A Method and Device for Feedforward Gas Supply Control of Robot Fuel Cell Based on Motion Intent Analysis and Model Prediction

Abstract

This application discloses a method and device for feedforward gas supply control of a robot fuel cell based on motion intent analysis and model prediction, relating to the interdisciplinary fields of new energy and robot control. This method moves the control trigger source from the underlying physical state feedback to the upper-level motion planning stage. The system analyzes communication control information in real time, interprets the motion intent, and sets a future feedforward time window. Simultaneously, based on the motion intent, the future load prediction sequence generated by dynamic mapping is introduced as a feedforward disturbance into the discrete state-space prediction model. Rolling optimization is performed while satisfying the lower limit constraint of the cathode peroxide ratio, thereby achieving an advanced response of gas supply regulation to high-dynamic load changes. This realizes a fuel cell gas supply method combining active feedforward and model predictive control based on the "information prior—physical lag" time difference, improving the robot's power continuity and the operational stability of the fuel cell system under high-dynamic conditions.

Description

Technical Field

[0001] This application relates to the interdisciplinary fields of new energy and robot control, and in particular to a method and device for feedforward gas supply control of a robot fuel cell based on motion intention analysis and model prediction. Background Technology

[0002] Proton exchange membrane fuel cells (PEMFCs) have shown promising application prospects in mobile robot power systems due to their high energy density and fast refueling speed. However, when these robots perform high-dynamic actions such as jumping, starting, climbing, and obstacle crossing, the load power typically increases rapidly within a short period, causing the fuel cell system to experience significant transient load shocks. If the cathode gas supply cannot keep up with the load changes, insufficient oxygen supply to the cathode can easily occur, leading to a drop in the stack output voltage, increased local concentration polarization, and in severe cases, triggering low-voltage protection or affecting the long-term stable operation of the stack.

[0003] Most existing fuel cell air supply control methods rely on feedback parameters such as stack current, voltage, pressure, and flow rate for adjustment. This means that the air compressor or related actuators are only controlled to increase the air supply after the load has changed. While these methods can maintain system operation under steady-state or general dynamic conditions, they often lag under extreme dynamic robot movements. This is because of the mechanical inertia of the air compressor and the flow establishment process in the air path, making it difficult to promptly match the rapid increase in robot load. Summary of the Invention

[0004] Therefore, it is necessary to provide a robot fuel cell feedforward gas supply control method and device based on motion intention analysis and model prediction to address the above-mentioned technical problems, which can improve the operational stability under high dynamic conditions.

[0005] Firstly, this application provides a method for feedforward gas supply control of a robot fuel cell based on motion intention analysis and model prediction. The method includes:

[0006] Real-time acquisition of robot control information;

[0007] The control information is parsed to identify the type of action within the future feedforward time window. Based on the action type, the load of the corresponding joint actuator of the robot is predicted to obtain the expected load sequence. The power source of the joint actuator includes fuel cells.

[0008] A discrete state-space prediction model for the gas supply actuator is constructed, using the expected load sequence as the feedforward disturbance. An objective function is constructed that integrates the tracking error of the cathode oxygen ratio of the fuel cell, the change in control input, and the parasitic power of the gas supply actuator. With the constraint that the cathode oxygen ratio of the fuel cell should not be lower than the safety lower limit, the objective control quantity is solved by rolling optimization, so that the gas supply actuator adjusts the oxygen supply of the fuel cell according to the objective control quantity.

[0009] In one embodiment, the feedforward time window is determined based on the lumped response delay of the air supply actuator. The lumped response delay is related to the communication and control calculation delay, the response delay to overcome mechanical inertia and complete the acceleration, and the flow delay of air establishing effective pressure and flow rate through the intake pipe and cathode channel.

[0010] In one embodiment, predicting the load on the robot's corresponding joint actuators based on the action type to obtain the expected load sequence includes:

[0011] Based on the action type, extract the expected torque and expected angular velocity of the corresponding joint actuator;

[0012] Based on the power ratio factor of the fuel cell in the joint actuator, the fuel cell stack operating voltage, the total power of auxiliary components, the overall efficiency of the electric drive system, and the desired torque and angular velocity, the expected load sequence is calculated.

