Hydrogen fuel cell unmanned aerial vehicle energy management and trajectory tracking cooperative control method

Abstract

The invention discloses a hydrogen fuel cell unmanned aerial vehicle energy management and trajectory tracking cooperative control method, and belongs to the field of hydrogen fuel cell long-endurance unmanned aerial vehicle energy management and trajectory tracking control. According to the method, the influence of external disturbance and model uncertainty is studied, fixed time control and active disturbance rejection control principles are introduced, a double-closed-loop unmanned aerial vehicle trajectory tracking method based on fixed time active disturbance rejection control is designed, and effective unmanned aerial vehicle working condition demand information and power control output in a time domain are obtained; based on demand information generated by trajectory tracking control, factors such as hydrogen fuel cell temperature difference, lithium battery SoC maintenance, hydrogen fuel consumption and energy source service life are comprehensively considered, a life attenuation cost function and an energy consumption cost function based on minimum equivalent fuel consumption are established, and an intelligent energy management cooperative control method fusing trajectory tracking control is designed. Hydrogen consumption is reduced, and the cruising ability of the whole machine is improved.

Description

Technical Field

[0001] The present invention relates to the field of energy management and trajectory tracking control of long-endurance unmanned aerial vehicles (UAVs) powered by hydrogen fuel cells, and specifically to a collaborative control method for energy management and trajectory tracking of hydrogen fuel cell UAVs. Background Art

[0002] Long-endurance UAVs can complete uninterrupted inspections of railway lines dozens of kilometers long in a single flight, avoiding blind spots caused by traditional segmented manual inspections, and are an important means to improve the efficiency of hidden danger detection. Railway lines often pass through high-risk areas such as mountains, tunnels, and bridges. UAVs need to continuously cope with interferences such as strong winds and low temperatures. High-endurance models can ensure the completion of key tasks in bad weather. Hydrogen fuel cells have outstanding low-temperature resistance, and the energy density of hydrogen fuel cells can reach several times or even hundreds of times that of lithium batteries, which can significantly expand the inspection range of a single flight. Energy management of the hydrogen fuel cell hybrid power system is one of the core technologies to improve the endurance of the power system. Research on the collaborative control strategy of energy management and trajectory tracking for long-endurance hydrogen fuel cell UAVs for railway inspection, and overall energy management of the system at the whole-machine level has become the future development trend of hydrogen fuel cell UAVs.

[0003] During the flight of a hydrogen fuel cell UAV, there are unpredictable interferences, and there is a certain error between the UAV mathematical model and the actual machine. During actual flight, in order to ensure that the UAV flies along the expected predetermined trajectory, it is necessary to introduce a control method to ensure the accuracy and precision of trajectory tracking. At the same time, when the UAV performs complex trajectory tracking tasks such as sharp turns and obstacle avoidance, the sudden change in power demand is likely to cause the power supply system to be instantaneously overloaded, resulting in trajectory deviation. How to make the energy management system predict the power demand of trajectory control commands in real time and dynamically adjust the collaborative output of hydrogen fuel cells and energy storage devices is a challenge. Summary of the Invention

[0004] The present invention aims to provide a collaborative control method for energy management and trajectory tracking of hydrogen fuel cell UAVs. By studying the influence of external disturbances and model uncertainties, introducing the fixed-time control and active disturbance rejection control principles, designing a double-closed-loop UAV trajectory tracking method based on fixed-time active disturbance rejection control, obtaining effective UAV operating condition demand information and dynamic control output in the time domain; based on the demand information generated by trajectory tracking control, comprehensively considering factors such as the temperature difference of hydrogen fuel cells, the maintenance of the state of charge (SoC) of lithium batteries, hydrogen fuel consumption, and the service life of energy sources, establishing a life attenuation cost function and an energy consumption cost function based on the minimum equivalent fuel consumption, and designing an intelligent energy management collaborative control method that integrates trajectory tracking control to reduce hydrogen consumption and improve the endurance of the whole machine.

[0005] To solve the above technical problems, the present invention provides a collaborative control method for energy management and trajectory tracking of hydrogen fuel cell UAVs, including the following steps: S1. Build a UAV dynamics model. According to the fixed-time control principle and the active disturbance rejection control principle, design a double-loop UAV trajectory tracking method based on fixed-time active disturbance rejection control to obtain effective UAV working condition demand information and dynamic control output in the time domain;

[0006] S2. Design an intelligent energy management framework for a hybrid power system based on trajectory tracking control. Build a UAV load condition fusion prediction model that integrates historical data-based time-series load condition prediction and trajectory tracking motion control information. Based on factors such as the temperature difference of the hydrogen fuel cell, the maintenance of the lithium battery SoC, hydrogen fuel consumption, and the service life of the energy source, establish a life attenuation cost function and an energy consumption cost function based on minimizing equivalent fuel consumption.

