Hybrid excitation operator-based hydrogen energy unmanned aerial vehicle health monitoring and management method

Abstract

The invention provides a hydrogen energy unmanned aerial vehicle health monitoring and management method based on a hybrid excitation operator, and belongs to the field of hydrogen energy unmanned aerial vehicles, and the method comprises the steps: building a hydrogen fuel cell voltage estimator based on online parameter identification, estimating voltage output in real time, generating a voltage residual error with a voltage measurement value, and obtaining the maximum output power of a fuel cell; a hybrid excitation operator is designed, high-frequency excitation is injected into the fuel cell through a DC / DC converter, low-frequency excitation which does not influence flight stability is injected through maneuvering of the active unmanned aerial vehicle, characteristic frequency impedance is calculated online by using an injected excitation signal and a response voltage signal, and the characteristic frequency impedance is used as a fault diagnosis characteristic; designing a fault classifier, and classifying the faults into water logging, film dryness and oxygen deficiency based on the characteristic frequency impedance; and based on the diagnosed specific fault type, controlling components such as an air inlet fan and an exhaust valve to perform fault recovery. According to the invention, online health management of the hydrogen energy unmanned aerial vehicle is realized, and the method is suitable for a hydrogen energy unmanned aerial vehicle system requiring online health management.

Description

Technical Field

[0001] The present invention belongs to the field of hydrogen energy drones, and particularly relates to a health monitoring and management method for hydrogen energy drones based on a hybrid excitation operator. Background Art

[0002] Hydrogen energy drones have the advantages of high energy density, long endurance, and pollution-free emissions, and play an important role in zero-carbon aviation. However, their safety and reliability still need to be improved urgently. As the power core of hydrogen-powered drones, the health management of fuel cells is crucial for improving system safety and extending its service life. The high-dynamic and multi-scenario characteristics of hydrogen energy drones bring many challenges to system health management: existing fuel cell fault diagnosis methods can be divided into two categories based on the health indicators used, namely voltage-based and impedance-based. The high-dynamic characteristics of drone applications determine that the impedance-based method cannot operate continuously online because it relies on excitation injection, and continuous injection of excitation will interfere with drone flight and even pose a threat to flight safety. The voltage-based method contains less health information, is difficult to distinguish different fault types, and cannot take targeted recovery measures. Therefore, how to identify and classify fuel cell faults online is a major challenge. In addition, the change in environmental conditions brought about by the multi-scenario application of drones will cause changes in the characteristics of fuel cells, making the residual index based on the static fuel cell model no longer reliable, which will increase the false alarm rate of faults and pose a challenge to the environmental adaptability of the management method.

[0003] At present, there is no dedicated health management method for hydrogen energy drones in the publicly available materials, while research has been conducted on fault diagnosis methods for isolated fuel cells or fuel cell / lithium battery hybrid systems. Chinese Patent Application CN113447843A analyzes the electrochemical impedance spectrum of fuel cells by establishing an equivalent circuit model, extracts impedance spectrum characteristic parameters, and classifies the health loss degree of fuel cells as high, medium, and low using fuzzy rules. However, this method cannot operate continuously online and cannot take targeted recovery measures according to different health states. Chinese Patent Application CN112117475A diagnoses flooding and membrane drying faults of fuel cells by training a locally preserving projection and learning vector quantization neural network based on voltage experimental data. Chinese Patent Application CN119165369A realizes voltage-based fault diagnosis by training a fuel cell fault diagnosis model. However, these methods rely on a large amount of experimental data, have high requirements for the deployed hardware platform, and are difficult to apply to different fuel cells or cope with the fuel cell characteristic drift caused by changes in the drone environment. "C. Yan, J. Chen, H. Liu, L. Kumar, and H. Lu, “Health management for pem fuel cells based on an active fault tolerant control strategy,” IEEE Transactions on Sustainable Energy, vol. 12, no. 2, pp. 1311 1320, 2020" proposes an algorithm for online fault detection based on a fuel cell voltage model and fault diagnosis using characteristic frequency impedance spectrum information. However, the fuel cell model adopted is a static model, which does not consider the inherent characteristics of fuel cell characteristic drift deviating from the preset model in the multi-scenario application of drones and lacks environmental adaptability. At the same time, the above methods do not consider the impact of fault monitoring or diagnosis algorithms on the flight of hydrogen energy drones, nor do they consider the problem of excessive energy supply pressure in the energy system during the excitation injection process, and cannot be directly applied to hydrogen energy drones.