[0013] In one embodiment, calculating the expected load sequence includes:

[0014] Based on the expected torque and expected angular velocity of each joint actuator corresponding to the action type, calculate the expected power of each joint actuator, and sum the expected power to obtain the total expected power;

[0015] The power component that the fuel cell needs to output is obtained by multiplying the power ratio coefficient and the expected total power, and then dividing the product by the overall efficiency of the electric drive system.

[0016] The total power output of the fuel cell is obtained by combining the power component required to be output by the fuel cell with the total power of the auxiliary components.

[0017] Divide the total power output required by the fuel cell by the fuel cell stack operating voltage to obtain the expected load sequence.

[0018] In one embodiment, the method further includes:

[0019] The target control quantity is adjusted based on the hydrogen supply from the anode of the fuel cell.

[0020] In one embodiment, the method further includes:

[0021] Real-time acquisition of fuel cell status feedback data, and real-time adjustment of target control variables based on the status feedback data.

[0022] Secondly, this application provides a fuel cell control device, the device comprising:

[0023] The data acquisition module is used to acquire the robot's control information in real time;

[0024] The intent recognition module is used to parse the control information, identify the type of action within the future feedforward time window, predict the load on the corresponding joint actuators of the robot based on the action type, and obtain the expected load sequence; wherein, the power source of the joint actuators includes fuel cells.

[0025] The feedforward MPC algorithm module is used to construct a discrete state-space prediction model of the gas supply actuator with the expected load sequence as the feedforward perturbation. It constructs an objective function that integrates the tracking error of the cathode oxygen ratio of the fuel cell, the change in control input, and the parasitic power of the gas supply actuator. With the constraint that the cathode oxygen ratio of the fuel cell is not lower than the safety lower limit, it continuously optimizes and solves the target control quantity, so that the gas supply actuator can adjust the oxygen supply of the fuel cell according to the target control quantity.

[0026] Thirdly, this application provides a fuel cell power system, which includes: a fuel cell, a gas supply actuator, and a sensor unit;

[0027] The gas supply actuator receives and executes the control command corresponding to the target control quantity obtained in the above method embodiment to adjust the oxygen supply of the fuel cell.

[0028] Fourthly, this application provides a robot, including: a robot platform, a fuel control device, and a fuel cell power system;

[0029] The robot platform includes an upper-level motion controller and a drive and execution module. The upper-level motion controller outputs control information about the drive and execution module.

[0030] The fuel control device is connected to the upper motion controller, executes the steps of the above method embodiment, outputs the target control quantity, and issues the control command corresponding to the target control quantity;

[0031] The fuel cell power system is connected to the fuel control unit to receive and execute control commands to meet the power requirements of the drive and actuation modules.

[0032] The aforementioned robot fuel cell feedforward gas supply control method and device based on motion intent parsing and model prediction shifts the control trigger source from the underlying physical state feedback to the upper-level motion planning stage. The system analyzes communication control information in real time, interprets the motion intent, and sets a future feedforward time window. Simultaneously, based on the motion intent, the future load prediction sequence generated by dynamic mapping is imported as a feedforward perturbation into the discrete state-space prediction model. Rolling optimization is performed while satisfying the lower limit constraint of the cathode oxygen ratio, thereby achieving an advanced response of gas supply regulation to high-dynamic load changes. This realizes a fuel cell gas supply method combining active feedforward and model predictive control based on the "information prior—physical lag" time difference, improving the robot's power continuity and the operational stability of the fuel cell system under high-dynamic conditions. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating a robot fuel cell feedforward gas supply control method based on motion intent parsing and model prediction in one embodiment.

[0034] Figure 2 This is a working logic diagram of a robot fuel cell feedforward gas supply control method based on motion intent parsing and model prediction in one embodiment.

[0035] Figure 3 This is a comparison diagram of the results of the present invention and the traditional feedback air supply method under typical variable load conditions;

[0036] Figure 4 This is a hardware framework diagram for robot fuel cell control in one embodiment. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0038] This application provides a robot fuel cell feedforward gas supply control method based on motion intent parsing and model prediction. The application scenario involves a robot, a fuel cell control device, and a gas supply actuator. The fuel cell control device calculates the fuel cell's oxygen demand and instructs the gas supply actuator to provide the corresponding amount of oxygen, ensuring that the energy generated by the fuel cell meets the robot's activity requirements. The main components of the gas supply actuator are an air compressor and valves. The fuel cell control device regulates the fuel cell's oxygen supply by issuing commands such as air compressor start / stop control, frequency converter control, and intake valve opening.

