Unmanned aerial vehicle path planning method based on hybrid improvement strategy

Abstract

The invention belongs to the technical field of unmanned aerial vehicle path planning, and particularly relates to an unmanned aerial vehicle path planning method based on a hybrid improvement strategy, which comprises the following steps of: 1, establishing a three-dimensional model of an unmanned aerial vehicle flight space; 2, determining a flight cost function of the unmanned aerial vehicle according to path length, collision, flight height and path smoothness; step 3, initializing parameters; step 4, using a Harlisia eagle algorithm HHO and an Aquila optimizer in a heuristic algorithm to finally provide a brand-new hybrid improved strategy algorithm named as an EAOHHO algorithm, and then using the EAOHHO algorithm to optimize the flight cost function; and 5, taking a path obtained by searching according to an EAOHHO algorithm as a trajectory path of the unmanned aerial vehicle. According to the method, the fast search capability of the HHO and the global exploration advantage of the AO are deeply fused, the fitness-distance balance FDB mechanism is innovatively introduced, and the convergence efficiency and robustness of the algorithm are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of UAV path planning technology, specifically relating to a UAV path planning method based on a hybrid improvement strategy. Background Technology

[0002] Unmanned aerial vehicles (UAVs) have been widely used in civilian and other fields due to their significant cost-effectiveness and excellent flexibility, covering multiple scenarios such as traffic monitoring, maritime rescue, disaster prevention and mitigation, search and rescue, forest fire prevention, and auxiliary communication. As the complexity of application scenarios continues to increase, the demand for intelligent UAVs is becoming increasingly urgent. However, most current UAVs rely on battery power, which has a limited energy capacity, greatly restricting mission execution time. Against this backdrop, exploring low-energy, high-coverage, and safe collision-free path planning solutions has become a core problem that urgently needs to be solved in the field of UAV path planning—namely, the UAV 3D path planning problem. This problem belongs to the NP-hard category, and its core objective is to automatically find the optimal or near-optimal path from the starting point to the target point.

[0003] Currently, UAV path planning algorithms are mainly divided into two categories. One category is traditional path planning algorithms, including A algorithm, D algorithm, and artificial potential field method; the other category is path planning algorithms based on swarm intelligence optimization, with typical examples being particle swarm optimization, gray wolf optimization algorithm, and whale optimization algorithm. Among them, the Harris Eagle Optimization Algorithm (HHO) and Sky Eagle Optimization Algorithm (AO), as novel algorithms proposed in the field of swarm intelligence optimization in the last five years, have been widely used in path planning. However, it is undeniable that both algorithms have certain limitations, such as slow convergence speed and a tendency to get trapped in local optima, affecting the efficiency and quality of path planning.

[0004] In the prior art, for example, Chinese patent publication number CN112230678B discloses a 3D UAV path planning method and system based on particle swarm optimization algorithm, but it has the following drawbacks:

[0005] It is prone to getting trapped in local optima: Particle swarm optimization may converge to a local optimum prematurely during the search process, rather than the global optimum.

[0006] The ability to handle complex constraints is limited. In 3D path planning problems, there are often multiple complex constraints, such as flight altitude limits and obstacle avoidance rules. Particle swarm optimization (PSO) may exhibit certain limitations when dealing with these complex constraints, and it is difficult to guarantee that the generated path fully satisfies all constraints.

[0007] Therefore, this invention provides a UAV path planning method based on a hybrid improvement strategy. Summary of the Invention

[0008] The purpose of this invention is to provide a UAV path planning method based on a hybrid improvement strategy, which addresses the problems of slow convergence speed and easy getting trapped in local optima in the Harris Eagle Optimization Algorithm (HHO) and Sky Eagle Optimization Algorithm (AO) in UAV path planning. This invention deeply integrates the fast search capability of HHO and the global exploration advantage of AO, and innovatively introduces the fitness-distance balance (FDB) mechanism, which significantly improves the convergence efficiency and robustness of the algorithm.

