Unmanned aerial vehicle three-dimensional path planning method based on improved whale optimization algorithm
By introducing chaotic mapping, reverse learning, nonlinear convergence factors and adaptive inertial weights into the whale optimization algorithm, combined with dynamic mutant perturbation, the three-dimensional path planning of the drone is improved, and the search efficiency of traditional algorithms in complex environments is solved. It is easy to fall into the local optimal problem, achieving faster convergence speed and more efficient path planning.
Patent Information
- Application Number
- CN202411057793.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-02
- Publication Date
- 2024-11-08
AI Technical Summary
Traditional whale optimization algorithms have low search efficiency in multivariate complex problems, poor convergence ability, easy to fall into local optimality, and uneven distribution of initialized populations, resulting in low path planning efficiency in forest scenes with large terrain and complex environments.
By introducing chaotic mapping and reverse learning, the initialized population is generated, the nonlinear convergence factor and adaptive inertial weights are introduced, and the whale optimization algorithm is improved to enhance its global and local search capabilities and avoid local optimal traps.
It significantly improves the convergence speed and global optimal solution of the drone's three-dimensional path planning, and can plan safe and efficient routes in multi-obstruction forest scenarios, improving the efficiency and quality of path planning.
Smart Images

Figure CN118915810A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) path planning, and in particular to a three-dimensional path planning method for UAV based on an improved whale optimization algorithm. Background Art
[0002] At present, unmanned aerial vehicles (UAVs) are widely used in search and patrol, fire rescue, reconnaissance and surveillance, disaster relief, logistics distribution, power inspection, agricultural irrigation and other fields due to their high flexibility, strong maneuverability, low safety risk factor, easy remote control and low cost. At present, the path planning methods for UAVs can be mainly divided into two categories: traditional methods and intelligent algorithms. Traditional path planning methods mainly include A* algorithm, rapid-exploring random trees (RRT), artificial potential field method, etc., but these algorithms are mostly used in urban areas with relatively simple terrain. For forest scenes with large terrain fluctuations and complex environments, the search efficiency is low and the convergence is slow. The main intelligent algorithms include particle swarm algorithm, Harris eagle algorithm, gray wolf algorithm, ant colony algorithm, etc. These intelligent algorithms have the advantages of simple implementation and good real-time performance. The Whale Optimization Algorithm (WOA) is a new type of swarm intelligence optimization search method proposed by Mirjalili and Lewis in 2016. It is derived from the simulation of the hunting behavior of humpback whale groups in nature. The WOA algorithm assumes that the prey caught by the whale is the optimal solution and the location of the whale is a potential solution. The algorithm contains three mathematical models: search for food, shrinkage and spiral update. In each iteration, the position update strategy of each whale is determined by the value of the random number p and the modulus of the coefficient vector A. As the iteration proceeds, the whale population continues to move closer to the optimal solution.
[0003] Since the three population update mechanisms of the whale optimization algorithm are independent of each other, the global exploration and local development processes in the optimization phase can be run and controlled separately. In addition, the whale optimization algorithm does not require manual setting of various control parameter values, which improves the efficiency of the algorithm and reduces the difficulty of application. Compared with other swarm intelligence optimization algorithms, the WOA algorithm has a novel structure and fewer control parameters. It shows good optimization performance in solving many numerical optimization and engineering problems, and is superior to intelligent optimization algorithms such as ant colony algorithm and particle swarm algorithm.
[0004] However, traditional WOA is usually generated by random initialization, which sometimes causes the distribution of whales in the search domain to be uneven. This may cause the population to move away from the global optimal solution, making the search time to converge to the optimal solution longer, resulting in a lower convergence speed, easy to fall into the local optimum, and has the disadvantage of a single search method. Summary of the invention
[0005] In view of the above problems, the purpose of the present invention is to provide a three-dimensional path planning method for unmanned aerial vehicles based on an improved whale optimization algorithm, which generates an initialization population through chaotic mapping and reverse learning, introduces nonlinear convergence factors and dynamic inertia weights, and integrates the optimization method of adaptive mutation enhancement to optimize the WOA algorithm in multiple aspects, so that it has a better convergence speed and avoids falling into the local optimum. The technical solution is as follows:
[0006] A three-dimensional path planning method for unmanned aerial vehicles based on an improved whale optimization algorithm comprises the following steps:
[0007] Step 1: Set up the drone constraints:
[0008] Assuming that the UAV maintains a predetermined flight speed, the flight distance constraint is implemented by taking the Euclidean distance between two flight nodes as the length of each flight segment; the flight altitude constraint is implemented by setting a value that penalizes values exceeding the flight altitude range; the obstacle avoidance constraint is implemented by setting the range of the flight path entering the danger zone or collision zone around obstacles; the smoothness constraint is implemented by constraining the horizontal steering angle and the vertical pitch angle;
[0009] Step 2: Use a spherical coordinate system to describe the flight trajectory of the UAV. Use the current radial distance, horizontal steering angle, and vertical pitch angle of the UAV to represent its position at the current moment, and perform environmental modeling on the flight trajectory of the UAV according to the flight environment.
