Solar unmanned aerial vehicle energy-path collaborative management method based on hierarchical PSO algorithm

Abstract

The invention relates to the technical field of unmanned aerial vehicle autonomous control and intelligent decision making, in particular to a solar unmanned aerial vehicle energy-path collaborative management method based on a hierarchical PSO algorithm, and the method comprises the steps: collecting the dynamic parameters of an unmanned aerial vehicle through a sensor, and obtaining environment prediction data; taking the candidate flight path as a particle, initializing a particle state and generating a plurality of particles; determining and recording an optimal working voltage sequence of each particle based on the attitude and the irradiance; determining the joint fitness of each particle according to the energy change, the flight time and the path smoothness of each particle; determining an optimal path based on the joint fitness, and updating the state of each particle by adopting a particle swarm optimization constraint condition; when the termination condition is met, acquiring the optimal working voltage sequence of the current optimal path as a power control signal, and controlling the flight of the unmanned aerial vehicle according to the current optimal path and the power control signal; according to the method, the energy-path global energy efficiency optimization effect can be improved.

Description

Technical Field

[0001] This invention relates to the field of autonomous control and intelligent decision-making technology for unmanned aerial vehicles (UAVs), specifically to a method for energy-path collaborative management of solar-powered UAVs based on a hierarchical PSO algorithm. Background Technology

[0002] High-aspect-ratio, long-endurance solar-powered unmanned aerial vehicles (UAVs) have become core platforms for strategic missions such as persistent stratospheric reconnaissance and communication relay, thanks to their ultra-high lift-to-drag ratio and large wing surface photovoltaic (PV) deployment potential. However, the energy systems of these UAVs face unique challenges: the bending deformation of lightweight wings causes dynamic shifts in the PV panel's solar radiation angle, while the energy balance requirement for day-and-night flight necessitates that the UAV's wing surface solar panels absorb as much solar energy as possible. To achieve continuous loiter time of several months, it is urgent to establish a dynamic closed-loop decision-making theory that integrates flight attitude, trajectory, and PV energy capture.

[0003] To address the path planning problem for unmanned aerial vehicles (UAVs), the technical solution provided in Chinese invention patent application CN119088048A employs a path optimization method based on an improved particle swarm optimization (PSO) algorithm. After initializing the particle swarm, the current position of each particle is set as its individual historical best solution (individual extreme value), and the optimal solution for the entire swarm (global extreme value) is selected. During iteration, particles are evaluated using a fitness function (integrating route and energy consumption constraints): if the current position has better fitness, its individual extreme value is updated; if the optimal individual extreme value for the entire swarm is better than the global extreme value, the global extreme value is updated. A hybrid probability mechanism is introduced to select particles to enter a hybrid pool for collaborative optimization. The iteration terminates when a preset upper limit is reached, and the path corresponding to the global extreme value is output as the optimal solution.

[0004] The core flaw of this existing technology lies in the fact that, for high aspect ratio UAVs, the wing bending deformation (pitch angle θ / roll angle) has not been established. The quantitative mapping model between the photovoltaic panel and the effective light-receiving area leads to problems during the turning phase (typical roll-off). While still operating at theoretical planar efficiency, the irradiance loss of the outer photovoltaic panels exceeds 35%; dynamic response is fragmented, path planning ignores the spatiotemporal propagation characteristics of cloud cover (e.g., cumulus cloud movement speed > 5 m / s), and static flight paths cannot avoid sudden shadow areas. Therefore, developing a real-time collaborative decision-making theory that integrates wing deformation attitude, dynamic illumination prediction, and multi-scale energy consumption modeling is of great significance for overcoming the energy management bottleneck of ultra-long-endurance high-aspect-ratio solar-powered UAVs. Summary of the Invention

[0005] In view of the above problems, this invention provides a method for energy-path collaborative management of solar-powered UAVs based on hierarchical PSO algorithm, which solves the technical problems of lack of attitude-light dynamic coupling modeling and lag in global energy efficiency optimization of high aspect ratio long endurance solar-powered UAVs in energy-path collaborative optimization.

[0006] This invention provides a method for energy-path collaborative management of solar-powered UAVs based on a hierarchical PSO algorithm, comprising the following steps:

[0007] Step S1: Collect dynamic parameters of the UAV by sensors and obtain environmental prediction data; the UAV dynamic parameters include position, attitude and speed; the environmental prediction data includes irradiance and three-dimensional wind speed field;

[0008] Step S2: Treat the candidate flight path as a particle, initialize the particle state with the position in the UAV dynamic parameters, and generate multiple particles;

[0009] Step S3: Determine and record the optimal working voltage sequence for each particle based on the attitude and irradiance.

[0010] The joint fitness of each particle is determined by its energy change, flight time, and path smoothness.

[0011] Step S4: Determine the optimal path based on the joint fitness, and update the state of each particle using particle swarm optimization constraints;

[0012] Step S5: Return to step S3 until the termination condition is met. Obtain the optimal operating voltage sequence of the current optimal path as the power control signal, and control the drone flight with the current optimal path and the power control signal.

[0013] Preferably, in step S1, the position is the three-dimensional coordinate of the UAV in the geocentric coordinate system, the attitude includes the roll angle, pitch angle and yaw angle of the UAV, the speed is the speed of the UAV in the geocentric coordinate system, the irradiance represents the solar irradiance intensity within a preset future time period, and the three-dimensional wind speed field represents the wind speed in the geocentric coordinate system within a preset future time period.

