Hydrogen fuel cell hybrid power unmanned aerial vehicle energy self-adaptive optimization system and method

Abstract

The application provides a hydrogen fuel cell hybrid power unmanned aerial vehicle energy adaptive optimization system and method, which aims to realize adaptive optimization when multiple uncertain factors exist in the flight process of the unmanned aerial vehicle. The method includes the following steps: through mechanical analysis on a typical flight profile of the unmanned aerial vehicle, an estimation model of load demand power in each stage in the flight process of the unmanned aerial vehicle is constructed. The core of the method is to model the flight process of the unmanned aerial vehicle as a Markov decision process, and to reasonably construct an adaptive optimization model composed of a power energy consumption state space, an action space of power supply power distribution, a reward function, an electric energy supply constraint condition and the like. For the reward function of the electric energy distribution, a proximal policy optimization algorithm is applied to train an intelligent agent to continuously update an actor network, and iterative learning is performed to approximate an expected control strategy, so that the hydrogen fuel hybrid power unmanned aerial vehicle energy management system realizes adaptive intelligent optimal distribution of hydrogen energy and electric energy, and a system implementation scheme is provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy management, in particular to a hydrogen fuel cell hybrid power unmanned aerial vehicle energy adaptive optimization system and method. BACKGROUND

[0002] As an advanced energy conversion device, the core of hydrogen fuel cell is to directly convert the chemical energy of hydrogen and oxygen into electrical energy, which follows the reverse reaction principle of water electrolysis. Its most significant advantage is environmental friendliness; secondly, hydrogen fuel cell has high energy conversion efficiency, usually more than 50%, much higher than traditional internal combustion engine; in addition, hydrogen fuel cell runs quietly with extremely low noise level; finally, the fuel of hydrogen fuel cell-hydrogen has a wide range of sources and is renewable, and through renewable energy (such as photovoltaic and wind energy) hydrogen production, a completely clean energy cycle can be formed.

[0003] In recent years, with the transformation of global energy structure and the improvement of environmental awareness, the application research of hydrogen fuel cell in the field of unmanned aerial vehicles has been increasingly in-depth. Many domestic and foreign scientific research institutions and enterprises have invested resources to develop efficient and reliable hydrogen fuel cell unmanned aerial vehicle energy management systems. However, the current technology development still faces many challenges, such as optimization of energy distribution strategy, real-time response to various uncertain factors (such as flight environment, load change, battery state, etc.). Traditional energy management methods often have difficulty in fully considering these complex factors, resulting in difficulty in achieving optimal performance of unmanned aerial vehicles in actual application.

[0004] Therefore, in view of the above problems, the present application proposes a hydrogen fuel cell hybrid power unmanned aerial vehicle energy adaptive optimization method and system. This technology can realize real-time sensing and processing of various uncertain factors in the flight process of unmanned aerial vehicles through proximal strategy optimization algorithm, dynamically adjust the energy distribution strategy, and maximize the endurance, flight efficiency and safety of unmanned aerial vehicles. SUMMARY

[0005] The present application is to solve the problems of the prior art, and proposes a hydrogen fuel cell hybrid power unmanned aerial vehicle energy adaptive optimization method and system. Firstly, it can more accurately predict and respond to various changes in the flight process, improve the efficiency and stability of energy utilization; secondly, through the self-optimization ability of proximal strategy optimization algorithm, it can continuously learn and adapt to new flight environment and task requirements, improve the intelligent level and adaptability of unmanned aerial vehicles; thirdly, it realizes the mixed use of hydrogen fuel cell and lithium battery, and further improves the overall performance and reliability of unmanned aerial vehicles through adaptive intelligent optimal distribution.

[0006] The present application provides a hydrogen fuel cell hybrid power unmanned aerial vehicle energy adaptive optimization system, which comprises:

[0007] The sub-modeling module is used to build the load demand power model, hydrogen fuel cell model, and lithium battery model of the hydrogen fuel cell hybrid UAV based on the basic equipment parameters of the UAV.

[0008] The model analysis module is used to build the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid UAV based on the load demand power model, the hydrogen fuel cell model and the lithium battery model, as well as to build the flight reward function and power supply constraints of the hydrogen fuel cell hybrid UAV.

[0009] The algorithm optimization module is used to construct the actor neural network and commentator neural network of the hydrogen fuel cell hybrid UAV using the preset PPO algorithm and the basic equipment parameters. The actor neural network is used to optimize the power consumption state space and the power supply allocation space respectively to obtain several optimization conditions. The commentator neural network is used to perform feedback analysis on the optimization conditions according to the flight reward function and the power supply constraints to obtain several hyperparameters.

[0010] An optimized training module is used to construct an energy management agent based on the hyperparameters, and to map the load demand power model, the hydrogen fuel cell model, and the lithium battery model to the energy management agent to generate a training agent for the hydrogen fuel cell hybrid UAV. The training agent is then run to obtain and display the optimal strategy presented by the hydrogen fuel cell hybrid UAV when it converges to the flight reward function.

[0011] In one feasible approach

[0012] The process of building the load demand power model for the aforementioned hydrogen fuel cell hybrid drone includes:

[0013] The induced power of the hydrogen fuel cell hybrid drone is calculated according to formula (1);

[0014] P ik =T d ·v ik (1)

[0015] Among them, P ik The induced power, v, represents the induced power of the hydrogen fuel cell hybrid drone. ik This indicates the actual induced velocity of the hydrogen fuel cell hybrid drone;

[0016] The drag power of the hydrogen fuel cell hybrid UAV is calculated according to formula (2);

[0017]

[0018] Among them, P pr The form drag power of the hydrogen fuel cell hybrid UAV is represented by As, the cross-sectional area of ​​the UAV propeller blade, ω, the angular velocity of the blade rotation, R, the radius of the turntable, and δ, the tensile strength coefficient of the profile.