[0010] Step 3: Based on the set drone constraints and environment modeling, the drone path planning is performed using the IWOA algorithm:
[0011] Step 3.1: Set the initial parameters of IWOA, including population size, maximum number of iterations, and search space range;
[0012] Step 3.2: Initialize the population using Tent chaotic mapping and reverse learning strategy;
[0013] Step 3.3: Calculate the fitness values of individuals in the population, and select individuals whose fitness values are greater than the set threshold to enter the next generation population for further iteration;
[0014] Step 3.4: Determine whether the maximum number of iterations has been reached. If so, output the optimal solution. If not, generate a random number p between [-1, 1] and update the inertia weight ω.
[0015] Step 3.5: Define the mathematical model of search and foraging and its coefficient vector and Introduce a nonlinear convergence factor, update the control parameter a, and calculate the coefficient vector in the mathematical model of search and foraging based on the updated control parameter and
[0016] Step 3.6: When the random number p<0.5, if |A|<1, the shrinking and encircling update is performed according to the mathematical model of shrinking and encircling; if |A|>1, the searching and foraging update is performed according to the mathematical model of searching and foraging; when the random number p>0.5, the spiral mechanism is updated according to the mathematical model of spiral update position;
[0017] Step 3.7: Perform dynamic mutation perturbation update on the updated position and enter the next iteration.
[0018] Furthermore, in step 1, the drone constraint function is expressed as:
[0019]
[0020] Where: X i is the decision variable; k is the weight of each constraint function; g k (X i ) are the constraint functions;
[0021] The UAV constraints specifically include:
[0022] 1) Flight distance constraints
[0023] The Euclidean distance between two nodes is taken as the length of each segment flight, and the overall flight length is the sum of the lengths of each segment. Then the flight length g 1 (X i ) is expressed as:
[0024]
[0025] Where: x i ,y i 、z i are the horizontal coordinate, vertical coordinate and altitude value of the ith flight node respectively; n is the total number of flight nodes;
[0026] 2) Flight altitude constraints
[0027] Select an appropriate flight altitude. If the altitude exceeds the range, set the penalty value g. 2 (X i ):
[0028]
[0029] Where: h ij is the current flight altitude, h max is the maximum allowed flight altitude, h min is the minimum allowed flight altitude; Hij It is the difference between the current flight altitude and the allowed average flight altitude;
[0030] 3) Obstacle avoidance constraints
[0031] Simplify the obstacle into a cylinder with a fixed radius, set the total number of obstacles to K, and the center of the obstacle to C k , the obstacle radius is R k , the diameter of the drone is D, then define R k +D is the radius of the collision area; S is the width of the annular danger zone outside the collision area; Assume that the vertical distance between the line connecting two adjacent flight nodes and the center of the obstacle circle is d k , when d k >S+D+R k When D+R k <d k ≤S+D+R k When the drone is in the danger zone, and d k The smaller the value, the greater the collision risk. k ≤D+R k When , the UAV collides and the obstacle avoidance constraint is infinite; it is expressed as follows:
[0032]
[0033] Where: represents the obstacle avoidance constraint of the kth obstacle, and Indicates two adjacent flight nodes;
[0034] 4) Smoothness Constraint
[0035] Horizontal steering angle φ ij is the angle between two consecutive path segments projected on the horizontal plane Oxy. The horizontal steering angle is calculated as:
[0036]
[0037] In the formula, It is the flight path Projection on the horizontal plane Oxy; It is the flight path Projection on the horizontal plane Oxy;
[0038] Vertical pitch angle ψ ij is the angle between two consecutive flight path segments projected onto the vertical axis, then the vertical pitch angle is calculated as:
[0039]
[0040] Where: zi,j and z i,j+1 are the flight altitudes of the UAVs;
[0041] The smoothness constraint is expressed as:
[0042]
[0043] Where: η 1 and η 2 are the constraint coefficients of the horizontal steering angle and the vertical pitch angle respectively; ψ i,j-1 It is the vertical pitch angle of the previous flight node.