[0039] like Figure 1 and 2As shown, the method is applied to a control device for a combustion battery and includes the following steps:

[0040] Step S1: Obtain the robot's control information in real time.

[0041] Control information includes the robot's upper-level motion planning instructions, joint states, and fuel cell operating parameters.

[0042] When the robot system is in standby mode, the air compressor maintains idle speed to balance low parasitic work and basic air supply needs.

[0043] Step S2: Perform intent parsing on the control information, identify the action type within the future feedforward time window, predict the load on the corresponding joint actuator of the robot based on the action type, and obtain the expected load sequence; wherein, the power source of the joint actuator includes a fuel cell.

[0044] This step involves parsing the control information to identify the types of actions that may cause a rapid increase in load within the future feedforward time window, and extracting the expected torque, expected angular velocity, and action execution time information of the target joint motor (i.e., the joint actuator related to the action type) to obtain the expected load sequence.

[0045] In one embodiment, the feedforward time window is determined by combining the lumped response delay of the gas supply actuator, which can be expressed as: .in, For feedforward time window, For communication and control, calculate the delay. To compensate for the response delay of the air compressor in overcoming mechanical inertia and accelerating, The flow delay is used to establish effective pressure and flow rate for air through the intake pipe and cathode channel. In this embodiment, the overall response delay range of the air supply actuator is 160~240ms, preferably 200ms. Therefore, the system triggers feedforward air supply regulation approximately 200ms before the expected rapid load increase. Generally, the bus communication and controller calculation delays are in the millisecond range; the mechanical acceleration delay of the air compressor and the air flow establishment delay together determine the feedforward time window.

[0046] After obtaining the feedforward time window, the feedforward time window includes For each time step, the system calculates the expected load current for the future feedforward time window based on the expected torque, expected angular velocity, and operating condition correction parameters of each target joint motor. Its schematic expression can be written as:

[0047]

[0048] In the formula, Let be the desired torque of the motor of the j-th target joint; Let be the desired angular velocity of the motor of the j-th target joint; The power ratio factor allocated by the energy management module to the fuel cell; This refers to the operating voltage of the fuel cell stack. For the overall efficiency of the electric drive system; Power for auxiliary components such as gas supply actuators; The target number of joints; Indicates the sampling time.

[0049] The expected load current corresponding to each time step of the feedforward time window is used to form the expected load sequence.

[0050] This formula converts the mechanical side motion requirements into the predicted load current of the fuel cell side by combining the operating condition correction parameters, and is used to characterize the load change trend within the future feedforward time window.

[0051] Step S3: The predicted load sequence is imported as a feedforward disturbance into the discrete state-space prediction model of the gas supply actuator. The model after importation can be expressed as: In the formula, For discrete sampling times; This represents the system state vector, such as air compressor speed, cathode pressure, etc. This refers to the control input vector, i.e., the air compressor control command; This is the feedforward perturbation vector, i.e., the aforementioned expected load sequence; , , These are the system state matrix, control input matrix, and disturbance input matrix, respectively.

[0052] Step S4: Under the constraint that the cathode oxygen ratio is not lower than the safety lower limit, the air compressor control quantity within the future feedforward time window is solved by rolling optimization. Its objective function can comprehensively consider the oxygen ratio tracking error, control input changes, and air compressor parasitic work; its schematic expression can be written as:

[0053]

[0054] In the formula, This represents the number of steps within the feedforward time window. To allow changing the number of steps in the control command, ; To predict the cathode oxygen ratio; This is the set safety lower limit reference value for the superoxide ratio; To control the amount of input variation; This refers to the parasitic power of the air compressor. and These are the penalty weight matrices for the oxygen ratio tracking error and the control increment, respectively; This is the weighting coefficient for parasitic power.

[0055] Step S5 involves sending the optimal control value corresponding to the current moment as the first execution command to the air compressor drive, achieving proactive adjustment. When the actual load increases rapidly, the air compressor has already completed part or all of the speed-up process in advance, thereby better matching the cathode air supply capacity with future load demands.

[0056] To verify the control effect, a numerical simulation environment was constructed based on the mathematical model of the fuel cell mechanism and the discrete control logic of the system. In this simulation environment, the output voltage of a single cell can be expressed as: .in, It is a thermodynamically reversible electromotive force, which is affected by transient temperature and pressure. To activate the overpotential; For Ohm's overpotential; This represents the concentration overpotential.

