Hydrogen energy unmanned aerial vehicle logistics path planning method and system

Abstract

The application relates to the technical field of hydrogen energy unmanned aerial vehicle control, in particular to a hydrogen energy unmanned aerial vehicle logistics path planning method and system, which comprises the following steps: acquiring multi-dimensional information of an unmanned aerial vehicle; based on the multi-dimensional information and characteristic data, a multi-factor coupling energy consumption model is constructed to determine real-time demand power of the unmanned aerial vehicle in each flight stage and calculate total flight energy consumption; taking the minimum total flight energy consumption as an optimization target, a mixed integer convex programming path optimization model is solved to obtain an optimal energy consumption path; according to the real-time demand power corresponding to the optimal path, energy supply batteries of corresponding types are distributed; real-time energy closed-loop management is performed on a direct current bus of the unmanned aerial vehicle, and according to the difference between total system output power and real-time demand power of the unmanned aerial vehicle, redundant energy is stored or short energy is compensated and output. The application can improve the path planning and energy optimization effect of the hydrogen energy unmanned aerial vehicle.

Description

Technical Field

[0001] This application relates to the technical field of hydrogen-powered drone control, and in particular to a method and system for planning logistics routes for hydrogen-powered drones. Background Technology

[0002] With the rapid development of smart logistics systems and the continuous increase in demand for unmanned delivery, drone logistics delivery has moved from pilot demonstrations to large-scale implementation, showing great application potential in fields such as e-commerce express delivery, medical supplies, fresh food cold chain, and emergency rescue.

[0003] Currently, most mainstream logistics drones use lithium batteries as their sole power source. However, this results in generally short flight times, with most multi-rotor logistics drones having a flight time of only 20 to 40 minutes, making it difficult to cover long-distance delivery tasks exceeding 5 kilometers. Furthermore, lithium batteries have long charging times, typically requiring 1 to 3 hours for a full charge. The battery swapping process is also complex, making it difficult to meet the demands of high-density, high-frequency logistics scheduling. Performance also degrades significantly under low-temperature and high-load conditions, further limiting their application scope.

[0004] Hydrogen fuel cells, as a clean and efficient new energy device, have significant advantages such as high energy density, fast refueling speed, long service life, good low-temperature adaptability, and zero emissions. They can fundamentally solve the pain points of short flight time and slow refueling of lithium battery drones, becoming the core power solution for the next generation of long-endurance logistics drones. In recent years, small hydrogen fuel cell systems have gradually achieved lightweighting, modularization, and high power density, perfectly adapting to various logistics drone platforms such as multi-rotor and VTOL fixed-wing drones, providing reliable energy support for long-distance, high-load, and multi-node logistics delivery.

[0005] However, hydrogen fuel cells still have significant technical shortcomings in practical applications: their energy conversion relies on a series of processes including hydrogen supply, oxygen diffusion, electrochemical reactions, and electron transport, resulting in a slow dynamic response speed, typically ranging from hundreds of milliseconds to seconds. This makes them unable to quickly adapt to the high-frequency, large-amplitude power fluctuations during drone takeoff, climb, sharp turns, and wind resistance, leading to problems such as delayed power response, insufficient power output, decreased system efficiency, and accelerated lifespan degradation. In actual flight, the power demand during takeoff and climb can be many times higher than during level flight, and millisecond-level power spikes occur during turns and wind resistance. Hydrogen fuel cells often cannot keep up with these sudden changes, causing bus voltage drops and sluggish power response, which can, in severe cases, lead to attitude instability and flight interruptions. Furthermore, in terms of bus energy management, a lack of reasonable coordination strategies can easily result in energy redundancy and waste or momentary power shortages, affecting flight safety and delivery stability.

[0006] At the path planning level, traditional UAV path planning methods often prioritize the shortest path length and minimum flight time, failing to adequately consider the impact of key factors such as flight attitude, wind disturbances, flight altitude, air density, load variations, and energy characteristics on energy consumption. This results in significant discrepancies between planned paths and actual energy consumption, leading to low energy utilization. Particularly for hydrogen-electric hybrid UAVs, if path planning and energy management are independent, it's impossible to achieve synergistic optimization of optimal energy consumption and stable power, hindering the full realization of the long-endurance advantage of hydrogen energy. For example, the traditional shortest path might traverse areas with strong headwinds, causing a sharp increase in energy consumption and even insufficient endurance to complete the mission. Some path planning methods fail to consider air density differences and rotor energy consumption variations caused by altitude changes, resulting in a large discrepancy between theoretical and actual flight energy consumption.

[0007] In existing technologies, some solutions for power allocation research on hydrogen-electric hybrid drones mostly employ simple filtering or rule-based control without incorporating path planning for global energy consumption optimization. Some path planning methods only consider geometric constraints and do not embed accurate energy consumption models and hybrid energy characteristics. A few solutions involve energy consumption optimization, but do not employ convex programming to ensure global optimality, resulting in slow solution speeds and difficulty in meeting real-time requirements. Furthermore, most systems lack real-time endurance monitoring and dynamic path replanning mechanisms, making them prone to insufficient endurance and mission interruption risks in long-distance delivery. In complex urban environments, mountainous terrain, and variable wind fields, existing technologies struggle to simultaneously meet multiple requirements such as minimum energy consumption, flight safety, rapid response, and reliable delivery, thus hindering the large-scale commercial application of hydrogen-powered logistics drones. Summary of the Invention

[0008] To improve the path planning and energy optimization of hydrogen-powered drones, this application provides a method and system for planning logistics paths for hydrogen-powered drones.

[0009] Firstly, this application provides a method for planning logistics routes for hydrogen-powered unmanned aerial vehicles (UAVs), employing the following technical solution:

[0010] A method for planning logistics routes for hydrogen-powered drones includes the following steps:

[0011] Acquire multi-dimensional information about the drone, including real-time flight status data, environmental perception data, and logistics delivery task information;

[0012] Based on the multi-dimensional information and characteristic data, a multi-factor coupled energy consumption model is constructed to determine the real-time power demand of the UAV in each flight phase, and the total energy consumption of the entire flight is calculated based on the real-time power demand.

[0013] With the goal of minimizing the total energy consumption of the entire flight, a mixed integer convex programming path optimization model is solved to obtain the energy-optimal path;

[0014] Based on the real-time power demand corresponding to the optimal path, a low-pass filter is used to decompose the real-time power demand corresponding to the optimal path and allocate it to the corresponding type of power supply battery.