[0007] Further, the S1 includes:

[0008] S1-1. Analyze the typical working modes and energy characteristics of the hydrogen fuel cell hybrid system. Based on kinematic and dynamic methods, build the dynamics model of the hydrogen fuel cell hybrid system;

[0009] S1-2. Use the double-loop UAV trajectory tracking method based on fixed-time active disturbance rejection control to track the target trajectory.

[0010] Further, the S1-1 includes:

[0011] The hydrogen fuel cell hybrid system is a parallel hybrid power system with a hydrogen fuel cell as the main energy source and a lithium battery as the auxiliary energy source. The hydrogen fuel cell is connected to the bus through a boost converter, and the output of the lithium battery is directly connected to the bus. The energy equation of the hydrogen fuel cell hybrid system is as follows:

[0012] P FC η + P B =P D + P AE

[0013] Wherein, P FC is the output power of the hydrogen fuel cell, η is the efficiency of the hydrogen fuel cell, P B is the output power of the lithium battery, P D is the power consumed by the electric propulsion system, and P AE is the power consumed by auxiliary equipment.

[0014] Further, in the S1-2, the construction process of the double-loop UAV trajectory tracking method based on fixed-time active disturbance rejection control is as follows:

[0015] First, design a fixed-time UAV position tracking controller based on a state observer. The trajectory tracking motion control of the UAV includes two closed loops: an outer loop and an inner loop. The outer loop uses a position tracking control that combines model predictive control and an extended state observer, and the inner loop uses an active disturbance rejection controller. The position tracking control of the UAV dynamics model is as follows:

[0016]

[0017] where represent the accelerations in the x, y, and z directions respectively; k dx , k dy , k dz represent the gain coefficients in the x, y, and z directions respectively; u x , u y , u z represent the inputs of the controllers in the x, y, and z directions respectively, d x , d y , d z represent the disturbances in the x, y, and z directions respectively, m represents the mass of the UAV, represents the velocity in the x direction, represents the velocity in the y direction, represents the velocity in the z direction;

[0018] Secondly, use the active disturbance rejection control strategy as a means for the UAV to achieve attitude tracking in a disturbed environment. The active disturbance rejection control strategy includes a tracking differentiator, an extended state observer, and an error compensation controller: The tracking differentiator is responsible for extracting the differential information of the reference trajectory, and the extended state observer is used to estimate and weaken the influence of external disturbances, and a fixed-time controller is used to replace the nonlinear error feedback mechanism;

[0019] Finally, based on the reference tracking value obtained from the position tracking control of the UAV dynamics model, and combined with the inequality scaling technique and the Cauchy-Schwarz inequality technique, design a fixed-time active disturbance rejection control and make the fixed-time active disturbance rejection control satisfy the Lyapunov fixed-time lemma

[0020] Furthermore, the S2 includes:

[0021] S2-1. Set an optimization objective function according to the hydrogen fuel consumption, the lithium battery SOC, and the durability of the hydrogen fuel cell. Set the power output as the action, and set the load state and the lithium battery SOC as the states to construct an intelligent energy management space;

[0022] S2-2. Introduce a reward mechanism model based on the Q-network, design a loss function to guide the learning direction of the reward mechanism model based on the Q-network, and design a state transition function to enable the intelligent energy management space to understand the dynamic behavior of the environment.

[0023] Further, S2-1 includes:

[0024] First, determine the power demand data according to the dual closed-loop UAV trajectory tracking method based on fixed-time active disturbance rejection control, and adopt a reinforcement learning method that combines the Dyna algorithm and the DQN algorithm to optimize the intelligent energy management framework of the hybrid power system. The intelligent energy management framework of the hybrid power system includes an offline learning stage and an online learning stage. In the offline learning stage, use the typical load condition data of the UAV to train the gated recurrent unit model, so that the intelligent energy management framework of the hybrid power system learns the environmental and condition characteristics; in the online learning stage, the intelligent energy management framework of the hybrid power system continuously accumulates interaction experience with the environment through the dual closed-loop UAV trajectory tracking method based on fixed-time active disturbance rejection control to update the reinforcement learning model of the reinforcement learning method in real time, and then continuously reduce the cumulative error of the reinforcement learning model during operation.

