating function is expressed as follows:

[0078] in, For gain coefficient, when much smaller When the function outputs a large value, it does not restrict the motor's dynamic response; when Approaching When the function output approaches 0, the system forcibly limits it. The change in shaft current brings it to a steady state, preventing further temperature rise. This function establishes a monotonically increasing mapping relationship between the temperature difference and the upper limit of the rate of change of current.

[0079] This invention achieves its purpose by adding an adaptive limiting circuit to the output of the current loop controller. The controller uses a thermally limited derating function. The maximum allowable rate of change slope is calculated in real time. If the rate of change of the current command required by the control law exceeds this slope, the system will force the current to change according to the specified slope. The defined slope is used to perform current updates. This is achieved by limiting... This can significantly reduce the transient losses of inverter power transistors under high-frequency switching conditions, as well as the thermal effects of motor windings during sudden current changes, thereby achieving closed-loop temperature protection.

[0080] This step ensures that the joint module does not suffer thermal failure due to high-frequency charging and discharging while achieving energy recovery.

[0081] In another embodiment, the present invention provides an integrated joint module with torque sensing and energy self-recovery, which is designed using the above-described design method.

[0082] Next, a robot joint with a rated power of 400W and a reduction ratio of 1:100 was selected as the test object. It integrates a supercapacitor bank (nominal voltage 48V, capacitance 25F) and a bidirectional Buck-Boost drive circuit. The specific steps include: S1. Constructing the physical architecture and multi-sensor layout of the module: Integrating PMSM, harmonic reducer, and dual encoders, and integrating a full-bridge strain gauge on the rigid output disc of the harmonic reducer.

[0083] S2. Constructing a nonlinear elastic and dynamic friction model in the controller kernel: First, input the stiffness coefficient of the harmonic reducer. , Through the formula Calculate the theoretical elastic torque.

[0084] Next, initialize the dynamic friction model parameters: static friction force. Coulomb friction Stribeck speed The state variables of the mane are solved by numerical integration. Thus, the dynamic friction compensation value can be calculated. .

[0085] S3, Physical sampling and temperature field decoupling calibration system reads the voltage of the full-bridge strain circuit in real time. In this embodiment, the sensitivity coefficient is set. Built-in thermistor provides feedback on the current core temperature of the module. Compare with reference temperature The corrected physical torque is calculated using the quadratic compensation equation. At this time, the temperature drift compensation term is approximately This effectively avoids zero-position drift caused by motor overheating.

[0086] S4. Moment fusion based on adaptive extended Kalman filter: combining model torques and physical sampling torque For the observed input, define the state variables. By dynamically adjusting the process noise covariance When the joint is subjected to impact load, the weight of the physical sampling torque is increased, and the weight of the model torque is increased during smooth motion, thereby obtaining a smooth and high-bandwidth torque signal.

[0087] S5. Real-time prediction of regenerative power limit: When the joint module decelerates, the controller predicts the upper limit of regenerative power based on the number of pole pairs in the motor parameters. stator resistance Calculate the current instantaneous feedback power. Assume the current... , Substitute into the formula:

[0088] The theoretical recoverable power is derived as follows: This value serves as the input constraint for the MPC algorithm in S6.

[0089] S6. Reliability Quantization and Switching of Energy Feedback Signals: To address the issue of unstable switching caused by unverified transient signals in existing technologies, this step quantifies the bus voltage rise rate. Robustness. If the detected voltage exceeds [a certain value] for three consecutive sampling periods. If the confidence assessment shows that the signal is not a high-frequency noise fluctuation, then the recovery mechanism will be smoothly activated.

[0090] At this point, the local energy recovery MPC controller based on the MPC algorithm solves for the optimal duty cycle sequence in each control cycle. This algorithm enables the supercapacitor to respond to the motor's braking demand in the shortest possible time, achieving efficient kinetic energy capture.

