Trajectory optimization method for electric unmanned aerial vehicle based on fuzzy neural network sequence convex optimization
By optimizing the flight trajectory of a hybrid electric UAV based on a fuzzy neural network sequence convex optimization method, the problem of high energy consumption during long-endurance flight of the hybrid electric UAV is solved, and the output power of solar cells is efficiently utilized, thereby improving flight time.
Patent Information
- Application Number
- CN202211524903.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-11-30
- Publication Date
- 2026-02-03
- Estimated Expiration
- 2042-11-30
AI Technical Summary
Existing trajectory optimization methods are difficult to apply effectively to long-endurance flights of hybrid electric UAVs, and cannot maximize the use of solar cell output power, resulting in high energy consumption and insufficient flight time.
A fuzzy neural network-based sequential convex optimization method is adopted. By introducing the residual energy equation into the UAV mass dynamics equation, the state equation of the hybrid energy system is constructed, and convexification is performed within the trust region. The fuzzy neural network is used to adaptively adjust the size of the trust region to optimize the flight trajectory and reduce energy consumption.
It significantly improves the energy efficiency of hybrid electric drones, increases flight time, maximizes the output power of solar cells, and reduces energy consumption during flight.
Smart Images
Figure CN116126011B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a trajectory optimization method for electric unmanned aerial vehicles (UAVs) based on fuzzy neural network sequence convex optimization, and belongs to the field of UAVs. Background Technology
[0002] Drones have been widely used in various fields such as environmental monitoring, disaster relief, pesticide spraying, communication relay, and express delivery. Electric drones, in particular, combine the advantages of being clean and pollution-free, low-noise, and highly reliable, leading to significant advancements in electric drone technology in recent years. Among these, hybrid electric drones powered by solar cells / fuel cells / lithium batteries offer an effective solution for achieving long-endurance flight at low and medium altitudes. However, the output power of the solar cells on the wing surface is affected by changes in the drone's attitude, resulting in a complex coupling relationship between the hybrid energy system state and flight motion. Therefore, optimizing the flight trajectory of hybrid electric drones can maximize the output power of the solar cells, effectively reduce energy consumption during flight, and increase the drone's flight time.
[0003] Trajectory optimization for unmanned aerial vehicles (UAVs) is a typical nonlinear optimal control problem, characterized by high constraint dimensionality and strong nonlinearity. It typically requires parameterization, transforming it into a nonlinear optimization problem, and then employing optimization methods such as pseudospectral methods, convex optimization, particle swarm optimization, and genetic algorithms to solve it and obtain the UAV's flight trajectory. Existing trajectory optimization methods mainly focus on how UAVs can quickly reach their target positions with relatively low control input. However, for hybrid electric UAVs, the hybrid energy system state is coupled with the UAV's flight trajectory, and the flight trajectory directly determines the power demand profile of the hybrid energy system. Existing trajectory optimization methods are difficult to directly apply to the long-endurance flight trajectory optimization of hybrid electric UAVs. Therefore, it is essential to plan a low-energy-consumption flight trajectory for long-endurance hybrid electric UAVs at the flight trajectory level. This paper designs a hybrid electric UAV trajectory optimization method based on fuzzy neural network sequence convex optimization to reduce the energy consumption of hybrid electric UAVs and improve flight time at the flight trajectory level. Summary of the Invention
[0004] To address the challenge of high-energy-efficiency, long-endurance autonomous flight of hybrid electric unmanned aerial vehicles (UAVs), this invention optimizes the power demand profile of the UAV at the flight trajectory level. The main objective is to provide a trajectory optimization method for electric UAVs based on fuzzy neural network sequential convex optimization. This method first introduces a residual energy equation into the UAV's particle dynamics equations to construct the state equations of the hybrid energy system. Within the trust region, the state equations and obstacle avoidance constraints are made convex, constructing a convex optimization model for the hybrid electric UAV's flight trajectory problem. For the trust region size adjustment problem, a fuzzy neural network is designed to achieve adaptive adjustment of the trust region, improving the optimality of the sequential convex optimization method and accelerating its convergence speed. Through iterative solution, a low-energy-consumption flight trajectory for the hybrid electric UAV is obtained. Flying along the optimized trajectory maximizes the utilization of solar cells, reduces the energy consumption of the hybrid energy system, and increases the UAV's flight time.