[0019] The exhaust resistance power of the hydrogen fuel cell hybrid drone is calculated according to formula (3);

[0020]

[0021] Among them, P pa This indicates the exhaust resistance power of the hydrogen fuel cell hybrid drone;

[0022] The interference power of the hydrogen fuel cell hybrid UAV is calculated according to formula (4);

[0023] P dis (t)~N(0,1)×c,P dis_min <P dis (t) <P dis_max (4)

[0024] Where c represents the disturbance amplitude coefficient of the hydrogen fuel cell hybrid drone;

[0025] The load mass of the hydrogen fuel cell hybrid UAV at different flight times is calculated according to formula (5);

[0026]

[0027] Where m(t) represents the total weight of the hydrogen fuel cell hybrid drone at time t. The value represents the amount of hydrogen consumed, M represents the molar mass of hydrogen, P represents the operating pressure of the hydrogen fuel cell stack, R represents Avogadro's constant, and T represents the operating thermodynamic temperature of the stack.

[0028] Based on the calculation results of formula (5), the load mass corresponding to the hydrogen fuel cell hybrid UAV at different flight times is determined, and the load demand power model of the fuel cell hybrid UAV is built.

[0029] In one feasible approach

[0030] Also includes:

[0031] Calculate the thrust required by the hydrogen fuel cell hybrid UAV during flight according to formula (6);

[0032]

[0033] Among them, T da represents the thrust required by the hydrogen fuel cell hybrid drone during flight. c v represents the acceleration of the hydrogen fuel cell hybrid drone, and v represents the airspeed of the hydrogen fuel cell hybrid drone. f Let β represent the air velocity, β represent the angle between the air velocity direction and the flight direction of the hydrogen fuel cell hybrid drone, θ represent the angle between the flight direction of the hydrogen fuel cell hybrid drone and the vertical direction, ρ represent the air density, and A represent the air velocity. f The vertical projected area of ​​the hydrogen fuel cell hybrid drone is represented by m, the total mass of the hydrogen fuel cell hybrid drone is represented by g, and g is represented by gravitational acceleration.

[0034] The force values ​​corresponding to each propeller disk on the hydrogen fuel cell hybrid drone were obtained separately.

[0035] When the force value corresponding to each of the propeller disks is the same, it is determined that the hydrogen fuel cell hybrid drone is in normal flight state.

[0036] When the hydrogen fuel cell hybrid drone is in normal flight, the theoretical speed of the hydrogen fuel cell hybrid drone is calculated according to formula (7);

[0037]

[0038] Among them, S T A represents the calculation of the real-time speed of the hydrogen fuel cell hybrid drone. p γ represents the horizontal projected area of ​​the hydrogen fuel cell hybrid drone, γ represents the tilt angle of the hydrogen fuel cell hybrid drone, and v in Indicates the theoretical induced velocity;

[0039] The actual speed of the hydrogen fuel cell hybrid drone is obtained, and the actual speed is substituted into the formula (7) to obtain the actual induced speed of the hydrogen fuel cell hybrid drone.

[0040] In one feasible approach

[0041] The process of building the hydrogen fuel cell model and lithium battery model of the aforementioned hydrogen fuel cell hybrid UAV includes:

[0042] The hydrogen fuel consumption of the hydrogen fuel cell hybrid UAV per unit time is calculated according to formula (8).

[0043]

[0044] in, P represents the hydrogen fuel consumption per unit time of the hydrogen fuel cell hybrid drone. fc This indicates the output power of the hydrogen fuel cell;

[0045] A hydrogen fuel cell model for the hydrogen fuel cell hybrid UAV is established based on the hydrogen fuel consumption.

[0046] The discharge cycle output power of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (9), the output current of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (10), and the state of charge of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (11).

[0047]

[0048] Among them, V OC R0 represents the open-circuit voltage, I represents the internal resistance, and R0 represents the open-circuit voltage. b P represents the output current. batt State of Charge (SOC) represents the output power during a charge-discharge cycle. t Indicates the state of charge, Q0 represents the initial battery capacity, and Q represents the rated battery capacity;

[0049] The lithium battery model of the hydrogen fuel cell hybrid UAV is established based on the calculation results of formulas (9), (10) and (11).

[0050] In one feasible approach

[0051] The process of constructing the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid UAV includes:

[0052] The load demand power P of the hydrogen fuel cell hybrid UAV at the current moment is determined using the aforementioned load demand power model. t ;

[0053] The hydrogen fuel cell model is used to determine the remaining hydrogen amount of the hydrogen fuel cell hybrid UAV at the current moment.

[0054] The remaining state of charge (SOC) of the hydrogen fuel cell hybrid drone at the current moment is determined using the lithium battery model. t ;

[0055] Based on the current load power demand P of the hydrogen fuel cell hybrid drone. t Remaining amount of hydrogen Remaining power SOC tEstablish the power consumption state space of the hydrogen fuel cell hybrid UAV;

[0056] The hydrogen fuel cell output power of the hydrogen fuel cell hybrid UAV is determined based on the hydrogen fuel cell model, and the hydrogen fuel cell output power is set within a preset continuous range [P]. fc_min ,P fc_max The power set is obtained by discretizing the [] [] [] [] [].

[0057] Among them, P fc_min P represents the preset minimum output power of the hydrogen fuel cell. fc_max This indicates the preset maximum output power of the hydrogen fuel cell;

[0058] The distribution characteristics of hydrogen fuel in the power set are analyzed to establish the power allocation space for the hydrogen fuel cell hybrid UAV.