[0009] The specific technical solution adopted by this invention is as follows:

[0010] A method for UAV path planning based on a hybrid improvement strategy includes the following steps:

[0011] Step 1: A three-dimensional model of the UAV's flight space was established;

[0012] Step 2: Determine the flight cost function of the UAV based on path length, collision occurrence, flight altitude, and path smoothness;

[0013] Step 3: Initialize parameters, including population size, maximum number of iterations, dimension of the objective function, upper bound of the initial value, and lower bound of ub (lb).

[0014] Step 4: Apply the UAV trajectory planning method based on the hybrid improvement strategy to the path planning problem of UAVs. Specifically, the Harris Hawk algorithm (HHO) and Aquila optimizer (AO) in the heuristic algorithm are used to improve the two algorithms in a hybrid manner. The two improvement strategies are used to finally propose a brand-new hybrid improvement strategy algorithm, named EAOHHO algorithm (Ensemble of Aquila Optimizer and Harris Hawks Optimization, EAOHHO). Then, the EAOHHO algorithm is used to optimize the flight cost function.

[0015] Step 5: Use the path obtained by the EAOHHO algorithm as the trajectory path of the UAV.

[0016] The technical effects achieved by this invention are as follows:

[0017] Traditional particle swarm optimization (PSO), Harris Eagle algorithm (HHO), and Eagle Algorithm (AO) converge slowly during the iterative process, requiring significant computational resources and time to find a satisfactory solution. This invention deeply integrates the advantages of both HHO and AO algorithms and introduces a fitness-distance balance (FDB) selection mechanism. During the exploration phase, it more intelligently selects the direction of the individual that contributes most to the search, greatly accelerating the convergence process and reducing the required number of iterations and computation time.

[0018] This invention constructs a composite cost function that comprehensively considers path length, collision detection, flight altitude variance, and path smoothness (turning angle and climb angle). The EAOHHO algorithm optimizes this function and can automatically satisfy these complex constraints, ultimately generating a high-quality trajectory that is not only short in length but also stable, safe, reliable, and energy efficient. Attached Figure Description

[0019] Figure 1 This is a schematic diagram illustrating the calculation of the turning and climb angle of this invention;

[0020] Figure 2 This is a flowchart of the EAOHHO algorithm of the present invention. Detailed Implementation

[0021] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0022] like Figure 1 As shown, a UAV path planning method based on a hybrid improvement strategy is presented, which is particularly suitable for solving the problems of slow convergence speed and easy getting trapped in local optima in existing intelligent optimization algorithms in UAV path planning. The method includes the following steps.

[0023] Step 1: A three-dimensional model of the UAV's flight space was established;

[0024] In step one, the UAV flight space considered in this invention is mountainous terrain; therefore, the mathematical expression for the mountain model in environmental modeling is:

[0025]

[0026] Where (x,y) are the coordinates of the mountain peak on the horizontal plane; (x0,y0) are the coordinates of the center point of the mountain peak on the horizontal plane; m1 and m2 represent the height; m1 and m2 reflect the steepness of the peak.

[0027] Step 2: Determine the flight cost function of the UAV based on path length, collision occurrence, flight altitude, and path smoothness;

[0028] Step 3: Initialize parameters, including population size, maximum number of iterations, dimension of the objective function, upper bound of the initial value, and lower bound of ub (lb).

[0029] In step three, the maximum path length is a key constraint in the practical application of UAV 3D path planning, directly determining the reachability of a single UAV mission. If the UAV collides with mountainous terrain, the path length is set to infinity. The path length expression is:

[0030]

[0031] l i =||(x i+1 ,y i+1 ,z i+1 )-(x i ,y i ,z i )|| 2

[0032] Among them l i This represents the distance between the i-th waypoint and the (i+1)-th waypoint. Furthermore, (x... i ,y i ,z i ) and (x i+1 ,y i+1 ,z i+1 ) are the coordinates of the i-th waypoint and the (i+1)-th waypoint, respectively.