[0015] Real-time closed-loop energy management is performed on the DC bus of the UAV. Based on the difference between the total output power of the system and the real-time power demand of the UAV, redundant energy is stored or insufficient energy is compensated for.

[0016] In some embodiments, a multi-factor coupled energy consumption model is constructed to determine the real-time power demand of the UAV at each flight phase, and the total energy consumption for the entire flight is calculated based on the real-time power demand, including the following steps:

[0017] The energy consumption per unit time determined based on the multi-factor coupled energy consumption model is expressed as follows:

[0018] ,

[0019] Where v is the airspeed and h is the altitude. The pitch angle, For roll angle, Yaw angle Let m be the three-dimensional wind field component, and m be the total mass of the UAV. Air density at different altitudes;

[0020] The total energy consumption of the entire flight of the UAV is expressed as follows:

[0021] ,

[0022] in, Real-time energy consumption at each stage, The start and end times for each stage.

[0023] In some embodiments, with the goal of minimizing the total energy consumption of the entire flight, a mixed-integer convex programming path optimization model is solved, including the following steps:

[0024] Set the order of delivery node access and the selection of key points in the path to integer variables of 0-1;

[0025] Flight speed, flight altitude, attitude angle, fuel cell output power, and lithium battery charging and discharging power are set as continuous variables;

[0026] By combining several constraints, the mixed integer convex programming model is solved to obtain the energy-optimal path;

[0027] The flight parameter sequence and energy output sequence are output based on the energy-optimal path.

[0028] In some embodiments, based on the real-time power demand corresponding to the optimal path, a low-pass filter is used to decompose the real-time power demand corresponding to the optimal path and allocate it to the corresponding type of power supply battery, including the following steps:

[0029] The real-time demand power is separated into a low-frequency stable component and a high-frequency abrupt component using a low-pass filter.

[0030] The low-frequency stable component is allocated to the hydrogen fuel cell as the basic output, and the high-frequency abrupt component is allocated to the lithium battery for dynamic compensation.

[0031] The power allocation satisfies the real-time power balance relationship, expressed as:

[0032] ,

[0033] in, The real-time power requirement of the drone. For the output power of hydrogen fuel cells, This refers to the output power of the lithium battery.

[0034] In some embodiments, the following steps are also included:

[0035] When allocating power, a model is used to predict future power demand, and the output of the corresponding hydrogen fuel cell is adjusted in advance based on the future power demand to compensate for its dynamic response lag.

[0036] In some embodiments, the real-time power demand corresponding to the optimal path is decomposed using a low-pass filter and allocated to the corresponding type of power supply battery, and the method further includes the following steps:

[0037] A discrete sliding window is introduced and the sample variance corresponding to the recent wind speed measurement is calculated. The cutoff frequency of the low-pass filter is dynamically adjusted based on the sample variance.

[0038] A proportional-derivative compensation term is applied to the low-frequency stable component based on the rate of change of the cutoff frequency to obtain the optimized low-frequency component.

[0039] The optimized low-frequency component and the real-time demand power are used to calculate the optimized high-frequency component.

[0040] In some embodiments, real-time closed-loop energy management is performed on the DC bus of the UAV. Based on the difference between the total system output power and the real-time power demand of the UAV, redundant energy is stored or insufficient energy is compensated for, including the following steps:

[0041] When the total output power of the system is greater than the real-time demand power, the energy difference is used to charge the lithium battery through a bidirectional DC / DC converter to achieve the storage of redundant energy.

[0042] When the total output power of the system is less than the real-time power demand, the lithium battery is controlled to discharge in order to compensate for the energy shortage.

[0043] In some embodiments, the following steps are also included:

[0044] The remaining available energy of the hydrogen-powered drone is monitored in real time and compared with the energy consumption required to complete the remaining delivery route;

[0045] When it is determined that the remaining available energy is less than the sum of the energy required to complete the remaining delivery route and the preset safety backup energy, the dynamic route replanning is initiated to reconstruct and solve the mixed integer convex programming route optimization model in order to update the optimal route and energy consumption allocation.

[0046] In some embodiments, the following steps are also included:

[0047] Calculate the energy difference between the remaining available energy and the energy required to complete the remaining delivery route;

[0048] Calculate the rate of change of the wind field corresponding to the three-dimensional wind field vector at adjacent time intervals;

[0049] A decision sliding window is introduced, and the first discrete difference entropy corresponding to the energy difference and the second discrete difference entropy corresponding to the wind field change rate are calculated within the decision sliding window.

[0050] Set event triggering conditions and determine whether the first discrete difference entropy and / or the second discrete difference entropy meet the event triggering conditions;

[0051] The dynamic path replanning is initiated only when the event triggering condition is met, and the mixed integer convex programming path optimization model is reconstructed and solved to update the optimal path and energy consumption allocation.

[0052] Secondly, this application provides a hydrogen-powered drone logistics path planning system, which adopts the following technical solution:

[0053] A hydrogen-powered drone logistics route planning system is provided to implement the above-mentioned method.

[0054] The technical solutions provided by the embodiments of this application have the following technical effects:

[0055] This invention achieves optimal energy consumption, efficient energy utilization, and stable and reliable flight throughout the entire process of hydrogen-powered drone logistics delivery by deeply integrating a multi-factor precise energy consumption model, mixed-integer convex programming path optimization, model-predicted power allocation, bus energy closed-loop management, and dynamic endurance assurance. Under the same load and flight distance conditions, it has significant energy consumption and stability advantages compared to traditional path planning methods. It can effectively adapt to various logistics delivery scenarios such as cities, suburbs, mountains, and islands, meeting the needs of long-distance, multi-node, and high-frequency delivery, and providing reliable technical support for the large-scale and commercial deployment of hydrogen-powered logistics drones. Attached Figure Description

[0056] Figure 1 This is a schematic diagram illustrating the steps of a hydrogen-powered drone logistics route planning method provided in this embodiment;

[0057] Figure 2 This is a schematic diagram of the hydrogen-powered drone delivery route (from A to C to B) in an embodiment of this application.