[0014] Preferably, step S2 specifically includes:

[0015] Step S2-1: Treat the candidate flight path as a particle, initialize the position of the particle with the position in the UAV dynamic parameters, and generate multiple particles;

[0016] Step S2-2: Initialize the velocity of the particles to zero, and determine the position of each particle as the initial value of the individual optimal solution of each particle.

[0017] Preferably, step S2-1 specifically includes:

[0018] The candidate flight path includes a sequence of waypoints at multiple discrete moments. Multiple UAV positions are randomly generated within a cube-shaped region centered on the current UAV position. These multiple UAV positions are used as the coordinates of the first waypoint to form multiple candidate flight paths. The waypoint sequence of each candidate flight path is encoded as the position of a particle, generating multiple particles. The expression for initializing the particle positions is:

[0019]

[0020] in, Let represent the initial position of the k-th particle, Pos(t) represent the position of the UAV at time t, and ΔR k Let ΔR represent the random perturbation of the k-th particle. k Obey [-R] max ,R max A uniform distribution over the interval, R max For the maximum search radius, N p This represents the total number of particles.

[0021] Preferably, step S3 specifically includes:

[0022] Step S3-1: Determine the effective irradiance based on the attitude and irradiance, and determine the maximum photovoltaic power sequence and the corresponding optimal operating voltage sequence for each particle based on the effective irradiance;

[0023] Step S3-2: For each particle, the energy change is determined by the maximum photovoltaic power sequence, the airspeed determined by the speed of the UAV and the three-dimensional wind speed field;

[0024] Step S3-3: Determine the flight time from waypoint coordinates and airspeed, and calculate the path smoothness; determine the joint fitness of each particle from energy change, flight time, and path smoothness.

[0025] Preferably, step S3-1 specifically includes calculating the effective irradiance using the following expression:

[0026] G eff (t i ) = G pred (t i )·cos(θ sun (t i )-φ roll (t i ))·η shading

[0027] Among them, G pred (t i ) for t i Predicted irradiance at time; θsun (t i ) represents the angle of incidence of the sun; φ roll (t i η is the roll angle of the drone; shading The wing self-shielding coefficient;

[0028] Maximum power point tracking optimization was performed to obtain the maximum photovoltaic power of the aircraft photovoltaic energy at each waypoint. And the optimal operating voltage corresponding to maximum power. The maximum photovoltaic power sequence and the optimal operating voltage sequence are formed; the constraint expressions are:

[0029]

[0030] in, Indicates the search working voltage V mppt To maximize the objective value, P pv G represents photovoltaic power. eff η represents irradiance. cell β is the intrinsic conversion efficiency of the photovoltaic cell, and T is the temperature degradation coefficient. cell T represents the actual temperature of the battery. amb For ambient temperature, T std For standard temperature, η mppt (V mppt (V) represents the DC-DC conversion efficiency curve; oc,min V oc,max These are the minimum and maximum operating voltages, respectively.

[0031] Preferably, step S3-2 specifically includes:

[0032] The energy change is calculated using the following expression:

[0033]

[0034] Where, ΔE k (t i ) represents the k-th particle at time t i Energy changes over time P represents the maximum photovoltaic power at time τ. prop (τ) represents the propulsion power at time τ, P avionics This represents the constant power consumption of avionics. C represents the photovoltaic power at maximum operating voltage. D ρ is the aerodynamic drag coefficient; ρ is the air density; A is the wing reference area; Vel is the airspeed; ||·|| represents the modulus; C D0 K is the zero lift coefficient. C Here, α is the induced drag coefficient, and α is the angle of attack.

[0035] Step S3-3 specifically includes:

[0036] The joint fitness is calculated using the following expression:

[0037]

[0038] Wherein, F(X) path,k Let ∑ΔE be the joint fitness of the k-th particle. k Let t represent the sum of the energy changes of the k-th particle at each discrete time point. total κ is the time efficiency term, N is the path smoothing term, and N is the path efficiency term. t WP represents the total number of discrete time points. i+1 ,WP i The waypoint coordinates at times i+1 and i are respectively, Vel i Let ω1 be the airspeed at time i, r(s) be the path curve parameterized with arc length s, and ω1 / ω2 / ω3 be the weighting coefficients.

[0039] Preferably, step S4 specifically includes:

[0040] Step S4-1: For each particle, obtain the current joint fitness and historical joint fitness of the particle, and update the individual optimal solution with the particle corresponding to the maximum joint fitness.

[0041] Obtain the current joint fitness and historical joint fitness of all particles in the population, and update the global optimal solution with the particle corresponding to the maximum joint fitness.

[0042] Step S4-2: Update the state of each particle using particle swarm optimization constraints based on the individual optimal solution and the global optimal solution.

[0043] Preferably, step S4-2 specifically includes updating the state of each particle using the following expression:

[0044]

[0045] in, These represent the velocities of particle k at the (j+1)th and jth iterations, respectively. P represents the position of particle k at the (j+1)th and jth iterations, respectively. best,k G represents the individual optimal solution for particle k. best Let χ represent the global optimal solution, w be the inertia weight, c1 be the cognitive learning factor, c2 be the social learning factor, and r1 and r2 be random perturbation terms.