[0033] The flight altitude of a drone should be maintained within a certain range. The specific formula for the altitude cost of a drone is as follows:

[0034]

[0035] In UAV 3D path planning, path smoothness cost is a key indicator for measuring flight path quality. Its core lies in the quantitative evaluation of parameters such as path turning angle and climb rate, which directly affect the feasibility and efficiency of the flight path. For example... Figure 1 As shown, the turning angle φ ij Defined as two consecutive path segments, i.e., P' ij P' i,j+1 With P' i,j+1 P' i,j+2 The angle between the projections onto the horizontal plane Oxy. Let k be a unit vector pointing towards the z-axis, then the projection vector can be calculated as:

[0036] P' ij P' i,j+1 =k×(P ij P i,j+1 ×k)

[0037] Therefore, the turning angle φ ij It can be calculated using the following formula:

[0038]

[0039] Ascent angle ψ ij Represents path segment P ij P i,j+1 Its projection P' on the horizontal plane ijP' i,j+1 The angle between them is calculated using the following formula:

[0040]

[0041] The path smoothness cost consists of two parts: a penalty for turning angles and a penalty for the rate of change of continuous climb angles, expressed as follows:

[0042]

[0043] In summary, the flight cost function of the UAV is defined as:

[0044] F(x) = L + H + S;

[0045] By optimizing the flight cost function, the flight path planning of the UAV is completed.

[0046] The goal of UAV flight path planning is to optimize the flight cost function to its minimum to meet scenario requirements and achieve reasonable flight path planning for the UAV. According to the definition of the flight cost function, the terms within it are not linearly related; it is a nonlinear, non-convex combinatorial optimization problem. Since the objective function involves multiple discrete decision variables, it means that multiple local optima may be encountered during the optimization process. Based on these characteristics of the objective function, traditional optimization methods are usually ineffective in solving this problem for the following reasons: In non-convex optimization problems, traditional optimization methods are prone to getting trapped in local optima. This is because these methods rely on gradients or other local information and cannot effectively escape local optima.

[0047] Step 4: Apply the UAV trajectory planning method based on the hybrid improvement strategy to the path planning problem of UAVs. Specifically, the Harris Hawk algorithm (HHO) and Aquila optimizer (AO) in the heuristic algorithm are used to improve the two algorithms in a hybrid manner. The two improvement strategies are used to finally propose a brand-new hybrid improvement strategy algorithm, named EAOHHO algorithm (Ensemble of Aquila Optimizer and Harris Hawks Optimization, EAOHHO). Then, the EAOHHO algorithm is used to optimize the flight cost function.

[0048] like Figure 2 As shown, the EAOHHO algorithm in step four includes five stages:

[0049] Phase 1: Initialization Phase. Individuals are initialized using the Circle chaotic mapping, and individuals with higher fitness are retained to form the final initial population. The formula is as follows:

[0050]

[0051] In the Circle chaotic mapping formula, mod is the modulo function; using the Circle chaotic mapping instead of random initialization can make the population more evenly distributed in the search space.

[0052] Phase Two: Update the value E, and execute either a global exploration strategy or a local development strategy based on the value E. The formula is:

[0053]

[0054] Where: E represents the escape energy of the prey; rand is a random number between 0 and 1, z is a random number between -1 and 1, t and T are the current iteration number and the maximum iteration number, respectively, h is a random number in the interval [-2, 2], and w represents a constant. The variable E lays a good foundation for a smooth transition from exploration to exploitation. When |E| ≥ 1, the exploration phase begins. On the other hand, when |E| < 1, the development phase begins.