[0058] Figure 3 This is a schematic diagram of the optimal path and energy consumption distribution in the embodiments of this application. Detailed Implementation

[0059] To better understand the purpose, technical solutions, and advantages of this application, it has been described and illustrated below with reference to the accompanying drawings and embodiments. However, those skilled in the art should understand that this application can be implemented without these details. In some cases, to avoid obscuring various aspects of this application due to unnecessary description, well-known methods, processes, systems, components, and / or circuits already described at a higher level will not be elaborated upon. It will be apparent to those skilled in the art that various modifications can be made to the embodiments disclosed in this application, and the general principles defined in this application can be applied to other embodiments and application scenarios without departing from the principles and scope of this application. Therefore, this application is not limited to the illustrated embodiments, but conforms to the broadest scope consistent with the scope of protection claimed in this application.

[0060] It should be noted that the descriptions of these embodiments are for the purpose of aiding understanding the present invention, but do not constitute a limitation thereof. Furthermore, 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.

[0061] In the description of this application, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0062] In the description of this application, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples.

[0063] like Figure 1 As shown in the figure, this application discloses a method for planning logistics routes for hydrogen-powered drones, including the following steps:

[0064] The S100 acquires multi-dimensional information about the drone, including real-time flight status data, environmental perception data, and logistics delivery task information.

[0065] Multi-dimensional information is mainly collected through the flight controller, accelerometer, gyroscope, barometric pressure sensor, GPS module, wind field sensor, wireless data transmission module, ground logistics scheduling platform, obstacle perception module, etc. included in the hydrogen energy drone logistics path planning system.

[0066] Accelerometers, gyroscopes, barometric pressure sensors, GPS modules, wind field sensors, and obstacle perception modules simultaneously collect UAV flight attitude, three-dimensional position, flight altitude, airspeed, ground speed, wind speed and direction, pitch angle, roll angle, yaw angle, and airspace obstacle distribution information as flight status data and environmental perception data. At the same time, Kalman filtering algorithm is used to reduce noise, fuse, and compensate sensor data to improve the accuracy of state calculation.

[0067] Meanwhile, the wireless data transmission module loads constraint information such as the logistics delivery starting point, multiple intermediate delivery points, destination coordinates, delivery time window, load weight, task priority, and restricted airspace from the ground logistics dispatch platform to serve as logistics delivery task information, completing task initialization and information verification.

[0068] In this embodiment, the sensor sampling frequency is set to 200Hz, the GPS update frequency is 10Hz, and the wind field sensor output frequency is 50Hz. All data are synchronized in time and outlier removal is performed before entering the computing unit to ensure the reliability and consistency of the input information.

[0069] The S200 constructs a multi-factor coupled energy consumption model based on multi-dimensional information and characteristic data to determine the real-time power demand of the UAV in each flight phase, and calculates the total energy consumption of the entire flight based on the real-time power demand.

[0070] The flight controller also includes built-in APM flight control firmware and energy consumption optimization calculation module. The APM flight control firmware obtains information such as environmental wind field disturbance, flight attitude change, load mass change, and altitude air density change based on the UAV's flight status, and combines it with the UAV's dynamic characteristics, rotor aerodynamic characteristics, and motor ESC characteristics to obtain comprehensive characteristic data.

[0071] By combining characteristic data with multi-dimensional information, a multi-factor coupled energy consumption model is constructed through the energy consumption optimization calculation module to calculate the real-time power demand, segmented energy consumption, and total energy consumption for the entire flight required to formulate a logistics route.

[0072] For S300, with the goal of minimizing total energy consumption throughout the flight, a mixed integer convex programming path optimization model is solved to obtain the energy-optimal path.

[0073] The flight controller also has a built-in mixed integer convex programming solution unit, which uses the minimum total energy consumption of the entire flight as the optimization objective to construct and solve the mixed integer convex programming path optimization model. Based on the variable processing of each parameter and the construction of constraints, it completes the rapid solution of the optimal delivery route.

[0074] S400 uses low-pass filtering to decompose the real-time power demand corresponding to the optimal path and allocate it to the corresponding type of power supply battery based on the real-time power demand corresponding to the optimal path.

[0075] The flight controller also has a built-in model prediction power control unit. After obtaining the energy-optimal path for the delivery mission, it uses a low-pass filter to distribute the power between the hydrogen fuel cell and the lithium battery based on the real-time power demand corresponding to the energy-optimal path. This stabilizes the bus output voltage, suppresses power fluctuations, and improves the response speed and output stability of the energy system.

[0076] The S500 performs real-time closed-loop energy management of the UAV's DC bus. Based on the difference between the total system output power and the UAV's real-time power demand, it stores redundant energy or compensates for insufficient energy output.

[0077] The system collects and analyzes the voltage and current on the DC bus in real time. Based on the difference between the total output power of the system and the real-time power demand of the UAV, it executes redundant energy storage or missing energy replenishment strategies to achieve efficient energy utilization and stable system operation.

[0078] Through the above steps, by constructing a multi-factor coupled precise energy consumption model, a hybrid integer convex programming path optimization model with the goal of minimizing total energy consumption is established. Combined with the model's prediction of hydrogen-electric hybrid power allocation and DC bus energy closed-loop management, the global optimal logistics path, the lowest energy consumption, and the highest energy utilization rate are achieved. At the same time, it solves technical problems such as slow dynamic response of hydrogen fuel cells, large power fluctuations, and insufficient long-distance delivery range, thereby improving the efficiency, stability, and reliability of hydrogen energy drone logistics delivery.

[0079] In other embodiments, a multi-factor coupled energy consumption model is constructed to determine the real-time power demand of the UAV at each flight phase, and the total energy consumption for the entire flight is calculated based on the real-time power demand, including the following steps:

[0080] The energy consumption per unit time determined based on the multi-factor coupled energy consumption model is expressed as follows:

[0081] ,

[0082] Where v is the airspeed and h is the altitude. The pitch angle, For roll angle, Yaw angle denoted as the three-dimensional wind field component, and m as the total mass of the UAV (including fuselage, payload, battery, and hydrogen tank). This represents the air density at different altitudes.

[0083] The above function is a multi-input single-output nonlinear coupled function. The input is all the flight states, environmental parameters and self-attributes of the UAV at time t, and the output is the instantaneous electrical power required for the UAV to maintain its current flight at that time.

[0084] Among these factors, airspeed determines the rotor induced power and drag power, with an "optimal cruise airspeed" (approximately 10⁻¹² m / s) that minimizes total power. Higher altitudes result in lower air density, requiring higher rotor speeds to generate the same lift, thus increasing induced power. Larger pitch angles require greater centripetal force during turns, necessitating greater rotor thrust and significantly increasing energy consumption. Similarly, larger roll angles require greater centripetal force during turns, requiring greater rotor thrust and significantly increasing energy consumption. Higher yaw rates increase drag and rotor differential power during turns. In the three-dimensional wind field, headwinds significantly increase relative airspeed and drag power, crosswinds increase attitude correction power, and tailwinds reduce energy consumption. For UAV mass, greater mass requires greater rotor lift, with induced power proportional to the 1.5th power of mass. Air density decreases exponentially with altitude, making it a key mediating variable affecting energy consumption.