[0057] The test condition was set as a rapid load change process corresponding to the robot performing a limit jump. The comparison was between the feedforward model predictive control strategy of this invention and the traditional PID feedback control strategy based on adjustment after a sudden change in fuel cell current. Both were numerically solved and compared under the same 1 kW rated power fuel cell model and the same low idle initial state. Detailed timing response results are as follows: Figure 3 As shown. Figure 3 The vertical axis of the upper coordinate system represents the normalized current and flow parameters, the vertical axis of the lower coordinate system represents the single-cell voltage, and the horizontal axis represents time.

[0058] like Figure 3 As shown in the upper curve, the test system experiences a step jump from the normalized current of 0 to 1 at t=0. In the traditional feedback control group, the air compressor only begins to accelerate after detecting the actual current jump. Due to mechanical inertia and the establishment process of airflow, its air supply flow rate exhibits a significant lag, leading to a decrease in the effective oxygen partial pressure at the cathode during the initial rapid load increase. The decrease in cathode oxygen concentration reduces the limiting current density, and the schematic relationship can be expressed as follows:

[0059]

[0060] In the formula, For the number of transferred electrons, It is Faraday's constant. This is the effective diffusion coefficient of oxygen in the gas diffusion layer. The thickness of the diffusion layer, This represents the effective oxygen concentration on the surface of the cathode catalyst layer. The sudden drop leads to the limiting current density of the fuel cell stack. It decays rapidly and approaches the actual load current. As the effective oxygen concentration decreases, the limiting current density decreases, which in turn increases the concentration overpotential. The schematic relationship of the concentration overpotential can be expressed as follows:

[0061]

[0062] In the formula, RT represents molar heat energy. When the operating current... Approaching the limiting current density after decay At this time, the concentration overpotential will increase rapidly, which will lead to a significant drop in the single cell voltage.

[0063] In contrast, the feedforward control group of this invention utilizes the motion intention information output in advance by the upper-level controller to issue the air compressor acceleration command before the actual step load occurs. When the feedforward advance is set reasonably, when the actual load impact arrives at t=0, the air compressor's air supply flow has already completed its ramp-up, and an air supply margin adapted to the future load has been established in the cathode air path. Thanks to the increased cathode oxygen supply capacity in advance, the limiting current density boundary is improved, the sudden increase in concentration overpotential is suppressed, and therefore the voltage drop of a single cell is significantly reduced. At the same time, the cathode oxygen ratio is always maintained above the safe lower limit, and the system maintains good operational stability throughout the entire load variation cycle.

[0064] Simulation results show that the single-cell voltage drops significantly under traditional feedback control, while the minimum single-cell voltage is significantly improved after adopting the method of the present invention. This indicates that the method of the present invention helps to reduce the voltage drop caused by gas supply lag during rapid load changes and improves the operational stability under high dynamic conditions.

[0065] In practical implementation, the fuel cell control device analyzes the intercepted control information and extracts multi-dimensional motion feature data in real time. The system can not only identify motion categories such as jumping, starting, and climbing, but also extract the expected angular velocity, target torque, and motion execution time information for each target joint. Based on this, the system calls upon a preset motion-power mapping relationship to generate a basic load prediction sequence within a preset future time window. The basic data for this mapping relationship can be obtained through prior dynamic analysis, bench testing, or whole-machine calibration. To improve the consistency between the prediction results and actual operating conditions, the system can also combine load estimates, slope information, terrain resistance information, or task mode information to correct the basic prediction sequence, obtaining a final load prediction curve that better matches real operating conditions.

[0066] After obtaining the final load prediction curve, the system imports it as a feedforward disturbance into the discrete state-space prediction model. Considering that conventional feedforward control may cause excessive compressor acceleration and system oscillation or surge risks when dealing with large transient disturbances, this invention adopts a model predictive control algorithm for multi-step deduction and rolling optimization. The algorithm sets the cathode superoxide ratio not lower than the safety lower limit as the core constraint condition, and comprehensively considers factors such as changes in control input, compressor parasitic work, and system stability to solve for the optimal control sequence that satisfies the constraint condition in each control cycle.

[0067] Finally, the actuator increases the air compressor speed in advance according to the optimal control command before the actual load current rises rapidly, so that the cathode gas circuit establishes a gas supply margin adapted to the future load. In this way, the cathode oxygen deficiency, voltage drop and related operational risks caused by gas supply lag during rapid load changes can be reduced, thereby improving the robot's power continuity and the operational stability of the fuel cell system under highly dynamic conditions.