[0005] The objective of this invention is achieved through the following technical solution:
[0006] The trajectory optimization method for electric unmanned aerial vehicles (UAVs) based on fuzzy neural network sequence convex optimization disclosed in this invention includes the following steps:
[0007] Step 1: Set initial condition s0 and terminal condition s f Boundary conditions s min ,s max ,u min ,u max Obstacle Information Λ m Initialize the initial trust region ε and convergence tolerance ξ. Use a straight-line flight trajectory from the starting point to the ending point as the reference trajectory. Initialize k = 0, 1, ..., K;
[0008] The hybrid electric drone is a hybrid electric drone powered by solar cells, fuel cells, and lithium batteries.
[0009] Step 2: Based on the drone's flight status and remaining energy demand E sd Let T be the thrust T generated by the propeller and n be the vertical overload. v Horizontal overload n h To control the variables, a state equation is constructed for the UAV trajectory optimization problem; the reference trajectory obtained in step one is used for initialization. For the UAV trajectory optimization problem, the state equations are made convex, i.e., within the reference trajectory, k = 0, 1, ..., K. A first-order Taylor polynomial expansion within the trust region k = 0, 1, ..., K yields the state equation for the convexized UAV trajectory optimization problem. Simultaneously, the initial reference trajectory is used... The obstacle avoidance constraints of the UAV trajectory optimization problem are made convex by k = 0, 1, ..., K; furthermore, the remaining energy demand to meet the convex optimization conditions is used as the optimization objective function to construct a convex optimization model of the hybrid electric UAV flight trajectory, which is conducive to solving the problem using convex optimization methods; in addition, since the solar cell has the highest output priority, using the remaining energy demand as the optimization objective can maximize the utilization of solar energy, reduce the fuel consumption of the hybrid energy system, and improve the flight time of the UAV.
[0010] The flight status of the UAV includes its spatial position, flight speed, track tilt angle, and heading angle; the remaining energy requirement is the energy required for the UAV to fly minus the energy generated by the solar cells on the wing surface.
[0011] The state variables s of the UAV trajectory optimization problem include the UAV's three-dimensional position (x, y, h), altitude h, flight speed V, heading angle χ, trajectory tilt angle γ, and remaining energy requirement E. sd The control variable u in the UAV trajectory optimization problem includes the propeller thrust T and the vertical overload n. v Horizontal overload n h The state equation for the UAV trajectory optimization problem is:
[0012]
[0013] In equation (1), m is the mass of the UAV; g is the gravitational acceleration; ρ is the air density; S is the reference wing area of the UAV; C D0 k is the parasitic drag coefficient. e η is the aerodynamic coefficient. p For the efficiency of the UAV power system; P pv The output power of the solar cells on the wing surface;
[0014] The convex optimization model for the hybrid electric unmanned aerial vehicle (UAV) flight trajectory problem is as follows:
[0015]
[0016] In equation (2), K is the number of discrete flight trajectories; Δt is the time interval between discrete flight trajectories, i.e., Δt = t f / K,t f E represents the time it takes for the drone to reach the target location. sd [K] represents the remaining energy required at time K; s[0] represents the state variables of the UAV trajectory optimization problem at the initial time; s[K] represents the state variables of the UAV trajectory optimization problem at time K; s[k] represents the state variables of the UAV trajectory optimization problem at time k; u[k] represents the control variables of the UAV trajectory optimization problem at time k. The reference trajectory is ε, and the trust region size is A. k+1A k B k+1 B k C k+1 C k D k+1 D k Λ are constant matrices related to the reference trajectory, which can be obtained by performing a first-order Taylor polynomial expansion of the state equation near the reference trajectory; H is the UAV horizontal position extraction matrix; Λ m Let r be the position of the m-th obstacle; m Let M be the radius of the m-th obstacle; obs The number of obstacles;
[0017] Step 3: Solve the convex optimization model of the hybrid electric UAV flight trajectory constructed in Step 2 using the convex optimization method to obtain the hybrid electric UAV flight trajectory, and update the reference trajectory described in Step 1 using the optimized hybrid electric UAV flight trajectory. k = 0, 1, ..., K;
[0018] Step 4: Using constraint violation ΔC and objective function increment ΔJ as the outputs of the fuzzy neural network, and the size of the trust region ε as the output of the fuzzy neural network, a fuzzy rule base is constructed, building a fuzzy neural network mainly composed of a fuzzy inference engine and a neural network. By adaptively adjusting the trust region through the fuzzy neural network, the optimality of the hybrid electric UAV flight trajectory obtained by the sequential convex optimization method is improved, and the convergence speed of the sequential convex optimization method is accelerated.