[0085] Meanwhile, the instantaneous electric power calculated above is based on several descriptions in the embodiments of this application, and it can be decomposed into 5 independent energy consumption components, which consist of induced power, type resistance power, parasitic power, attitude adjustment power, and auxiliary power.

[0086] Induced power is the basic power required for the rotor to generate lift, accounting for 60-70% of energy consumption; drag power is the air resistance power generated by the rotation of the rotor blades themselves; parasitic power is the wind resistance power of components such as the fuselage, battery, and hydrogen tank; attitude adjustment power is the differential power required to maintain pitch, roll, and yaw attitudes, which is proportional to the square of the attitude angle and angular velocity; auxiliary power is the fixed energy consumption of airborne equipment such as sensors, flight controllers, and data transmission modules.

[0087] The total energy consumption of a drone during its entire flight is the sum of the energy consumption integrals of each flight phase, specifically expressed as follows:

[0088] ,

[0089] in, Real-time energy consumption at each stage, The start and end times of each stage are represented by n, which represents the total number of flight stages. Different stages include takeoff, climb, level flight, turn, descent, hovering, etc.

[0090] Based on the dynamic model of a six-rotor UAV, a model relating thrust, torque, and rotational speed is established. Combined with equivalent circuit models of a brushless motor and electronic speed controller, the motor drive power is calculated. Factors such as flight speed, attitude angle, three-dimensional wind field, air density, load mass, and flight altitude are coupled to establish a wind resistance energy consumption model and an induced power loss model. Finally, by integrating the output characteristics of a hydrogen fuel cell, lithium battery charging and discharging losses, and DC / DC conversion efficiency, a complete energy consumption model for a hydrogen-electric hybrid energy system is formed. This model can accurately calculate real-time power requirements, segmented energy consumption, and total energy consumption during different flight states, including takeoff, climb, level flight, turning, descent, and hovering, providing precise objective function input for path planning.

[0091] In this embodiment, the rotor radius is calculated as 0.35m, the nominal voltage of the motor is 48V, the rated power of the fuel cell is 1500W, the capacity of the lithium battery is 10Ah, the DC / DC conversion efficiency is not less than 96%, and the energy consumption calculation error of the model under different operating conditions is controlled within 5%.

[0092] like Figure 2 As shown, in some other embodiments, with the goal of minimizing the total energy consumption throughout the flight, the mixed-integer convex programming path optimization model is solved, including the following steps:

[0093] S310 sets the delivery node access order and path key point selection to integer variables of 0-1.

[0094] The S320 sets flight speed, flight altitude, attitude angle, fuel cell output power, and lithium battery charging and discharging power as continuous variables.

[0095] S330, by combining several constraints, solves the mixed integer convex programming model to obtain the energy-optimal path.

[0096] S340 outputs the corresponding flight parameter sequence and energy output sequence based on the energy-optimal path.

[0097] The order of delivery node access, the selection of key points on the path (inflection points to avoid obstacles / restricted flight zones), and whether to hover for delivery are defined as integer 0-1 variables, which form the basis of the optimized topology. After solving the integer variables, the topology structure of "starting point - optimal intermediate node access order - ending point" is obtained, as well as the coordinates of the collision-free key points connecting each node, which are combined to form the globally optimal logistics delivery path.

[0098] Flight speed, altitude, roll angle, and pitch angle are defined as continuous variables, representing the execution details of the path. These parameters are obtained by arranging the results of flight control parameter solutions along a time axis. The optimal flight parameters for each flight phase—takeoff, climb, level flight, turn, descent, and hover—are arranged chronologically to form a time-step sequence of flight commands, which serves as the flight parameter sequence. For example, in areas with strong headwinds, the flight altitude is automatically reduced (lower wind speeds at lower altitudes); during level flight, the cruise speed with the lowest induced power is selected; and during turns, the attitude angle is controlled within a safe range while minimizing energy consumption.

[0099] Fuel cell power and lithium battery power are defined as continuous variables, obtained by arranging the energy power parameter solutions along a time axis. Combined with the real-time power demand corresponding to the flight parameter sequence, the optimal hydrogen-electric power allocation ratio for each time step is calculated, forming a time-step energy output sequence. For example, during level flight, the fuel cell provides all stable power, while during climb / turn / wind-resistance phases, the lithium battery's compensation power is planned in advance.

[0100] By setting constraints such as path connectivity, flight altitude, flight speed, attitude angle, collision avoidance, logistics time window, and energy output, the interior point method is used to quickly solve the convex programming model, obtaining the globally optimal logistics delivery path, flight parameter sequence, and energy output sequence, ensuring that the path is feasible, safe, and has the lowest energy consumption.

[0101] The objective function corresponding to the mixed integer convex programming path optimization model is:

[0102] .

[0103] The constraints include:

[0104] (1) Path connectivity constraints: ,in Assuming node access status, ensure the path is uninterrupted, and form a continuous closed loop from the starting point to the intermediate node to the ending point;

[0105] (2) Flight altitude constraints: ;

[0106] (3) Flight speed constraints: ;

[0107] (4) Attitude angle constraints: ;

[0108] (5) Collision avoidance constraints: ,in The distance between the drone and the obstacle. To ensure a safe operating radius, the distance between the drone and obstacles must be greater than the safe operating radius, and the drone must automatically avoid buildings, restricted flight zones, etc.

[0109] (6) Logistics time window constraints: By combining the delivery priority of nodes, we can ensure that the access order of each node meets the delivery time requirements.

[0110] (7) Energy output constraints: , Limit the maximum / minimum output power of hydrogen fuel cells and the upper limit of charging and discharging power of lithium batteries to avoid overloading the energy system.

[0111] In this embodiment, the flight altitude is constrained to 60m to 120m, the flight speed is constrained to 8m / s to 15m / s, the attitude angle is constrained to ±15°, the safe distance from obstacles is not less than 15m, the allowable error range of the delivery time window is ±3 minutes, and the time to solve a single path does not exceed 200ms, which meets the requirements of airborne real-time planning.

