Modular integration of robotic joint actuators with servo control system

Abstract

This invention discloses a modular integration and servo control system for robot joint actuators, relating to the field of robot control technology. It aims to solve the technical problem that existing technologies cannot proactively and dynamically coordinate mechanical and electrical energy paths, leading to difficulties in balancing hardware protection and energy efficiency optimization, and resulting in high overall energy loss. The system includes: a modular joint module comprising a motor, a variable stiffness elastomer, a regenerative braking circuit, and an energy storage capacitor; an energy distribution unit connected to the regenerative braking circuit and an adjustment mechanism; and a central controller communicatively connected to the energy distribution unit. The central controller is configured to acquire the operating parameters of the joint module, including the charge of the energy storage capacitor, the cumulative number of actions of the variable stiffness elastomer, and the temperature of the regenerative braking circuit. This invention achieves proactive management and path optimization of impact energy, effectively solving the problem of hardware damage caused by reactive control in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, and more specifically, to a modular integration and servo control system for robot joint actuators. Background Technology

[0002] Robot joint actuators are the core execution components for robot motion control, and their performance directly determines the robot's dynamic response capability, operational accuracy, and system reliability. As robot technology advances towards higher dynamic response, higher integration, longer endurance, and safer human-robot interaction, higher demands are placed on the control performance of joint actuators.

[0003] Existing robot joint actuators typically employ a single energy absorption or recovery strategy when facing complex conditions such as impact loads and continuous high-frequency movements. For example, they may rely solely on elastomers for passive buffering or solely on regenerative braking circuits for energy recovery. However, such single strategies have significant drawbacks: relying solely on elastomers to absorb impact energy will cause the elastomers to be in a high-stress state for extended periods, accelerating material fatigue and shortening joint lifespan; relying solely on regenerative braking circuits to recover energy may lead to repeated charging and discharging of the energy storage capacitors under continuous impact conditions, causing overcharging damage or energy leakage. At the same time, the heat generated during the regenerative braking process will exacerbate the thermal aging of the actuator's power devices.

[0004] More critically, existing control systems typically employ reactive control strategies based on the current state, passively executing protective actions only after detecting fault signals such as over-limit capacitor voltage or excessive temperature. This control method cannot predict future operational conditions. When impact energy changes rapidly on a millisecond timescale, the control system often suffers hardware damage or energy waste due to response lag. Furthermore, existing systems lack comprehensive management of the coupling relationships between multiple physical quantities, such as elastomer fatigue life, capacitor health, and actuator thermal characteristics. Each actuator operates independently, failing to dynamically adjust the ratio of mechanical absorption to electrical recovery based on real-time operational conditions, resulting in low energy utilization efficiency and a trade-off between hardware lifespan and system reliability. Therefore, constructing a robot joint drive control system capable of proactively sensing operational condition evolution trends, dynamically coordinating mechanical absorption and electrical recovery paths, and minimizing overall losses under multi-physical quantity coupling constraints is a pressing technical problem to be solved in this field. Summary of the Invention

[0005] The purpose of this invention is to provide a modular integration and servo control system for robot joint actuators, in order to solve the technical problem that the existing technology cannot proactively and dynamically coordinate mechanical and electrical energy paths, resulting in difficulty in balancing hardware protection and energy efficiency optimization, and large overall losses.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a modular integration and servo control system for robot joint actuators, comprising:

[0007] A modular joint module includes a motor, a variable stiffness elastomer, a regenerative braking circuit, and an energy storage capacitor; wherein the variable stiffness elastomer includes an elastomer and an adjustment mechanism acting on the elastomer.

[0008] An energy distribution unit, which is connected to the regenerative braking circuit and the regulating mechanism;

[0009] A central controller, which is communicatively connected to the energy distribution unit, is configured to:

[0010] The operating parameters of the joint module are obtained, including the charge of the energy storage capacitor, the cumulative number of actions of the variable stiffness elastomer, and the temperature of the regenerative braking circuit.

[0011] Based on the operating parameters, a mechanical absorption ratio sequence of the variable stiffness elastomer and an electrical absorption ratio sequence of the regenerative braking circuit are generated in the prediction time domain, wherein the sum of the mechanical absorption ratio and the electrical absorption ratio is one.

[0012] With the goal of minimizing the overall loss in the prediction time domain, the target values ​​for mechanical absorption and electrical absorption in the current control cycle are determined from the mechanical absorption ratio sequence and the electrical absorption ratio sequence.

[0013] The mechanical absorption target value is sent to the regulating mechanism, and the electrical absorption target value is sent to the regenerative braking circuit.

[0014] This invention achieves proactive management and path optimization of impact energy by constructing a collaborative control architecture that dynamically allocates the ratio of mechanical absorption to electrical recovery within the prediction time domain, effectively solving the problem of hardware damage caused by reactive control in existing technologies. Specifically, the central controller in this invention acquires key operating parameters in real time, such as the energy storage capacitor charge, the cumulative number of actions of the variable stiffness elastomer, and the temperature of the regenerative braking circuit, and generates mechanical absorption ratio sequences and electrical absorption ratio sequences within the prediction time domain. This prediction mechanism enables the control system to anticipate future energy input trends and constraint boundaries, proactively adjusting the allocation strategy before constraint violations occur. Compared to traditional control methods that only passively respond after a fault occurs, this invention achieves "predictive absorption" of impact energy by solving the optimal allocation target value for the current control cycle through rolling optimization. When it is predicted that the capacitor charge will reach its upper limit, the system reduces the electrical absorption ratio in advance to avoid overcharging damage; when it is predicted that the elastomer fatigue accumulation will reach its upper limit, the system reduces the mechanical absorption ratio in advance to extend the elastomer's lifespan; when it is predicted that the actuator temperature will reach its upper limit, the system disables regenerative braking in advance to prevent thermal failure. This predictive collaborative control mechanism fundamentally solves the hardware damage problem caused by response lag in traditional solutions, significantly improving the operational reliability and service life of the joint drive system.

[0015] Preferably, the energy distribution unit includes:

[0016] A programmable logic array that integrates a state machine module and a lookup table module;

[0017] A current shunt is connected between the regenerative braking circuit and the motor to detect the regenerative braking current and the motor drive current.

[0018] A hardware comparator, whose input is connected to the voltage detection terminal of the energy storage capacitor and the temperature detection terminal of the regenerative braking circuit, is used to output an interrupt signal to the programmable logic array when the voltage of the energy storage capacitor or the temperature of the regenerative braking circuit exceeds a preset threshold.

[0019] In response to the interrupt signal, the programmable logic array forcibly adjusts the mechanical absorption target value and the electrical absorption target value.

[0020] Preferably, the programmable logic array further includes a hardware PID adjustment module. The input terminal of the hardware PID adjustment module is connected to the voltage detection terminal of the energy storage capacitor, and the output terminal is connected to the current reference terminal of the regenerative braking circuit. It is used to dynamically correct the electrical absorption target value according to the deviation between the voltage of the energy storage capacitor and the target voltage.