[0019] Trust region constraints reduce the approximate convexity error of nonlinear state equations and obstacle avoidance constraints by limiting the deviation range between the reference trajectory and the optimal trajectory in each iteration. Therefore, the smaller the trust region, the higher the accuracy of approximate convexity. However, a small trust region can also lead to an excessively small feasible region during iteration, resulting in the failure to find a feasible solution and thus the solution failure. A fixed trust region size is difficult to meet the requirements of the trajectory optimization problem of hybrid electric UAVs. Using constraint violation ΔC and objective function increment ΔJ as the output of the fuzzy neural network, and the trust region size as the output of the fuzzy neural network, a fuzzy rule base is constructed, and a fuzzy inference engine and neural network are built to adaptively adjust the trust region. A fuzzy logic neural network based on constraint violation and objective function increment is used to adaptively adjust the trust region size and update the trust region.
[0020] The input and output variables of the fuzzy neural network are divided into five states: very high (VH), high (H), medium (M), low (L), and very low (VL).
[0021] The fuzzy inference method in the fuzzy inference engine adopts the Mamdani method, and the defuzzification method in the fuzzy inference engine adopts the centroid method.
[0022] The constraint violation degree ΔC and objective function increment ΔJ are calculated as follows:
[0023]
[0024]
[0025] In equation (4), the superscript q represents the number of iterations.
[0026] Step 5: If the optimized flight trajectory satisfies the preset iteration convergence criterion during the iteration process, the iteration process terminates, and the currently optimized hybrid electric UAV flight trajectory is taken as the optimal flight trajectory, then proceed to Step 6. Otherwise, return to Step 2 and continue iterating;
[0027] The trajectory optimization iterative convergence criterion for the hybrid electric unmanned aerial vehicle (UAV) is described.
[0028] s q [k]-s q-1 [k]|≤ξ,k=0,1,...,K (5)
[0029] Step Six: Output the flight trajectory obtained in Step Five, which includes state variables, control variables, and discretized time steps. The hybrid electric UAV flies according to the optimized trajectory, which can maximize the use of solar energy, reduce the energy consumption of the hybrid electric UAV, and increase the flight time of the UAV.
[0030] Beneficial effects:
[0031] 1. This invention discloses a trajectory optimization method for electric unmanned aerial vehicles (UAVs) based on fuzzy neural network sequence convex optimization. While optimizing the flight trajectory, it considers the characteristics of the hybrid energy system, optimizes the power demand profile of the hybrid electric UAV at the flight trajectory level, maximizes the output power of the solar cells, reduces energy consumption during flight, improves the energy efficiency of the hybrid electric UAV, and increases the flight time of the UAV.
[0032] 2. This invention discloses a trajectory optimization method for electric unmanned aerial vehicles (UAVs) based on fuzzy neural network sequence convex optimization. It introduces a residual energy equation on the basis of the UAV mass dynamics equation, and makes the state equation and obstacle avoidance constraints convex within the trust region of the reference trajectory to construct a convex optimization model for the flight trajectory of a hybrid electric UAV. This transforms the non-convex optimization problem into a convex optimization problem and solves it using convex optimization methods, which significantly improves the optimization efficiency of the trajectory optimization problem of hybrid electric UAVs.