[0059] In one feasible approach

[0060] The process of constructing the flight reward function and power supply constraints for the hydrogen fuel cell hybrid UAV includes:

[0061] The flight reward function of the hydrogen fuel cell hybrid UAV is calculated according to formula (12);

[0062]

[0063] Where F represents the flight reward function of the hydrogen fuel cell hybrid UAV, a1 and a2 represent reward weights, and a3 and a4 represent penalty weights. State of Charge (SOC) indicates the maximum hydrogen consumption. ref This indicates the current remaining battery level, while SOC indicates the initial battery level.

[0064] The power supply constraints of the hydrogen fuel cell hybrid UAV are calculated according to formula (13);

[0065]

[0066] Among them, P load P represents the load power requirement of the hydrogen fuel cell hybrid drone. fc P represents the output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone. batt P represents the lithium battery output power of the hydrogen fuel cell hybrid drone. fc_max P fc_min P represents the maximum output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone and the minimum output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone, respectively. batt_max Pbatt_min These represent the maximum power of the lithium battery in the hydrogen fuel cell hybrid drone and the minimum power of the lithium battery in the hydrogen fuel cell hybrid drone, respectively.

[0067] In one feasible approach

[0068] The process of running the trained agent to obtain and display the optimal strategy presented by the hydrogen fuel cell hybrid UAV when it converges to the flight reward function includes:

[0069] Run the training agent and input the running data into (14) to calculate the flight characteristics of the hydrogen fuel cell hybrid drone when performing flight actions;

[0070]

[0071] Where z represents the actor network weights, φ represents the commentator network weights, λ represents the smoothing coefficient, γ represents the discount factor, V(φ) represents the state value function, and ts represents the start time of the round.

[0072] The flight reward function is used to analyze the flight energy consumption of the hydrogen fuel cell hybrid UAV under each of the aforementioned flight characteristics.

[0073] The energy consumption of the flight is optimized to generate and display the optimal strategy for the hydrogen fuel cell hybrid drone.

[0074] This invention provides an energy adaptive optimization method for hydrogen fuel cell hybrid unmanned aerial vehicles, comprising:

[0075] Step 1: Build the basic equipment parameters of the hydrogen fuel cell hybrid drone, including the load demand power model, hydrogen fuel cell model, and lithium battery model.

[0076] Step 2: Based on the load demand power model, the hydrogen fuel cell model, and the lithium battery model, construct the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid UAV, as well as the flight reward function and power supply constraints of the hydrogen fuel cell hybrid UAV.

[0077] Step 3: Construct the actor neural network and commentator neural network of the hydrogen fuel cell hybrid UAV using the preset PPO algorithm and the basic equipment parameters. Use the actor neural network to perform strategy optimization on the power consumption state space and the power supply allocation space respectively to obtain several optimization conditions. Use the commentator neural network to perform feedback analysis on the optimization conditions according to the flight reward function and the power supply constraints to obtain several hyperparameters.

[0078] Step 4: Construct an energy management agent based on the hyperparameters, map the load demand power model, hydrogen fuel cell model, and lithium battery model to the energy management agent to generate a training agent for the hydrogen fuel cell hybrid UAV, run the training agent to obtain and display the optimal strategy presented by the hydrogen fuel cell hybrid UAV when it converges to the flight reward function.

[0079] The beneficial effects that can be achieved by the above technical solution are: (1) The near-end strategy optimization algorithm can continuously learn and adapt to changes in the flight environment, such as natural factors like wind speed, temperature, and air pressure, as well as internal factors like load changes and battery status. This adaptive capability enables the UAV to maintain an efficient and stable flight state under complex and ever-changing flight conditions, reducing energy waste and unnecessary flight risks.

[0080] (2) By using the near-end strategy optimization algorithm to respond to various uncertainties during flight in real time, it can adaptively and intelligently allocate energy to ensure that hydrogen fuel cells and lithium batteries work together in the best state, thereby maximizing the flight time and mission range of the UAV.

[0081] (3) This patent applies deep reinforcement learning technology to the energy management system of UAVs, realizing a leap from data-driven to intelligent decision-making. UAVs can autonomously adjust their energy allocation strategies according to their own state and external environment without human intervention or preset rules, thereby improving the intelligence level and autonomous decision-making ability of UAVs.

[0082] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0083] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0084] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0085] Figure 1 The load demand power curve for a fixed flight mission without considering uncertainties;

[0086] Figure 2 Power demand curve for UAV load considering uncertainties under fixed flight missions;

[0087] Figure 3 Here is a flowchart of the PPO algorithm training process;

[0088] Figure 4 Diagram of drone energy management hardware implementation;

[0089] Figure 5 Energy distribution curves for test cases;

[0090] Figure 6 The SOC curve of the lithium battery for the test case;

[0091] Figure 7 The hydrogen consumption curve of the hydrogen fuel cell for the test case;

[0092] Figure 8 This is a schematic diagram of the components of the present invention. Detailed Implementation

[0093] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0094] Example 1

[0095] This embodiment provides an energy adaptive optimization system for hydrogen fuel cell hybrid unmanned aerial vehicles, such as... Figure 8 As shown, it includes:

[0096] The sub-modeling module is used to build the load demand power model, hydrogen fuel cell model, and lithium battery model of the hydrogen fuel cell hybrid UAV based on the basic equipment parameters of the UAV.

[0097] The model analysis module is used to build the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid UAV based on the load demand power model, the hydrogen fuel cell model and the lithium battery model, as well as to build the flight reward function and power supply constraints of the hydrogen fuel cell hybrid UAV.