[0112] In other embodiments, based on the real-time power demand corresponding to the optimal path, a low-pass filter is used to decompose the real-time power demand corresponding to the optimal path and allocate it to the corresponding type of power supply battery, including the following steps:

[0113] The S410 separates the real-time demand power into a low-frequency stable component and a high-frequency abrupt component using a low-pass filter.

[0114] The S420 distributes the low-frequency stable component to the hydrogen fuel cell as the basic output, and distributes the high-frequency abrupt component to the lithium battery for dynamic compensation.

[0115] Based on the real-time power demand corresponding to the optimal path, a low-pass filter is used to decompose the demanded power into a low-frequency stable component and a high-frequency abrupt component; the low-frequency component is allocated to the hydrogen fuel cell as the basic output, and the high-frequency component is allocated to the lithium battery for dynamic compensation.

[0116] The power allocation satisfies the real-time power balance relationship, expressed as:

[0117] ,

[0118] in, For the real-time power requirements of the drone, For the output power of hydrogen fuel cells, This refers to the output power of the lithium battery (positive for discharging and negative for charging).

[0119] The power components separated by the low-pass filter are: , .

[0120] The purpose of power splitting through a low-pass filter is because the energy conversion design of hydrogen fuel cells involves multiple physicochemical processes such as hydrogen diffusion, electrochemical reaction, and electron transport, resulting in a long dynamic response time. This makes it impossible to respond to millisecond-level power spikes in scenarios such as takeoff / climb, sharp turns, and gust interference for drones. By splitting the power, the rate of change of low-frequency components is within the response capability range of the fuel cell, and there will be no power lag. In contrast, lithium batteries have a shorter response time, and their allocation of high-frequency components can follow any power spike.

[0121] Meanwhile, if the hydrogen fuel cell is directly made to follow the total power demand, the lag in its output power will cause large fluctuations in the bus voltage. For example, when the power suddenly increases, the fuel cell output cannot keep up, and the bus voltage drops instantly, which may cause the flight control to restart or the motor to stall. When the power suddenly drops, the fuel cell output cannot be reduced in time, and the bus voltage overshoots, which may damage the airborne electronic equipment.

[0122] Through low-pass filtering, the fuel cell always outputs stable low-frequency power, reducing sudden power changes, while the lithium battery quickly compensates for high-frequency gaps or absorbs redundant power to maintain stable bus voltage.

[0123] The current energy loss is expressed as follows:

[0124] .

[0125] Among them, This refers to the efficiency of a bidirectional DC / DC converter.

[0126] Meanwhile, the lifespan of hydrogen fuel cells is strongly negatively correlated with the number of power cycles and the magnitude of load changes. Frequent and large power changes will accelerate the aging of proton exchange membranes and catalyst loss.

[0127] By using low-pass filtering, the hydrogen fuel cell outputs an almost constant low-frequency power, avoiding large power cycles. All power cycles are handled by the lithium battery, and the power cycle life of the lithium battery is much longer than that of the hydrogen fuel cell.

[0128] like Figure 2 , 3 As shown, in some other embodiments, the following steps are also included:

[0129] The S430 uses a model to predict future power demand during power allocation and adjusts the output of the corresponding hydrogen fuel cell in advance based on future power demand to compensate for its dynamic response lag.

[0130] By combining model prediction algorithms, power demand in the future can be predicted based on historical power data and state transition patterns, and the output of hydrogen fuel cells can be adjusted in advance to reduce power lag and fluctuations.

[0131] Furthermore, subsequent steps involve a power pre-allocation strategy based on the model prediction algorithm, while the decomposition result of the low-pass filter is the basic prerequisite data for the model prediction algorithm.

[0132] The low-frequency component serves as the basic output command for the hydrogen fuel cell. The model prediction algorithm, based on historical power data, predicts the changes in the high-frequency component in the next 50 steps (100ms per step), and fine-tunes the reference output of the fuel cell 300ms in advance, further reducing the corresponding lag.

[0133] Without pre-decomposition using low-pass filtering, the model prediction algorithm needs to predict both low-frequency and high-frequency components simultaneously, which significantly increases the prediction error and fails to effectively compensate for the response delay of the fuel cell.

[0134] In this embodiment, the low-pass filter cutoff frequency is set to 0.5Hz, the fuel cell response delay compensation time is 300ms, the prediction step size is set to 50 steps, and the time corresponding to each step is 100ms, which can effectively cover the dynamic response lag range of the fuel cell.

[0135] In other embodiments, the method of using low-pass filtering to decompose the real-time power demand corresponding to the optimal path and allocate it to the corresponding type of power supply battery further includes the following steps:

[0136] S440 introduces a discrete sliding window and calculates the sample variance corresponding to recent wind speed measurements, and dynamically adjusts the cutoff frequency of the low-pass filter based on the sample variance.

[0137] S450 applies a proportional-derivative compensation term to the low-frequency stable component based on the rate of change of the cutoff frequency to obtain the optimized low-frequency component.

[0138] S460 optimizes high-frequency components based on low-frequency component optimization and real-time demand power calculation.

[0139] Furthermore, in order to address the issue that a low-pass filter with a fixed cutoff frequency cannot effectively separate power components when there are sudden changes in the noise intensity of the wind field sensor, resulting in an uncoordinated power distribution between the hydrogen fuel cell and the lithium battery and increased power fluctuations on the bus, this approach is proposed.

[0140] First, based on the real-time demand power separation, a discrete sliding window of length N is introduced to calculate recent wind speed measurements in real time. The sample variance is used to dynamically adjust the cutoff frequency of the low-pass filter. When variance increases (noise intensifies), reduce the cutoff frequency. To filter out more noise; when the variance decreases (noise stabilizes), appropriately increase the cutoff frequency. To reduce signal delay.

[0141] Meanwhile, based on the rate of change of the current cutoff frequency, a proportional-derivative compensation term is applied to the low-frequency stable component to suppress the power jump caused by the sudden change in cutoff frequency.

[0142] The formula for calculating the variance of the sliding window is as follows:

[0143] ;

[0144] .

[0145] Where t represents the discrete-time index, with the unit being the sampling period; N is the sliding window length, ranging from 10 to 50 (corresponding to historical data of about 0.1s to 0.5s), and the window step size in this embodiment is 20; Characterized as the first The wind speed measurement at any given time (the magnitude of the three-dimensional wind speed vector), in m / s; This represents the arithmetic mean of the wind speed within the sliding window. This represents the sample variance of wind speed within the sliding window.