[0055] Phase 3: Exploration Phase. When |E| ≥ 1, the exploration phase begins. In the exploration phase, the fitness-distance balance mechanism (FDB) is used to improve the optimization strategy, as is typical for the Aquila optimizer. The general formula for the exploration phase of the Aquila optimizer is as follows:

[0056] If rand ≤ 0.5

[0057]

[0058] If rand > 0.5

[0059] X i (t+1)=X B (t)×Levy(D)+X r (t)+(yx)×rand (2)

[0060] In formula (1), X i (t+1) refers to the update position of the i-th individual in the next iteration t, X best (t) represents the optimal solution found so far, X m (t) represents the average position of all individuals in the current population, where t and T are the current iteration number.

[0061] In formula (2), X r This represents the random position of an individual selected from the current population [1, N]. Levy(D) represents the Levy flight function, which is expressed as follows:

[0062]

[0063] u and v are random numbers in the interval [0,1], Γ(.) represents the gamma function, and β is a constant value equal to 1.5. In formula (2), y and x represent the contour spiral shape in the search process, and their formulas are as follows:

[0064]

[0065] Where r represents the number of search cycles between 1 and 20, U is a constant of 0.00565, D1 is an integer vector from 1 to the dimension size (D), and ω is also a small value equivalent to 0.005.

[0066] The exploration phase of the general Aquila optimizer described above is improved using the Fitness-Distance Balance (FDB) strategy. The goal of FDB is to discover one or more candidate solutions that contribute most to the algorithm's search process. FDB differs from other selection methods in that the selection process is performed based on the scores of candidate solutions, not just their fitness values. The implementation steps of FDB are as follows:

[0067] First, assume the optimization problem has dimension D, and N is the total number of candidate solutions in the population. The i-th candidate solution can be defined as X. i =(x i,1 ,x i,2 ,…,x i,D ), i = 1, 2, ..., N. Therefore, each solution is related to the highest solution X in the population. best The Euclidean distance between them is calculated using the following formula:

[0068]

[0069] Then the distance vector DX for each candidate solution is given by the following formula:

[0070]

[0071] Then, after normalization, the fitness and distance values ​​of the candidate solutions are used to calculate the score, as shown below:

[0072]

[0073] Where γ is a constant equal to 0.5, normF Xi Normalized fitness of the solution, normD Xi This represents the normalized distance value.

[0074] Finally, the score vector X represents the FDB score of the entire population.FDB (t) is shown in the formula below:

[0075]

[0076] To improve search efficiency, we employ the FDB selection strategy to determine a candidate solution X that contributes the most to the search process. FDB (t), to replace X in formula (2) r (t), the formula is as follows:

[0077] X i (t+1)=X B (t)×Levy(D)+X FDB (t)+(yx)×rand (4)

[0078] Phase Four: Development Phase. The development phase begins when |E| < 1. The development phase consists of four mechanisms. Assume r is a random number and r ∈ [0, 1].

[0079] Mechanism 1: When r ≥ 0.5 and |E| ≥ 0.5, the individual's current position is updated using the following formula.

[0080] X i (t+1)=ΔX i (tE|JX best (t)-X i (t)| (5)

[0081] ΔX i (t)=X best (t)-X i (t)

[0082] J = 2(1 - r8)

[0083] Where: ΔX i (t) is the difference between the prey's position and the current position of the i-th eagle in iteration t, r8 represents a random number between 0 and 1, and J represents the random jump intensity of the prey.

[0084] Mechanism 2: If r ≥ 0.5 and |E| < 0.5, the individual's current position is updated using the following formula.

[0085] X i (t+1)=X best (t)-E∣ΔX i (t)∣ (6)

[0086] Mechanism 3: When r < 0.5 and |E| ≥ 0.5, the individual's current position is updated using the following formula.

[0087] Y = X best (t)-E|JXbese (t)-X i (t)|

[0088] Z = Y + S × Levy(D)

[0089]

[0090] In the formula: D is the dimension of the problem; S is a random vector of size 1×D; u and v are random numbers in the interval [0,1]; Γ(·) represents the standard gamma function; β is a constant value fixed at 1.5. Levy(D) represents Formula 3.