[0004] In summary, there is currently a lack of a health monitoring and management method designed for hydrogen energy drones. In the application of drones, existing fuel cell fault diagnosis methods cannot achieve continuous monitoring and diagnosis, affect flight, have high energy supply pressure, and lack environmental adaptability, and cannot be directly applied to the high-dynamic and multi-scenario applications of hydrogen energy drones. It is urgent to overcome the closed-loop health management problem applicable to hydrogen energy drones, which has the capabilities of online monitoring, diagnosis, and recovery. Summary of the Invention

[0005] In view of the multi-scenario and high-dynamic characteristics of hydrogen energy drones, there is currently no health management strategy specifically designed for hydrogen energy drones. Moreover, the existing fuel cell health management strategies have problems such as being unable to achieve continuous monitoring and diagnosis, affecting flight, having a large power supply pressure, and insufficient environmental adaptability, and cannot be directly applied to hydrogen energy drones. The present invention provides a health monitoring and management method for hydrogen energy drones based on a hybrid excitation operator. First, a hydrogen fuel cell voltage estimator based on online parameter identification is established to estimate the voltage output in real time and generate a voltage residual with the measured voltage value. At the same time, the maximum output power of the fuel cell is obtained to conduct health monitoring and take emergency landing for the special case of insufficient maximum output power of the fuel cell. Then, a hybrid excitation operator is designed to inject high-frequency excitation into the fuel cell through a DC / DC converter and inject low-frequency excitation that does not affect flight stability through the active maneuver of the drone. The characteristic frequency impedance is calculated online using the injected excitation signals and the response voltage signals and used as a fault diagnosis feature. Secondly, a fault classifier is designed to classify faults into three categories: flooding, membrane drying, and oxygen deficiency based on the characteristic frequency impedance. Finally, based on the specific fault type diagnosed, components such as the intake fan and exhaust valve are controlled to recover from the fault. The present invention realizes the online health management of hydrogen energy drones, realizes functions such as online fault monitoring and recovery and online fault diagnosis of the fuel cell and the drone system, has advantages such as strong environmental adaptability, not damaging flight stability, and reducing the pressure of the energy subsystem, and is applicable to hydrogen energy drone systems that require online health management.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A health monitoring and management method for hydrogen energy drones based on a hybrid excitation operator, comprising the following steps:

[0008] First step, establish a hydrogen fuel cell voltage estimator based on online parameter identification to estimate the voltage output in real time, generate a voltage residual with the measured voltage value, and at the same time obtain the maximum output power of the fuel cell to conduct health monitoring and take emergency landing for the situation of insufficient maximum output power of the fuel cell;

[0009] Second step, design a hybrid excitation operator to inject high-frequency excitation into the fuel cell through a DC / DC converter and inject low-frequency excitation that does not affect flight stability through the active and maneuver of the drone, calculate the characteristic frequency impedance online using the injected high-frequency excitation, low-frequency excitation and the response voltage signal, and use the characteristic frequency impedance as a fault diagnosis feature;

[0010] Third step, design a fault classifier to classify the fault types into three categories: flooding, membrane drying, and oxygen deficiency based on the characteristic frequency impedance;

[0011] Fourth step, control the exhaust components to recover from the fault based on the specific fault type diagnosed.