[0025] Second, the intelligent energy management space is as follows:

[0026]

[0027]

[0028] In the formula: rd represents the abnormal state negative constant, S represents the state space, s represents the state sequence, A represents the action space, a represents the action sequence, P FC is the output power of the hydrogen fuel cell, r represents the reward function, R(s,a) represents the reward value space, P Load (k) represents the demand power at time k, SoC(k) represents the state of charge of the lithium battery at time k, SoC(k) T represents the transpose of the state of charge of the lithium battery at time k, m H2 (k) represents the hydrogen consumption mass at time k, SoC ref (k) represents the reference state of charge of the lithium battery at time k, I FC (k) represents the output current of the fuel cell at time k, r k , ξ1, ξ2 respectively represent the hydrogen consumption mass coefficient, the state of charge difference coefficient of the lithium battery, and the output current coefficient of the fuel cell.

[0029] Further, S2-2 includes:

[0030] First, an environment model is established. The environment model includes a state transition function and a loss function in the intelligent energy management framework of the hybrid power system. The loss function is as follows:

[0031]

[0032] where Loss(θ i ) represents the loss function, Q RT (s(k), a(k), θ i ) represents the target network return function Q RE (s(k), a(k), θ i ) represents the evaluation network return function; Q * (s(k), a(k), θ i ) is the optimal action value function, Q(s(k), a(k), θ i ) represents the action value function, ε-greedy is the greedy policy function; θ i is the parameter at the iteration number i, s(k) represents the state at time k, a(k) represents the action at time k, s(k + 1) represents the state at time k + 1, and a(k + 1) represents the action at time k + 1;

[0033] Secondly, the state transition function is defined by using the UAV load condition fusion prediction model. The state transition function is used to calculate the future state of charge of the lithium battery in the hydrogen fuel cell hybrid system according to the current load condition, the output power of the hydrogen fuel cell, and the parameters of the UAV load condition fusion prediction model.

[0034] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0035] A fixed-time active disturbance rejection controller is designed to track the target trajectory, making the trajectory tracking more accurate and fast, with strong anti-interference ability, and capable of obtaining more reasonable power demand information. Then, by using the obtained power demand information, considering factors such as the temperature difference of the hydrogen fuel cell, the maintenance of the lithium battery SoC, hydrogen fuel consumption, and the service life of the energy source, a life decay cost function and an equivalent fuel consumption-based minimum energy consumption cost function are established, and a fusion trajectory tracking control intelligent energy management collaborative control method is designed, resulting in less hydrogen consumption and stronger endurance. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the collaborative control method for energy management and trajectory tracking of a hydrogen fuel cell UAV;

[0037] Figure 2 It is a schematic diagram of the trajectory tracking control method;

[0038] Figure 3Schematic diagram of intelligent energy management for a hybrid power system based on trajectory tracking control. Detailed implementation manners

[0039] The following provides a further detailed description of the present invention in conjunction with the accompanying drawings.

[0040] As Figure 1 shown is a schematic diagram of a method for collaborative control of energy management and trajectory tracking of a hydrogen fuel cell unmanned aerial vehicle provided by an embodiment of the present invention. It can be seen from the figure that the method includes the following steps:

[0041] S1: Construct a dynamic model of the unmanned aerial vehicle, study the influence of external disturbances and model uncertainties, introduce the fixed-time control and active disturbance rejection control principles, and design a double-closed-loop unmanned aerial vehicle trajectory tracking method based on fixed-time active disturbance rejection control. As Figure 2 shown, obtain effective unmanned aerial vehicle operating condition demand information and power control output within the time domain;

[0042] S2: Design an intelligent energy management framework for a hybrid power system based on trajectory tracking control, as Figure 3 shown. Study a complex load condition fusion prediction model for the unmanned aerial vehicle that fuses the prediction of historical data time-series load conditions and trajectory tracking motion control information. Considering factors such as the temperature difference of the hydrogen fuel cell, the maintenance of the lithium battery SoC, hydrogen fuel consumption, and the service life of the energy source, establish a life attenuation cost function and an energy consumption cost function based on the minimum equivalent fuel consumption.