[0091] Mechanism 4: When r < 0.5 and |E| < 0.5, the individual's current position is updated using the following formula.

[0092] Y = X best (t)-E|JX best (t)-X m (t)|

[0093] Z = Y + S × Levy(D)

[0094]

[0095] Mechanism 5: By comparing the fitness values ​​of all individuals in the entire search space, the current global optimal solution is continuously updated, and it is determined whether the maximum number of iterations T has been reached. If the condition is met, the fitness value and position of the optimal individual are output; otherwise, the population is reinitialized.

[0096] Step 5: Use the path obtained by the EAOHHO algorithm as the trajectory path of the UAV.

[0097] In step five, cubic spline interpolation is used to smooth the UAV's flight trajectory. The path points that the UAV needs to pass through during its flight are P = (P1, P2...Pn), and the coordinates of each point pk in the three-dimensional coordinate system are (xk, yk, zk). Cubic spline interpolation is performed on (x1, x2...xn), (y1, y2...yn), and (z1, z2...zn) respectively to finally form a smooth flight trajectory curve.

[0098] In the field of UAV path planning, this invention deeply integrates the Harris Eagle Optimization Algorithm (HHO) and the Sky Eagle Optimization Algorithm (AO), giving full play to the fast local search capability of HHO and the strong global exploration capability of AO, thus significantly improving the overall path planning capability of the EAOHHO algorithm.

[0099] This invention introduces a fitness-distance balancing mechanism to select the candidate solution that contributes the most to the search process during the exploration phase, which significantly improves the convergence speed of the algorithm and its ability to escape local optima.

[0100] This invention introduces Circle chaotic mapping to initialize the population, which significantly improves the diversity and distribution uniformity of the initial population, enhances the global exploration capability of the algorithm, avoids the population aggregation problem that may be caused by traditional random initialization, and lays a better foundation for the subsequent optimization process.

[0101] This invention introduces a novel phased hybrid mechanism and defines an intelligent switching criterion based on escape energy (E). Instead of simply alternating between HHO and AO, this mechanism dynamically and intelligently selects one of four development strategies—either the improved exploration phase of AO or HHO—based on the escape energy value calculated during iteration, characterized by |E|≥1 or |E|<1, and its comparison with 0.5. This significantly improves the algorithm's adaptability.

[0102] Traditional particle swarm optimization (PSO), Harris Eagle algorithm (HHO), and Eagle Algorithm (AO) converge slowly during the iterative process, requiring significant computational resources and time to find a satisfactory solution. This invention deeply integrates the advantages of both HHO and AO algorithms and introduces a fitness-distance balance (FDB) selection mechanism. During the exploration phase, it more intelligently selects the direction of the individual that contributes most to the search, greatly accelerating the convergence process and reducing the required number of iterations and computation time.

[0103] This invention constructs a composite cost function that comprehensively considers path length, collision detection, flight altitude variance, and path smoothness (turning angle and climb angle). The EAOHHO algorithm optimizes this function and can automatically satisfy these complex constraints, ultimately generating a high-quality trajectory that is not only short in length but also stable, safe, reliable, and energy efficient.

[0104] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A UAV path planning method based on a hybrid improvement strategy, characterized in that: Includes the following steps: Step 1: A three-dimensional model of the UAV's flight space was established; Step 2: Determine the flight cost function of the UAV based on path length, collision occurrence, flight altitude, and path smoothness; Step 3: Initialize parameters, including population size, maximum number of iterations, dimension of the objective function, upper bound of the initial value, and lower bound of ub (lb). Step 4: Apply the UAV trajectory planning method based on the hybrid improvement strategy to the path planning problem of UAVs. Specifically, the Harris Eagle Algorithm (HHO) and the Aquila optimizer are used in the heuristic algorithm to improve the two algorithms in a hybrid manner. Two improvement strategies are used to finally propose a brand-new hybrid improvement strategy algorithm, named the EAOHHO algorithm. Then, the EAOHHO algorithm is used to optimize the flight cost function. Step 5: Use the path obtained by the EAOHHO algorithm as the trajectory path of the UAV.