[0146] The formula for calculating the dynamic cutoff frequency is:

[0147] .

[0148] in, The reference cutoff frequency is used, and this application defaults to 0.5Hz; The reference variance is used for normalization and is set to the typical variance value of the historical wind speed stable period. In this application, it is set to 1.0. It is a very small positive number, used to prevent the denominator from being zero; To adjust the exponent and control the sensitivity of the cutoff frequency to noise, the value is set between 0.3 and 0.7. Let be the dynamic cutoff frequency at time t.

[0149] The compensated low-frequency power component is:

[0150] .

[0151] in, It is characterized as the low-frequency power component obtained after low-pass filtering at the dynamic cutoff frequency; It is represented by a proportional compensation coefficient, with a value ranging from 0.2 to 1.0; This is the differential compensation coefficient, with a value ranging from 0.1 to 0.5; The compensated low-frequency power component is used as the output reference for the hydrogen fuel cell.

[0152] Meanwhile, the high-frequency components are still obtained by subtracting the compensated low-frequency power components from the real-time demand power.

[0153] In other embodiments, real-time closed-loop energy management is performed on the UAV's DC bus. Based on the difference between the total system output power and the UAV's real-time power demand, redundant energy is stored or insufficient energy is compensated for, including the following steps:

[0154] When the total output power of the system exceeds the real-time power demand, the S510 charges the lithium battery with the energy difference through a bidirectional DC / DC converter to achieve redundant energy storage.

[0155] S520 controls the lithium battery to discharge when the total output power of the system is less than the real-time power demand, in order to compensate for the energy shortage.

[0156] The system performs real-time sampling and data analysis of the voltage and current on the DC bus. Based on the difference between the total system output power and the real-time power demand of the UAV, it executes redundant energy storage or missing energy replenishment strategies. When the total system output power is greater than the real-time power demand, the excess energy is used to charge the lithium battery through a bidirectional DC / DC converter to achieve energy recovery. When the total system output power is less than the real-time power demand, the system controls the lithium battery to discharge rapidly to replenish the energy gap, ensuring sufficient power for the UAV and improving energy utilization and system stability.

[0157] In this embodiment, the rated bus voltage is 48V, with an allowable fluctuation range of ±2V, a sampling frequency of 1kHz, and a control cycle of 10ms to ensure real-time energy regulation. During the stable level flight phase, the system can recover excess power to the lithium battery in the form of charging; during high-power demand phases such as takeoff, climb, and wind resistance, the lithium battery quickly provides compensation, and the bus voltage does not drop significantly.

[0158] In other embodiments, the following steps are also included:

[0159] The S600 monitors the remaining available energy of hydrogen-powered drones in real time and compares it with the energy required to complete the remaining delivery route.

[0160] S600: When it is determined that the remaining available energy is less than the sum of the energy required to complete the remaining delivery route and the preset safety backup energy, dynamic route replanning is initiated. The mixed integer convex programming route optimization model is reconstructed and solved to update the optimal route and energy allocation.

[0161] The system also includes a logistics node management unit, a remaining range monitoring unit, a dynamic route replanning unit, a hydrogen energy reserve monitoring unit, and a communication encryption unit. The logistics node management unit is used to store, update, and sort information such as the starting point, intermediate delivery points, destination, delivery time window, and delivery priority in delivery tasks. The remaining range monitoring unit is used to calculate in real time the maximum flight distance and delivery task range that can be supported by the current remaining energy. The dynamic route replanning unit is used to quickly generate new optimal routes when there is insufficient range, sudden changes in wind field, obstacle intrusion, or changes in delivery tasks. The hydrogen energy reserve monitoring unit is used to collect hydrogen storage tank pressure, hydrogen flow rate, and fuel cell output efficiency in real time and convert them into remaining usable energy. The communication encryption unit is used to ensure the security and stability of data transmission between the UAV and the ground logistics scheduling platform.

[0162] The system calculates the remaining available energy of the drone in real time, including the remaining hydrogen energy and the remaining lithium battery power, and converts it into the total available flight range. Based on historical data, it synchronously calculates the theoretical energy consumption required to complete the remaining delivery tasks and reserves backup energy. When the remaining energy is insufficient to complete the current path, strong headwinds cause a surge in energy consumption, new obstacles appear, or the delivery task changes, the dynamic path replanning module is immediately activated. Based on mixed integer convex programming, a new path is quickly generated, and the flight altitude, speed, and node sequence are optimized to ensure that the drone returns safely or completes the delivery task, thereby improving adaptability to complex environments and mission reliability.

[0163] In this embodiment, the safety backup energy is set at 15% of the total energy, the remaining battery life assessment cycle is 1 second, the replanning trigger threshold is when the remaining energy is lower than the sum of the theoretical requirement and the backup energy, and the replanning response time is less than 300ms.

[0164] Furthermore, when assessing remaining battery life, some transient spikes or abnormal fluctuations may exist in historical energy consumption data, causing jumps in the calculation of remaining battery life, leading to frequent or delayed path replanning, resulting in wasted computing power or task interruption.

[0165] Therefore, in some embodiments, instead of directly using the instantaneous energy consumption value of the latest cycle, an exponentially weighted moving average algorithm with a time decay constant is introduced to smooth the historical unit-time energy consumption sequence. The smoothed energy consumption value is used to predict the energy consumption required for the remaining tasks and compared with the remaining available energy. At the same time, the condition for triggering replanning is changed from the original "insufficient remaining energy" to "the smoothed expected remaining energy is lower than the threshold and lasts for more than one time window", adding a debouncing mechanism.

[0166] Specifically, the exponentially weighted moving average energy consumption is first calculated using the following formula:

[0167] .

[0168] Where t represents the discrete-time index. Characterized as the actual energy consumption per unit time within the t-th second. The energy consumption is represented by the exponentially weighted average energy consumption at second t. The forgetting factor, used to control the rate of decay of historical data, is set to a value of 0.8-0.99.

[0169] The smoothed remaining task's expected energy consumption is:

[0170] .

[0171] in, The remaining flight distance of the path at second t (calculated from the current optimal path); It is represented by the current average ground speed, which is the arithmetic mean of GPS speeds over the most recent 10 seconds; This is represented by the energy consumption required to predict and complete the remaining tasks.

[0172] Meanwhile, the triggering condition for the time window is:

[0173] .