[0098] The algorithm optimization module is used to construct the actor neural network and commentator neural network of the hydrogen fuel cell hybrid UAV using the preset PPO algorithm and the basic equipment parameters. The actor neural network is used to optimize the power consumption state space and the power supply allocation space respectively to obtain several optimization conditions. The commentator neural network is used to perform feedback analysis on the optimization conditions according to the flight reward function and the power supply constraints to obtain several hyperparameters.

[0099] An optimized training module is used to construct an energy management agent based on the hyperparameters, and to map the load demand power model, the hydrogen fuel cell model, and the lithium battery model to the energy management agent to generate a training agent for the hydrogen fuel cell hybrid UAV. The training agent is then run to obtain and display the optimal strategy presented by the hydrogen fuel cell hybrid UAV when it converges to the flight reward function.

[0100] In this example, hyperparameters include sampling time, discount factor, number of steps to calculate the advantage function, learning rate of the neural networks for actors and commentators, gradient threshold, etc.

[0101] In this example, the default PPO algorithm represents a reinforcement learning algorithm.

[0102] The working principle and beneficial effects of the above technical solution are as follows: To solve the problems in traditional technologies, when a hydrogen fuel cell hybrid drone is in flight, a corresponding load demand power model, a hydrogen fuel cell model, and a lithium battery model are built. Then, a corresponding power consumption state space and power supply allocation space are further built, which can more clearly show the power consumption of the hydrogen fuel cell hybrid drone during flight. In order to adapt the system to more different types of hydrogen fuel cell hybrid drones, a flight reward function and power supply constraints unique to the hydrogen fuel cell hybrid drone are built. Then, actor neural networks and commentator neural networks are used to optimize the power consumption state space and power supply allocation space and perform feedback processing to generate corresponding hyperparameters. Finally, the hyperparameters are used to build a training agent. By running the training agent, the optimal strategy of the hydrogen fuel cell hybrid drone is analyzed, realizing the adaptive intelligent optimal allocation of hydrogen energy and electric energy by the hydrogen fuel cell hybrid drone energy management system.

[0103] Example 2

[0104] Based on Example 1, the process of establishing the load demand power model, the hydrogen fuel cell model, and the lithium battery model is as follows:

[0105] The process of building the load demand power model for the aforementioned hydrogen fuel cell hybrid drone includes:

[0106] The induced power of the hydrogen fuel cell hybrid drone is calculated according to formula (1);

[0107] P ik =T d ·v ik (1)

[0108] Among them, P ik The induced power, v, represents the induced power of the hydrogen fuel cell hybrid drone. ikThis indicates the actual induced velocity of the hydrogen fuel cell hybrid drone;

[0109] The drag power of the hydrogen fuel cell hybrid UAV is calculated according to formula (2) (representing the viscous drag force encountered by the rotor rotating in the air).

[0110]

[0111] Among them, P pr The form drag power of the hydrogen fuel cell hybrid UAV is represented by As, the cross-sectional area of ​​the UAV propeller blade, ω, the angular velocity of the blade rotation, R, the radius of the turntable, and δ, the tensile strength coefficient of the profile.

[0112] The exhaust resistance power of the hydrogen fuel cell hybrid drone is calculated according to formula (3);

[0113]

[0114] Among them, P pa This indicates the exhaust resistance power of the hydrogen fuel cell hybrid drone;

[0115] The interference power of the hydrogen fuel cell hybrid UAV is calculated according to formula (4);

[0116] This is used to quantify other disturbances that a drone may encounter during its flight mission. These disturbances may include, but are not limited to, changes in ambient temperature, fluctuations in air density, minor mechanical vibrations of the drone itself, and the effects caused by changes in external magnetic fields.

[0117] P dis (t)~N(0,1)×c,P dis_min <P dis (t) <P dis_max (4)

[0118] Where c represents the disturbance amplitude coefficient of the hydrogen fuel cell hybrid drone;

[0119] The load mass of the hydrogen fuel cell hybrid UAV at different flight times is calculated according to formula (5);

[0120] The total payload mass is not constant. As the drone continuously consumes hydrogen during flight, the total mass of the drone will continuously decrease.

[0121]

[0122] Where m(t) represents the total weight of the hydrogen fuel cell hybrid drone at time t. The value represents the amount of hydrogen consumed, M represents the molar mass of hydrogen, P represents the operating pressure of the hydrogen fuel cell stack, R represents Avogadro's constant, and T represents the operating thermodynamic temperature of the stack.

[0123] Based on the calculation results of formula (5), determine the load mass of the hydrogen fuel cell hybrid UAV at different flight times, and build a load demand power model of the fuel cell hybrid UAV.

[0124] The hydrogen fuel consumption of the hydrogen fuel cell hybrid UAV per unit time is calculated according to formula (8).

[0125]

[0126] in, P represents the hydrogen fuel consumption per unit time of the hydrogen fuel cell hybrid drone. fc This indicates the output power of the hydrogen fuel cell;

[0127] A hydrogen fuel cell model for the hydrogen fuel cell hybrid UAV is established based on the hydrogen fuel consumption.

[0128] The discharge cycle output power of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (9), the output current of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (10), and the state of charge of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (11).

[0129]

[0130] Among them, V OC R0 represents the open-circuit voltage, I represents the internal resistance, and R0 represents the open-circuit voltage. b P represents the output current. batt State of Charge (SOC) represents the output power during a charge-discharge cycle. t Indicates the state of charge, Q0 represents the initial battery capacity, and Q represents the rated battery capacity;

[0131] The lithium battery model of the hydrogen fuel cell hybrid UAV is established based on the calculation results of formulas (9), (10) and (11).