[0046] The state of each particle is constrained by the following expression:

[0047] like but

[0048]

[0049] ||WP i -WP i-1 ||≥d min

[0050] Among them, X min ,X max Let be the lower and upper bounds of the search space for the particle's position, respectively, where proj(·) represents the projection function, and d min This indicates the minimum distance between waypoints.

[0051] Preferably, in step S5, the expression for the termination condition is:

[0052]

[0053] in, This represents the globally optimal solution in the j-th iteration. Let ε represent the global optimal solution in the (j-Δj)th iteration. pos For positional tolerance, ε represents the average joint fitness of each particle in the j-th and (j-1)-th iterations, respectively. fit For fitness tolerance, J max This represents the maximum number of iterations.

[0054] Compared with the prior art, the present invention has at least the following beneficial effects:

[0055] (1) This invention introduces a solar-powered UAV energy-path collaborative management method based on hierarchical particle swarm optimization (PSO) algorithm, realizing dynamic coupling optimization between flight status and energy system. This invention fully considers the dynamic feedback relationship between UAV attitude, wing deformation, and energy harvesting efficiency under different ambient light conditions, and more comprehensively models energy capture during flight, improving the accuracy and effectiveness of energy management, and effectively enhancing the UAV's mission sustainability in complex environments.

[0056] (2) By dynamically adjusting the maximum power point tracking (MPPT) strategy during path planning, this invention solves the problem of global energy utilization lag caused by the independent operation of path planning and power control systems. In particle swarm optimization, path and energy utilization are correlated within the fitness function, enabling flight path selection to not only consider range and safety but also respond in real-time to environmental changes and energy supply and demand, achieving closed-loop optimization of the entire energy capture and utilization process, effectively extending the UAV's flight time and improving system reliability.

[0057] (3) The present invention can process flight and environmental information collected by sensors in real time and perform large-scale particle swarm optimization calculations to improve the autonomous decision-making and emergency response capabilities of UAVs in ultra-long flight missions. Attached Figure Description

[0058] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0059] Figure 1 The flowchart of the solar-powered UAV energy-path collaborative management method based on the hierarchical PSO algorithm provided by the present invention is shown.

[0060] Figure 2 The flowchart of the energy-path collaborative management method for solar-powered UAVs based on the hierarchical PSO algorithm provided by this invention is shown in the figure. Detailed Implementation

[0061] To better understand the above-described objectives, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other. Furthermore, the present invention can be implemented in other ways different from those described herein; therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0062] This invention provides an energy-path collaborative management method for solar-powered UAVs based on a hierarchical particle swarm optimization (PSO) algorithm. It couples the dynamic interaction mechanism between the path planning system and the maximum power point tracking (MPPT) control system. By establishing a two-layer prediction architecture of a flight attitude-illuminance mapping model and an energy consumption-trajectory prediction model, and designing a hierarchical collaborative mechanism between the outer PSO (path planning layer) and the inner PSO (MPPT optimization layer), a closed-loop energy decision-making system under a time-domain optimization objective function is constructed. This system can consider the dynamic coupling relationship between flight attitude (especially attitude changes caused by wing bending deformation) and illumination distribution / reception efficiency, obtaining the optimal energy management scheme and planned path, thereby improving the performance of UAVs performing ultra-long endurance missions.

[0063] like Figure 1 As shown, this invention provides a method for energy-path collaborative management of solar-powered UAVs based on a hierarchical PSO algorithm, comprising the following steps:

[0064] Step S1: Collect dynamic parameters of the UAV by sensors and obtain environmental prediction data; the UAV dynamic parameters include position, attitude and speed; the environmental prediction data includes irradiance and three-dimensional wind speed field;

[0065] This invention first reads the dynamic parameters of the UAV and environmental prediction data to provide data support for subsequent decision-making. The UAV dynamic parameters include the UAV position vector, UAV attitude angle, UAV velocity vector, and UAV battery state of charge, etc., which are described in detail below.

[0066] UAV position vector The unit is meters (m), and t is time. This represents the 3D real number field, i.e., the three-dimensional coordinates in the geocentric coordinate system. It represents the three-dimensional coordinates (x, y, z) of the UAV in the geocentric coordinate system; the UAV attitude angle Θ(t) = [φ(t), θ(t), ψ(t)]. T The units are radians (rad), namely roll, pitch, and yaw angles; the UAV velocity vector (v) x ,v y ,v z The unit is m / s, which is the speed of the UAV in the geocentric coordinate system; the UAV battery state of charge (SOC(t)) ∈ [0,1] is the percentage of the battery's current remaining usable capacity to its maximum usable capacity when fully charged (0 = empty, 1 = fully charged). These parameters need to be collected in real time by sensor systems (such as IMU, GPS, ammeter, etc.) and the update frequency is marked with a timestamp t.