[0021] Preferably, the central controller includes:

[0022] The constrained prediction module is configured to generate a predicted charge level for the energy storage capacitor at a future time based on the capacitor's historical charging power and predicted future load power. Based on the historical compression counts of the variable stiffness elastomer and the mechanical absorption ratio, a predicted cumulative number of actions of the variable stiffness elastomer is generated. Based on the historical regenerative current and heat dissipation conditions of the regenerative braking circuit, a predicted temperature for the regenerative braking circuit is generated. ;

[0023] The rolling optimization module is configured to solve the mechanical absorption ratio sequence and the electrical absorption ratio sequence within the prediction time domain, with the constraints that the predicted power consumption does not exceed a first upper limit, the predicted cumulative number of actions does not exceed a second upper limit, and the predicted temperature does not exceed a third upper limit.

[0024] Preferably, the energy storage capacitor's predicted charge at future times... Calculated using the following formula:

[0025] ;

[0026] In the formula, This represents the actual charge level of the energy storage capacitor at the current moment. For regenerative braking power; Predict power for the load;

[0027] The predicted cumulative number of actions of the variable stiffness elastomer Calculated using the following formula:

[0028] ;

[0029] In the formula, This represents the actual cumulative number of actions of the variable stiffness elastic body at the current moment. This is the fatigue accumulation coefficient; The proportion of mechanical absorption; Impact energy; Indicates the rate of change of impact energy;

[0030] The predicted temperature of the regenerative braking circuit Calculated using the following formula:

[0031] ;

[0032] In the formula, The actual temperature of the regenerative braking circuit at the current moment; This is the heating function; The proportion of electrical absorption; For regenerative braking current; The heat dissipation coefficient; The ambient temperature.

[0033] Preferably, the rolling optimization module is further configured to:

[0034] When the predicted power consumption will reach the first upper limit within the prediction time domain, the electrical absorption ratio is forcibly reduced in the latter part of the prediction time domain.

[0035] When the predicted cumulative number of actions will reach the second upper limit within the prediction time domain, the mechanical absorption ratio will be forcibly reduced in the latter part of the prediction time domain.

[0036] When the predicted temperature will reach the third upper limit within the prediction time domain, the electrical absorption ratio will be forcibly reduced to zero in the latter part of the prediction time domain.

[0037] The constraints in the prediction time domain are expressed as follows:

[0038] ;

[0039] In the formula, To predict the number of control cycles in the time domain; This indicates the first upper limit of the energy storage capacitor's capacity; This represents the second upper limit of the cumulative number of actions of a variable stiffness elastic body; This indicates the third upper limit of the temperature for the regenerative braking circuit.

[0040] Preferably, when any predicted value is about to reach its corresponding upper limit within the prediction time domain, in the latter part of the prediction time domain... Forced adjustment of allocation ratios, among which The starting control cycle number for mandatory adjustment is represented by the following adjustment method:

[0041] , ;

[0042] , ;

[0043] In the formula, and The electrical absorption ratios before and after the forced adjustment are respectively; and The percentages of mechanical absorption before and after the forced adjustment are respectively; This is a forced reduction factor for the proportion of electrical absorption, applied when the predicted electricity consumption is about to reach its upper limit. When the predicted temperature will reach its upper limit ; This is a forced reduction coefficient for the proportion of mechanical absorption, applied when the predicted cumulative number of actions will reach its upper limit. .

[0044] Preferably, the regenerative braking circuit includes:

[0045] A three-phase inverter bridge is connected to the motor;

[0046] The braking resistor is connected to the DC bus of the three-phase inverter bridge via a switching device;

[0047] A bidirectional DC-DC converter is connected between the DC bus and the energy storage capacitor;

[0048] A current reference generation unit, whose input terminal is connected to the output terminal of the energy distribution unit, is used to generate a current reference value based on the electrical absorption target value.

[0049] A current loop controller, whose input is connected to the current reference generation unit and the current shunt, and whose output is connected to the three-phase inverter bridge, is used to make the regenerative braking current follow the current reference value.

[0050] The electrical absorption target value is used to adjust the duty cycle of the bidirectional DC-DC converter to control the proportion of regenerative energy fed back to the energy storage capacitor.

[0051] Preferably, the central controller is configured to determine the mechanical absorption target value and the electrical absorption target value for the current control cycle in the following manner:

[0052] Within the prediction time domain, the mechanical absorption ratio sequence and the electrical absorption ratio sequence are used as optimization variables, and the optimal allocation sequence within the prediction time domain is solved with the goal of minimizing the overall loss.

[0053] Extract the mechanical absorption ratio and electrical absorption ratio of the first control cycle in the optimal allocation sequence, and use them as the mechanical absorption target value and the electrical absorption target value;

[0054] In the next control cycle, the optimal allocation sequence is re-solved based on the updated operating parameters.

[0055] Preferably, the optimal allocation sequence is obtained by solving the following rolling optimization problem:

[0056] ;

[0057] The constraints are:

[0058] ;

[0059] In the formula, The objective function is... For control cycle number; , , These are the weighting coefficients; This is the mechanical absorption loss function; This is the electrical absorption loss function; For the risk function of capacitor overcharging;

[0060] The mechanical absorption target value and electrical absorption target value for the current control cycle are expressed as follows:

[0061] ;

[0062] In the formula, , These represent the mechanical absorption ratio and the electrical absorption ratio in the first control cycle of the optimal allocation sequence, respectively. , These are the mechanical absorption target value and the electrical absorption target value for the current control cycle, respectively.

[0063] Compared with the prior art, the beneficial effects of the present invention are:

[0064] 1. This invention achieves proactive management and path optimization of impact energy by constructing a collaborative control architecture that dynamically allocates the ratio of mechanical absorption to electrical recovery within the prediction time domain, effectively solving the problem of hardware damage caused by reactive control in existing technologies. Specifically, the central controller in this invention acquires key operating parameters in real time, such as the energy storage capacitor charge, the cumulative number of actions of the variable stiffness elastomer, and the temperature of the regenerative braking circuit, and generates mechanical absorption ratio sequences and electrical absorption ratio sequences within the prediction time domain. This prediction mechanism enables the control system to anticipate future energy input trends and constraint boundaries, proactively adjusting the allocation strategy before constraint violations occur. Compared to traditional control methods that only passively respond after a fault occurs, this invention achieves "predictive absorption" of impact energy by solving the optimal allocation target value for the current control cycle through rolling optimization. When it is predicted that the capacitor charge will reach its upper limit, the system reduces the electrical absorption ratio in advance to avoid overcharging damage; when it is predicted that the elastomer fatigue accumulation will reach its upper limit, the system reduces the mechanical absorption ratio in advance to extend the elastomer's lifespan; when it is predicted that the actuator temperature will reach its upper limit, the system disables regenerative braking in advance to prevent thermal failure. This predictive collaborative control mechanism fundamentally solves the hardware damage problem caused by response lag in traditional solutions, significantly improving the operational reliability and service life of the joint drive system.