[0174] in, This is represented by the remaining usable energy of the drone at second t (calculated using hydrogen fuel and lithium battery). The energy is set as a safety backup, based on 15% of the total energy; M is the anti-shake time window length, ranging from 2 to 5 seconds.

[0175] In other embodiments, the following steps are also included:

[0176] S620 calculates the energy difference between the remaining available energy and the energy required to complete the remaining delivery route.

[0177] S630 calculates the rate of change of the wind field corresponding to the three-dimensional wind field vector at adjacent time intervals.

[0178] S640 introduces a decision sliding window and calculates the first discrete difference entropy corresponding to the energy difference within the decision sliding window and the second discrete difference entropy corresponding to the wind field change rate.

[0179] S650, set the event triggering conditions, and determine whether the first discrete difference entropy and / or the second discrete difference entropy meet the event triggering conditions.

[0180] S660 initiates dynamic path replanning only when the event triggering conditions are met, reconstructs and solves the mixed integer convex programming path optimization model to update the optimal path and energy consumption allocation.

[0181] Furthermore, when performing path replanning, the original scheme was set to evaluate at a fixed period, which would lead to a waste of computing power during periods of stable environment and an insufficient evaluation frequency during periods of drastic environmental change.

[0182] To address the aforementioned issues, this embodiment further modifies the fixed-period replanning decision-making into an event-driven model. Two key signals are defined: the difference between remaining energy and the energy required to complete the remaining tasks, and the rate of change of the three-dimensional wind field components. The discrete differential entropy of each signal within a sliding time window is calculated to measure the signal's fluctuation complexity. Only when the differential entropy of either signal exceeds a preset threshold is a complete endurance assessment and replanning decision calculation triggered; otherwise, the original path is maintained, and only low-overhead energy consumption monitoring is performed. By dynamically adjusting the calculation triggering conditions, computational power consumption is significantly reduced while ensuring safety.

[0183] Therefore, in some other embodiments, the energy difference signal and the wind field change rate signal are calculated as follows:

[0184] ;

[0185] .

[0186] in, Characterized as remaining available energy, This is represented as the remaining task energy consumption predicted based on the previous planning results (unupdated path). The three-dimensional wind field vector is represented as the wind field at second t. t is the Euclidean norm, used to calculate the magnitude of wind speed change, and t is the discrete-time index.

[0187] The method for calculating discrete difference entropy is as follows:

[0188] ϵ .

[0189] Within a sliding window of length L (L=5), The maximum and minimum values ​​among the L historical values ​​are divided into K intervals (K=4). This is represented as falling within the i-th interval at second t. The empirical probability is the number of historical values ​​divided by L. If there are no values ​​within the interval, then... ϵ is 0; ϵ is a very small integer; This is the first discrete difference value of the energy difference signal, in bits.

[0190] The calculation method for the differential entropy of the wind field change rate signal is similar, to calculate the second discrete differential entropy. .

[0191] The final event triggering conditions are set as follows:

[0192] .

[0193] The threshold value for energy difference entropy is 1.5 bits in this embodiment. The threshold value for the differential entropy of the wind field change rate is 1.8 bits in this embodiment.

[0194] based on Figure 3 To illustrate the complete technical steps of the present invention: This embodiment is applied to a multi-node last-mile logistics delivery scenario in cities. The drone takes off from the urban logistics distribution center N0, passes through four community delivery nodes (N1, N2, N3, N4) in sequence, and finally arrives at the last-mile delivery station N5. The total path length is about 7.8 kilometers, the rated load is 2.5 kilograms, the operating environment is a typical urban low-altitude environment with crosswinds and gusts of 2-5 m / s, and obstacles such as buildings, roads, and green belts along the way. The delivery time window requires all delivery tasks to be completed within 40 minutes. The optimal delivery path was finally calculated, and the total energy consumption was calculated to be 21.33.

[0195] After the system starts, the S100 step is executed first to complete the acquisition of multi-source sensor data and loading of logistics task information, clarify the take-off point, four intermediate nodes, destination coordinates, delivery time window, load weight, airspace restricted flight zone and other constraints, and at the same time acquire urban wind field and obstacle distribution information in real time.

[0196] The process then proceeds to step S200, where a precise energy consumption model for all operating conditions is constructed based on the six-rotor dynamics, motor ESC model, wind resistance model, and hybrid energy characteristics. The power and energy consumption of each stage, including takeoff, climb, level flight, turn, descent, and hovering delivery, are calculated to form the overall energy consumption objective function.

[0197] In step S300, the system aims to minimize total energy consumption. It constructs a hybrid integer convex programming model, sets the access order of the four intermediate nodes as integer variables, and sets flight speed, altitude, attitude, and power output as continuous variables. It incorporates constraints such as path connectivity, flight safety, logistics time, and energy limitations, and uses the interior point method to solve the problem quickly, obtaining the optimal delivery path and node access order. The overall path avoids areas with strong winds and high obstacles, ensuring optimal energy consumption and flight safety.

[0198] The S400 step allocates power based on the real-time power changes corresponding to the optimal path, using a combination of low-pass filtering and model prediction. The hydrogen fuel cell is responsible for providing stable base power, while the lithium battery is responsible for compensating for instantaneous power spikes caused by gusts, turns, and climbs. By predicting future power demand and adjusting the fuel cell output in advance, the response lag problem is effectively alleviated, and the bus power is kept stable.

[0199] The S500 process performs high-frequency real-time sampling of DC bus voltage and current, continuously assesses the relationship between total output power and demand power, recovers redundant energy to lithium batteries during level flight energy-saving phases, and rapidly replenishes energy gaps by lithium batteries during high-power demand phases, achieving efficient energy utilization and stable system operation, with bus voltage fluctuations controlled within a very small range throughout the entire process.

[0200] Throughout the S600-S660 process, the system monitors the remaining hydrogen energy, lithium battery power, and remaining route energy consumption in real time. When encountering increased local gusts during flight, the system quickly assesses the changes in energy consumption and determines that the range requirements are still met, so there is no need to replan. When approaching the third delivery node, the energy consumption increases slightly due to wind disturbances that are slightly higher than expected. However, the system still maintains sufficient range margin and continues to execute the mission according to the optimal route. Finally, it completes the delivery to all nodes in sequence and successfully arrives at the destination.