2. The UAV path planning method based on a hybrid improvement strategy according to claim 1, characterized in that: In step one, the mathematical expression for the mountain model in environmental modeling is: Where (x,y) are the coordinates of the mountain peak on the horizontal plane; (x0,y0) are the coordinates of the center point of the mountain peak on the horizontal plane; m1 and m2 represent the height; m1 and m2 reflect the steepness of the peak.

3. The UAV path planning method based on a hybrid improvement strategy according to claim 1, characterized in that: In step three, the path length expression is: l i =||(x i+1 ,y i+1 ,z i+1 )-(x i ,y i ,z i )|| 2 Among them l i This represents the distance between the i-th waypoint and the (i+1)-th waypoint; furthermore, (x i ,y i ,z i ) and (x i+1 ,y i+1 ,z i+1 ) are the coordinates of the i-th waypoint and the (i+1)-th waypoint, respectively; The flight altitude of a drone should be maintained within a certain range. The specific formula for the altitude cost of a drone is as follows: In UAV 3D path planning, path smoothness cost is a key indicator for measuring flight path quality. Its core lies in the quantitative evaluation of parameters such as turning angle and climb rate, with the turning angle φ being a crucial factor. ij Defined as two consecutive path segments, i.e., P' ij P' i,j+1 With P' i,j+1 P' i,j+2 The angle between the projections onto the horizontal plane Oxy; let k be a unit vector pointing towards the z-axis, then the projection vector can be calculated as: P' ij P' i,j+1 =k×(P ij P i,j+1 ×k) Therefore, the turning angle φ ij It can be calculated using the following formula: Ascent angle ψ ij Represents path segment P ij P i,j+1 Its projection P' on the horizontal plane ij P' i,j+1 The angle between them is calculated using the following formula: The path smoothness cost consists of two parts: a penalty for turning angles and a penalty for the rate of change of continuous climb angles, expressed as follows: In summary, the flight cost function of the UAV is defined as: F(x) = L + H + S; By optimizing the flight cost function, the flight path planning of the UAV is completed.

4. The UAV path planning method based on a hybrid improvement strategy according to claim 1, characterized in that: The EAOHHO algorithm in step four includes five stages: Phase 1: Initialization Phase. Individuals are initialized using the Circle chaotic mapping, and individuals with higher fitness are retained to form the final initial population. The formula is as follows: In the Circle chaotic mapping formula, mod is the modulo function; using the Circle chaotic mapping instead of random initialization can make the population more evenly distributed in the search space. Phase Two: Update the value E, and execute either a global exploration strategy or a local development strategy based on the value E. The formula is: Where: E represents the escape energy of the prey; rand is a random number between 0 and 1, z is a random number between -1 and 1, t and T are the current iteration number and the maximum iteration number, respectively, h is a random number in the interval [-2,2], and w represents a constant; the variable E lays a good foundation for a smooth transition from exploration to utilization; when |E|≥1, the exploration phase begins; on the other hand, when |E|<1, the development phase begins. Phase 3: Exploration Phase; When |E|≥1, the exploration phase begins; In the exploration phase, the fitness-distance balancing mechanism is used to improve policy optimization, similar to the general Aquila optimizer; The formula for the Aquila optimizer's exploration phase is as follows: If rand ≤ 0.5 If rand > 0.5 X i (t+1)=X B (t)×Levy(D)+X r (t)+(y-x)×rand (2) In formula (1), X i (t+1) refers to the update position of the i-th individual in the next iteration t, X best (t) represents the optimal solution found so far, X m (t) represents the average position of all individuals in the current population, where t and T are the current iteration number; In formula (2), X r This represents the random position of an individual selected from the current population [1,N]; Levy(D) represents the Levy flight function, which is expressed as follows: u and v are random numbers in the interval [0,1], Γ(.) represents the gamma function, and β is a constant value equal to 1.5; in formula (2), y and x represent the contour spiral shape in the search process, and its formula is as follows: Where r represents the number of search cycles between 1 and 20, U is a constant of 0.00565, D1 is an integer vector from 1 to the dimension size (D), and ω is also a small value equivalent to 0.