[0067] Environmental prediction data includes irradiance and three-dimensional wind speed field, expressed as:

[0068]

[0069] in, It represents the time from t to t+T. h Environmental prediction data at time T h For the time domain of the prediction, the unit is seconds, i.e., the length of future time covered by the prediction, N t =T h / Δt is the number of discrete time steps, i.e., T h The total number of steps obtained after discretization with a fixed time step Δt; (x,t) i (x) represents the spacetime coordinates, x represents the spatial location, and t represents the time coordinates. i =t+i·Δt represents the i-th discrete time; G pred (x,t i ) represents the predicted position x at t i Solar irradiance at a given time, expressed in W / m². 2 W pred (x,t i)=[w x ,w y ,w z ] T To predict the position x at t i The three-dimensional wind speed field at any given moment, [·] T Indicates transpose, w x ,w y ,w z These represent the wind speed components of the three-dimensional wind speed field along each coordinate axis in the geocentric coordinate system.

[0070] In some embodiments, the environmental prediction data of the present invention can be obtained by real-time monitoring of the radiation illuminance, wind speed and wind direction of the current area by meteorological sensors, or by obtaining relevant meteorological data from external data sources to determine the radiation illuminance, wind speed and wind direction; the collected data can be simulated and extrapolated by combining environmental models to obtain the prediction results of the radiation illuminance and three-dimensional wind speed field in the future specified time domain.

[0071] Step S2: Using the candidate flight path as a particle, initialize the particle state with the position in the UAV dynamic parameters to generate multiple particles;

[0072] This invention addresses the energy management and path planning problems of unmanned aerial vehicles (UAVs) by employing a particle swarm optimization (PSO) algorithm to search the feasible solution space. The invention uses candidate flight paths as particles in the PSO algorithm, and each candidate path includes the coordinates of waypoints at multiple discrete time points. The particles are represented by the following expression:

[0073]

[0074] Among them, X path,k This represents the sequence of waypoint coordinates for the k-th particle, where k = 1, 2, ..., N. p N p WP is the total number of particles in the swarm. i =[x i ,y i ,z i ] T N represents the coordinates of a three-dimensional waypoint at the i-th discrete time step, in meters; WP This represents the total number of waypoints along the route. The degree of freedom of the space is 3N. WP The real number field.

[0075] Each particle represents a route to be optimized. This route is encoded as the particle's "position," and the optimal route is eventually obtained through iterative searching using the particle swarm optimization algorithm.

[0076] This invention generates multiple particles and initializes the state of each particle. That is, it generates the initial position and velocity of the particles according to rules.

[0077] The expression for initializing the particle's position is:

[0078]

[0079] in, Let ΔR represent the initial position of the k-th particle. k Let ΔR represent the random perturbation of the k-th particle. k Obey [-R] max ,R max A uniform distribution over the interval, R max >0 represents the maximum search radius in meters, determined by physical scene constraints (such as sensor detection range or airspace boundaries); the geometric meaning of particle position initialization is that the first waypoint of the initial path of the particle swarm is centered on the current position of the UAV within the cubic region [-R max ,R max ] 3 Randomly generated points within the area.

[0080] The expression for initializing the particle's velocity is:

[0081]

[0082] in, This represents the initial velocity of the k-th particle. An initial velocity of 0 indicates that the particle is initially in a "stationary" state. Subsequent iterations rely on the social cognition term in the particle swarm algorithm to guide the motion.

[0083] This invention also initializes the individual optimal solution, expressed as:

[0084]

[0085] Among them, P best,k This represents the individual optimal solution for the k-th particle. The individual optimal solution is initialized with the initial position of the particle. In subsequent steps, the individual optimal solution will be iteratively updated.

[0086] Through the above steps, this invention generates and initializes particles representing candidate flight paths by randomly scattering points within a cube-shaped area, centered on the current location of the UAV, as the first waypoint.

[0087] Step S3: Determine and record the optimal working voltage sequence for each particle based on the attitude and irradiance.

[0088] The joint fitness of each particle is determined by its energy change, flight time, and path smoothness.

[0089] In this step, the present invention first determines the quantitative relationship between the effective irradiance and radiation irradiance acting on the wing and the attitude of the UAV, and then determines the maximum photovoltaic power and the corresponding optimal operating voltage sequence based on the effective irradiance.

[0090] The effective irradiance acting on the wings of a high aspect ratio, long-endurance solar-powered UAV during flight is affected by the current flight attitude of the UAV. This invention models the effective irradiance and incorporates it into subsequent optimization steps.

[0091] The relationship between effective irradiance, radiation irradiance, and UAV attitude is expressed as follows:

[0092] G eff (t i ) = G pred (t i )·cos(θ sun (t i )-φ roll (t i ))·η shading

[0093] Among them, G pred (t i ) for t i The predicted solar irradiance at any given time, which is the irradiance from the environmental prediction data obtained in step S1; θ sun (t i ) represents the angle of incidence of the sun (in radians); φ roll (t i ) represents the roll angle of the drone (in radians), obtained in step S1; η shading This is the wing self-blocking coefficient, with a value of [0,1].