[0033] 3. This invention discloses a trajectory optimization method for electric unmanned aerial vehicles based on fuzzy neural network sequential convex optimization. According to the constraint violation and objective function increment, a fuzzy neural network is designed to adaptively adjust the trust region size, thereby enhancing the optimality of the sequential convex optimization method and accelerating its convergence speed. Attached Figure Description
[0034] Figure 1 The flowchart shows the trajectory optimization method for electric unmanned aerial vehicles based on fuzzy neural network sequence convex optimization.
[0035] Figure 2 To optimize the obtained flight trajectory; Figure 2 (a) 3D view of the UAV flight trajectory Figure 2 (b) is a top view of the UAV's flight path;
[0036] Figure 3 This is a comparison chart of remaining energy demand. Detailed Implementation
[0037] To better illustrate the purpose and advantages of this invention, the following description, in conjunction with the accompanying drawings and simulation examples, further explains the content of this invention.
[0038] Example 1:
[0039] The initial state of the UAV is s0 = (0, 0, 300m, 17m / s, 0.014, π / 4), and the target state of the UAV is s f = (10km, 10km, 500m, 17m / s, 0.014, π / 4), ε = [300, 300, 10, 5, π / 18, π / 18, 200], ξ = [0.5, 0.5, 0.5, 0.01, 0.05, 0.05, 1], and K = 40; the wing surface of the Dandelion I UAV has 120 solar panels, and the radiation intensity on the day of flight is 1200W / m 2 The drone took off at 9:00 AM that day. The simulation environment was a desktop computer running MATLAB 2019b, configured with Windows 10, an Intel(R) Core(TM) CPU i7-7500 2.93GHz, and 16GB of RAM.
[0040] Table 1Main parameters of the UAV
[0041]
[0042] To verify the feasibility and beneficial effects of the hybrid electric UAV trajectory optimization method based on fuzzy neural network sequence convex optimization disclosed in this invention, the technical solution of this invention is clearly and thoroughly described below in a case study, as shown in the flowchart. Figure 1 As shown.
[0043] The trajectory optimization method for electric unmanned aerial vehicles (UAVs) based on fuzzy neural network sequence convex optimization disclosed in this example has the following specific implementation steps:
[0044] Step 1: Set initial condition s0 and terminal condition s f Boundary conditions s min ,s max ,u min ,u max Obstacle Information Λ m Initialize the initial trust region ε and convergence tolerance ξ. Use a straight-line flight trajectory from the starting point to the ending point as the reference trajectory. Initialize k = 0, 1, ..., K;
[0045] The hybrid electric drone is a hybrid electric drone powered by solar cells, fuel cells, and lithium batteries.
[0046] Step 2: Based on the drone's flight status and remaining energy demand E sd Let T be the thrust T generated by the propeller and n be the vertical overload. v Horizontal overload n h To control the variables, a state equation is constructed for the UAV trajectory optimization problem; the reference trajectory obtained in step one is used for initialization. For the UAV trajectory optimization problem, the state equations are made convex, i.e., within the reference trajectory, k = 0, 1, ..., K. A first-order Taylor polynomial expansion within the trust region k = 0, 1, ..., K yields the state equation for the convexized UAV trajectory optimization problem. Simultaneously, the initial reference trajectory is used... The obstacle avoidance constraints of the UAV trajectory optimization problem are made convex by k = 0, 1, ..., K; furthermore, the remaining energy demand to meet the convex optimization conditions is used as the optimization objective function to construct a convex optimization model of the hybrid electric UAV flight trajectory, which is conducive to solving the problem using convex optimization methods; in addition, since the solar cell has the highest output priority, using the remaining energy demand as the optimization objective can maximize the utilization of solar energy, reduce the fuel consumption of the hybrid energy system, and improve the flight time of the UAV.
[0047] The flight status of the UAV includes its spatial position, flight speed, track tilt angle, and heading angle; the remaining energy requirement is the energy required for the UAV to fly minus the energy generated by the solar cells on the wing surface.