[0132] The working principle and beneficial effects of the above technical solution are as follows: By building multiple models to represent the various parts of the hydrogen fuel cell hybrid drone, the amount of computation can be reduced and various data can be obtained intuitively.

[0133] Example 3

[0134] Based on Example 2, the process for determining the actual induced velocity of the hydrogen fuel cell hybrid drone is as follows:

[0135] Calculate the thrust required by the hydrogen fuel cell hybrid UAV during flight according to formula (6);

[0136]

[0137] Among them, T d a represents the thrust required by the hydrogen fuel cell hybrid drone during flight. c v represents the acceleration of the hydrogen fuel cell hybrid drone, and v represents the airspeed of the hydrogen fuel cell hybrid drone. f Let β represent the air velocity, β represent the angle between the air velocity direction and the flight direction of the hydrogen fuel cell hybrid drone, θ represent the angle between the flight direction of the hydrogen fuel cell hybrid drone and the vertical direction, ρ represent the air density, and A represent the air velocity. f The vertical projected area of ​​the hydrogen fuel cell hybrid drone is represented by m, the total mass of the hydrogen fuel cell hybrid drone (including the fixed weight of the drone body, lithium battery, hydrogen fuel cell, hydrogen storage tank, etc., as well as the weight of hydrogen), and g is the acceleration due to gravity.

[0138] (The drone uses a quadcopter configuration);

[0139] The force values ​​corresponding to each propeller disk on the hydrogen fuel cell hybrid drone were obtained separately.

[0140] When the force value corresponding to each of the propeller disks is the same, it is determined that the hydrogen fuel cell hybrid drone is in normal flight state.

[0141] When the hydrogen fuel cell hybrid drone is in normal flight, the theoretical speed of the hydrogen fuel cell hybrid drone is calculated according to formula (7);

[0142]

[0143] Among them, S T A represents the calculation of the real-time speed of the hydrogen fuel cell hybrid drone. p γ represents the horizontal projected area of ​​the hydrogen fuel cell hybrid drone, γ represents the tilt angle of the hydrogen fuel cell hybrid drone, and v in Indicates the theoretical induced velocity;

[0144] The actual speed of the hydrogen fuel cell hybrid drone is obtained, and the actual speed is substituted into the formula (7) to obtain the actual induced speed of the hydrogen fuel cell hybrid drone.

[0145] The working principle and beneficial effects of the above technical solution are as follows: using tension to derive the induced velocity can reduce the amount of calculation required for normal calculation, improve the calculation efficiency of induced power, and improve the optimization efficiency.

[0146] Example 4

[0147] Based on Example 1, the process of constructing the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid drone includes:

[0148] The load demand power P of the hydrogen fuel cell hybrid UAV at the current moment is determined using the aforementioned load demand power model. t ;

[0149] The hydrogen fuel cell model is used to determine the remaining hydrogen amount of the hydrogen fuel cell hybrid UAV at the current moment.

[0150] The remaining state of charge (SOC) of the hydrogen fuel cell hybrid drone at the current moment is determined using the lithium battery model. t ;

[0151] Based on the current load power demand P of the hydrogen fuel cell hybrid drone. t Remaining amount of hydrogen Remaining power SOC t Establish the power consumption state space of the hydrogen fuel cell hybrid UAV;

[0152] The hydrogen fuel cell output power of the hydrogen fuel cell hybrid UAV is determined based on the hydrogen fuel cell model, and the hydrogen fuel cell output power is set within a preset continuous range [P]. fc_min ,P fc_max The power set is obtained by discretizing the [] [] [] [] [].

[0153] Among them, P fc_min P represents the preset minimum output power of the hydrogen fuel cell. fc_max This indicates the preset maximum output power of the hydrogen fuel cell;

[0154] The distribution characteristics of hydrogen fuel in the power set are analyzed to establish the power allocation space for the hydrogen fuel cell hybrid UAV.

[0155] The working principle and beneficial effects of the above technical solution are as follows: By building a power consumption state space and a power supply allocation space, the flight operation of the UAV can be further analyzed in detail, which improves the accuracy of optimization.

[0156] Example 5

[0157] Based on Example 1, the process of constructing the flight reward function and power supply constraints for the hydrogen fuel cell hybrid UAV includes:

[0158] The flight reward function of the hydrogen fuel cell hybrid UAV is calculated according to formula (12);

[0159]

[0160] Where F represents the flight reward function of the hydrogen fuel cell hybrid UAV, a1 and a2 represent reward weights, and a3 and a4 represent penalty weights. State of Charge (SOC) indicates the maximum hydrogen consumption. ref This indicates the current remaining battery level, while SOC indicates the initial battery level.

[0161] The power supply constraints of the hydrogen fuel cell hybrid UAV are calculated according to formula (13);

[0162]

[0163] Among them, P load P represents the load power requirement of the hydrogen fuel cell hybrid drone. fc P represents the output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone. batt P represents the lithium battery output power of the hydrogen fuel cell hybrid drone. fc_max P fc_min P represents the maximum output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone and the minimum output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone, respectively. batt_max P batt_min These represent the maximum power of the lithium battery in the hydrogen fuel cell hybrid drone and the minimum power of the lithium battery in the hydrogen fuel cell hybrid drone, respectively.

[0164] The working principle and beneficial effects of the above technical solution are as follows: Since different drones have different characteristics, by analyzing the flight reward function and power supply constraints of the hydrogen fuel cell hybrid drone, the unique flight conditions of the drone can be determined, which facilitates subsequent optimization work.