[0091] S7, Peak Auxiliary Power Release: Total Current Required When the Robot Executes a Jump or Heavy Load Start Command. achieve Upon detecting the load, the Local Energy Management Unit (LEMU) immediately switches to Boost mode, with the supercapacitor array providing 5A of current to reduce the external bus current requirement to 10A, significantly mitigating the instantaneous voltage drop of the entire power supply.

[0092] S8. Real-time assessment of the supercapacitor's remaining state of charge (SoE). Furthermore, the module temperature was detected to be approaching a critical value. The controller automatically adjusts the proportional coefficient. This is achieved by increasing the proportion of active energy dissipation in the motor windings to reduce the supercapacitor charging current, ensuring the module operates within its thermal safety boundaries. The relevant formula is:

[0093] in, This refers to excess feedback power that supercapacitors cannot absorb. As a regulating factor, it determines how much excess energy should be dissipated by increasing the motor's internal resistance losses. This is achieved through adjustment. The controller can change in real time The magnitude of the shaft current, so as not to affect Precisely control the heating power of the winding under shaft torque current conditions.

[0094] When the state of charge (SoE) of the supercapacitor approaches saturation or the module temperature reaches the safety boundary, the controller actively injects a non-power current component that does not generate torque (i.e., (Current). According to Joule's law These currents generate heat as they pass through the windings. Essentially, this method briefly switches the motor from a "power source" to an "energy-consuming load," converting the electrical energy that would otherwise be fed back to the bus into heat energy within the motor beforehand. This protects the DC bus voltage and prevents the supercapacitor from overcharging.

[0095] Final output: the final fused sensing torque The signal is transmitted to the main controller for control. Experimental results show that the robustness of torque sensing was improved by 35% throughout the implementation process, and the DC bus voltage fluctuation amplitude decreased from the original value. Down to The supercapacitor's energy recycling function reduced the power consumption of a single mission by 13.2%.

[0096] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0097] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A design method for an integrated joint module with torque sensing and energy self-recovery, characterized in that, Includes the following steps: S1. Construct the physical architecture of the integrated joint module, which includes a permanent magnet synchronous motor, a harmonic reducer, a motor-end encoder, an output-end encoder, and a full-bridge strain gauge array. S2. Construct the nonlinear elastic torsional model and dynamic friction model of the harmonic reducer, and calculate and derive the torque; S3. Acquire the voltage signal output by the full-bridge strain gauge array, and calculate the physical sampling torque considering temperature changes; S4. The derived torque and the physical sampled torque are fused using an adaptive Kalman filter to obtain the final torque signal; S5. Construct an energy feedback model under motor braking mode. When the integrated joint module decelerates or is dragged by external force, it enters the energy recovery mode. The instantaneous flow of recoverable electrical energy is calculated in real time based on the energy feedback model. S6. Introduce supercapacitors and DC-DC converters for energy recovery; S7. When the integrated joint module starts up or enters the acceleration phase, it enters the power assist mode and releases the energy stored in the supercapacitor to the DC bus.

2. The design method of the integrated joint module with torque sensing and energy self-recovery as described in claim 1, characterized in that, In S2, the nonlinear elastic torsional model considers the elastic deformation of the flexure of the harmonic reducer, and establishes a nonlinear relationship between the output torque and the angular displacement of the motor end and the angular displacement of the output end. The expression of the nonlinear elastic torsional model is as follows: in, The output torque of the harmonic reducer For the reduction ratio, For linear stiffness, It is the third-order hardening coefficient. This refers to the angular displacement of the motor end. This represents the angular displacement at the output end.

3. The design method of the integrated joint module with torque sensing and energy self-recovery as described in claim 2, characterized in that, The dynamic friction model first solves for the microscopic state variables of the hairline deformation of the contact surface inside the harmonic reducer through numerical integration, and then calculates the dynamic friction compensation value. The specific expression is as follows: in, The microscopic state variable is the deformation of the hair-like structure on the internal contact surface of the harmonic reducer. Let z be the rate of change; This refers to the rotational speed of the permanent magnet synchronous motor. This refers to the microscopic stiffness coefficient; A continuous function describing the Stribek effect; The Coulomb friction torque; This is the static friction torque; This is the Stribeck critical characteristic velocity; The micro damping coefficient; The macroscopic coefficient of viscous friction; This is the dynamic friction compensation value; This is the final derived torque.