[0094] A constrained optimization problem is constructed, and the Maximum Power Point Tracking (MPPT) method is used for optimization to obtain the maximum power point of the aircraft's photovoltaic energy and the corresponding optimal operating voltage at each waypoint. The constraint expression for the MPPT optimization is as follows:

[0095]

[0096] in, Indicates the search working voltage V mppt To maximize the objective value, P pv G represents photovoltaic power. eff η represents irradiance. cell The intrinsic conversion efficiency of a photovoltaic cell is typically a calibrated value, β is the temperature degradation coefficient (in % / ℃), and T is the temperature coefficient. cell The actual temperature of the battery, satisfying T cell =T amb +0.03·Geff T amb The ambient temperature, i.e., the real-time atmospheric temperature of the drone's flight environment, is expressed in degrees Celsius (°C). std Standard temperature, representing the reference temperature of a photovoltaic cell under standard laboratory test conditions (STC), is fixed at 25°C. η mppt (V mppt The DC-DC conversion efficiency curve is obtained by piecewise polynomial fitting. Specifically, quadratic polynomials can be used to calculate the DC-DC conversion efficiency curve for multiple voltage ranges; V oc,min V oc,max These are the minimum and maximum operating voltages, V. oc,min =0.7V oc,STC V oc,max =0.95V oc,STC V oc,sTC This is the open-circuit voltage under standard test conditions.

[0097] Through the above methods, this invention optimizes the MPPT method based on the relationship between effective irradiance, radiation irradiance, and UAV attitude to obtain the maximum photovoltaic power of aircraft photovoltaic energy at each waypoint. And the optimal operating voltage corresponding to maximum power. And form the optimal operating voltage sequence.

[0098] To evaluate the joint fitness of each path, this invention constructs an energy dynamic balance equation and establishes a net energy change model to calculate the energy change, expressed as:

[0099]

[0100] Where, ΔE k (t i ) represents the k-th particle at time t i Energy changes over time P represents the maximum photovoltaic power at time τ. prop (τ) represents the propulsion power at time τ, P avionics This represents the constant power consumption of avionics. C represents the photovoltaic power at maximum operating voltage. D ρ is the aerodynamic drag coefficient; ρ is the air density (unit: kg / m³). 3 A is the wing reference area (in m²). 2 Vel represents airspeed (in m / s), and ||·|| represents the modulus. This invention determines the airspeed modulus based on the UAV's velocity and the three-dimensional wind field in step S1. D0 The zero-lift coefficient represents the basic aerodynamic drag coefficient of the UAV when it is in a zero-lift state, K.C The induced drag coefficient represents the additional drag coefficient caused by lift generation, and α is the angle of attack, which is the angle between the wing chord and the direction of relative airflow (unit: degrees).

[0101] The joint fitness of each path in this invention comprehensively considers energy variation, flight time, and path smoothness. The energy variation has already been obtained through the above steps. The joint fitness expression for each path is as follows:

[0102]

[0103] Wherein, F(X) path,k Let ∑ΔE be the joint fitness of the k-th particle. k Let t represent the sum of the energy changes of the k-th particle at each discrete time point. total κ represents the time efficiency term, and κ represents the path smoothing term.

[0104] N t WP represents the total number of discrete time points. i+1 ,WP i The waypoint coordinates at times i+1 and i are respectively, Vel i Let ω1 be the airspeed at time i, and r(s) be the path curve parameterized with arc length s. ω1 / ω2 / ω3 are weighting coefficients, which should satisfy ω1+ω2+ω3=1.

[0105] Through the above calculations, this invention obtains the energy change based on the maximum photovoltaic power of the current candidate path, the speed of the UAV, and the airspeed determined by the three-dimensional wind field; it obtains the flight time based on the waypoint coordinates and airspeed; and it calculates the path smoothness and weights it to obtain the joint fitness.

[0106] Step S4: Determine the optimal path based on the joint fitness, and update the state of each particle using particle swarm optimization constraints;

[0107] In this step, the individual optimal solution P for each particle and the particle swarm is first obtained from the joint fitness of each candidate path, which is also the joint fitness of each particle. best,k and the global optimal solution G best .include:

[0108] For each particle, obtain the particle's current joint fitness and historical joint fitness, and update the individual optimal solution with the particle corresponding to the maximum joint fitness.

[0109] Obtain the current joint fitness and historical joint fitness of all particles in the population, and update the global optimal solution with the particle corresponding to the maximum joint fitness.

[0110] After obtaining the individual optimal solution and the global optimal solution, this invention uses particle swarm optimization constraints to update the state of each particle. The state update expression is as follows:

[0111]

[0112] in, These represent the velocities of particle k at the (j+1)th and jth iterations, respectively. Let χ represent the positions of particle k in the (j+1)th and jth iterations, respectively. χ is the compression factor, w is the inertia weight, c1 is the cognitive learning factor, c2 is the social learning factor, and r1 and r2 are random perturbation terms, following a Gaussian distribution. The value of w adopts a linear decay strategy. max w represents the initial value of the inertia weight in the particle swarm optimization algorithm, indicating the strength of the particle's tendency to maintain its original motion in the early stages of iteration. min In particle swarm optimization (PSO) algorithm, the termination value of the inertia weight represents the minimum strength at which particles maintain their motion tendency when the algorithm converges. J max This represents the maximum number of iterations and is one of the termination conditions.

[0113] The constraints are:

[0114] like but

[0115]

[0116] ||WP i -WP i-1 ||≥d min

[0117] Among them, X min ,X max Here, represents the lower and upper bounds of the search space for particle positions (i.e., waypoint sequences), respectively, and proj(·) is the projection function. When the updated particle positions... When proj(·) projects it into the boundary of the search space, d min This indicates the minimum distance between waypoints.

[0118] Through the above steps, each particle updates its velocity and position based on inertia, self-awareness, and social learning, while satisfying constraints such as not exceeding boundaries, not exceeding speed limits, and not being too close between waypoints. After completing the above steps, the process returns to step S3, where the adjusted particles evaluate each candidate flight path, and the optimization is performed iteratively.