[0044] Furthermore, in step 3.2, the population is initialized using the Tent chaotic map and the reverse learning strategy, specifically:
[0045] In order to prevent the Tent chaotic sequence from falling into small periodic points and unstable periodic points during iteration, random numbers are added to the expression of the typical Tent mapping, namely:
[0046]
[0047] In the formula, is the initial vector of the whale's individual position; N is the population size, rand(·) is a random number between (0, 1); X is the initial population;
[0048] Using the chaotic sequence Z kj Generate the corresponding new initial population
[0049]
[0050] In the formula, and are the maximum initial vector and the minimum initial vector of the individual whale position respectively;
[0051] Generate reverse population
[0052]
[0053] Furthermore, the dynamic adjustment of the inertia weight ω in step 3.4 adopts an adaptive inertia weight formula:
[0054]
[0055] Where: w max and w min are the maximum and minimum values of weight change, T is the maximum number of iterations, and t is the current number of iterations.
[0056] Furthermore, in step 3.5, the control parameter a is updated, and the coefficients A and C in the mathematical model of searching for food are calculated, specifically:
[0057] Step 3.5.1: Define the mathematical model of search and foraging:
[0058]
[0059] Where: is the position vector of a whale individual randomly selected from the current whale group, is the current individual whale position vector, Represents the distance vector between the current whale individual and the randomly selected whale individual, coefficient vector and is defined as:
[0060]
[0061] Where: is the control parameter vector, is a random vector distributed between [0,1];
[0062] Step 3.5.2: Introduce a nonlinear convergence factor and update the control parameter a to:
[0063]
[0064] Where: T is the maximum number of iterations, t is the current number of iterations, is the initial control parameter vector, It is the modulus of the new control parameter vector generated by the change of the initial control parameter vector with the number of iterations t;
[0065] Step 3.5.3: Based on the updated control parameter vector Calculate the coefficient vector in the mathematical model of search and foraging and
[0066]
[0067] Furthermore, in step 3.6, the mathematical model of the shrinkage enclosure is:
[0068]
[0069] Where: is the individual whale position vector with the best objective function value in the current whale group,
[0070] is the encirclement step length of the current whale individual, The smaller the whale is, the smaller its stride length will be;
[0071] The mathematical model of the spiral update position is as follows:
[0072]
[0073] Where: Represents the distance vector between the current individual whale and the whale in the best position; the constant coefficient b determines the shape of the spiral when the individual whale spirals forward. When b takes the value of 1, it is an ordinary logarithmic spiral. l is a random number between [-1,1].
[0074] Furthermore, in step 3.7, the update position is dynamically updated by a disturbance change, specifically:
[0075] Gaussian mutation weights and Cauchy mutation weights are introduced. In the early stage of iteration, the weight of Cauchy mutation is greater than that of Gaussian mutation; in the later stage of iteration, the weight of Gaussian mutation is greater than that of Cauchy mutation. The formula is as follows:
[0076]
[0077] Where: is the current iteration weight, is the Gaussian variation weight, is the Cauchy mutation weight, T is the maximum number of iterations, and t is the current number of iterations.
[0078] The beneficial effects of the present invention are:
[0079] 1) In view of the problems of low search efficiency, poor convergence ability and easy to fall into the optimal deficiency in traditional whale optimization algorithm when facing multi-variable complex problems, the present invention generates the initialization population through chaotic mapping and reverse learning, introduces nonlinear convergence factor and adaptive inertia weight balance to optimize the global and local search capabilities, and integrates the search ability and convergence speed of the dynamic variation disturbance enhancement algorithm; the present invention has a strong convergence speed and global optimal ability, can plan safe and efficient routes for drones in multi-obstacle forest scenes, and improve the three-dimensional path planning ability of drones.
[0080] 2) The present invention adopts a spherical coordinate system to describe the trajectory of the UAV. Compared with the rectangular coordinate system, the generated motion trajectory meets the requirements of the steering angle and pitch angle in the smoothing cost, and the solution quality that can be generated is higher and it is easier to generate a feasible solution.