[0065] 2. This invention also achieves a dynamic balance between energy efficiency and hardware protection by constructing a comprehensive loss model that includes elastomer hysteresis loss, regenerative braking conversion loss, and capacitor overcharge risk loss, thus solving the problem of low energy utilization efficiency caused by a single protection strategy. Specifically, the central controller in this invention aims to minimize the comprehensive loss in the prediction time domain, incorporating elastomer hysteresis loss, regenerative braking conversion loss, and capacitor overcharge risk loss into a unified optimization framework. Elastomer hysteresis loss reflects the energy loss characteristics of the mechanical absorption path, regenerative braking conversion loss reflects the efficiency characteristics of the electrical recovery path, and capacitor overcharge risk loss reflects the safety risk when the energy storage element approaches saturation. By adjusting the weighting coefficients, the system can dynamically adjust its optimization preferences according to real-time operating conditions: when the capacitor has sufficient charge, the system prioritizes reducing the proportion of mechanical absorption to reduce elastomer hysteresis loss and improve energy recovery efficiency; when the elastomer has accumulated significant fatigue, the system prioritizes reducing the proportion of electrical absorption to reduce regenerative braking conversion loss and extend the elastomer's lifespan; when the capacitor is close to saturation, the system prioritizes increasing the weight of overcharge risk loss, actively reducing the electrical recovery ratio, and prioritizing the use of mechanical absorption or energy discharge. This multi-objective collaborative optimization mechanism enables the system to dynamically pursue the optimal balance of energy utilization efficiency while ensuring hardware security, overcoming the inherent contradiction of "emphasizing protection over energy efficiency" or "emphasizing energy efficiency over protection" in traditional solutions.

[0066] 3. This invention also achieves adaptive management of long-term evolution such as elastomer fatigue, capacitor aging, and actuator thermal state by constructing a dynamic time-varying constraint prediction model and a rolling optimization framework, solving the problem of long-term performance degradation caused by parameter fixation in traditional control systems. Specifically, the constraint prediction module in this invention generates the predicted charge of the energy storage capacitor at future times based on the historical charging power and the predicted power of the future load; it generates the predicted cumulative number of actions of the elastomer based on the historical compression count and mechanical absorption ratio of the variable stiffness elastomer; and it generates the predicted temperature of the regenerative braking circuit based on the historical regenerative current and heat dissipation conditions. The above prediction results are not only used for allocation decisions in the current control cycle, but are also updated cycle by cycle under the rolling optimization framework, enabling the control system to perceive the long-term evolution trends such as elastomer fatigue accumulation, capacitor aging, and actuator thermal aging in real time, and dynamically adjust the constraint boundaries and optimization weights of the allocation strategy accordingly. When the predicted cumulative number of actions approaches the upper limit of the lifespan, the system actively reduces the mechanical absorption ratio, allocating more energy to the electrical recovery path to delay elastomer fatigue; when the predicted temperature remains high, the system reduces the electrical absorption ratio in advance to reduce regenerative braking heat generation and protect power devices. Furthermore, this predictive data can provide a basis for predictive maintenance decisions, enabling the system to issue early warnings before the end of the hardware's lifespan. This control architecture, with its long-term evolutionary and adaptive capabilities, fundamentally solves the problem of long-term performance degradation and reliability decline in traditional control systems due to fixed parameters and a lack of foresight, providing technical assurance for the stable operation of robot joint drive systems throughout their entire lifecycle. Attached Figure Description

[0067] Figure 1 This is a schematic diagram of the overall system framework of the present invention;

[0068] Figure 2 This is a schematic diagram of the modular joint module structure of the present invention;

[0069] Figure 3 This is a schematic diagram of the overall process of the core servo control method of the present invention;

[0070] Figure 4 This is a schematic diagram of the working process of the energy distribution unit of the present invention;

[0071] Figure 5 This is a detailed flowchart illustrating the constraint prediction and rolling optimization process of the present invention. Detailed Implementation

[0072] like Figures 1 to 5 As shown, the present invention relates to a modular integration and servo control system for a robot joint actuator, comprising:

[0073] A modular joint module includes a motor, a variable stiffness elastomer, a regenerative braking circuit, and an energy storage capacitor; wherein the variable stiffness elastomer includes an elastomer and an adjustment mechanism acting on the elastomer.

[0074] In an embodiment of the present invention, the modular joint module further includes:

[0075] The joint housing, the motor, the variable stiffness elastomer, the regenerative braking circuit and the energy storage capacitor are integrated inside the joint housing;

[0076] An input interface, located at the first end of the joint housing, is used to connect to the upper-level robotic arm or robot torso.

[0077] An output interface, located at the second end of the joint housing, is used to connect to a lower-level robotic arm or end effector.

[0078] The energy distribution unit is integrated on the power circuit board of the regenerative braking circuit.

[0079] In an embodiment of the present invention, the adjusting mechanism includes:

[0080] A locking component acts on the elastic body to enable the elastic body to exhibit rigid transmission in the locked state and to enable the elastic body to participate in transmission in the released state.

[0081] The locking assembly includes an electromagnetic brake or a piezoelectric brake; the stroke adjustment assembly includes an electric push rod or a shape memory alloy actuator.

[0082] A stroke adjustment assembly, connected to the elastomer, is used to adjust the maximum compression stroke of the elastomer according to the mechanical absorption target value.

[0083] In an embodiment of the present invention, the regenerative braking circuit includes:

[0084] A three-phase inverter bridge is connected to the motor;

[0085] The braking resistor is connected to the DC bus of the three-phase inverter bridge via a switching device;

[0086] A bidirectional DC-DC converter is connected between the DC bus and the energy storage capacitor;

[0087] A current reference generation unit, whose input terminal is connected to the output terminal of the energy distribution unit, is used to generate a current reference value based on the electrical absorption target value.

[0088] A current loop controller, whose input is connected to the current reference generation unit and the current shunt, and whose output is connected to the three-phase inverter bridge, is used to make the regenerative braking current follow the current reference value.

[0089] The electrical absorption target value is used to adjust the duty cycle of the bidirectional DC-DC converter to control the proportion of regenerative energy fed back to the energy storage capacitor.

[0090] An energy distribution unit, which is connected to the regenerative braking circuit and the regulating mechanism;

[0091] The energy distribution unit is connected to the regenerative braking circuit and the adjustment mechanism by hard wiring, and the energy distribution unit is connected to the central controller by bus communication.