[0165] Example 6

[0166] Based on Example 1, the process of running the training agent to obtain and display the optimal strategy presented by the hydrogen fuel cell hybrid UAV when it converges to the flight reward function includes:

[0167] Run the training agent and input the running data into (14) to calculate the flight characteristics of the hydrogen fuel cell hybrid drone when performing flight actions;

[0168]

[0169] Where z represents the actor network weights, φ represents the commentator network weights, λ represents the smoothing coefficient, γ represents the discount factor, V(φ) represents the state value function, and ts represents the start time of the round.

[0170] The flight reward function is used to analyze the flight energy consumption of the hydrogen fuel cell hybrid UAV under each of the aforementioned flight characteristics.

[0171] The energy consumption of the flight is optimized to generate and display the optimal strategy for the hydrogen fuel cell hybrid drone.

[0172] The working principle and beneficial effects of the above technical solution are: it improves the overall performance and reliability of the drone.

[0173] Example 7

[0174] This embodiment provides an energy adaptive optimization method for hydrogen fuel cell hybrid unmanned aerial vehicles, characterized by including:

[0175] Step 1: Build the basic equipment parameters of the hydrogen fuel cell hybrid drone, including the load demand power model, hydrogen fuel cell model, and lithium battery model.

[0176] Step 2: Based on the load demand power model, the hydrogen fuel cell model, and the lithium battery model, construct the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid UAV, as well as the flight reward function and power supply constraints of the hydrogen fuel cell hybrid UAV.

[0177] Step 3: Construct the actor neural network and commentator neural network of the hydrogen fuel cell hybrid UAV using the preset PPO algorithm and the basic equipment parameters. Use the actor neural network to perform strategy optimization on the power consumption state space and the power supply allocation space respectively to obtain several optimization conditions. Use the commentator neural network to perform feedback analysis on the optimization conditions according to the flight reward function and the power supply constraints to obtain several hyperparameters.

[0178] Step 4: Construct an energy management agent based on the hyperparameters, map the load demand power model, hydrogen fuel cell model, and lithium battery model to the energy management agent to generate a training agent for the hydrogen fuel cell hybrid UAV, run the training agent to obtain and display the optimal strategy presented by the hydrogen fuel cell hybrid UAV when it converges to the flight reward function.

[0179] In this example, hyperparameters include sampling time, discount factor, number of steps to calculate the advantage function, learning rate of the neural networks for actors and commentators, gradient threshold, etc.

[0180] In this example, the default PPO algorithm represents a reinforcement learning algorithm.

[0181] The working principle and beneficial effects of the above technical solution are as follows: To solve the problems in traditional technologies, when a hydrogen fuel cell hybrid drone is in flight, a corresponding load demand power model, a hydrogen fuel cell model, and a lithium battery model are built. Then, a corresponding power consumption state space and power supply allocation space are further built, which can more clearly show the power consumption of the hydrogen fuel cell hybrid drone during flight. In order to adapt the system to more different types of hydrogen fuel cell hybrid drones, a flight reward function and power supply constraints unique to the hydrogen fuel cell hybrid drone are built. Then, actor neural networks and commentator neural networks are used to optimize the power consumption state space and power supply allocation space and perform feedback processing to generate corresponding hyperparameters. Finally, the hyperparameters are used to build a training agent. By running the training agent, the optimal strategy of the hydrogen fuel cell hybrid drone is analyzed, realizing the adaptive intelligent optimal allocation of hydrogen energy and electric energy by the hydrogen fuel cell hybrid drone energy management system.

[0182] This method uses the PPO algorithm to train the energy management agent. The training process is as follows: ① Initialize the weights of the actor network and commentator network, and the experience replay pool. ② Start training. The energy management system will continuously interact with the hydrogen fuel cell hybrid drone and collect trajectory data {s(t), a(t), r(t+1), s(t+1), log p(a)}. t |s t The data is stored in the experience replay pool. ③ Once the experience replay pool is full, the energy management agent stops interacting with the hydrogen fuel cell hybrid drone and instead retrieves data from the experience replay pool. It then uses the commentator network to estimate the state value, uses formula xx to calculate the advantage function of the state-action pair to update the weights of the policy network, and uses formula xx to calculate the loss function to update the weights of the commentator network. This process is repeated a certain number of times. ④ After the update is complete, the experience replay pool is emptied, and steps 2, 3, and 4 are repeated.

[0183] The following examples illustrate this point:

[0184] To comprehensively simulate the power consumption characteristics of UAVs performing typical flight missions, UAV flight missions as shown in Table 1 were set. Based on the aforementioned UAV load demand power model, power demand curves were obtained, as shown below. Figure 1 , Figure 2 As shown.

[0185] Table 1 Description of UAV Flight Missions

[0186]

[0187] The agent was trained using the Simulink platform, and the relevant algorithm parameters are shown in Table 2. After training for a certain number of rounds and waiting for the agent to converge, the agent was saved. A new load curve was randomly generated as the input to the agent to test its generalization ability. The test results are shown in Table 2. Figure 5 , Figure 6 , Figure 7 As shown.

[0188] Table 2 PPO Training Parameter Settings

[0189]

[0190] As the results show, the fuel cell provided significant power during takeoff to meet high load demands. During cruise, the fuel cell power remained at a high level to sustain flight, while hydrogen was consumed steadily throughout the entire flight profile. The lithium-ion battery power increased rapidly in the early stages of flight, providing auxiliary power to support the fuel cell. Subsequently, the lithium-ion battery power fluctuated at a lower level, balancing load demands and fuel cell output. In the mid-to-late stages of flight, the lithium-ion battery power gradually decreased and tended towards negative values, as the fuel cell recharged the battery. The State of Charge (SOC) remained near the reference value of 0.8 throughout the flight, indicating that the energy management strategy effectively protected the lithium-ion battery, avoiding over-discharge and over-charge, and contributing to extended battery life.