[0081] 3) The digital elevation model of a forest area used in the present invention as the path planning environment of the UAV is more in line with the actual flight working environment of the UAV than the simulated environment. BRIEF DESCRIPTION OF THE DRAWINGS
[0082] Figure 1 Schematic diagram of the obstacle area.
[0083] Figure 2 This is the flight angle coordinate diagram.
[0084] Figure 3 A digital elevation model of a simple scene.
[0085] Figure 4 It is a digital elevation model of complex scenes.
[0086] Figure 5 This is the algorithm flow chart of IWOA (Improvement Whale Optimization Algorithm).
[0087] Figure 6 Resulting path graph for simple scenario path planning.
[0088] Figure 7 Path planning for complex scenarios.
[0089] Figure 8 Path planning results for a simple scenario.
[0090] Fig. 9 Path planning results for complex scenarios. DETAILED DESCRIPTION
[0091] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0092] 1 Constraints and environment modeling
[0093] 1.1 UAV Constraints
[0094] There are many constraints that need to be considered in the three-dimensional path planning of UAVs. Assuming that the UAV maintains a predetermined flight speed, it is necessary to consider flight distance constraints, flight altitude constraints, obstacle avoidance constraints, and smoothness constraints. The mathematical representation of the above constraint function is:
[0095]
[0096] Where: X i is the decision variable; k is the weight of each constraint function; g k (X i ) are the constraint functions.
[0097] (1) Flight distance constraints
[0098] The distance of the UAV's flight route should be as short as possible to reduce flight time and improve remote inspection efficiency. The Euclidean distance between two flight nodes is taken as the length of each flight segment, and the overall flight length is the sum of the lengths of each segment. The flight length can be expressed as:
[0099]
[0100] Where: x i ,y i 、z i are the horizontal coordinate, vertical coordinate and altitude value of the i-th flight node respectively.
[0101] (2) Flight altitude constraints
[0102] When flying a drone, it is necessary to select a suitable flight altitude. When the flight altitude exceeds the range, a penalty value is set. The flight altitude constraint is expressed as:
[0103]
[0104] Where: h ij is the current flight altitude, h max is the maximum allowed flight altitude, h min is the minimum allowed flight altitude.
[0105] (3) Obstacle avoidance constraints
[0106] During the drone's aerial survey, it may encounter obstacles such as communication towers, overhead transmission towers, tall trees, etc., which require emergency obstacle avoidance flight. To further simplify the model, the obstacle is simplified to a cylinder with a fixed radius, and the radius of the obstacle is the radius of the cylinder (see Figure 1 ). Figure 1 In the example, the total number of obstacles is set to K and the center of the obstacle circle is set to C. k , R k is the obstacle radius, D is the diameter of the drone, then R k +D is the radius of the collision area; S represents the width of the annular danger zone outside the collision area, which depends on the positioning accuracy of the drone and the flight environment; P′ i,j and P′ i,j+1 Two adjacent flight nodes and Projection on the horizontal plane Oxy. If the drone is in a static environment and the GPS signal is good, the ring width S is small; if there are moving objects in the environment or the GPS signal is poor, the ring width S will exceed 100 meters. Assume that the vertical distance between the line connecting two adjacent flight nodes and the center of the obstacle circle is d k , when d k >S+D+R kWhen D+R k <d k ≤S+D+R k When the drone is in the danger zone, and d k The smaller the value, the greater the collision risk. k ≤D+R k When , the UAV collides, and the constraint condition is infinite; the above obstacle avoidance constraint condition can be expressed as:
[0107]
[0108] (4) Smoothness constraint
[0109] The flight angle control parameters of the UAV mainly include the horizontal steering angle and the vertical pitch angle (see Figure 2 ), these two parameter variables must satisfy the actual angle constraints of the UAV, otherwise the trajectory planning model cannot generate a feasible flight path.
[0110] Horizontal steering angle φ ij is the angle between two consecutive path segments projected on the horizontal plane Oxy. The horizontal steering angle is calculated as:
[0111]
[0112] Vertical pitch angle ψ ij is the angle between two consecutive path segments projected on the vertical axis, then the vertical pitch angle is calculated as:
[0113]
[0114] The smoothness constraint can be expressed as:
[0115]
[0116] Where: η 1 and η 2 are the constraint coefficients of the horizontal steering angle and the vertical pitch angle respectively.