005. The exploration phase of the general Aquila optimizer described above uses a fitness-distance balancing mechanism to improve policy optimization. The purpose of FDB is to discover one or more candidate solutions that contribute most to the algorithm's search process. The fitness-distance balancing mechanism differs from other selection methods in that the selection process is performed based on the scores of candidate solutions, not just their fitness values. The implementation steps of the fitness-distance balancing mechanism are as follows: First, assume the optimization problem has dimension D, and N is the total number of candidate solutions in the population; the i-th candidate solution can be defined as X. i =(x i,1 ,x i,2 ,…,x i,D ), i = 1, 2, ..., N; therefore, each solution is related to the highest solution X in the population. best The Euclidean distance between them is calculated using the following formula: Then the distance vector DX for each candidate solution is given by the following formula: Then, after normalization, the fitness and distance values ​​of the candidate solutions are used to calculate the score, as shown below: Where γ is a constant equal to 0.5, normF Xi Normalized fitness of the solution, normD Xi This represents the normalized distance value; Finally, the score vector X represents the FDB score of the entire population. FDB The formula for (t) is as follows: A fitness-distance balancing mechanism is used to select a candidate solution X that contributes the most to the search process. FDB (t), to replace X in formula (2) r (t), as shown in the formula below: X i (t+1)=X B (t)×Levy(D)+X FDB (t)+(y-x)×rand (4) Phase Four: Development Phase; When |E|<1, the development phase begins; The development phase consists of four mechanisms; Assume r is a random number and r∈[0,1]; Mechanism 1: When r ≥ 0.5 and |E| ≥ 0.5, the individual's current position is updated using the following formula. X i (t+1)=ΔX i (t)-E|JX best (t)-X i (t)| (5) ΔX i (t)=X best (t)-X i (t) J＝2(1-r8) Where: ΔX i (t) is the difference between the prey's position and the current position of the i-th eagle in iteration t, r8 represents a random number between 0 and 1, and J represents the random jump intensity of the prey. Mechanism 2: If r ≥ 0.5 and |E| < 0.5, the individual's current position is updated using the following formula. X i (t+1)=X best (t)-E∣ΔX i (t)∣ (6) Mechanism 3: When r < 0.5 and |E| ≥ 0.5, the individual's current position is updated using the following formula. Y=X best (t)-E|JX best (t)-X i (t)| Z = Y + S × Levy(D) In the formula: D is the dimension of the problem; S is a random vector of size 1×D; u and v are random numbers in the interval [0,1]; Γ(·) represents the standard gamma function; β is a constant value fixed at 1.5; Levy(D) represents Formula 3; Mechanism 4: When r < 0.5 and |E| < 0.5, the individual's current position is updated using the following formula. Y=X best (t)-E|JX best (t)-X m (t)| Z = Y + S × Levy(D) Mechanism 5: By comparing the fitness values ​​of all individuals in the entire search space, the current global optimal solution is continuously updated, and it is determined whether the maximum number of iterations T has been reached. If the condition is met, the fitness value and position of the optimal individual are output; otherwise, the population is reinitialized.

5. The UAV path planning method based on a hybrid improvement strategy according to claim 4, characterized in that: In step five, cubic spline interpolation is used to smooth the flight trajectory of the UAV. The path points that the UAV needs to pass through during flight are P = (P1, P2...Pn), and the coordinates of each point pk in the three-dimensional coordinate system are (xk, yk, zk). Cubic spline interpolation is performed on (x1, x2...xn), (y1, y2...yn), and (z1, z2...zn) respectively to finally form a smooth flight trajectory curve.

Images