[0012] The beneficial effects of the present invention compared with the prior art are as follows:

[0013] In view of the problems in the design of hydrogen - energy UAVs that the existing fuel - cell health management strategies cannot achieve continuous monitoring and diagnosis, affect flight, have a large energy - supply pressure, and lack environmental adaptability, and cannot be directly applied to high - dynamic and multi - scenario hydrogen - energy UAV applications, a system - level health monitoring and management framework including four steps of health monitoring, hybrid excitation operator injection, fault diagnosis, and fault recovery is analyzed and designed; the proposed systematic management framework of continuous monitoring - excitation injection - fault diagnosis - fault recovery realizes online continuous health monitoring and trigger - type fault diagnosis of hydrogen - energy UAVs through a fuel - cell voltage estimator and a hybrid excitation operator, and classifies faults online using characteristic frequency impedance as a diagnosis feature and takes targeted recovery measures according to the fault category to achieve closed - loop management; meanwhile, the proposed hybrid excitation operator makes full use of the redundant control channels of quadrotors, injects low - frequency excitation by designing a periodic yaw maneuver signal to change the load power demand, and designs a sliding - mode flight controller to ensure the stability of UAV trajectory tracking under the condition of low - frequency excitation injection. Under the condition of ensuring flight performance, it greatly compensates for the power fluctuation borne by the lithium - battery during excitation injection, reduces the energy - supply pressure of the energy system, and extends its life. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is an algorithm framework diagram of a health monitoring and management method for hydrogen - energy UAVs based on a hybrid excitation operator according to an embodiment of the present invention;

[0015] Figure 2 It is a flow chart of the hybrid excitation operator in the present invention;

[0016] Figure 3 It is a lithium - battery output power diagram when injecting low - frequency excitation using a DC / DC converter;

[0017] Figure 4 It is a lithium - battery output power diagram when injecting low - frequency excitation using the hybrid excitation operator of the present invention;

[0018] Figure 5 It is a frequency diagram of the lithium - battery output power fluctuation when injecting low - frequency excitation using a DC / DC converter and when injecting low - frequency excitation using the hybrid excitation operator of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other. As Figure 1 shown, a health monitoring and management method for a hydrogen energy unmanned aerial vehicle based on a hybrid excitation operator according to an embodiment of the present invention includes the following steps:

[0020] First step, establish a hydrogen fuel cell voltage estimator based on online parameter identification, estimate the voltage output in real time and generate a voltage residual with the voltage measurement value, and at the same time obtain the maximum output power of the fuel cell, so as to conduct health monitoring and take emergency landing for special cases where the maximum output power of the fuel cell is insufficient;

[0021] Second step, design a hybrid excitation operator, inject high-frequency excitation into the fuel cell through a DC / DC converter, actively inject low-frequency excitation that does not affect flight stability through the maneuver of the unmanned aerial vehicle, and calculate the characteristic frequency impedance online using the injected excitation signal and the response voltage signal, and use it as a fault diagnosis feature;

[0022] Third step, design a fault classifier, and classify faults into three categories: water flooding, membrane drying, and oxygen deficiency based on the characteristic frequency impedance;

[0023] Fourth step, control the exhaust components such as the intake fan and exhaust valve based on the specific fault type diagnosed for fault recovery.

[0024] The present invention realizes functions such as "control for sensing", online fault monitoring and recovery, and online fault diagnosis of hydrogen energy unmanned aerial vehicles, has advantages such as strong environmental adaptability and low energy subsystem pressure, and is applicable to hydrogen energy unmanned aerial vehicle systems that require online health management.

[0025] Specifically, the first step includes:

[0026] Step 1.1 Establish a fuel cell voltage model:

[0027] (1)

[0028] ;

[0029] Wherein, is the voltage of a single fuel cell, is the Nernst potential determined by factors such as temperature, is the fuel cell current density, is the internal resistance of the fuel cell, is the limiting current density, , and is an unknown coefficient, is the stack temperature, and are the anode hydrogen pressure and the cathode oxygen pressure, respectively.

[0030] Define the vector of unknown parameters of the model , then Equation (1) can be expressed as , where represents the measurement equation, and the superscript T represents the transpose of the matrix. It should be noted that the parameters in the fuel cell voltage model will change with the operating conditions and environmental factors. Therefore, an online parameter identification algorithm is designed to perform online identification to achieve accurate voltage estimation in a dynamic environment.