[0117] 1.2 Environmental Modeling
[0118] This embodiment uses a digital elevation model (DEM) of a forest area as the path planning environment for the drone, and appropriately adds obstacles as obstacle avoidance factors. According to the complexity of the obstacles, two types of forest terrain environments are generated (see Figure 3 and Figure 4 ).
[0119] The present invention uses a spherical coordinate system to describe the trajectory of the UAV. By using the radial distance r ij, horizontal steering angle φ ij , vertical pitch angle ψ ij The position of the drone at a certain moment is represented by a spherical coordinate system. The motion trajectory of the drone meets the requirements of the steering angle and pitch angle in the smoothing cost. Compared with randomly generated rectangular coordinates, the solution quality of this spherical coordinate system is higher and it is easier to generate feasible solutions. The conversion relationship between spherical coordinates and rectangular coordinates can be expressed as:
[0120]
[0121] 2WOA Algorithm
[0122] The WOA algorithm assumes that the prey caught by the whale is the optimal solution, and the location of the whale is a potential solution. The algorithm contains three mathematical models: search and foraging, shrinkage and encirclement, and spiral update position. In each iteration, the position update strategy of each whale is determined by the value of the random number p and the modulus of the coefficient vector A. As the iteration proceeds, the whale population continues to move closer to the optimal solution.
[0123] The mathematical model of search and foraging is defined as:
[0124]
[0125] Where: is the position vector of a whale individual randomly selected from the current whale group, is the current individual whale position vector, Represents the distance vector between the current whale individual and the randomly selected whale individual, coefficient vector and is defined as:
[0126]
[0127] Where: Control parameter vector As the number of iterations increases, it decreases linearly from 2 to 0. is a random vector distributed between [0,1].
[0128] The mathematical model of shrinkage is defined as follows:
[0129]
[0130] Where: is the individual whale position vector with the best objective function value in the current whale group, is the encirclement step length of the current whale individual, The smaller the whale is, the smaller its stride length is.
[0131] The mathematical model of the spiral update position is as follows:
[0132]
[0133] Where: Represents the distance vector between the current individual whale and the whale in the best position; the constant coefficient b determines the shape of the spiral when the individual whale spirals forward. When b is 1, it is an ordinary logarithmic spiral, and l is a random number between [-1,1].
[0134] Humpback whales move around their prey in a shrinking circle while swimming along a spiral path. To simulate this simultaneous behavior, assume that there is a 50% probability of choosing between shrinking or spiraling, that is, there is a random number p between [-1,1]. When p≥0.5, the mathematical model of spiral updating is used for updating. When p<0.5, the modulus of the coefficient vector A is used to determine whether to search for food or shrink. When When , enter the shrinking and surrounding phase to update the position. When , it enters the search and foraging phase to update the position. In summary, the mathematical expression of the WOA algorithm is as follows:
[0135]
[0136] 3IWOA algorithm
[0137] When it comes to the three-dimensional path planning problem of UAVs, there are multiple constraints. The traditional WOA algorithm has the disadvantages of low search efficiency, poor convergence ability, and easy to fall into local optimality. This paper will improve the WOA algorithm by optimizing the initialization population, introducing nonlinear convergence factors and other optimization methods to enhance the algorithm's optimization ability.
[0138] 3.1 Initialization of population optimization
[0139] The present invention proposes to generate an initialization population based on the idea of chaotic mapping and reverse learning. Considering that Tent mapping is prone to problems at small cycle periods and fixed points, in order to prevent the Tent chaotic sequence from falling into small cycle points and unstable cycle points during iteration, random numbers are added to the expression of the typical Tent mapping, namely:
[0140]
[0141] Using the chaotic sequence Z kj Generate the corresponding initial population
[0142]
[0143] Generate reverse population
[0144]
[0145] Compare the initial population and the reverse population, and select N individuals with the best fitness to form the initial population. Through this initialization strategy, high-quality solutions can be further obtained in a more evenly distributed population, thereby accelerating the convergence speed.
[0146] 3.2 Nonlinear Convergence Factor
[0147] According to formula (12), the global search and local search of the whale algorithm are related to a. When a is large, the algorithm has a strong global search ability and is easy to escape the local optimum, but its local search ability is weak, resulting in a decrease in convergence speed. On the contrary, if a is small, its local search ability is strong and the convergence speed is fast, but it is easy to fall into the local optimum. In the traditional WOA algorithm, a decreases linearly from 2 to 0, but there are often multiple local optimal values in complex optimization problems. The linear decrease strategy will affect the ability of the algorithm to escape the local optimum.