[0068] In terms of technical solution, unlike existing technologies that rely on passive feedback mechanisms that respond to sudden changes in current or voltage, this invention employs an active feedforward triggering mechanism based on predictive information. This method moves the trigger source for gas supply control from the feedback signal of the fuel cell's physical state to the parsing stage of the upper-level motion planning instructions. Since communication signal transmission and control calculation are typically faster than the physical process of a servo motor overcoming mechanical inertia and actually outputting power, the system can utilize this inherent time difference to compensate for the acceleration delay of the centrifugal air compressor and the lag in air path transmission, thereby preemptively increasing the cathode gas supply capacity before the actual load rapidly increases.

[0069] In terms of technical effectiveness, this invention moves the trigger point of air supply control forward to the motion intention generation stage, realizing the transformation from "passive response after load occurs" to "active adjustment before load occurs." This method utilizes the time window of communication and control calculation preceding the actual occurrence of mechanical action to compensate for the hysteresis caused by the mechanical response of the air compressor and the establishment of airflow, so that the cathode oxygen ratio can still be maintained within a reasonable range under high dynamic load conditions, thereby reducing performance fluctuations and downtime risks caused by air shortage.

[0070] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0071] Based on the same inventive concept, this application also provides a fuel cell control device. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more fuel cell control device embodiments provided below can be found in the limitations of the robot fuel cell feedforward gas supply control method based on motion intention analysis and model prediction described above, and will not be repeated here.

[0072] A fuel cell control device includes: a data acquisition module for acquiring control information of a robot in real time;

[0073] The intent recognition module is used to parse the control information, identify the type of action within the future feedforward time window, predict the load on the corresponding joint actuators of the robot based on the action type, and obtain the expected load sequence; wherein, the power source of the joint actuators includes fuel cells.

[0074] The feedforward MPC algorithm module is used to construct a discrete state-space prediction model of the gas supply actuator with the expected load sequence as the feedforward perturbation. It constructs an objective function that integrates the tracking error of the cathode oxygen ratio of the fuel cell, the change in control input, and the parasitic power of the gas supply actuator. With the constraint that the cathode oxygen ratio of the fuel cell is not lower than the safety lower limit, it continuously optimizes and solves the target control quantity, so that the gas supply actuator can adjust the oxygen supply of the fuel cell according to the target control quantity.

[0075] Based on the same inventive concept, this application also provides a fuel cell power system, which includes: a fuel cell, a gas supply actuator, a sensor unit, and necessary auxiliary components;

[0076] For a hybrid power supply system combining fuel cells and batteries, the energy management module first allocates the target power to be supplied by the fuel cells, and then generates the corresponding predicted load sequence. The gas supply actuators may include a centrifugal air compressor, intake piping, and related actuators, used to respond to control commands issued by the controller and regulate the airflow entering the fuel cell stack cathode. If necessary, the system can also be coordinated with the anode hydrogen supply branch to improve the overall gas supply matching of the fuel cell system.

[0077] like Figure 4 As shown, based on the same inventive concept, this application also provides a robot system, including: a robot platform, a fuel control device, and a fuel cell power system;

[0078] The robot platform includes an upper-level motion controller and a drive and execution module. The upper-level motion controller outputs control information about the drive and execution module.

[0079] The fuel control device is connected to the upper-level motion controller via a CAN / EtherCAT bus, executes the steps of the above method embodiment, outputs the target control quantity, and issues the corresponding control command for the target control quantity.

[0080] The fuel cell power system is connected to the fuel control unit to receive and execute control commands to meet the power requirements of the drive and actuation modules.

[0081] The overall operation process of the robot system is as follows:

[0082] The upper-level motion controller on the robot platform side is responsible for planning motion trajectories, gait sequences, or task actions, and outputting corresponding motion intent information and joint state information. The fuel cell control unit, as the core computing unit for feedforward control, incorporates an intent parsing module and a feedforward MPC algorithm module. Its communication input is connected to the upper-level motion controller to acquire control information in real time; it also receives status feedback data such as pressure, flow rate, current, voltage, or speed from sensors on the fuel cell power system side. Its control output is electrically connected to the gas supply actuator within the fuel cell power system, used to output optimal feedforward control commands to the actuator.