[0119] Step S5: Return to step S3 until the termination condition is met. Obtain the optimal operating voltage sequence of the current optimal path as the power control signal, and control the drone flight with the current optimal path and the power control signal.

[0120] The optimization termination condition expression determined by this invention is:

[0121]

[0122] in, This represents the globally optimal solution in the j-th iteration. Let ε represent the global optimal solution in the (j-Δj)th iteration. pos For positional tolerance, ε represents the average joint fitness of each particle in the j-th and (j-1)-th iterations, respectively. fit For fitness tolerance, J max This represents the maximum number of iterations.

[0123] When the above termination conditions are met, obtain the optimal path corresponding to the current global optimal solution. * and Path * The corresponding optimal operating voltage sequence As a power control signal.

[0124] The system controls the flight of drones using optimal paths and power control signals, thus achieving coordinated energy path management for solar-powered drones.

[0125] To illustrate the effectiveness of the method proposed in this invention, the following detailed description of the above technical solution of this invention is provided through a specific embodiment.

[0126] Example 1

[0127] This embodiment takes a certain type of high aspect ratio solar-powered UAV as an example to demonstrate the energy-path collaborative planning method.

[0128] The example scenario is set as follows: the UAV flies at an altitude of 2000m, hovering over Beijing. Specific values ​​for each parameter are given below. Figure 2 As shown, this embodiment provides a method for collaborative energy management and path planning of solar-powered UAVs based on hierarchical PSO:

[0129] The first step is to read the drone's dynamic parameters and environmental prediction data to provide data support for subsequent decision-making.

[0130] The dynamic parameters of the UAV include the UAV position vector. The value is Pos(t) = [116.4, 39.9, 2000] T (m); UAV attitude angle Θ(t) = [0.12, 0.05, 1.57] T (rad), representing roll, pitch, and yaw angles; UAV velocity vector [25,0,0.2] T(m / s), which is the speed of the UAV in the geocentric coordinate system; the UAV's battery state of charge (SOC(t)) ∈ [0,1], assumed to be 0.65 at takeoff. These parameters need to be collected in real time by sensor systems (such as IMU, GPS, ammeter) and the update frequency is marked with a timestamp t.

[0131] Environmental prediction data obtains predictive information for the future time domain through environmental models or external data sources. The expression is:

[0132]

[0133] Where T h For the prediction time domain, the value is 7200 seconds; N t =T h / Δt is the number of discrete time steps, with a value of 120; (x,t) i (x) represents the spacetime coordinates, x represents the spatial location, and t represents the time coordinates. i =t+i·Δt represents the i-th discrete time; G pred (x,t i (W / m²) represents irradiance. 2 That is, the predicted position is x at t i Solar irradiance at a given time; W pred (x,t i )=[w x ,w y ,w z ] T Let W be the three-dimensional wind speed field. pred (x,t i = [2.5, -1.2, 0.1] T ±0.3.

[0134] The second step is to establish an outer loop (path planning layer) to set the basic framework for the optimization solver of the UAV 3D path planning problem, and to search for the globally optimal waypoint sequence in the feasible solution space using the particle swarm optimization algorithm.

[0135] The decision space is constructed using waypoint coordinates as optimization variables. Path particles are represented by the following formula:

[0136]

[0137] Among them WP i =[x i ,y i ,z i ] T N represents the coordinates of the i-th 3D waypoint, in meters; WP The total number of waypoints for the planned path is 5; particle dimension D = 3N WPThe corresponding degrees of freedom in the solution space are 18. Particle swarm size. The value is 100.

[0138] Position initialization is represented by the following formula:

[0139]

[0140] in, For uniformly distributed random disturbances, R max The value is 5000(m); U is defined in [-R max ,R max Uniform distribution over the interval.

[0141] Velocity initialization is expressed by the following formula:

[0142]

[0143] That is, the initial velocity of all particles is set to zero vector. In a dynamical interpretation, the particles are initially in a "stationary" state, and subsequent iterations rely on social cognition terms to guide the motion.

[0144] The initialization of an individual's optimal solution is expressed by the following formula:

[0145]

[0146] The initial position of the particle is set to the individual's historical optimal solution to ensure that the objective function value improves monotonically.

[0147] The third step is to establish the inner layer (MPPT optimization layer) loop. This involves performing illumination-attitude mapping modeling to establish a quantitative relationship between solar irradiance and UAV attitude.

[0148] G eff (t i ) = G pred (t i )·cos(θ sun (t i )-φ roll (t i ))·η shading

[0149] Among them, G pred (t i ) for t i The predicted solar irradiance at any given time comes from the environmental predictions in the first step; θ sun (t i Let θ be the angle of incidence of the sun (in radians), and the formula is: φ roll (t i η is the roll angle (in radians) of the UAV, generated by the flight control law; shadingThe wing self-shading coefficient has a value of 0.92.