[0031] Step 1.2 Design a voltage estimator based on the extended Kalman filter:

[0032] First, write the fuel cell voltage model in the form of a discrete state-space equation:

[0033] ;

[0034] where the subscript k represents the kth step, and are the process noise and measurement noise with known Gaussian distributions, and their covariance matrices are and , respectively, and their means are both 0, is the Nernst potential calculated based on the temperature, anode hydrogen pressure, and cathode oxygen pressure measured at the kth step, is the fuel cell current density at the kth step. The online parameter identification algorithm is designed as follows:

[0035] ;

[0036] where the annotation symbol ^ represents the estimated value of the variable it acts on, and the superscript - represents the prior estimate of the variable it acts on, is the state covariance matrix, is the output matrix.

[0037] Step 1.3 Online update the vector of unknown parameters of the model through the online parameter identification algorithm to obtain the real-time maximum power of the fuel cell:

[0038] ;

[0039] where is the number of single cells contained in the fuel cell, is the effective reaction area of a single cell, is the maximum power current point of the fuel cell, Satisfy:

[0040] ;

[0041] When occurs, an emergency landing will be executed and the algorithm of online parameter identification ends. Otherwise, the following steps are carried out, where is the power required for the UAV to hover.

[0042] At the same time, the voltage estimation residual is obtained through parameter identification :

[0043] ;

[0044] Among them, is the measured value of the fuel cell voltage, is the estimated value of the fuel cell voltage.

[0045] The voltage estimation residual is calculated online. When exceeds the set threshold it is determined that a failure has occurred.

[0046] Specifically, as Figure 2 shown, the second step includes:

[0047] Design a hybrid excitation operator. The operator includes a high-frequency excitation signal for detecting proton transport losses inside the fuel cell and a low-frequency excitation signal for detecting mass transfer losses, and is injected into the system in the form of a current signal. The signal generation process is as follows:

[0048] Step 2.1: The DC / DC converter issues a fuel cell current control signal for injecting high-frequency excitation, where is the fuel cell current, and the subscript des represents the desired signal, is the high-frequency excitation amplitude, is the high-frequency excitation signal frequency, is selected as the frequency corresponding to the intersection point of the fuel cell impedance spectrum and the real axis. The amplitude of the AC component of each excitation current is selected as 5-10% of the DC component of the fuel cell operating current, and the injection duration of each excitation signal should be no less than 8 cycles.

[0049] Step 2.2: Send a periodic yaw command signal to the UAV flight controller, generate a periodic load demand power through the periodic yaw maneuver of the UAV, and then generate a periodic fuel cell current for injecting low-frequency excitation, where is the periodic yaw angle amplitude, is the low-frequency excitation signal frequency, It is selected as the frequency corresponding to the maximum value of the imaginary part of the impedance spectrum.

[0050] This periodic yaw command brings strong non-linear coupling to the relationship between the UAV pose state and the control input. To ensure the stability of the original flight path tracking while injecting this excitation operator, a sliding mode-based flight controller is designed, and its control law is:

[0051] ;

[0052] where, and are the thrust and torque generated by the motor in the body coordinate system, respectively, that is, the control input, and are the position vector and velocity vector of the UAV, is the first-order time derivative of the desired trajectory, is the second-order time derivative of the desired trajectory, is the gravitational constant, is the angular velocity vector, is the transformation matrix from the body coordinate system to the inertial coordinate system, is the quaternion error determined by the current attitude angle, is the sliding mode vanishing term, , , and are adjustable parameters, is the integral sliding variable.

[0053] The motor speed can be determined by the control input:

[0054] ;

[0055] where, and are the thrust and torque coefficients of the motor, respectively, is the distribution matrix determined by the quadrotor configuration and physical dimensions, , , and are the rotational speeds of each motor. After the motor speeds are determined, the corresponding required power is:

[0056] ;

[0057] where, , and are the voltage, current and rotational speed of the i-th motor ( ), , , , are the internal resistance, KV value, no-load voltage, and no-load current of the motor, respectively. is the power of auxiliary equipment, including the fuel cell controller, etc. The load power demand injects a sinusoidal current into the fuel cell through the fuel cell voltage model (1).