[0201] The entire delivery process was characterized by stable flight attitude, smooth power output, rapid energy system response, and no power interruption or voltage anomalies. Actual flight energy consumption was effectively reduced compared to traditional shortest path algorithms, with ample range. All delivery nodes were completed within the time window, and the mission success rate and flight stability met the design targets. This fully verifies the feasibility, safety, and efficiency of the invention in complex urban environments and multi-node logistics delivery scenarios.

[0202] This application also discloses a hydrogen-powered drone logistics path planning system for implementing the above-described method.

[0203] The implementation principle is as follows:

[0204] By constructing a multi-factor coupled precision energy consumption model, a hybrid integer convex programming path optimization model with the goal of minimizing total energy consumption is established. Combining the model's prediction of hydrogen-electric hybrid power allocation and DC bus energy closed-loop management, the global optimal logistics path, the lowest energy consumption, and the highest energy utilization rate are achieved. At the same time, it solves technical problems such as slow dynamic response of hydrogen fuel cells, large power fluctuations, and insufficient long-distance delivery range, thereby improving the efficiency, stability, and reliability of hydrogen-powered drone logistics delivery.

[0205] It should be understood that although the steps in the flowcharts in the accompanying drawings are shown sequentially as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise expressly stated herein, there is no strict order in which these steps are performed, and they may be performed in other orders.

[0206] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for planning logistics routes for hydrogen-powered unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: Acquire multi-dimensional information about the drone, including real-time flight status data, environmental perception data, and logistics delivery task information; Based on the multi-dimensional information and characteristic data, a multi-factor coupled energy consumption model is constructed to determine the real-time power demand of the UAV in each flight phase, and the total energy consumption of the entire flight is calculated based on the real-time power demand. With the goal of minimizing the total energy consumption of the entire flight, a mixed integer convex programming path optimization model is solved to obtain the energy-optimal path; Based on the real-time power demand corresponding to the optimal path, a low-pass filter is used to decompose the real-time power demand corresponding to the optimal path and allocate it to the corresponding type of power supply battery, including the following steps: The real-time demand power is separated into a low-frequency stable component and a high-frequency abrupt component using a low-pass filter. The low-frequency stable component is allocated to the hydrogen fuel cell as the basic output, and the high-frequency abrupt component is allocated to the lithium battery for dynamic compensation. The power allocation satisfies the real-time power balance relationship, expressed as: ， in, The real-time power requirement of the drone. For the output power of hydrogen fuel cells, This refers to the output power of the lithium battery. When allocating power, a model is used to predict future power demand, and the output of the corresponding hydrogen fuel cell is adjusted in advance based on the future power demand to compensate for its dynamic response lag. The process of using low-pass filtering to decompose the real-time power demand corresponding to the optimal path and allocate it to the corresponding type of power supply battery also includes the following steps: A discrete sliding window is introduced and the sample variance corresponding to the recent wind speed measurement is calculated. The cutoff frequency of the low-pass filter is dynamically adjusted based on the sample variance. A proportional-derivative compensation term is applied to the low-frequency stable component based on the rate of change of the cutoff frequency to obtain the optimized low-frequency component. The optimized high-frequency component is calculated based on the optimized low-frequency component and the real-time demand power. Real-time closed-loop energy management is performed on the DC bus of the UAV. Based on the difference between the total output power of the system and the real-time power demand of the UAV, redundant energy is stored or insufficient energy is compensated for.

2. The hydrogen-powered drone logistics route planning method according to claim 1, characterized in that, A multi-factor coupled energy consumption model is constructed to determine the real-time power demand of the UAV in each flight phase. Based on the real-time power demand, the total energy consumption for the entire flight is calculated, including the following steps: The energy consumption per unit time determined based on the multi-factor coupled energy consumption model is expressed as follows: ， Where v is the airspeed and h is the altitude. The pitch angle, For roll angle, Yaw angle Let m be the three-dimensional wind field component, and m be the total mass of the UAV. The air density at different altitudes; The total energy consumption of the entire flight of the UAV is expressed as follows: ， in, Real-time energy consumption at each stage, The start and end times for each stage.

3. The hydrogen-powered drone logistics route planning method according to claim 1, characterized in that, With the goal of minimizing the total energy consumption of the entire flight, the mixed integer convex programming path optimization model is solved, including the following steps: Set the order of delivery node access and the selection of key points in the path to integer variables of 0-1; Flight speed, flight altitude, attitude angle, fuel cell output power, and lithium battery charging and discharging power are set as continuous variables; By combining several constraints, the mixed integer convex programming model is solved to obtain the energy-optimal path; The flight parameter sequence and energy output sequence are output based on the energy-optimal path.

4. The hydrogen-powered drone logistics route planning method according to claim 1, characterized in that, Real-time closed-loop energy management is performed on the DC bus of the UAV. Based on the difference between the total output power of the system and the real-time power demand of the UAV, redundant energy is stored or insufficient energy is compensated for. This includes the following steps: When the total output power of the system is greater than the real-time demand power, the energy difference is used to charge the lithium battery through a bidirectional DC / DC converter to achieve the storage of redundant energy. When the total output power of the system is less than the real-time power demand, the lithium battery is controlled to discharge in order to compensate for the energy shortage.

5. The hydrogen-powered drone logistics route planning method according to claim 1, characterized in that, It also includes the following steps: The remaining available energy of the hydrogen-powered drone is monitored in real time and compared with the energy consumption required to complete the remaining delivery route; When it is determined that the remaining available energy is less than the sum of the energy required to complete the remaining delivery route and the preset safety backup energy, dynamic route replanning is initiated to reconstruct and solve the mixed integer convex programming route optimization model in order to update the optimal route and energy consumption allocation.

6. The hydrogen-powered drone logistics route planning method according to claim 5, characterized in that, It also includes the following steps: Calculate the energy difference between the remaining available energy and the energy required to complete the remaining delivery route; Calculate the rate of change of the wind field corresponding to the three-dimensional wind field vector at adjacent time intervals; A decision sliding window is introduced, and the first discrete difference entropy corresponding to the energy difference and the second discrete difference entropy corresponding to the wind field change rate are calculated within the decision sliding window. Set event triggering conditions and determine whether the first discrete difference entropy and / or the second discrete difference entropy meet the event triggering conditions; The dynamic path replanning is initiated only when the event triggering condition is met, and the mixed integer convex programming path optimization model is reconstructed and solved to update the optimal path and energy consumption allocation.

7. A hydrogen-powered unmanned aerial vehicle (UAV) logistics route planning system, characterized in that, Used to implement the method as described in any one of claims 1-6.

Images