[0043] Specific implementation process in step S1: S1-1: Construct a parallel hybrid power system with a hydrogen fuel cell as the main energy source and a lithium battery as the auxiliary energy source. The hydrogen fuel cell power source is connected to the bus through a boost converter, and the output of the lithium battery is directly connected to the bus. The energy equation of the system is as follows P FC η + P B = P D + P AE

[0044] where P FC is the output power of the hydrogen fuel cell, η is the efficiency of the hydrogen fuel cell, P B is the output power of the lithium battery, P D is the power consumed by the electric propulsion system, and P AE is the power consumed by auxiliary equipment.

[0045] The process of the double-closed-loop unmanned aerial vehicle trajectory tracking method based on fixed-time active disturbance rejection control is as follows:

[0046] First, design a fixed-time UAV position tracking controller based on a state observer. The trajectory tracking motion control of the UAV consists of two closed loops. The outer loop adopts a position tracking control combining model predictive control and an extended state observer, and the inner loop adopts an active disturbance rejection controller. The position control of the UAV dynamics model is as follows:

[0047] where represent the accelerations in the x, y, and z directions respectively; k dx , k dy , k dz represent the gain coefficients in the x, y, and z directions respectively; u x , u y , u z represent the inputs of the controllers in the x, y, and z directions respectively. d x , d y , d z represent the disturbances in the x, y, and z directions respectively, m represents the mass of the UAV, represents the velocity in the x direction, represents the velocity in the y direction, represents the velocity in the z direction;

[0048] In the UAV flight scenario, if its predetermined flight path is clear, according to the system model reconstruction, design a state observer to estimate the disturbance error, and adjust the control law accordingly, aiming to reduce the disturbance error to almost zero, thereby effectively eliminating the interference factors affecting the UAV position.

[0049] Secondly, design a UAV attitude tracking controller based on active disturbance rejection fixed-time control. Since the UAV is continuously affected by external disturbances and model uncertainties, therefore, use the active disturbance rejection controller for the attitude tracking control of the UAV in a disturbed environment. The active disturbance rejection control includes a tracking differentiator, an extended state observer, and an error compensation controller. The tracking differentiator is used to obtain the differential signal of the reference trajectory, which can simplify the controller design process. The extended state observer is used to estimate and suppress external disturbances. In order to make the attitude tracking faster, introduce a fixed-time controller to replace the original non-linear error feedback.

[0050] Finally, based on the reference tracking values of the three angles obtained from the previous position tracking, introduce inequality scaling and the Cauchy-Schwarz inequality technique to design an active disturbance rejection fixed-time control to satisfy the Lyapunov fixed-time lemma

[0051] The specific implementation process in step S2 is as follows:

[0052] S2-1: First, based on the power demand data obtained from the trajectory tracking control system, a reinforcement learning method that combines the Dyna algorithm and the DQN algorithm, namely the Dyna-DQN method, is adopted to make full use of the advantages of the two algorithms, thereby optimizing the energy management process of the hydrogen fuel cell UAV. This intelligent energy management framework consists of two key learning stages: one is the offline learning stage, and the other is the online learning stage. In the offline learning stage, the typical load condition data of the UAV is used to train the GRU (Gated Recurrent Unit) model, enabling the energy management control strategy to pre-understand the environmental and working condition characteristics. And

[0053] In the online learning stage, through close cooperation with the trajectory tracking control system, this framework continuously accumulates interaction experience with the environment, updates the reinforcement learning model in real time, and gradually reduces the cumulative error of the model during long-term operation to achieve more accurate and efficient energy management.

[0054] Secondly, under safe conditions, comprehensively considering hydrogen fuel consumption, battery SOC, hydrogen fuel cell durability, etc., an optimization objective function is set, with the power output set as the action and the load state battery SOC, etc. set as the state, and the intelligent energy management space is constructed as follows

[0055] where: rd represents the abnormal state negative constant, S represents the state space, s represents the state sequence, A represents the action space, a represents the action sequence, P FC is the output power of the hydrogen fuel cell, r represents the reward function, R(s,a) represents the reward value space, P Load (k) represents the demand power at time k, SoC(k) represents the state of charge of the lithium battery at time k, SoC(k) T represents the transpose of the state of charge of the lithium battery at time k, m H2 (k) represents the mass of hydrogen consumed at time k, SoC ref (k) represents the reference state of charge of the lithium battery at time k, I FC (k) represents the output current of the fuel cell at time k, r k , ξ1, ξ2 respectively represent the hydrogen consumption mass coefficient, the difference coefficient of the state of charge of the lithium battery, and the output current coefficient of the fuel cell. S2-2: First, an environment model is established. The environment model includes the state transition function and the reward function of the above hydrogen fuel cell UAV system model. Thanks to the introduction of the environment model, the Dyna algorithm shows higher efficiency and cost-effectiveness compared to directly learning the strategy. By integrating DQN with the Dyna algorithm, the Dyna-DQN algorithm can improve performance and efficiency in the energy management of the hybrid power system. Within the Dyna-DQN framework, the design of the reward mechanism relies on the Q-network, which is beneficial for sharing the same architecture to promote the effective circulation and processing of information. Finally, the loss function of the Dyna-DQN algorithm is as follows:

[0056] where Loss(θ i ) represents the loss function, Q RT (s(k), a(k), θ i ) represents the target network return function Q RE (s(k), a(k), θ i ) represents the evaluation network return function; Q * (s(k), a(k), θ i ) is the optimal action value function, Q(s(k), a(k), θ i ) represents the action value function, and ε-greedy is the greedy policy function; θ i is the parameter at the iteration number i, s(k) represents the state at time k, a(k) represents the action at time k, s(k + 1) represents the state at time k + 1, and a(k + 1) represents the action at time k + 1;

[0057] Secondly, the load condition prediction model is used to define the state transition function. It can accurately calculate the future state of charge of the battery in the hydrogen fuel cell hybrid power system according to the current load condition and the selected action (i.e., the output power of the hydrogen fuel cell), in combination with the parameters of the system model.

[0058] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A collaborative control method for energy management and trajectory tracking of a hydrogen fuel cell unmanned aerial vehicle, characterized in that, The following steps are involved: S1: Construct a UAV dynamics model and design a dual-closed-loop UAV trajectory tracking method based on fixed-time ADRC according to the fixed-time control principle and ADRC principle to obtain effective UAV working condition demand information and power control output in the time domain; S2, design an intelligent energy management framework for hybrid power systems based on trajectory tracking control, build a UAV load condition fusion prediction model that integrates historical data-based time-series load condition prediction and trajectory tracking motion control information, and establish a life attenuation cost function and an energy consumption cost function based on minimum equivalent fuel consumption based on hydrogen fuel cell temperature difference, lithium battery SoC maintenance, hydrogen fuel consumption, and energy source service life factors.

2. The collaborative control method for energy management and trajectory tracking of a hydrogen fuel cell unmanned aerial vehicle according to claim 1, wherein, Said S1 comprises: S1-1, analyzing the typical operating modes of the hydrogen fuel cell hybrid system and the energy characteristics under different modes, and constructing a dynamic model of the hydrogen fuel cell hybrid system based on kinematics and dynamics methods; S1-2, track the target trajectory using the dual closed-loop UAV trajectory tracking method based on fixed-time ADRC.

3. A collaborative control method for energy management and trajectory tracking of a hydrogen fuel cell unmanned aerial vehicle according to claim 2, characterized in that, The S1-1 includes: The hydrogen fuel cell hybrid system is a parallel hybrid power system with hydrogen fuel cells as the main energy source and lithium batteries as the auxiliary energy source. The hydrogen fuel cells are connected to the busbar through a boost converter, and the output of the lithium battery is directly connected to the busbar. The energy equation of the hydrogen fuel cell hybrid system is as follows: P FC η+P B =P D +P AE Among them, P FC is the output power of the hydrogen fuel cell, η is the efficiency of the hydrogen fuel cell, P B is the output power of the lithium battery, P D is the power consumption of the electric propulsion system, P AE Consumes power for auxiliary devices.