[0058] Step 2.3 Calculate the characteristic frequency impedance online by measuring the characteristic frequency and the excitation current and the fuel cell voltage response at the characteristic frequency as the characteristics for fault diagnosis. The characteristic frequency impedance includes the real part of the high-frequency impedance and the imaginary part of the low-frequency impedance , and the calculation is as follows:

[0059] ;

[0060] where and are the real-axis and imaginary-axis components of the high-frequency excitation current signal, respectively, and are the real-axis and imaginary-axis components of the high-frequency response voltage signal, respectively, and are the real-axis and imaginary-axis components of the low-frequency excitation current signal, respectively, and are the real-axis and imaginary-axis components of the low-frequency response voltage signal, respectively. The calculation of each component is as follows:

[0061] ;

[0062] ;

[0063] ;

[0064] ;

[0065] where is the sampling start time, is the fuel cell voltage, is a positive integer. In this embodiment, and are selected as 1 kHz and 14 Hz respectively, and the excitation current amplitude is selected as 4 A.

[0066] Specifically, the third step includes:

[0067] First, classify into high and medium categories according to historical data and analysis, and classify into high, medium, and low categories; then design a fault classifier based on mechanism analysis and expert knowledge, including the following fuzzy logic rules:

[0068] (1) If is medium and is low, then the system status is normal;

[0069] (2) If is medium and is medium, then the system status is waterlogging;

[0070] (3) If is medium and is high, then the system status is oxygen deficiency;

[0071] (4) If is high and is low, then the system status is membrane dryness;

[0072] (5) If is high and is medium, then the system status is membrane dryness;

[0073] (6) If is high and is high, then the system status is oxygen deficiency.

[0074] This fault classifier is designed based on mechanism analysis and expert knowledge. According to the characteristic frequency impedance and the fuel cell faults are classified into three categories: waterlogging, membrane dryness, and oxygen deficiency, providing a basis for the targeted recovery measures in the fourth step.

[0075] Specifically, the fourth step includes:

[0076] After a fault is detected, the actuators such as fans and valves will be controlled according to the diagnostic results to restore the health of the fuel cell as much as possible and ensure the safety of the drone. When a membrane dryness fault is detected, the speed of the cooling fan will increase to reduce the fuel cell temperature and thus reduce the water evaporation rate; when a waterlogging fault is detected, the purge time of the cathode drain valve in one cycle will increase to accelerate the discharge of excess water products in the fuel cell; when an oxygen deficiency fault is detected, the speed of the cathode intake fan will increase to increase the oxygen supply per unit time. In this specific implementation case, when a membrane dryness occurs, the speed of the cooling fan will be increased to 80%, when waterlogging is detected, the discharge cycle of the cathode drain valve will be shortened from 4.5 s to 2.5 s to accelerate the discharge of product water, and when oxygen deficiency is detected, the speed of the intake fan will be increased to 90%.

[0077] The actual flight power curve is used for comparative simulation, and the lithium battery power results during the injection of the excitation are obtained as Figures 3 - 5 shown. Figure 3 The lines in Figure 4The lines in [ ] represent the output power of the lithium battery when injecting low-frequency excitation using the hybrid excitation operator of the present invention. When using the lithium battery of the present invention, the power fluctuation is significantly reduced, which is beneficial to extending its lifespan. Figure 5 The dotted line and the solid line in [ ] respectively represent the frequency diagrams of the output power fluctuation of the lithium battery when injecting low-frequency excitation using a DC / DC converter and when injecting low-frequency excitation using the hybrid excitation operator of the present invention. When using the strategy of the present invention, the mean square error of the output power fluctuation of the lithium battery is reduced by 37%.

[0078] The content not described in detail in the specification of the present invention belongs to the prior art well-known to those skilled in the art.