[0083] The modules in the aforementioned fuel cell control device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0084] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0085] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0086] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for feedforward gas supply control of a robot fuel cell based on motion intent parsing and model prediction, characterized in that, The method includes: Real-time acquisition of robot control information; The control information is parsed to identify the action type within the future feedforward time window. Based on the action type, the load of the corresponding joint actuator of the robot is predicted to obtain the expected load sequence. The power source of the joint actuator includes a fuel cell. A discrete state-space prediction model for the gas supply actuator is constructed, using the expected load sequence as the feedforward perturbation. An objective function is constructed that integrates the tracking error of the cathode oxygen ratio of the fuel cell, the change in control input, and the parasitic power of the gas supply actuator. With the constraint that the cathode oxygen ratio of the fuel cell is not lower than the safety lower limit, the objective control quantity is solved by rolling optimization, so that the gas supply actuator adjusts the oxygen supply of the fuel cell according to the objective control quantity.

2. The method according to claim 1, characterized in that: The feedforward time window is determined based on the lumped response delay of the air supply actuator. The lumped response delay is related to the communication and control calculation delay, the response delay to overcome mechanical inertia and complete the acceleration, and the flow delay of air establishing effective pressure and flow rate through the intake pipe and cathode channel.

3. The method according to claim 1, characterized in that, The step of predicting the load on the robot's corresponding joint actuators based on the action type and obtaining the expected load sequence includes: Based on the action type, extract the desired torque and desired angular velocity corresponding to the joint actuator; The expected load sequence is calculated based on the power ratio coefficient of the fuel cell in the joint actuator, the stack operating voltage of the fuel cell, the total power of auxiliary components, the overall efficiency of the electric drive system, and the expected torque and the expected angular velocity.

4. The method according to claim 3, characterized in that, The calculation of the expected load sequence includes: Based on the expected torque and expected angular velocity of each joint actuator corresponding to the action type, the expected power of each joint actuator is calculated, and the expected power is summed to obtain the total expected power; The power component that the fuel cell needs to output is obtained by multiplying the power ratio coefficient and the expected total power, and then dividing the product by the overall efficiency of the integrated electric drive system. The total power output of the fuel cell is obtained based on the power component that the fuel cell needs to output and the total power of the auxiliary components. The expected load sequence is obtained by dividing the total power that the fuel cell needs to output by the stack operating voltage of the fuel cell.

5. The method according to claim 1, characterized in that, The method further includes: The target control quantity is adjusted according to the hydrogen supply from the anode of the fuel cell.

6. The method according to claim 1, characterized in that, The method further includes: The state feedback data of the fuel cell is acquired in real time, and the target control variable is adjusted in real time based on the state feedback data.

7. A fuel cell control device, characterized in that, The device includes: The data acquisition module is used to acquire the robot's control information in real time; An intent recognition module is used to parse the control information, identify the action type within the future feedforward time window, predict the load on the corresponding joint actuator of the robot based on the action type, and obtain the expected load sequence; wherein, the power source of the joint actuator includes a fuel cell. The feedforward MPC algorithm module is used to construct a discrete state-space prediction model of the gas supply actuator with the expected load sequence as the feedforward perturbation. It constructs an objective function that integrates the tracking error of the cathode oxygen ratio of the fuel cell, the change in control input, and the parasitic power of the gas supply actuator. With the constraint that the cathode oxygen ratio of the fuel cell is not lower than the safety lower limit, it performs rolling optimization to solve for the target control quantity, so that the gas supply actuator adjusts the oxygen supply of the fuel cell according to the target control quantity.

8. A fuel cell power system, characterized in that, The system includes: a fuel cell, a gas supply actuator, and a sensor unit; The gas supply actuator receives and executes a control command corresponding to the target control quantity obtained by the method according to any one of claims 1 to 6, so as to adjust the oxygen supply of the fuel cell.

9. A robot, characterized in that, include: Robotic platforms, fuel control devices, and fuel cell power systems; The robot platform includes an upper-level motion controller and a drive and execution module. The upper-level motion controller outputs control information about the drive and execution module. The fuel control device is connected to the upper motion controller, executes the method of any one of claims 1 to 6, outputs the target control quantity, and issues a control command corresponding to the target control quantity; The fuel cell power system is connected to the fuel control device and is used to receive and execute the control commands to meet the power requirements of the drive and execution modules.

Images