[0092] In an embodiment of the present invention, the energy distribution unit includes:

[0093] A programmable logic array that integrates a state machine module and a lookup table module;

[0094] The programmable logic array further includes a hardware PID control module. The input of the hardware PID control module is connected to the voltage detection terminal of the energy storage capacitor, and the output terminal is connected to the current reference terminal of the regenerative braking circuit. It is used to dynamically correct the electrical absorption target value based on the deviation between the voltage of the energy storage capacitor and the target voltage.

[0095] A current shunt is connected between the regenerative braking circuit and the motor to detect the regenerative braking current and the motor drive current.

[0096] A hardware comparator, whose input is connected to the voltage detection terminal of the energy storage capacitor and the temperature detection terminal of the regenerative braking circuit, is used to output an interrupt signal to the programmable logic array when the voltage of the energy storage capacitor or the temperature of the regenerative braking circuit exceeds a preset threshold.

[0097] In response to the interrupt signal, the programmable logic array forcibly adjusts the mechanical absorption target value and the electrical absorption target value.

[0098] In this embodiment, the energy distribution unit and the central controller communicate via a CAN bus. The central controller performs rolling optimization every 10ms, sending the mechanical absorption target value and electrical absorption target value of the first control cycle in the calculated optimal distribution sequence to the energy distribution unit. After receiving the target value, the energy distribution unit writes it into its internal register and independently executes hardware-level distribution control every 1ms: a hardware comparator monitors the voltage of the energy storage capacitor and the temperature of the regenerative braking circuit in real time. When the detected value exceeds a preset threshold, an interrupt signal is sent to the programmable logic array within 100μs. The programmable logic array immediately overwrites the target value in the current register, forcibly adjusting the distribution ratio. Simultaneously, a watchdog circuit is set between the energy distribution unit and the central controller. When three consecutive communication timeouts occur (i.e., no instruction is received from the central controller for more than 50ms), the energy distribution unit automatically switches to independent operation mode, executes control according to the locally stored safety distribution strategy, and sends a fault signal to the central controller.

[0099] A central controller, which is communicatively connected to the energy distribution unit;

[0100] In an embodiment of the present invention, the central controller includes:

[0101] The constraint prediction module is configured to generate a predicted charge of the energy storage capacitor at a future time based on the historical charging power and predicted future load power of the energy storage capacitor; generate a predicted cumulative number of actions of the variable stiffness elastomer based on the historical compression count and mechanical absorption ratio of the variable stiffness elastomer; and generate a predicted temperature of the regenerative braking circuit based on the historical regenerative current and heat dissipation conditions of the regenerative braking circuit. The mechanical absorption ratio is the proportion of impact energy absorbed by the variable stiffness elastomer, ranging from 0 to 1; the electrical absorption ratio is the proportion of impact energy recovered by the regenerative braking circuit, also ranging from 0 to 1; the sum of these two ratios is always equal to 1, together forming a complete impact energy distribution path.

[0102] The rolling optimization module is configured to solve for the mechanical absorption ratio sequence and the electrical absorption ratio sequence within the prediction time domain, under the constraints that the predicted power consumption does not exceed a first upper limit, the predicted cumulative number of actions does not exceed a second upper limit, and the predicted temperature does not exceed a third upper limit. The prediction time domain refers to the number of control cycles predicted backward from the current moment. The corresponding time length, the length of each control cycle is In this embodiment, Take 1ms, The value is 10, meaning the prediction time domain length is 10ms.

[0103] The energy storage capacitor at a future time The predicted power consumption is calculated using the following formula:

[0104] ;

[0105] In the formula, For energy storage capacitors in the future Predicted power consumption; This represents the actual charge level of the energy storage capacitor at the current moment. Regenerative braking power refers to the instantaneous power fed back to the capacitor through the regenerative braking circuit; Forecasted load power refers to the predicted instantaneous power consumed by motor drive or subsequent impacts at future moments. In this embodiment, the central controller receives the future motion trajectory sent by the robot's host computer and calculates it using a feedforward calculation based on the joint dynamics model; when the future motion trajectory cannot be obtained, an autoregressive moving average model is used to predict the historical load power over time.

[0106] In this implementation, the predicted power of the future load The data is obtained as follows: The central controller receives the future motion trajectory (including joint angles, angular velocities, and angular acceleration sequences) from the robot's host computer. Combining this with the joint dynamics model, it calculates the driving torque and driving power at each future moment using a feedforward method. This feedforward calculation cycle is synchronized with the prediction time domain, updating every millisecond. When the future motion trajectory is unavailable, an autoregressive moving average model is used to predict the historical load power over time to maintain the continuity of the prediction.

[0107] The predicted cumulative number of actions of the variable stiffness elastic body is calculated using the following formula:

[0108] ;

[0109] In the formula, This indicates that the variable stiffness elastic body will change in the future. The predicted cumulative number of actions; This represents the actual cumulative number of actions of the variable stiffness elastic body at the current moment. The fatigue accumulation coefficient is the proportion of mechanical absorption. Impact energy The function characterizes the contribution of unit impact energy to the fatigue of a variable stiffness elastic body. The dimensions are , The dimension of is , and the dimension of the product is . After integration, we obtain a dimensionless number; The impact energy change rate represents the impact energy input to the joint per unit time. This variable reflects the instantaneous intensity of the impact energy, enabling fatigue accumulation prediction to distinguish the differential effects of steep and gentle pulses on the fatigue of the elastomer. The mechanical absorption ratio represents the proportion of impact energy absorbed by the variable stiffness elastic body. The total energy that the finger joints need to absorb during the impact process;

[0110] The predicted temperature of the regenerative braking circuit is calculated using the following formula:

[0111] ;

[0112] In the formula, For regenerative braking circuits at future moments The predicted temperature; The actual temperature of the regenerative braking circuit at the current moment; The heating function is the proportion of electrical absorption. With regenerative braking current The function characterizes the heat power generated during the regenerative braking process; The electrical absorption ratio indicates the proportion of impact energy recovered by the regenerative braking circuit. The regenerative braking current is the instantaneous current flowing through the regenerative braking circuit. The heat dissipation coefficient characterizes the efficiency of the regenerative braking circuit in dissipating heat to the environment. Ambient temperature;