[0148] By introducing the nonlinear convergence factor (see formula 19), it can be achieved that in the early stage of iteration, a takes a larger value and decays slowly, and the global exploration ability is strong, which is conducive to preventing the population from falling into the local optimal value; in the later stage of iteration, a decays rapidly to a smaller value, and the local development ability is strong, which is conducive to accelerating the convergence of the population. Therefore, this nonlinear convergence factor update strategy can be better applied to the solution of nonlinear complex optimization problems.
[0149]
[0150] Where: T is the maximum number of iterations, and t is the current iteration value.
[0151] 3.3 Adaptive inertia weight
[0152] In the iterative process of the WOA algorithm, the weight is dynamically adjusted according to the number of iterations; a larger weight is assigned in the early stage of the iteration, so that the whale algorithm has a stronger global search capability to prevent it from falling into a local extreme value; a smaller weight is assigned in the later stage of the iteration, at this time the whale algorithm has a stronger local search capability, which can accelerate the convergence of the algorithm to obtain the optimal solution. The formula for the adaptive inertia weight is as follows:
[0153]
[0154] Where: w max and w min are the maximum and minimum values of weight change, T is the maximum number of iterations, and t is the current number of iterations.
[0155] Substituting the dynamic weight formula into equation (14), the improved model of spiral update position is as follows:
[0156]
[0157] 3.4 Dynamic variation disturbance
[0158] In the later stage of iteration, due to the search strategy, all whale individuals in the population will gather towards the optimal individual, resulting in a decrease in population diversity. If the optimal individual is a local optimal solution at this time, the algorithm will converge prematurely. In order to prevent such problems, dynamic mutation perturbation is used. In the early stage of iteration, the weight of the Cauchy mutation is large, and a larger step size is obtained through the Cauchy mutation to prevent the algorithm from falling into the local optimal solution. As the algorithm continues to run, the weight of the Gaussian mutation is large. The outstanding local search ability of the Gaussian mutation enables the optimal solution to be accurately searched in a local range, improving the optimization accuracy of the algorithm. The formula is as follows:
[0159]
[0160] Where: is the current iteration weight, is the Gaussian variation weight, is the Cauchy mutation weight, T is the maximum number of iterations, and t is the current iteration value.
[0161] In summary, the specific steps of the IWOA algorithm are as follows:
[0162] Step 1. Set the initial parameters of IWOA, including population size, maximum number of iterations, and search space range;
[0163] Step 2. Initialize the population using Tent chaotic mapping (Formula (16)) and reverse learning strategy (Formula (17) and (18));
[0164] Step 3. Calculate the fitness values of individuals in the population, and select individuals with high fitness to enter the next generation of population to continue iteration;
[0165] Step 4. Determine whether the maximum number of iterations has been reached. If so, output the optimal solution. If not, generate a random number p and update the inertia weight ω according to the formula (Formula (20)).
[0166] Step 5. Update a according to formula (19), and calculate A and C according to formula (12);
[0167] Step 6. When the random number p<0.5, if |A|<1, perform shrinkage and encirclement update according to formula (13); if |A|>1, perform search and foraging update according to formula (11); when the random number p>0.5, perform spiral mechanism update according to formula (14);
[0168] Step 7. Perform dynamic mutation disturbance update on the updated position according to formula (22) and enter the next iteration.
[0169] The specific process is as follows Figure 5 .
[0170] 4 Simulation Experiment
[0171] In order to verify the effectiveness of the IWOA algorithm in path planning, simulation experiments were carried out in the digital elevation DEM map by setting obstacles of different complexity. At the same time, the particle swarm optimization algorithm (PSO), Grey Wolf Optimizer (GWO), Whale Optimization Algorithm (WOA) and Harris Hawks Optimization (HHO) were used for comparative analysis under the same conditions. The main setting parameters of the simulation experiment are shown in Table 1.
[0172] Table 1 Simulation experiment parameter settings
[0173]
[0174] Based on the above parameter settings, simulation is performed in MATLAB R2016b software. In a simple scenario, the simulation results of the three-dimensional path of the UAV generated by each algorithm are as follows: Figure 6 As shown; in complex scenarios, the three-dimensional path simulation results of the drone generated by each algorithm are as follows Figure 7 shown.