[0048] The state variables s of the UAV trajectory optimization problem include the UAV's three-dimensional position (x, y, h), altitude h, flight speed V, heading angle χ, trajectory tilt angle γ, and remaining energy requirement E. sdThe control variable u in the UAV trajectory optimization problem includes the propeller thrust T and the vertical overload n. v Horizontal overload n h The state equation for the UAV trajectory optimization problem is:
[0049]
[0050] In equation (6), m is the mass of the UAV; g is the acceleration due to gravity; ρ is the air density; S is the reference wing area of the UAV; C D0 k is the parasitic drag coefficient. e η is the aerodynamic coefficient. p For the efficiency of the UAV power system; P pv The output power of the solar cells on the wing surface;
[0051] The convex optimization model for the hybrid electric unmanned aerial vehicle (UAV) flight trajectory problem is as follows:
[0052]
[0053] In equation (7), K is the number of discrete flight trajectories; Δt is the time interval between discrete flight trajectories, i.e., Δt = t f / K,t f E represents the time it takes for the drone to reach the target location. sd [K] represents the remaining energy required at time K; s[0] represents the state variables of the UAV trajectory optimization problem at the initial time; s[K] represents the state variables of the UAV trajectory optimization problem at time K; s[k] represents the state variables of the UAV trajectory optimization problem at time k; u[k] represents the control variables of the UAV trajectory optimization problem at time k. The reference trajectory is ε, and the trust region size is A. k+1 A k B k+1 B k C k+1 C k D k+1 D k Λ are constant matrices related to the reference trajectory, which can be obtained by performing a first-order Taylor polynomial expansion of the state equation near the reference trajectory; H is the UAV horizontal position extraction matrix; Λ m Let r be the position of the m-th obstacle; m Let M be the radius of the m-th obstacle; obs The number of obstacles;
[0054] Step 3: Solve the convex optimization model of the hybrid electric UAV flight trajectory constructed in Step 2 using the convex optimization method to obtain the hybrid electric UAV flight trajectory, and update the reference trajectory described in Step 1 using the optimized hybrid electric UAV flight trajectory. k = 0, 1, ..., K;
[0055] Step 4: Using constraint violation ΔC and objective function increment ΔJ as the outputs of the fuzzy neural network, and the size of the trust region ε as the output of the fuzzy neural network, a fuzzy rule base is constructed, building a fuzzy neural network mainly composed of a fuzzy inference engine and a neural network. By adaptively adjusting the trust region through the fuzzy neural network, the optimality of the hybrid electric UAV flight trajectory obtained by the sequential convex optimization method is improved, and the convergence speed of the sequential convex optimization method is accelerated.
[0056] Trust region constraints reduce the approximate convexity error of nonlinear state equations and obstacle avoidance constraints by limiting the deviation range between the reference trajectory and the optimal trajectory in each iteration. Therefore, the smaller the trust region, the higher the accuracy of approximate convexity. However, a small trust region can also lead to an excessively small feasible region during iteration, resulting in the failure to find a feasible solution and thus the solution failure. A fixed trust region size is difficult to meet the requirements of the trajectory optimization problem of hybrid electric UAVs. Using constraint violation ΔC and objective function increment ΔJ as the output of the fuzzy neural network, and the trust region size as the output of the fuzzy neural network, a fuzzy rule base is constructed, and a fuzzy inference engine and neural network are built to adaptively adjust the trust region. A fuzzy logic neural network based on constraint violation and objective function increment is used to adaptively adjust the trust region size and update the trust region.
[0057] The input and output variables of the fuzzy neural network are divided into five states: very high (VH), high (H), medium (M), low (L), and very low (VL).
[0058] The fuzzy inference method in the fuzzy inference engine adopts the Mamdani method, and the defuzzification method in the fuzzy inference engine adopts the centroid method.
[0059] The constraint violation degree ΔC and objective function increment ΔJ are calculated as follows:
[0060]
[0061]
[0062] In equation (9), the superscript q indicates the number of iterations;
[0063] Step 5: If the optimized flight trajectory satisfies the preset iteration convergence criterion during the iteration process, the iteration process terminates, and the currently optimized hybrid electric UAV flight trajectory is taken as the optimal flight trajectory, then proceed to Step 6. Otherwise, return to Step 2 and continue iterating;
[0064] The trajectory optimization iterative convergence criterion for the hybrid electric unmanned aerial vehicle (UAV) is described.