Claims

1. A health monitoring and management method for hydrogen energy drones based on a hybrid excitation operator, characterized in that, It includes the following steps: The first step: Establish a hydrogen fuel cell voltage estimator based on online parameter identification to estimate the voltage output in real time, generate a voltage residual with the measured voltage value, and at the same time obtain the maximum output power of the fuel cell for health monitoring, and take an emergency landing in case of insufficient maximum output power of the fuel cell; The second step: Design a hybrid excitation operator to inject high-frequency excitation into the fuel cell through a DC / DC converter, and actively and maneuverably inject low-frequency excitation that does not affect flight stability by the UAV. Use the injected high-frequency excitation, low-frequency excitation and the response voltage signal to calculate the characteristic frequency impedance online, and use the characteristic frequency impedance as a fault diagnosis feature; The third step: Design a fault classifier. Based on the characteristic frequency impedance, classify the fault types into three categories: flooding, membrane drying, and oxygen deficiency; The fourth step: Control the exhaust component based on the specific fault type diagnosed for fault recovery.

2. The hydrogen energy UAV health monitoring and management method based on a hybrid incentive operator according to claim 1, characterized in that The first step includes: Establish a fuel cell voltage model as: (1) ； Among them, is the single fuel cell voltage, is the Nernst potential determined by factors such as temperature, is the fuel cell current density, is the internal resistance of the fuel cell, is the limiting current density, , and are unknown coefficients, is the stack temperature, and are the anode hydrogen pressure and the cathode oxygen pressure respectively; Define the vector of unknown parameters of the model , then Equation (1) can be expressed as , where represents the measurement equation, and the superscript T represents the transpose of a matrix.

3. The hydrogen energy UAV health monitoring and management method based on a hybrid excitation operator according to claim 2, wherein, The first step also includes: Design a fuel cell voltage estimator based on the extended Kalman filter. First, write the fuel cell voltage model in the form of a discrete state space equation: ； where the subscript k represents the k-th step, is the vector of unknown model parameters at the k-th step, is the voltage of a single fuel cell at the k-th step, and are the process noise and measurement noise of the known Gaussian distribution respectively, and their covariance matrices are and respectively, and their means are both 0. is the Nernst potential calculated based on the temperature, anode hydrogen pressure, and cathode oxygen pressure measured at the k-th step, is the fuel cell current density at the k-th step. The algorithm design for online parameter identification is as follows: ； where the annotation symbol ^ represents the estimated value of the variable being acted on, and the superscript - represents the prior estimate of the variable being acted on, is the state covariance matrix, is the output matrix.

4. A method for health monitoring and management of a hydrogen energy unmanned aerial vehicle based on a hybrid excitation operator according to claim 3, characterized in that, The first step also includes: Online update the unknown parameter vector of the model through an online parameter identification algorithm to obtain the real-time maximum power of the fuel cell :[[]]END]] ； Wherein, is the number of single cells contained in the fuel cell, is the effective reaction area of a single cell, is the maximum power current point of the fuel cell, satisfies: ； When emergency landing will be executed and the algorithm of online parameter identification ends, otherwise the following steps are carried out, where is the power required for the UAV to hover; Meanwhile, the voltage estimation residual is obtained through online parameter identification : ； Among them, is the measured value of the fuel cell voltage, is the estimated value of the fuel cell voltage; perform online calculation of the voltage estimation residual. When the voltage estimation residual exceeds the set threshold , it is determined that a fault has occurred.

5. The hydrogen energy UAV health monitoring and management method based on a hybrid incentive operator according to claim 2, wherein, In the second step, the hybrid excitation operator includes a signal for high-frequency excitation used to detect proton transport losses inside the fuel cell and a signal for low-frequency excitation used to detect mass transfer losses, and is injected in the form of a current signal. The signal generation process includes: The fuel cell current control signal is sent out by the DC / DC converter , which is used to inject high-frequency excitation, where is the fuel cell current, and the subscript des represents the desired signal, is the amplitude of the high-frequency excitation, is the frequency of the high-frequency excitation signal, is selected as the frequency corresponding to the intersection point of the fuel cell impedance spectrum and the real axis, and t represents time; the amplitude of the AC component of each excitation current is selected to be 5-10% of the DC component of the fuel cell operating current, and the injection duration of the high-frequency excitation signal and the low-frequency excitation signal is not less than 8 cycles.