[0113] Algorithm Formula Operation Logic: The three algorithm formulas mentioned above are used to predict three key state variables at future moments: the energy storage capacitor's charge, the cumulative number of actions of the variable stiffness elastomer, and the temperature of the regenerative braking circuit. These three predictions form the basis for subsequent rolling optimization and embody the core idea of ​​"dynamic time-varying constraints." The first formula predicts the energy storage capacitor's charge at future moments by integrating the net power flow (regenerative braking power minus load power consumption) between the current moment and the future moment. A positive regenerative braking power indicates charging of the capacitor, while a positive load power consumption indicates capacitor discharge; the algebraic sum of these two factors determines the trend of the capacitor's charge change. The second formula predicts the cumulative number of actions of the variable stiffness elastomer at future moments by integrating the product of the fatigue accumulation coefficient and the impact energy change rate. Fatigue accumulation coefficient. The proportion of mechanical absorption Impact energy The function reflects the nonlinear characteristics of fatigue accumulation in variable stiffness elastic bodies under different working conditions; the rate of change of impact energy. The first formula represents the impact energy input to the joint per unit time. Integrating the product of the first and second formulas yields the total number of fatigue accumulation increments, directly linking the prediction result to the instantaneous intensity of the impact energy. The third formula predicts the future temperature of the regenerative braking circuit by using the difference between the integral heating term and the heat dissipation term. (Heating term) With electrical absorption ratio and regenerative braking current Relatedly, the heat dissipation term is based on Newton's law of cooling, representing natural heat dissipation when the temperature is higher than the ambient temperature. Through these three prediction formulas, this system achieves proactive perception of key state variables, enabling energy allocation decisions to shift from reactive control based on the current instantaneous state to predictive control based on future evolutionary trends. This predictive capability allows the control system to proactively adjust its allocation strategy before constraint violations occur, avoiding hardware damage events such as capacitor overcharging, fatigue exceeding limits of variable stiffness elastomers, and driver overheating, significantly improving the long-term operational reliability and safety of the system.

[0114] In this embodiment, the fatigue accumulation coefficient Defined as:

[0115] ;

[0116] in, The baseline fatigue cumulative factor represents the cumulative fatigue energy at a reference impact energy level. Lower, mechanical absorption ratio The cumulative fatigue increment corresponding to a unit of impact energy per hour; For reference impact energy, It is a nonlinear exponent used to describe the accelerating effect of high-energy impacts on fatigue. and All were calibrated through fatigue tests on elastomer materials. When the time is right, it indicates that the acceleration effect of high-energy impact on fatigue is more significant.

[0117] Heating function Defined as:

[0118] ;

[0119] in, The equivalent thermal resistance coefficient of a power device characterizes the heat power generated per square unit regenerative braking current, and its dimensions are: The temperature rise rate of power devices can be measured under a known regenerative braking current through thermal simulation or experimental calibration, and obtained by fitting the data. The optimal estimate is given by the expression, which is based on Joule's law and reflects the linear relationship between the heat power generated by the regenerative braking current and the proportion of electrical absorption.

[0120] Baseline fatigue cumulative factor With nonlinear exponent Calibration was achieved by subjecting elastomer specimens to cyclic loading tests at different impact energy levels, recording the relationship curves between fatigue life and impact energy, and the proportion of mechanical absorption, and then obtaining the results through nonlinear regression fitting. and The optimal estimate. When the elastomer material is rubber. The value range is typically 1.2 to 2.0; when the elastic material is a metal spring, The value range is usually 1.0 to 1.5.

[0121] In an embodiment of the present invention, the rolling optimization module is further configured to:

[0122] When the predicted power consumption will reach the first upper limit within the prediction time domain, the electrical absorption ratio is forcibly reduced in the latter part of the prediction time domain.

[0123] When the predicted cumulative number of actions will reach the second upper limit within the prediction time domain, the mechanical absorption ratio will be forcibly reduced in the latter part of the prediction time domain.

[0124] When the predicted temperature will reach the third upper limit within the prediction time domain, the electrical absorption ratio will be forcibly reduced to zero in the latter part of the prediction time domain.

[0125] The constraints in the prediction time domain are expressed as follows:

[0126] ;

[0127] In the formula, The number of control cycles in the prediction time domain indicates the time range for forward prediction; This indicates the first upper limit of the energy storage capacitor's charge capacity; exceeding this value may damage the capacitor. This represents the second upper limit of the cumulative number of actions of a variable stiffness elastic body; exceeding this value may lead to fatigue fracture. This indicates the third upper limit of the temperature for the regenerative braking circuit; exceeding this value may cause thermal failure of power devices.

[0128] When any predicted value is about to reach its corresponding upper limit within the prediction time domain, in the latter part of the prediction time domain... Forced adjustment of allocation ratios, among which The starting control cycle number for mandatory adjustment indicates the moment in the prediction time domain when intervention begins, and the adjustment method is expressed as follows:

[0129] , ;

[0130] In the formula, and The electrical absorption ratios before and after the forced adjustment are respectively, and This is a temporarily adjusted variable, and the adjustment result will be used as an input constraint or initial value for subsequent rolling optimization, without affecting the optimization variable. The original definition; This is a forced reduction factor for the proportion of electrical absorption, with a value range of... The smaller the value, the greater the decrease; when the predicted power consumption is about to reach its limit... When the predicted temperature will reach its upper limit ;

[0131] , ;

[0132] In the formula, and The percentages of mechanical absorption before and after the forced adjustment are respectively, and This is a temporarily adjusted variable, and the adjustment result will be used as an input constraint or initial value for subsequent rolling optimization, without affecting the optimization variable. The original definition; This is a forced reduction coefficient for the proportion of mechanical absorption, with a value range of... The smaller the value, the greater the decrease; when the predicted cumulative number of actions is about to reach the upper limit... .

[0133] Algorithm Formula Operation Logic: Based on the aforementioned prediction results, this section defines the specific expressions of the constraints and the mandatory adjustment mechanism when constraints are violated. The first set of formulas compares the three predicted state variables with their corresponding safety upper limits, requiring that the constraints be satisfied in every control cycle within the prediction time domain. This is the core constraint of the rolling optimization problem, reflecting the mandatory enforcement characteristic of "dynamic time-varying constraints". The second formula indicates that when the predicted electricity or predicted temperature is about to reach the upper limit, the electrical absorption ratio is forcibly reduced in the latter part of the prediction time domain. Forced reduction of coefficient The value is less than 1 and can be dynamically adjusted based on how close it is to the upper limit; the closer it is to the upper limit, the greater the reduction. The third formula indicates that when the predicted cumulative number of actions is about to reach the upper limit, the mechanical absorption ratio is forcibly reduced in the latter part of the prediction time domain. Similarly, through coefficients Proportional adjustment is achieved. Specifically, when the predicted temperature is about to reach its upper limit, the electrical absorption ratio is forcibly reduced to zero, effectively disabling regenerative braking. This is the most stringent protection measure, as excessively high temperatures can damage power devices. This section transforms the safety boundary of the physical system into mathematical constraints and designs a graded response forced adjustment mechanism, realizing a complete control closed loop from "soft constraint warning" to "hard constraint protection." When a constraint violation is predicted, the system does not wait for the violation to occur before responding passively, but actively adjusts in advance within the prediction time domain. This "predictive protection" mechanism effectively avoids hardware damage caused by response delays, while graded adjustment (first reducing the proportional ratio, then forcibly reducing it to zero) ensures control smoothness and avoids system oscillations caused by sudden changes.