[0065] |sq [k]-s q-1 [k]|≤ξ,k=0,1,...,K (10)
[0066] Step Six: Output the flight trajectory obtained in Step Five, which includes state variables, control variables, and discretized time steps. The hybrid electric UAV flies according to the optimized trajectory, which can maximize the use of solar energy, reduce the energy consumption of the hybrid electric UAV, and increase the flight time of the UAV.
[0067] The results obtained based on the electric unmanned aerial vehicle trajectory optimization method based on fuzzy neural network sequence convex optimization disclosed in this invention are as follows: Figure 2 and Figure 3 As shown. Figure 2 The method of the present invention optimizes the flight trajectory of the hybrid electric drone. The results show that the method of the present invention can optimize the smooth flight trajectory of the drone from the starting point to the end point and maintain a safe distance from all obstacles. Figure 3 This image compares the remaining energy requirements obtained by the proposed method (FNNSCP) and sequential convex optimization (SCP). The SCP method does not consider the energy characteristics of the hybrid electric UAV. The proposed method yields a remaining energy requirement of 368.03 KJ and a flight time of 948.1 s; while the SCP method yields a remaining energy requirement of 769 KJ and a flight time of 716.4 s. Simulation results show that, compared to the SCP method, the proposed method can save 52.16% of energy consumption. By optimizing the flight trajectory, it reduces the UAV's energy consumption at the flight trajectory level, thereby increasing the flight time of the hybrid electric UAV.
[0068] The above detailed description is a further explanation of the purpose, technical solution and beneficial effects of the invention. It should be understood that the above description is only a specific implementation example of the present invention and is only used to explain the present invention. It is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A trajectory optimization method for electric unmanned aerial vehicles based on fuzzy neural network sequence convex optimization, characterized in that: Includes the following steps, Step 1: Set initial condition s0 and terminal condition s f Boundary conditions s min ,s max ,u min ,u max Obstacle Information Λ m Initialize the initial trust region ε and convergence tolerance ξ; use a straight flight trajectory from the starting point to the ending point as the reference trajectory. Initialization is performed, where The reference trajectory is K, which represents the number of discrete flight trajectories. The electric drone is a hybrid electric drone powered by solar cells, fuel cells, and lithium batteries. Step 2: Based on the drone's flight status and remaining energy demand E sd Let T be the thrust T generated by the propeller and n be the vertical overload. v Horizontal overload n h To control the variables, a state equation is constructed for the UAV trajectory optimization problem; the reference trajectory obtained in step one is used for initialization. The state equation of the UAV trajectory optimization problem is made convex, that is, in the reference trajectory A first-order Taylor polynomial expansion is performed within the trust region to obtain the state equation of the convex UAV trajectory optimization problem. Simultaneously, the reference trajectory obtained during initialization is used... The obstacle avoidance constraints of the UAV trajectory optimization problem are made convex; furthermore, the remaining energy demand to meet the convex optimization conditions is used as the optimization objective function to construct a convex optimization model for the flight trajectory of the hybrid electric UAV, which is conducive to solving the problem using convex optimization methods; in addition, since solar cells have the highest output priority, using the remaining energy demand as the optimization objective can maximize the utilization of solar energy, reduce the fuel consumption of the hybrid energy system, and improve the flight time of the UAV. The flight status of the UAV includes its spatial position, flight speed, track tilt angle, and heading angle; the remaining energy requirement is the energy required for the UAV to fly minus the energy generated by the solar cells on the wing surface. Step 3: Solve the convex optimization model of the hybrid electric UAV flight trajectory constructed in Step 2 using the convex optimization method to obtain the hybrid electric UAV flight trajectory, and update the reference trajectory described in Step 1 using the optimized hybrid electric UAV flight trajectory. Step 4: Using the constraint violation degree ΔC and the objective function increment ΔJ as inputs to the fuzzy neural network, and the trust region ε as the output of the fuzzy neural network, a fuzzy rule base is constructed, and a fuzzy neural network mainly composed of a fuzzy inference engine and a neural network is built. By adaptively adjusting the trust region through the fuzzy neural network, the optimality of the hybrid electric UAV flight trajectory obtained by the sequential convex optimization method is improved, and the convergence speed of the sequential convex optimization method is accelerated. Step 5: If the optimized flight trajectory satisfies the preset iteration convergence criterion during the iteration process, the iteration process terminates, the currently optimized hybrid electric UAV flight trajectory is taken as the optimal flight trajectory, and the process proceeds to Step 6; otherwise, return to Step 2 and continue iterating. Step six: Output the flight trajectory obtained in step five, which includes state variables, control variables, and discretized time steps. The hybrid electric UAV flies according to the optimized trajectory, which can maximize the use of solar energy, reduce the energy consumption of the hybrid electric UAV, and increase the flight time of the UAV.