6. The method for health monitoring and management of a hydrogen energy unmanned aerial vehicle based on a hybrid excitation operator according to claim 5, wherein The signal generation process further includes: sending a periodic yaw command signal to the UAV flight controller , generating a periodic load demand power through the periodic yaw maneuver of the UAV, and further generating a periodic fuel cell current for injecting a low-frequency excitation, where is the amplitude of the periodic yaw angle, is the frequency of the low-frequency excitation signal, is selected as the frequency corresponding to the maximum value of the imaginary part of the impedance spectrum; Design a sliding-mode-based UAV flight controller, and its control law is: ； Among them, and are the pulling force and torque generated by the motor in the body coordinate system, i.e., the control inputs, is the z-axis component, are the x, y, and z-axis components respectively, and are the position vector and velocity vector of the UAV, is the first-order time derivative of the desired trajectory, is the second-order time derivative of the desired trajectory, is the gravitational constant, is the angular velocity vector, is the transformation matrix from the body coordinate system to the inertial coordinate system, is the quaternion error determined by the current attitude angle, is the sliding mode vanishing term, , , and are adjustable parameters, is the integral sliding variable, is the desired angular velocity vector; Determine the motor speed by the control input: ； Among them, and are the pulling force and torque coefficient of the motor respectively, is the distribution matrix determined by the quadrotor configuration and physical dimensions, , , and are the rotational speeds of each motor; after the motor rotational speeds are determined, the corresponding required power is generated: ； Among them, , and are the voltage, current and rotational speed of the i-th motor respectively, , , , , are the internal resistance, KV value, no-load voltage and no-load current of the motor respectively, is the power of auxiliary equipment; the load power demand injects a sinusoidal current into the fuel cell through the fuel cell voltage model.

7. A health monitoring and management method for a hydrogen energy unmanned aerial vehicle based on a hybrid excitation operator according to claim 6, characterized in that The signal generation process further includes: by measuring the frequencies of the high-frequency excitation signal and the low-frequency excitation signal and the excitation current and fuel cell voltage response at the frequencies, calculating the characteristic frequency impedance online as a characteristic for fault diagnosis; the characteristic frequency impedance includes the real part of the high-frequency impedance and the imaginary part of the low-frequency impedance , and the calculation is as follows: ； Among them, and are the real-axis and imaginary-axis components of the high-frequency excitation current signal respectively, and are the real-axis and imaginary-axis components of the high-frequency response voltage signal respectively, and are the real-axis and imaginary-axis components of the low-frequency excitation current signal respectively, and are the real-axis and imaginary-axis components of the low-frequency response voltage signal respectively. The calculation of each component is as follows: ； ； ； ； wherein, is the sampling start time, is the fuel cell voltage, is a positive integer.

8. A health monitoring and management method for a hydrogen energy unmanned aerial vehicle based on a hybrid excitation operator according to claim 1, characterized in that The third step includes: First, according to historical data and analysis, the real part of the high-frequency impedance is divided into two categories: high and medium, and the imaginary part of the low-frequency impedance is divided into three categories: high, medium, and low. Then, a fault classifier is designed based on mechanism analysis and expert knowledge, and the fault classifier contains six fuzzy logic rules.

9. A health monitoring and management method for a hydrogen energy unmanned aerial vehicle based on a hybrid excitation operator according to claim 8, characterized in that The fuzzy logic rule is: (1) If is medium and is low, the system status is normal; (2) If is medium and is medium, then the system status is waterlogging; If (3) is medium and is high, the system status is oxygen-deficient; (4) If is high and is low, the system state is dry film; If is high and is medium, the system status is film dry; If is high and is high, the system state is oxygen-deficient.

10. A health monitoring and management method for a hydrogen energy unmanned aerial vehicle based on a hybrid excitation operator according to claim 1, characterized in that The fourth step includes: After a fault is detected, control the actuator of the exhaust component according to the diagnosis result to restore the health of the fuel cell and ensure the safety of the UAV; when a membrane drying fault is detected, increase the speed of the cooling fan to reduce the fuel cell temperature to reduce the water evaporation rate; when a flooding fault is detected, increase the purge time of the cathode drain valve in one cycle to accelerate the discharge of excess water products in the fuel cell; when an oxygen deficiency fault is detected, increase the speed of the cathode intake fan to increase the oxygen supply per unit time.

Images