4. The collaborative control method for energy management and trajectory tracking of a hydrogen fuel cell unmanned aerial vehicle according to claim 2, characterized in that In S1-2, the construction process of the dual closed-loop UAV trajectory tracking method based on fixed-time ADRC is as follows: First, a fixed-time UAV position tracking controller based on a state observer is designed. The trajectory tracking motion control of the UAV consists of two closed loops: an outer loop and an inner loop. The outer loop uses a position tracking control that combines model predictive control and an extended state observer, while the inner loop uses an active disturbance rejection controller. The position tracking control of the UAV dynamic model is as follows: in, Respectively represent the acceleration in the x, y, and z directions; k dx , k dy ,k dz Respectively represent the gain coefficients in the x, y, and z directions; u x ,u y ,u z Represents the input of the x, y, and z direction controllers respectively, d x , d y ,d z Represent the disturbance in the x, y, and z directions respectively, and m represents the mass of the drone. represents the velocity in the x direction, represents the velocity in the y direction, represents the velocity in the z direction; Secondly, an active disturbance rejection control strategy is used as a means to achieve attitude tracking of the UAV in a disturbed environment. The active disturbance rejection control strategy includes a tracking differentiator, an extended state observer, and an error compensation controller. The tracking differentiator is responsible for extracting the differential information of the reference trajectory, the extended state observer is used to estimate and weaken the influence of external disturbances, and a fixed-time controller is used to replace the nonlinear error feedback mechanism. Finally, based on the reference tracking values obtained from the position tracking control of the UAV dynamics model, and combined with the inequality scaling technique and the Cauchy-Schwarz inequality technique, a fixed-time active disturbance rejection control is designed and the fixed-time active disturbance rejection control satisfies the Lyapunov fixed-time lemma 5. The collaborative control method for energy management and trajectory tracking of a hydrogen fuel cell UAV according to claim 1, characterized in that The S2 includes: S2-1, set the optimization objective function based on hydrogen fuel consumption, lithium battery SOC and hydrogen fuel cell durability, set power output as action, and set load state and lithium battery SOC as state to build an intelligent energy management space; S2-2, introduce a Q-network-based reward mechanism model, design a loss function to guide the learning direction of the Q-network-based reward mechanism model, and design a state transfer function to enable the intelligent energy management space to understand the dynamic behavior of the environment.

6. The method for coordinated control of energy management and trajectory tracking of a hydrogen fuel cell drone according to claim 5, characterized in that: The S2-1 includes: First, determine the power demand data according to the double-closed-loop UAV trajectory tracking method based on fixed-time active disturbance rejection control, and adopt a reinforcement learning method combining the Dyna algorithm and the DQN algorithm to optimize the intelligent energy management framework of the hybrid power system. The intelligent energy management framework of the hybrid power system includes an offline learning stage and an online learning stage. In the offline learning stage, use the typical load condition data of the UAV to train the gated recurrent unit model, so that the intelligent energy management framework of the hybrid power system learns the environment and condition characteristics. In the online learning stage, the intelligent energy management framework of the hybrid power system continuously accumulates interaction experience with the environment through the double-closed-loop UAV trajectory tracking method based on fixed-time active disturbance rejection control, thereby updating the reinforcement learning model of the reinforcement learning method in real time, and further continuously reducing the cumulative error of the reinforcement learning model during operation. Secondly, the intelligent energy management space is as follows: Where: rd represents the abnormal state negative constant, S represents the state space, s represents the state sequence, A represents the action space, a represents the action sequence, P FC is the output power of the hydrogen fuel cell, r represents the reward function, R(s,a) represents the reward value space, P Load (k) represents the required power at time k, SoC(k) represents the state of charge of the lithium battery at time k, SoC(k) T represents the transpose of the lithium battery state of charge at time k, m H2 (k) represents the mass of hydrogen consumed at time k, SoC ref (k) represents the reference state of charge of the lithium battery at time k, I FC (k) represents the fuel cell output current at time k, r k , ξ1, ξ2 represent the hydrogen consumption mass coefficient, the lithium battery state of charge difference coefficient, and the fuel cell output current coefficient, respectively.

7. The collaborative control method for energy management and trajectory tracking of a hydrogen fuel cell unmanned aerial vehicle according to claim 5, characterized in that, The S2-2 includes: First, establish an environment model, and the environment model includes the state transition function and the loss function in the intelligent energy management framework of the hybrid power system. The loss function is as follows: Among them, Loss(θ i ) represents the loss function, Q RT (s(k),a(k),θ i ) represents the target network reward function Q RE (s(k),a(k),θ i ) represents the evaluation network reward function; Q * (s(k),a(k),θ i ) is the optimal action value function, Q(s(k),a(k),θ i ) represents the action value function, ε-greedy is the greedy policy function; θ i is the parameter at iteration number i, s(k) represents the state at time k, a(k) represents the action at time k, s(k+1) represents the state at time k+1, and a(k+1) represents the action at time k+1; Secondly, use the UAV load condition fusion prediction model to define the state transition function. The state transition function is used to calculate the future state of charge of the lithium battery in the hydrogen fuel cell hybrid system according to the current load condition, the output power of the hydrogen fuel cell, and the parameters of the UAV load condition fusion prediction model.

Images