2. The trajectory optimization method for electric unmanned aerial vehicles based on fuzzy neural network sequence convex optimization as described in claim 1, characterized in that: In step two, The state variables s of the UAV trajectory optimization problem include the UAV's three-dimensional position (x, y, h), altitude h, flight speed V, heading angle χ, trajectory tilt angle γ, and remaining energy requirement E. sd The control variable u in the UAV trajectory optimization problem includes the propeller thrust T and the vertical overload n. v Horizontal overload n h The state equation for the UAV trajectory optimization problem is: In equation (1), m is the mass of the UAV; g is the gravitational acceleration; ρ is the air density; S is the reference wing area of the UAV; C D0 k is the parasitic drag coefficient. e η is the aerodynamic coefficient. p For the power efficiency of drones; P pv The output power of the solar cells on the wing surface; The convex optimization model for the hybrid electric unmanned aerial vehicle (UAV) flight trajectory problem is as follows: In equation (2), K is the number of discrete flight trajectories; Δt is the time interval between discrete flight trajectories, i.e., Δt = t f / K,t f E represents the time it takes for the drone to reach the target location. sd [K] represents the remaining energy required at time K; s[0] represents the state variables of the UAV trajectory optimization problem at the initial time; s[K] represents the state variables of the UAV trajectory optimization problem at time K; s[k] represents the state variables of the UAV trajectory optimization problem at time k; u[k] represents the control variables of the UAV trajectory optimization problem at time k. The reference trajectory is ε, and the trust region size is A. k+1 A k B k+1 B k C k+1 C k D k+1 D k Λ are constant matrices related to the reference trajectory, which can be obtained by performing a first-order Taylor polynomial expansion of the state equation near the reference trajectory; H is the UAV horizontal position extraction matrix; Λ m Let r be the position of the m-th obstacle; m Let M be the radius of the m-th obstacle; obs The number of obstacles.
3. The trajectory optimization method for electric unmanned aerial vehicles based on fuzzy neural network sequence convex optimization as described in claim 2, characterized in that: Trust region constraints reduce the approximate convexity error of nonlinear state equations and obstacle avoidance constraints by limiting the deviation range between the reference trajectory and the optimal trajectory in each iteration. A fuzzy rule base is constructed using the constraint violation degree ΔC and the objective function increment ΔJ as the outputs of the fuzzy neural network, and the trust region size ε is used as the output of the fuzzy neural network. A fuzzy neural network mainly composed of a fuzzy inference engine and a neural network is built. The fuzzy neural network adaptively adjusts the trust region, improving the optimality of the hybrid electric UAV flight trajectory obtained by the sequential convex optimization method and accelerating the convergence speed of the sequential convex optimization method. A fuzzy logic neural network based on constraint violation and objective function increment is used to adaptively adjust the trust region size and update the trust region. The input and output variables of the fuzzy neural network are divided into five states: very high (VH), high (H), medium (M), low (L), and very low (VL). The fuzzy inference method in the fuzzy inference engine adopts the Mamdani method, and the defuzzification method in the fuzzy inference engine adopts the centroid method. The constraint violation degree ΔC and objective function increment ΔJ are calculated as follows: In equation (4), the superscript q represents the number of iterations.
4. The trajectory optimization method for electric unmanned aerial vehicles based on fuzzy neural network sequence convex optimization as described in claim 1, characterized in that: In step five, The convergence criterion for trajectory optimization of the hybrid electric unmanned aerial vehicle is |s q [k]-s q-1 [k]|≤ξ,k=0,1,...,K (5) In equation (5), the superscript q indicates the number of iterations.