[0150] To optimize the MPPT output voltage, a constrained optimization problem is constructed:

[0151]

[0152] Where, η cell β is the intrinsic conversion efficiency (calibrated value) of the photovoltaic cell, which is 0.22; β is the temperature degradation coefficient (% / ℃), which is -0.0045K. -1 K represents Kelvin temperature; T cell The actual temperature of the battery, satisfying T cell =T amb +0.03·G eff η mppt (V mppt The DC-DC conversion efficiency curve is fitted using a piecewise polynomial: η mppt =a0+a1V+a2V 2 (V∈[V k V k+1 ]), a0 = 0.92, a1 = 0.002, a2 ​​= -0.0001. The constraint boundary is: V oc,min =0.7V oc,STC The value is 21; V oc,max =0.95V oc,STC The value is 28.5.

[0153] Construct an energy dynamic balance equation and establish a net energy change model:

[0154]

[0155] in, P represents the maximum photovoltaic power. prop To drive power; C D C is the aerodynamic drag coefficient. D0 ρ is the aerodynamic drag coefficient at zero lift, with a value of 0.02; ρ is the air density, with a value of 1 (kg / m³). 3 A represents the wing reference area, which is 4.5 (m²). 2 ); ||Vell|| is the airspeed modulus (m / s); P avionics This is the constant power consumption for avionics, with a value of 30 (W).

[0156] Perform joint fitness evaluation and construct the fitness function:

[0157]

[0158] Where, ∑ΔE k For energy margin; t totalκ is the time efficiency term; r(s) is the path smoothing term; ω1 / ω2 / ω3 are the path curve parameterized by arc length s; ω1 = 0.5, ω2 = 0.3, ω3 = 0.2.

[0159] A particle state update mechanism is established, using the Clerc-constrained PSO model:

[0160]

[0161]

[0162] Where χ is the compression factor with a value of 0.729; ω is the inertia weight, which adopts a linear decay strategy, i.e. c1 is the cognitive learning factor; c2 is the social learning factor, c1 = c2 = 2.05; r1 and r2 are random perturbation terms, belonging to a standard Gaussian distribution. The constraints are: location boundary constraints, if... Then X (j +1) =proj(X (j+1) ); velocity constraint, |V (j+1) |≤0.2(X max -X min ); minimum waypoint spacing, ||WP i -WP i-1 ||≥d min .

[0163] The fourth step is to establish a termination condition determination system, the specific expression of which is as follows:

[0164]

[0165] Where, ε pos For positional tolerance; ε fit For fitness tolerance; J max This represents the maximum number of iterations.

[0166] Step 5: Output path and power control signals. The path output signal is represented as follows:

[0167]

[0168] Where, Path * The optimal path; It is a spatial coordinate sequence; d min Minimum flight segment constraint.

[0169] The power control signal is represented as follows:

[0170]

[0171] in, For power control sequence; For the Nth t The optimal operating voltage at each time point.

[0172] While the specific embodiments of the present invention depict actions or steps in a particular order, this should be understood as requiring such actions or steps to be performed in the specific order shown or in sequential order, or requiring all illustrated actions or steps to be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.

[0173] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for energy-path collaborative management of solar-powered unmanned aerial vehicles based on a hierarchical PSO algorithm, characterized in that, Includes the following steps: Step S1: Collect dynamic parameters of the UAV by sensors and obtain environmental prediction data; the UAV dynamic parameters include position, attitude and speed; the environmental prediction data includes irradiance and three-dimensional wind speed field; Step S2: Treat the candidate flight path as a particle, initialize the particle state with the position in the UAV dynamic parameters, and generate multiple particles; Step S3: Determine and record the optimal working voltage sequence for each particle based on the attitude and irradiance. The joint fitness of each particle is determined by its energy change, flight time, and path smoothness. Step S4: Determine the optimal path based on the joint fitness, and update the state of each particle using particle swarm optimization constraints; Step S5: Return to step S3 until the termination condition is met. Obtain the optimal operating voltage sequence of the current optimal path as the power control signal, and control the drone flight with the current optimal path and the power control signal.

2. The energy-path collaborative management method for solar-powered UAVs based on the hierarchical PSO algorithm according to claim 1, characterized in that, In step S1, the position is the three-dimensional coordinate of the UAV in the geocentric coordinate system, the attitude includes the roll angle, pitch angle and yaw angle of the UAV, the speed is the speed of the UAV in the geocentric coordinate system, the irradiance represents the solar irradiance intensity within a preset future time period, and the three-dimensional wind speed field represents the wind speed in the geocentric coordinate system within a preset future time period.

3. The energy-path collaborative management method for solar-powered UAVs based on the hierarchical PSO algorithm according to claim 2, characterized in that, Step S2 specifically includes: Step S2-1: Treat the candidate flight path as a particle, initialize the position of the particle with the position in the UAV dynamic parameters, and generate multiple particles; Step S2-2: Initialize the velocity of the particles to zero, and determine the position of each particle as the initial value of the individual optimal solution of each particle.

4. The energy-path collaborative management method for solar-powered UAVs based on the hierarchical PSO algorithm according to claim 3, characterized in that, Step S2-1 specifically includes: The candidate flight path includes a sequence of waypoints at multiple discrete moments. Multiple UAV positions are randomly generated within a cube-shaped region centered on the current UAV position. These multiple UAV positions are used as the coordinates of the first waypoint to form multiple candidate flight paths. The waypoint sequence of each candidate flight path is encoded as the position of a particle, generating multiple particles. The expression for initializing the particle positions is: in, Let represent the initial position of the k-th particle, Pos(t) represent the position of the UAV at time t, and ΔR k Let ΔR represent the random perturbation of the k-th particle. k Obey [-R] max ,R max A uniform distribution over the interval, R max For the maximum search radius, N p This represents the total number of particles.