[0191] The energy management system topology corresponding to this example is as follows: Figure 4As shown. The system mainly includes an energy management system, a drone, a proton exchange membrane fuel cell (PEMFC), and lithium batteries. In the entire system, the PEMFC provides the primary flight power and is connected to the DC bus load via a unidirectional DC / DC converter. To meet the load power requirements under rapidly changing operating conditions, lithium batteries are used as auxiliary energy storage devices and are connected to the DC bus via a bidirectional DC / DC converter. The energy management system, as the core control unit, is responsible for the energy management and execution of the entire system's control strategy. It receives reference signals (SOC, SH2, P...). t ), and generate the energy output (P) of each part of the UAV functional system according to the algorithm. FC P Batt The power control module is based on P FC P Batt Calculate the switching frequencies (Feq) input to the LM5118 and TPS40210. FC 、Feq Batt ) and duty cycle (D FC D Batt This allows for the control of the power output of the PEMFC and lithium battery, thus enabling intelligent allocation of the drone's energy system.

[0192] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A hydrogen fuel cell hybrid unmanned aerial vehicle (UAV) energy adaptive optimization system, characterized in that, include: The sub-modeling module is used to build the load demand power model, hydrogen fuel cell model, and lithium battery model of the hydrogen fuel cell hybrid UAV based on the basic equipment parameters of the UAV. The model analysis module is used to build the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid UAV based on the load demand power model, the hydrogen fuel cell model and the lithium battery model, as well as to build the flight reward function and power supply constraints of the hydrogen fuel cell hybrid UAV. The algorithm optimization module is used to construct the actor neural network and commentator neural network of the hydrogen fuel cell hybrid UAV using the preset PPO algorithm and the basic equipment parameters. The actor neural network is used to optimize the power consumption state space and the power supply allocation space respectively to obtain several optimization conditions. The commentator neural network is used to perform feedback analysis on the optimization conditions according to the flight reward function and the power supply constraints to obtain several hyperparameters. The optimization training module is used to construct an energy management agent based on the hyperparameters, map the load demand power model, the hydrogen fuel cell model, and the lithium battery model to the energy management agent to generate the training agent of the hydrogen fuel cell hybrid UAV, run the training agent to obtain the optimal strategy presented by the hydrogen fuel cell hybrid UAV when it converges to the flight reward function, and display it. The process of constructing the flight reward function and power supply constraints for the hydrogen fuel cell hybrid UAV includes: The flight reward function of the hydrogen fuel cell hybrid UAV is calculated according to formula (12); （12）， in, This represents the flight reward function of the hydrogen fuel cell hybrid drone. Indicates the reward weight. Indicates the penalty weight. This indicates the amount of hydrogen fuel consumed by the hydrogen fuel cell hybrid drone per unit time. This indicates the maximum hydrogen consumption. Indicates the current remaining battery power. Indicates the initial battery level; The power supply constraints of the hydrogen fuel cell hybrid UAV are calculated according to formula (13); （13）， in, This indicates the load power requirement of the hydrogen fuel cell hybrid drone. This indicates the output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone. This indicates the lithium battery output power of the hydrogen fuel cell hybrid drone. , These represent the maximum output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone and the minimum output power of the hydrogen fuel cell in the hydrogen fuel cell hybrid drone, respectively. , These represent the maximum power of the lithium battery in the hydrogen fuel cell hybrid drone and the minimum power of the lithium battery in the hydrogen fuel cell hybrid drone, respectively.

2. The hydrogen fuel cell hybrid unmanned aerial vehicle energy adaptive optimization system as described in claim 1, characterized in that, The process of building the load demand power model for the aforementioned hydrogen fuel cell hybrid drone includes: The induced power of the hydrogen fuel cell hybrid drone is calculated according to formula (1); （1）， in, This indicates the induced power of the hydrogen fuel cell hybrid drone. This indicates the thrust required by the hydrogen fuel cell hybrid drone during flight. This indicates the actual induced velocity of the hydrogen fuel cell hybrid drone; The drag power of the hydrogen fuel cell hybrid UAV is calculated according to formula (2); （2）， in, This indicates the drag power of the hydrogen fuel cell hybrid drone. Indicates air density, R represents the cross-sectional area of ​​the UAV propeller blade, ω represents the angular velocity of the blade rotation, and R represents the radius of the turntable. Indicates the tensile strength coefficient of the profile. This indicates the airspeed of the hydrogen fuel cell hybrid unmanned aerial vehicle. The exhaust resistance power of the hydrogen fuel cell hybrid drone is calculated according to formula (3); （3）， in, This indicates the exhaust resistance power of the hydrogen fuel cell hybrid drone. Indicates air velocity. This represents the angle between the direction of airflow and the flight direction of the hydrogen fuel cell hybrid drone. This represents the vertical projected area of ​​a hydrogen fuel cell hybrid drone. The interference power of the hydrogen fuel cell hybrid UAV at time t is calculated according to formula (4). ; （4）， in, It follows a standard normal distribution with a mean of 0 and a standard value of 1. This represents the disturbance amplitude coefficient of the hydrogen fuel cell hybrid drone. and These are the minimum and maximum values ​​of the interference power set. The load mass of the hydrogen fuel cell hybrid UAV at different flight times is calculated according to formula (5); （5）， in, express The total weight of the hydrogen fuel cell hybrid drone at any given moment. This indicates the amount of hydrogen fuel consumed by the hydrogen fuel cell hybrid drone per unit time. This indicates the molar mass of hydrogen. T represents Avogadro's constant, and T represents the operating thermodynamic temperature of the fuel cell stack. For time intervals; Based on the calculation results of formula (5), the load mass corresponding to the hydrogen fuel cell hybrid UAV at different flight times is determined, and the load demand power model of the fuel cell hybrid UAV is built.