[0134] In this embodiment, the specific method for determining the forced adjustment coefficient is as follows:

[0135] Let the first The predicted power consumption for each control cycle is The maximum capacity of the capacitor is Define the proximity of battery levels .when When this happens, a forced reduction mechanism is activated, forcibly reducing the coefficient. Determine using the following formula:

[0136] ;

[0137] That is, when the predicted power consumption reaches 80% of the upper limit, the electrical absorption ratio begins to decrease linearly, and when the upper limit is reached, it is completely disabled. Similarly, for the predicted cumulative number of actions and the predicted temperature, the corresponding proximity is defined using the same principle. and And determine the corresponding forced reduction coefficient. The condition for the electrical absorption ratio to return to zero.

[0138] The central controller is configured as follows:

[0139] The operating parameters of the joint module are obtained, including the charge of the energy storage capacitor, the cumulative number of actions of the variable stiffness elastomer, and the temperature of the regenerative braking circuit.

[0140] Based on the operating parameters, a mechanical absorption ratio sequence of the variable stiffness elastomer and an electrical absorption ratio sequence of the regenerative braking circuit are generated in the prediction time domain, wherein the sum of the mechanical absorption ratio and the electrical absorption ratio is one.

[0141] With the goal of minimizing the overall loss in the prediction time domain, the target values ​​for mechanical absorption and electrical absorption in the current control cycle are determined from the mechanical absorption ratio sequence and the electrical absorption ratio sequence. The overall loss refers to the weighted sum of the elastic hysteresis loss generated by the mechanical absorption path, the regenerative braking conversion loss generated by the electrical recovery path, and the overcharge risk loss of the energy storage capacitor due to the increase in charge within a control cycle. Its specific expression will be given in detail later.

[0142] The mechanical absorption target value is sent to the regulating mechanism, and the electrical absorption target value is sent to the regenerative braking circuit.

[0143] In an embodiment of the present invention, a watchdog circuit is provided between the energy distribution unit and the central controller. When the communication interruption between the energy distribution unit and the central controller lasts for more than a communication threshold, the energy distribution unit enters an independent operation mode. In the independent operation mode, the energy distribution unit executes a preset safety distribution strategy based on the charge of the energy storage capacitor and the temperature of the regenerative braking circuit.

[0144] The safety allocation strategy is a preset conservative control scheme: when the energy allocation unit enters independent operation mode, based on the current energy storage capacitor charge and regenerative braking circuit temperature, the following rules are executed: if the capacitor charge is lower than the safety threshold, electrical recovery is given priority. If the capacitor charge is higher than the safety threshold, mechanical absorption will be used preferentially. If the temperature exceeds a preset threshold, regenerative braking will be completely disabled. The impact energy is absorbed solely by the elastomer.

[0145] In an embodiment of the present invention, the central controller is further configured as follows:

[0146] Obtain operating parameters for multiple joint modules;

[0147] Based on the operating parameters of the multiple joint modules, a global optimization model is established, with the goal of minimizing the overall loss of the multiple joint modules.

[0148] Based on the solution results of the global optimization model, the mechanical absorption target value and electrical absorption target value of each joint are issued to the energy distribution unit of each joint module.

[0149] In an embodiment of the present invention, a shared DC bus is also included, and the energy storage capacitors of each joint module are connected to the shared DC bus through a bidirectional DC-DC converter; the central controller is also configured to control the bidirectional DC-DC converter to transfer energy between the plurality of joint modules according to the solution results of the global optimization model.

[0150] In an embodiment of the present invention, the system further includes:

[0151] An encoder, which is mounted on the motor, is used to detect the rotor position and speed of the motor;

[0152] A torque sensor, which is located at the output end of the modular joint module, is used to detect the joint output torque;

[0153] The central controller is also configured to adjust the mechanical absorption target value and the electrical absorption target value based on the rotor position, the rotational speed and the output torque.

[0154] In an embodiment of the present invention, the central controller is configured to determine the mechanical absorption target value and the electrical absorption target value of the current control cycle in the following manner:

[0155] Within the prediction time domain, the mechanical absorption ratio sequence and the electrical absorption ratio sequence are used as optimization variables, and the optimal allocation sequence within the prediction time domain is solved with the goal of minimizing the overall loss.

[0156] Extract the mechanical absorption ratio and electrical absorption ratio of the first control cycle in the optimal allocation sequence, and use them as the mechanical absorption target value and the electrical absorption target value;

[0157] In the next control cycle, the optimal allocation sequence is re-solved based on the updated operating parameters.

[0158] The optimal allocation sequence is obtained by solving the following rolling optimization problem:

[0159] ;

[0160] The constraints are:

[0161] ;

[0162] In the formula, Let be the objective function, representing the total overall loss in the prediction time domain; To control the cycle number, the value ranges from 1 to... ; , , These are weighting coefficients used to balance the importance of mechanical losses, electrical losses, and overcharging risk in the objective function; For the first Mechanical absorption loss function for each control cycle; For the first Electrical absorption loss function for each control cycle; For the first Capacitor overcharge risk function for each control cycle;

[0163] The mechanical absorption target value and electrical absorption target value for the current control cycle are expressed as follows:

[0164] ;

[0165] In the formula, The proportion of mechanical absorption in the first control cycle of the optimal allocation sequence; The electrical absorption ratio for the first control cycle in the optimal allocation sequence; The target value for mechanical absorption in the current control cycle; This is the target value for electrical absorption in the current control cycle.

[0166] In this embodiment, the weighting coefficient , , The capacity can be dynamically adjusted according to the system's operating phase: during the initial startup phase, when the capacitor charge is low, it can be increased appropriately. (Prioritize reducing mechanical losses and improving energy recovery efficiency); in the later stages of system operation, when the elastomer accumulates significant fatigue, the [efficiency / efficiency] can be appropriately increased. (Prioritize reducing electrical losses and extending the lifespan of the elastomer); when the capacitor is close to saturation, the capacitance can be appropriately increased. (Focus on preventing overcharging risks). The weighting coefficients can be adjusted using a preset fuzzy rule table or an adaptive algorithm.