5. The solar-powered UAV energy-path collaborative management method based on the hierarchical PSO algorithm according to claim 4, characterized in that, Step S3 specifically includes: Step S3-1: Determine the effective irradiance based on the attitude and irradiance, and determine the maximum photovoltaic power sequence and the corresponding optimal operating voltage sequence for each particle based on the effective irradiance; Step S3-2: For each particle, the energy change is determined by the maximum photovoltaic power sequence, the airspeed determined by the speed of the UAV and the three-dimensional wind speed field; Step S3-3: Determine the flight time from waypoint coordinates and airspeed, and calculate the path smoothness; determine the joint fitness of each particle from energy change, flight time, and path smoothness.

6. The energy-path collaborative management method for solar-powered UAVs based on the hierarchical PSO algorithm according to claim 5, characterized in that, Step S3-1 specifically includes calculating the effective irradiance using the following expression: G eff (t i )=G pred (t i )·cos(θ sun (t i )-φ roll (t i ))·or shading Among them, G pred (t i ) for t i Predicted irradiance at time; θ sun (t i ) represents the angle of incidence of the sun; φ roll (t i η is the roll angle of the drone; shading The wing self-shielding coefficient; Maximum power point tracking optimization was performed to obtain the maximum photovoltaic power of the aircraft photovoltaic energy at each waypoint. And the optimal operating voltage corresponding to maximum power. The maximum photovoltaic power sequence and the optimal operating voltage sequence are formed; the constraint expressions are: in, Indicates the search working voltage V mppt To maximize the objective value, P pv G represents photovoltaic power. eff η represents irradiance. cell β is the intrinsic conversion efficiency of the photovoltaic cell, and T is the temperature degradation coefficient. cell T represents the actual temperature of the battery. amb For ambient temperature, T std For standard temperature, η mppt (V mppt (V) represents the DC-DC conversion efficiency curve; oc,min V oc,max These are the minimum and maximum operating voltages, respectively.

7. The solar-powered UAV energy-path collaborative management method based on the hierarchical PSO algorithm according to claim 6, characterized in that, Step S3-2 specifically includes: The energy change is calculated using the following expression: Where, ΔE k (t i ) represents the k-th particle at time t i Energy changes over time P represents the maximum photovoltaic power at time τ. prop (τ) represents the propulsion power at time τ, P avionics This represents the constant power consumption of avionics. C represents the photovoltaic power at maximum operating voltage. D ρ is the aerodynamic drag coefficient; ρ is the air density; A is the wing reference area; Vel is the airspeed; ||·|| represents the modulus; C D0 K is the zero lift coefficient. C Here, α is the induced drag coefficient, and α is the angle of attack. Step S3-3 specifically includes: The joint fitness is calculated using the following expression: Wherein, F(X) path,k Let ∑ΔE be the joint fitness of the k-th particle. k Let t represent the sum of the energy changes of the k-th particle at each discrete time point. total κ is the time efficiency term, N is the path smoothing term, and N is the path efficiency term. t WP represents the total number of discrete time points. i+1 ,WP i The waypoint coordinates at times i+1 and i are respectively, Vel i Let ω1 be the airspeed at time i, r(s) be the path curve parameterized with arc length s, and ω1 / ω2 / ω3 be the weighting coefficients.

8. The energy-path collaborative management method for solar-powered UAVs based on the hierarchical PSO algorithm according to claim 5, characterized in that, Step S4 specifically includes: Step S4-1: For each particle, obtain the current joint fitness and historical joint fitness of the particle, and update the individual optimal solution with the particle corresponding to the maximum joint fitness. Obtain the current joint fitness and historical joint fitness of all particles in the population, and update the global optimal solution with the particle corresponding to the maximum joint fitness. Step S4-2: Update the state of each particle using particle swarm optimization constraints based on the individual optimal solution and the global optimal solution.

9. The energy-path collaborative management method for solar-powered UAVs based on the hierarchical PSO algorithm according to claim 8, characterized in that, Step S4-2 specifically includes updating the state of each particle using the following expression: in, These represent the velocities of particle k at the (j+1)th and jth iterations, respectively. P represents the position of particle k at the (j+1)th and jth iterations, respectively. best,k G represents the individual optimal solution for particle k. best Let χ represent the global optimal solution, w be the inertia weight, c1 be the cognitive learning factor, c2 be the social learning factor, and r1 and r2 be random perturbation terms. The state of each particle is constrained by the following expression: like but ||WP i -WP i-1 ||≥d min Among them, X min ,X max Let be the lower and upper bounds of the search space for the particle's position, respectively, where proj(·) represents the projection function, and d min This indicates the minimum distance between waypoints.

10. The energy-path collaborative management method for solar-powered UAVs based on the hierarchical PSO algorithm according to claim 9, characterized in that, In step S5, the expression for the termination condition is: in, This represents the globally optimal solution in the j-th iteration. Let ε represent the global optimal solution in the (j-Δj)th iteration. pos For positional tolerance, ε represents the average joint fitness of each particle in the j-th and (j-1)-th iterations, respectively. fit For fitness tolerance, J max This represents the maximum number of iterations.

Images