3. The hydrogen fuel cell hybrid unmanned aerial vehicle energy adaptive optimization system as described in claim 2, characterized in that, Also includes: Calculate the thrust required by the hydrogen fuel cell hybrid UAV during flight according to formula (6); （6）， in, This indicates the acceleration of the hydrogen fuel cell hybrid drone. This indicates the angle between the flight direction and the vertical direction of the hydrogen fuel cell hybrid unmanned aerial vehicle. This represents the vertical projected area of ​​a hydrogen fuel cell hybrid drone. The mass of the hydrogen fuel cell hybrid drone is represented by g, where g represents the acceleration due to gravity. The force values ​​corresponding to each propeller disk on the hydrogen fuel cell hybrid drone were obtained separately. When the force value corresponding to each of the propeller disks is the same, it is determined that the hydrogen fuel cell hybrid drone is in normal flight state. When the hydrogen fuel cell hybrid drone is in normal flight, the theoretical speed of the hydrogen fuel cell hybrid drone is calculated according to formula (7); （7）， in, This indicates the calculation of the real-time speed of the hydrogen fuel cell hybrid drone. This represents the horizontal projected area of ​​the hydrogen fuel cell hybrid drone. This indicates the tilt angle of the hydrogen fuel cell hybrid drone. Indicates the theoretical induced velocity; Obtain the actual speed of the hydrogen fuel cell hybrid drone, and substitute the actual speed into the formula (7) to obtain the actual induced speed of the hydrogen fuel cell hybrid drone.

4. The hydrogen fuel cell hybrid unmanned aerial vehicle energy adaptive optimization system as described in claim 1, characterized in that, The process of building the hydrogen fuel cell model and lithium battery model of the aforementioned hydrogen fuel cell hybrid UAV includes: The hydrogen fuel consumption per unit time of the hydrogen fuel cell hybrid drone is calculated according to formula (8). （8）， in, This indicates the amount of hydrogen fuel consumed by the hydrogen fuel cell hybrid drone per unit time. This indicates the output power of the hydrogen fuel cell; A hydrogen fuel cell model for the hydrogen fuel cell hybrid UAV is established based on the hydrogen fuel consumption. The discharge cycle output power of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (9), the output current of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (10), and the state of charge of the lithium battery of the hydrogen fuel cell hybrid drone is calculated according to formula (11). （9）， （10）， （11）， in, This represents the open-circuit voltage at time t. Indicates internal resistance. This represents the output current at time t. This represents the output power during the charge-discharge cycle at time t. This represents the state of charge at time t. Indicates the initial battery capacity. Indicates the rated battery capacity; The lithium battery model of the hydrogen fuel cell hybrid UAV is established based on the calculation results of formulas (9), (10) and (11).

5. The hydrogen fuel cell hybrid unmanned aerial vehicle energy adaptive optimization system as described in claim 1, characterized in that, The process of constructing the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid UAV includes: The load demand power model is used to determine the load demand power of the hydrogen fuel cell hybrid UAV at the current moment. ; The hydrogen fuel cell model is used to determine the remaining hydrogen amount of the hydrogen fuel cell hybrid UAV at the current moment. ; The remaining power of the hydrogen fuel cell hybrid drone at the current moment is determined using the lithium battery model. ; Based on the current load power requirement of the hydrogen fuel cell hybrid drone. Remaining amount of hydrogen Remaining battery power Establish the power consumption state space of the hydrogen fuel cell hybrid UAV; The hydrogen fuel cell output power of the hydrogen fuel cell hybrid UAV is determined based on the hydrogen fuel cell model, and the hydrogen fuel cell output power is set within a preset continuous range. The power set is obtained by discretization. in, This indicates the preset minimum output power of the hydrogen fuel cell. This indicates the preset maximum output power of the hydrogen fuel cell; The distribution characteristics of hydrogen fuel in the power set are analyzed to establish the power allocation space for the hydrogen fuel cell hybrid UAV.

6. A method for adaptive energy optimization of a hydrogen fuel cell hybrid unmanned aerial vehicle (UAV), employing the adaptive energy optimization system for a hydrogen fuel cell hybrid UAV as described in any one of claims 1-5, characterized in that, include: Step 1: Build the basic equipment parameters of the hydrogen fuel cell hybrid drone, including the load demand power model, hydrogen fuel cell model, and lithium battery model. Step 2: Based on the load demand power model, the hydrogen fuel cell model, and the lithium battery model, construct the power consumption state space and power supply allocation space of the hydrogen fuel cell hybrid UAV, as well as the flight reward function and power supply constraints of the hydrogen fuel cell hybrid UAV. Step 3: Construct the actor neural network and commentator neural network of the hydrogen fuel cell hybrid UAV using the preset PPO algorithm and the basic equipment parameters. Use the actor neural network to perform strategy optimization on the power consumption state space and the power supply allocation space respectively to obtain several optimization conditions. Use the commentator neural network to perform feedback analysis on the optimization conditions according to the flight reward function and the power supply constraints to obtain several hyperparameters. Step 4: Construct an energy management agent based on the hyperparameters, map the load demand power model, hydrogen fuel cell model, and lithium battery model to the energy management agent to generate a training agent for the hydrogen fuel cell hybrid UAV, run the training agent to obtain and display the optimal strategy presented by the hydrogen fuel cell hybrid UAV when it converges to the flight reward function.

Images