4. The design method of the integrated joint module with torque sensing and energy self-recovery as described in claim 1, characterized in that, In S3, the formula for calculating the physical sampling torque is: in, This is the physical sampling torque after temperature compensation; This is the voltage signal output by the full-bridge strain gauge array; This is the sensitivity constant; Temperature coefficient; The real-time temperature is collected by the embedded thermistor in the integrated joint module; For reference temperature; It is a function of zero-point offset as a function of temperature.

5. The design method of the integrated joint module with torque sensing and energy self-recovery as described in claim 1, characterized in that, In S4, the process of fusing the derived torque and the physical sampling torque using an adaptive Kalman filter includes: The derived torque calculated using S2 is used as the state prediction value, and the prior estimate value is generated through the state transition equation. The prior estimate is corrected using the physical sampling torque obtained from S3 as the observed value; The Kalman filter optimally fuses the derived torque and the physical sampled torque by adaptively adjusting the weights between the predicted and observed values, and outputs the final torque signal.

6. The design method of the integrated joint module with torque sensing and energy self-recovery as described in claim 1, characterized in that, In S5, the expression for the energy feedback model is: in, The instantaneous flow rate of recyclable electrical energy; The electrical angular velocity of the permanent magnet synchronous motor. It is a permanent magnet flux chain. For stator resistance, This refers to the d-axis current of the permanent magnet synchronous motor. This refers to the q-axis current of the permanent magnet synchronous motor. This refers to the iron loss power of the permanent magnet synchronous motor.

7. The design method of the integrated joint module with torque sensing and energy self-recovery as described in claim 1, characterized in that, S6 include: S61. A local energy management unit consisting of a supercapacitor and a bidirectional DC-DC converter is integrated on the integrated joint module driver board. The local energy management unit is connected in parallel on both sides of the DC bus inside the integrated joint module. S62. Monitor the DC bus voltage rise slope via the controller. When it is predicted that the DC bus voltage will exceed the safety threshold in the next control cycle At that time, calculate the optimal duty cycle of the bidirectional DC-DC converter: in, The optimal duty cycle for a bidirectional DC-DC converter. The DC bus voltage inside the integrated joint module; This represents the real-time voltage of the supercapacitor array. For energy storage inductance in bidirectional DC-DC converters; This is the upper limit of the inductor current ripple. The switching cycle of the bidirectional DC-DC converter; S63. Based on the optimal duty cycle, control the instantaneous flow rate of supercapacitors absorbing recyclable electrical energy.

8. The design method of the integrated joint module with torque sensing and energy self-recovery as described in claim 1, characterized in that, In the S7, the Boost mode is used to release the energy stored in the supercapacitor to the DC bus, and the external power supply current... for: in, This represents the total load power of the current joint module; This refers to the output power of the supercapacitor. For the conversion efficiency of the bidirectional DC-DC converter, This refers to the DC bus voltage inside the integrated joint module.

9. The design method of the integrated joint module with torque sensing and energy self-recovery as described in claim 1, characterized in that, Also includes: S8. Establish thermal balance constraint equations, monitor the temperature of power transistors and supercapacitors within the joint module in real time, and when the local temperature exceeds a preset threshold, limit the permanent magnet synchronous motor. Maximum rate of change of shaft current To reduce losses, the thermal balance constraint equation is expressed as follows: in, This is the preset maximum allowable operating temperature for the joint module; This is the current real-time temperature; This is the thermal limitation derating function. This is the gain coefficient.

10. An integrated joint module with torque sensing and energy self-recovery, characterized in that, It is designed using the design method described in any one of claims 1-9.

Images