[0167] Algorithm Formula Calculation Logic: This section defines the core algorithm of rolling optimization, integrating the aforementioned prediction model and constraints into a unified optimization problem, and clarifying the solution and execution methods. The first set of formulas defines the objective function and constraints of the rolling optimization problem. Objective Function It is a weighted sum of the comprehensive losses of each control cycle within the prediction time domain, where the weighting coefficients are... , , This reflects the relative importance of different losses in the optimization. Constraints include the sum of the allocation proportions being one, the range of values, and the state variable constraints defined above. The decision variable for the optimization problem is the proportion of mechanical absorption in each control cycle within the prediction time domain. With electrical absorption ratio The optimal allocation sequence is obtained by minimizing the objective function under constraints. The second formula indicates that the allocation ratio of the first control cycle is extracted from the optimal allocation sequence as the execution instruction for the current control cycle. This is a typical strategy of model predictive control—rolling optimization and successive implementation. The third formula indicates that in the next control cycle, the above optimization process is repeated based on the updated operating parameters, reflecting the rolling characteristic of control. This section transforms the energy allocation problem into a constrained dynamic optimization problem through the rolling optimization framework, realizing closed-loop control of "prediction-optimization-execution-re-prediction". Compared with traditional single-step optimization, rolling optimization can anticipate the changes in operating conditions for multiple future control cycles, avoiding global performance degradation caused by local optima. Through a weighted objective function, the system can dynamically adjust optimization preferences according to different operating conditions: prioritizing the reduction of mechanical losses when the capacitor has sufficient charge, prioritizing the reduction of electrical losses when the variable stiffness elastic body has accumulated a lot of fatigue, and focusing on preventing overcharging risks when the capacitor is close to saturation. This flexible multi-objective optimization capability enables the system to achieve a dynamic balance between hardware protection, energy efficiency, and control accuracy.

[0168] In this embodiment, since the objective function is linear, the constraints are linear inequalities and equality equations, and the decision variables are linear... and Since all variables are scalars, this rolling optimization problem can be solved using linear programming in milliseconds. Specifically, the prediction time domain... The decision variables within are expanded as follows There are scalar variables, the objective function is a linear weighted sum, and the constraints include equality constraints. Inequality constraints And predictive state constraints. This linear programming problem can be solved in real time in an embedded controller using the simplex method or interior-point method. When predicting in the time domain... When the value is greater than 5, in order to reduce the computational burden, a rolling time-domain contraction strategy can be adopted, that is, the prediction constraint is only used as a hard constraint in the three periods around the current control period, and only the allocation ratio and the range of values ​​are retained in subsequent periods.

[0169] In an embodiment of the present invention, the overall loss includes:

[0170] The hysteresis loss of the variable stiffness elastomer increases with the increase of the mechanical absorption ratio;

[0171] The energy conversion loss of the regenerative braking circuit increases with the increase of the electrical absorption ratio;

[0172] The risk of overcharging loss of the energy storage capacitor increases as the charge of the energy storage capacitor increases.

[0173] The expression for the overall loss in a single control cycle is:

[0174] ;

[0175] In the formula, For the first Total comprehensive loss for each control cycle; For the first Hysteresis loss of variable stiffness elastomer per control cycle; For the first Energy conversion loss of regenerative braking circuit per control cycle; For the first The risk of overcharging loss of the energy storage capacitor in each control cycle;

[0176] Among them, the Hysteresis loss of variable stiffness elastomer per control cycle The calculation formula is:

[0177] ;

[0178] In the formula, The hysteresis loss coefficient of a variable stiffness elastomer characterizes the proportion of energy loss of a variable stiffness elastomer material during compression and rebound. The proportion of mechanical absorption; Indicates the first The cumulative value of impact energy within each control cycle, i.e. , and continuous-time expression Corresponding through integral relationships;

[0179] Among them, the Energy conversion loss of regenerative braking circuit per control cycle The calculation formula is:

[0180] ;

[0181] In the formula, Regenerative braking efficiency is the efficiency by which a regenerative braking circuit converts mechanical energy into electrical energy.

[0182] Among them, the Overcharge risk loss of energy storage capacitor in each control cycle The calculation formula is:

[0183] ;

[0184] In the formula, This is the overfill risk coefficient, used to adjust the weight of overfill risk in the objective function; This represents the current charge level of the energy storage capacitor. This is the safe charge threshold for the capacitor; below this value, the risk of overcharging is negligible. This is an overcharge risk factor. It is zero when the capacitor charge is below the safety threshold, and increases with the square of the charge when it is above the threshold.

[0185] Algorithm Formula Calculation Logic: This section defines the specific expressions for each component of the overall loss, which is an expansion and refinement of the objective function mentioned above. The first formula decomposes the total overall loss into three parts: variable stiffness elastic body hysteresis loss, regenerative braking conversion loss, and capacitor overcharge risk loss. These three parts correspond to the aforementioned objective function. , , The second formula states that the hysteresis loss of a variable stiffness elastic body equals the hysteresis loss coefficient multiplied by the mechanical absorption ratio multiplied by the impact energy. Hysteresis loss is the energy lost by a variable stiffness elastic body due to internal friction during compression-rebound, and is proportional to the absorbed energy. The third formula states that the regenerative braking conversion loss equals the reciprocal of the conversion efficiency minus one multiplied by the electrical absorption ratio multiplied by the impact energy. Conversion efficiency The value is less than 1, therefore the loss term is positive, and the lower the efficiency and the more energy recovered, the greater the loss. The fourth formula states that the capacitor overcharge risk loss equals the overcharge risk coefficient multiplied by the square of the capacitor charge exceeding the safety threshold. This expression uses a square term, making the risk increase faster as the charge approaches the upper limit, reflecting a conservative control strategy for overcharge risk. This section transforms the abstract "optimal energy allocation" into a quantifiable mathematical optimization problem by establishing an explicit expression for comprehensive losses. The introduction of variable stiffness elastomer hysteresis loss allows the system to automatically weigh energy efficiency when using mechanical absorption, avoiding over-reliance on variable stiffness elastomers; the introduction of regenerative braking conversion loss allows the system to consider efficiency factors during electrical recovery, avoiding forced recovery under low-efficiency conditions; the introduction of capacitor overcharge risk loss allows the system to actively reduce the electrical recovery ratio when the capacitor is close to saturation, prioritizing mechanical absorption or energy dissipation. The combined effect of these three factors enables the system to achieve an optimal balance between energy efficiency, hardware protection, and system reliability. This multi-objective collaborative optimization capability is the core advantage of this technical solution that distinguishes it from traditional single-objective control.

[0186] The embodiments disclosed in this invention are preferred embodiments, but are not limited thereto. Those skilled in the art can easily understand the spirit of this invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of this invention, they are all within the protection scope of this invention.

Claims

1. A modular integration and servo control system for robot joint actuators, characterized in that, include: A modular joint module includes a motor, a variable stiffness elastomer, a regenerative braking circuit, and an energy storage capacitor; wherein the variable stiffness elastomer includes an elastomer and an adjustment mechanism acting on the elastomer. An energy distribution unit, which is connected to the regenerative braking circuit and the regulating mechanism; A central controller, which is communicatively connected to the energy distribution unit, is configured to: The operating parameters of the joint module are obtained, including the charge of the energy storage capacitor, the cumulative number of actions of the variable stiffness elastomer, and the temperature of the regenerative braking circuit. Based on the operating parameters, a mechanical absorption ratio sequence of the variable stiffness elastomer and an electrical absorption ratio sequence of the regenerative braking circuit are generated in the prediction time domain, wherein the sum of the mechanical absorption ratio and the electrical absorption ratio is one. With the goal of minimizing the overall loss in the prediction time domain, the target values ​​for mechanical absorption and electrical absorption in the current control cycle are determined from the mechanical absorption ratio sequence and the electrical absorption ratio sequence. The mechanical absorption target value is sent to the regulating mechanism, and the electrical absorption target value is sent to the regenerative braking circuit.