[0175] Table 2 Optimal fitness values of different algorithms
[0176]
[0177] The iterative convergence curves of different algorithms in simple scenarios are shown in Figure 8 , the iterative convergence curve in complex scenes is shown in Fig. 9 The optimal fitness values of each algorithm in the two types of scenarios are shown in Table 2. Through the above comparative analysis, it can be seen that, whether in simple or complex scenarios, the IWOA algorithm has advantages over other algorithms in robustness and escaping local optimality, can effectively avoid dangerous sources, has a faster convergence speed, and is superior in UAV path planning.
Claims
1. A three-dimensional path planning method for unmanned aerial vehicles based on an improved whale optimization algorithm, characterized in that: The following steps are involved: Step 1: Set up the drone constraints: Assuming that the UAV maintains a predetermined flight speed, the flight distance constraint is implemented by taking the Euclidean distance between two flight nodes as the length of each flight segment; the flight altitude constraint is implemented by setting a value that penalizes values exceeding the flight altitude range; the obstacle avoidance constraint is implemented by setting the range of the flight path entering the danger zone or collision zone around obstacles; the smoothness constraint is implemented by constraining the horizontal steering angle and the vertical pitch angle; Step 2: Use a spherical coordinate system to describe the flight trajectory of the UAV. Use the current radial distance, horizontal steering angle, and vertical pitch angle of the UAV to represent its position at the current moment, and perform environmental modeling on the flight trajectory of the UAV according to the flight environment. Step 3: According to the set UAV constraints and environment modeling, the UAV path planning is performed through the IWOA algorithm: Step 3.1: Set the initial parameters of IWOA, including population size, maximum number of iterations, and search space range; Step 3.2: Initialize the population using Tent chaotic mapping and reverse learning strategy; Step 3.3: Calculate the fitness values of individuals in the population, and select individuals whose fitness values are greater than the set threshold to enter the next generation population for further iteration; Step 3.4: Determine whether the maximum number of iterations has been reached. If so, output the optimal solution. If not, generate a random number p between [-1, 1] and update the inertia weight ω. Step 3.5: Define the mathematical model of search and foraging and its coefficient vector and Introduce a nonlinear convergence factor, update the control parameter a, and calculate the coefficient vector in the mathematical model of search and foraging based on the updated control parameter and Step 3.6: When the random number p<0.5, if |A|<1, the shrinking and encircling update is performed according to the mathematical model of shrinking and encircling; if |A|>1, the searching and foraging update is performed according to the mathematical model of searching and foraging; when the random number p>0.5, the spiral mechanism is updated according to the mathematical model of spiral update position; Step 3.7: Perform dynamic mutation perturbation update on the updated position and enter the next iteration.
2. The method for three-dimensional path planning of unmanned aerial vehicle based on improved whale optimization algorithm according to claim 1 is characterized in that: In step 1, the drone constraint function is expressed as: Where: X i is the decision variable; k is the weight of each constraint function; g k (X i ) are the constraint functions; The UAV constraints specifically include: 1) Flight distance constraints The Euclidean distance between two nodes is taken as the length of each segment flight, and the total flight length is the sum of the lengths of each segment. Then the flight length g1(X i ) is expressed as: Where: x i ,y i 、z i are the horizontal coordinate, vertical coordinate and altitude value of the ith flight node respectively; n is the total number of flight nodes; 2) Flight altitude constraints Select an appropriate flight altitude. If the altitude exceeds the range, set the penalty value g2(X i ): Where: h ij is the current flight altitude, h max is the maximum allowed flight altitude, h min is the minimum allowed flight altitude; H ij It is the difference between the current flight altitude and the allowed average flight altitude; 3) Obstacle avoidance constraints Simplify the obstacle into a cylinder with a fixed radius, set the total number of obstacles to K, and the center of the obstacle to C k , the obstacle radius is R k , the diameter of the drone is D, then define R k +D is the radius of the collision area; S is the width of the annular danger zone outside the collision area; Assume that the vertical distance between the line connecting two adjacent flight nodes and the center of the obstacle circle is d k , when d k >S+D+R k When D+R k <d k ≤S+D+R k When the drone is in the danger zone, and d k The smaller the value, the greater the collision risk. k ≤D+R k When , the UAV collides and the obstacle avoidance constraint is infinite; it is expressed as follows: Where: represents the obstacle avoidance constraint of the kth obstacle, and Indicates two adjacent flight nodes; 4) Smoothness Constraint Horizontal steering angle φ ij is the angle between two consecutive path segments projected on the horizontal plane Oxy. The horizontal steering angle is calculated as: In the formula, It is the flight path Projection on the horizontal plane Oxy; It is the flight path Projection on the horizontal plane Oxy; Vertical pitch angle ψ ij is the angle between two consecutive flight path segments projected onto the vertical axis, then the vertical pitch angle is calculated as: Where: z i,j and z i,j+1 are the flight altitudes of the UAVs; The smoothness constraint is expressed as: Where: η1 and η2 are the constraint coefficients of the horizontal steering angle and the vertical pitch angle respectively; ψ i,j-1 It is the vertical pitch angle of the previous flight node.