2. The modular integration and servo control system for a robot joint actuator according to claim 1, characterized in that, The energy distribution unit includes: A programmable logic array that integrates a state machine module and a lookup table module; A current shunt is connected between the regenerative braking circuit and the motor to detect the regenerative braking current and the motor drive current. A hardware comparator, whose input is connected to the voltage detection terminal of the energy storage capacitor and the temperature detection terminal of the regenerative braking circuit, is used to output an interrupt signal to the programmable logic array when the voltage of the energy storage capacitor or the temperature of the regenerative braking circuit exceeds a preset threshold. In response to the interrupt signal, the programmable logic array forcibly adjusts the mechanical absorption target value and the electrical absorption target value.

3. The modular integration and servo control system for a robot joint actuator according to claim 2, characterized in that, The programmable logic array also includes a hardware PID control module. The input of the hardware PID control module is connected to the voltage detection terminal of the energy storage capacitor, and the output is connected to the current reference terminal of the regenerative braking circuit. It is used to dynamically correct the electrical absorption target value according to the deviation between the voltage of the energy storage capacitor and the target voltage.

4. The modular integration and servo control system for a robot joint actuator according to claim 1, characterized in that, The central controller includes: The constrained prediction module is configured to generate a predicted charge level for the energy storage capacitor at a future time based on the capacitor's historical charging power and predicted future load power. Based on the historical compression counts of the variable stiffness elastomer and the mechanical absorption ratio, a predicted cumulative number of actions of the variable stiffness elastomer is generated. Based on the historical regenerative current and heat dissipation conditions of the regenerative braking circuit, a predicted temperature for the regenerative braking circuit is generated. ; The rolling optimization module is configured to solve the mechanical absorption ratio sequence and the electrical absorption ratio sequence within the prediction time domain, with the constraints that the predicted power consumption does not exceed a first upper limit, the predicted cumulative number of actions does not exceed a second upper limit, and the predicted temperature does not exceed a third upper limit.

5. The modular integration and servo control system for a robot joint actuator according to claim 4, characterized in that, The predicted charge of the energy storage capacitor at future times Calculated using the following formula: ； In the formula, This represents the actual charge level of the energy storage capacitor at the current moment. For regenerative braking power; Predict power for the load; The predicted cumulative number of actions of the variable stiffness elastomer Calculated using the following formula: ； In the formula, This represents the actual cumulative number of actions of the variable stiffness elastic body at the current moment. This is the fatigue accumulation coefficient; The proportion of mechanical absorption; Impact energy; Indicates the rate of change of impact energy; The predicted temperature of the regenerative braking circuit Calculated using the following formula: ； In the formula, The actual temperature of the regenerative braking circuit at the current moment; This is the heating function; The proportion of electrical absorption; For regenerative braking current; The heat dissipation coefficient; The ambient temperature.

6. The modular integration and servo control system for a robot joint actuator according to claim 5, characterized in that, The rolling optimization module is also configured to: When the predicted power consumption will reach the first upper limit within the prediction time domain, the electrical absorption ratio is forcibly reduced in the latter part of the prediction time domain. When the predicted cumulative number of actions will reach the second upper limit within the prediction time domain, the mechanical absorption ratio will be forcibly reduced in the latter part of the prediction time domain. When the predicted temperature will reach the third upper limit within the prediction time domain, the electrical absorption ratio will be forcibly reduced to zero in the latter part of the prediction time domain. The constraints in the prediction time domain are expressed as follows: ； In the formula, To predict the number of control cycles in the time domain; This indicates the first upper limit of the energy storage capacitor's capacity; This represents the second upper limit of the cumulative number of actions of a variable stiffness elastic body; This indicates the third upper limit of the temperature for the regenerative braking circuit.

7. The modular integration and servo control system for a robot joint actuator according to claim 6, characterized in that, When any predicted value is about to reach its corresponding upper limit within the prediction time domain, in the latter part of the prediction time domain... Forced adjustment of allocation ratios, among which The starting control cycle number for mandatory adjustment is represented by the following adjustment method: ， ； ， ； In the formula, and The electrical absorption ratios before and after the forced adjustment are respectively; and The percentages of mechanical absorption before and after the forced adjustment are respectively; This is a forced reduction factor for the proportion of electrical absorption, applied when the predicted electricity consumption is about to reach its upper limit. ; When the predicted temperature will reach its upper limit ; This is a forced reduction coefficient for the proportion of mechanical absorption, applied when the predicted cumulative number of actions will reach its upper limit. .

8. The modular integration and servo control system for a robot joint actuator according to claim 2, characterized in that, The regenerative braking circuit includes: A three-phase inverter bridge is connected to the motor; The braking resistor is connected to the DC bus of the three-phase inverter bridge via a switching device; A bidirectional DC-DC converter is connected between the DC bus and the energy storage capacitor; A current reference generation unit, whose input terminal is connected to the output terminal of the energy distribution unit, is used to generate a current reference value based on the electrical absorption target value. A current loop controller, whose input is connected to the current reference generation unit and the current shunt, and whose output is connected to the three-phase inverter bridge, is used to make the regenerative braking current follow the current reference value. The electrical absorption target value is used to adjust the duty cycle of the bidirectional DC-DC converter to control the proportion of regenerative energy fed back to the energy storage capacitor.

9. The modular integration and servo control system for a robot joint actuator according to claim 1, characterized in that, The central controller is configured to determine the mechanical absorption target value and the electrical absorption target value for the current control cycle in the following manner: Within the prediction time domain, the mechanical absorption ratio sequence and the electrical absorption ratio sequence are used as optimization variables, and the optimal allocation sequence within the prediction time domain is solved with the goal of minimizing the overall loss. Extract the mechanical absorption ratio and electrical absorption ratio of the first control cycle in the optimal allocation sequence, and use them as the mechanical absorption target value and the electrical absorption target value; In the next control cycle, the optimal allocation sequence is re-solved based on the updated operating parameters.

10. A modular integration and servo control system for a robot joint actuator according to claim 9, characterized in that, The optimal allocation sequence is obtained by solving the following rolling optimization problem: ； The constraints are: ； In the formula, The objective function is... For control cycle number; , , These are the weighting coefficients; This is the mechanical absorption loss function; This is the electrical absorption loss function; For the risk function of capacitor overcharging; The mechanical absorption target value and electrical absorption target value for the current control cycle are expressed as follows: ； In the formula, , These represent the mechanical absorption ratio and the electrical absorption ratio in the first control cycle of the optimal allocation sequence, respectively. , These are the mechanical absorption target value and the electrical absorption target value for the current control cycle, respectively.

Images