3. The method for three-dimensional path planning of unmanned aerial vehicle based on improved whale optimization algorithm according to claim 1 is characterized in that: In step 3.2, the population is initialized using the Tent chaotic map and reverse learning strategy, specifically: In order to prevent the Tent chaotic sequence from falling into small periodic points and unstable periodic points during iteration, random numbers are added to the expression of the typical Tent mapping, namely: In the formula, is the initial vector of the whale's individual position; N is the population size, rand(·) is a random number between (0, 1); X is the initial population; Using the chaotic sequence Z kj Generate the corresponding new initial population In the formula, and are the maximum initial vector and the minimum initial vector of the individual whale position respectively; Generate reverse population 4. The method for three-dimensional path planning of unmanned aerial vehicle based on improved whale optimization algorithm according to claim 1 is characterized in that: In step 3.4, the inertia weight ω is updated using the adaptive inertia weight formula: Where: w max and w min are the maximum and minimum values of weight change, T is the maximum number of iterations, and t is the current number of iterations.
5. The method for three-dimensional path planning of unmanned aerial vehicle based on improved whale optimization algorithm according to claim 4 is characterized in that: In step 3.5, the control parameter a is updated and the coefficient vector in the mathematical model of search and foraging is calculated. and Specifically: Step 3.5.1: Define the mathematical model of search and foraging: Where: is a whale individual position vector randomly selected from the current whale group, X(t) is the current whale individual position vector, Represents the distance vector between the current whale individual and the randomly selected whale individual, coefficient vector and is defined as: Where: is the control parameter vector, is a random vector distributed between [0,1]; Step 3.5.2: Introduce a nonlinear convergence factor and update the control parameter a to: Where: T is the maximum number of iterations, t is the current number of iterations, is the initial control parameter vector, It is the modulus of the new control parameter vector generated by the change of the initial control parameter vector with the number of iterations t; Step 3.5.3: Based on the updated control parameter vector Calculate the coefficient vector in the mathematical model of search and foraging and 6. The method for three-dimensional path planning of unmanned aerial vehicle based on improved whale optimization algorithm according to claim 5 is characterized in that: In step 3.6, the mathematical model of shrinkage and encirclement is: Where: is the individual whale position vector with the best objective function value in the current whale group, is the encirclement step size of the current whale individual, the coefficient vector The smaller the whale is, the smaller its stride length becomes; The mathematical model of the spiral update position is as follows: Where: Represents the distance vector between the current whale individual and the best position whale; the constant coefficient b determines the shape of the spiral line when the whale individual spirals forward. When b is 1, it is a common logarithmic spiral, and l is [-1,1] A random number between ω(t) and ω(t) is the adaptive inertia weight.
7. The method for three-dimensional path planning of unmanned aerial vehicle based on improved whale optimization algorithm according to claim 5 is characterized in that: In step 3.7, the update position is updated by dynamic mutation disturbance, specifically: Gaussian mutation weights and Cauchy mutation weights are introduced. In the early stage of iteration, the weight of Cauchy mutation is greater than that of Gaussian mutation; in the later stage of iteration, the weight of Gaussian mutation is greater than that of Cauchy mutation. The formula is as follows: Where: is the current iteration weight, is the Gaussian variation weight, is the Cauchy mutation weight, T is the maximum number of iterations, and t is the current number of iterations.
Citation Information
Cited By
Digital intelligent substation parallel control management system based on artificial intelligence
CN119093602A
Unmanned aerial vehicle refined route planning method and device and electronic equipment
CN119512182A
Precise regulation and control optimization method for respiratory flow based on PID
CN119909276A
Three-dimensional path multi-stage planning method for low-altitude unmanned aerial vehicle
CN120403660A
Multi-unmanned aerial vehicle path planning method and system based on quantum song eagle mechanism, and storage medium
CN120780020A
