Three-dimensional space unmanned aerial vehicle path planning method based on swarm intelligence algorithm

Abstract

The invention discloses a three-dimensional space unmanned aerial vehicle path planning method based on a swarm intelligence algorithm. The method comprises the following steps: S1, defining an initial boundary and sampling in the initial boundary to generate an initial population containing a plurality of candidate paths; s2, constructing a target function according to the optimization target; fitness values of the candidate paths are calculated based on the target function and the position function, and the candidate path with the high fitness value is selected as a global optimal solution; s3, the maximum number of iterations, the maximum number of iterations and the number of iterations of dynamic updating are set, and the specific number of iterations and the target probability are preset; introducing two dynamic parameters and designing a boundary adjustment coefficient, an adaptability factor and a dynamic probability which change along with the current iteration number; according to the method, autonomous navigation and path optimization of the unmanned aerial vehicle in a complex environment are realized by simulating the self-organization and distributed decision-making mechanism of group behaviors in nature, the path planning calculation speed can be effectively improved, the distance is reduced, and the adaptive capacity to environmental changes is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle path planning, in particular to a three-dimensional space unmanned aerial vehicle path planning method based on swarm intelligence algorithm. BACKGROUND

[0002] In recent years, with the continuous progress and maturity of intelligent algorithms and deep learning, we have witnessed its wide application in the fields of logistics distribution, agricultural monitoring, disaster relief, etc., marking the trend of swarm intelligence optimization algorithm more and more obvious. In particular, in the field of unmanned aerial vehicle path planning, the application of swarm intelligence algorithm has become a popular trend. However, in the field of swarm intelligence algorithm for realizing unmanned aerial vehicle path planning, especially in complex environments, some path planning methods often face problems such as high computational complexity, poor adaptability, etc., which are still weak and have not yet formed an effective technical system.

[0003] There are three main problems in the existing method for unmanned aerial vehicle path planning: (1) In natural environment, the obstacles and mountainous environment are fluctuating, the urban residential environment is complex and changeable, and the existing algorithm is difficult to effectively handle the dynamic changes of obstacles and changing trajectory conditions in the mountainous environment, resulting in inaccurate generated routes; (2) The existing algorithm converges slowly, which will make the unmanned aerial vehicle work in a short time, resulting in unsafe generated routes; (3) In the case of shortening the route and fast convergence, the route safety is multi-aspectally constrained, and the problems need to be considered comprehensively. There is an urgent need for a more effective path planning method to solve the above problems. SUMMARY

[0004] The purpose of the present application is to provide a three-dimensional space unmanned aerial vehicle path planning method based on swarm intelligence algorithm, which simulates the self-organization and distributed decision mechanism of group behavior in nature, realizes the autonomous navigation and path optimization of unmanned aerial vehicle in complex environment, and can effectively improve the path planning calculation speed, reduce the route distance and enhance the adaptability to environmental changes.

[0005] The technical scheme adopted by the present application is: a three-dimensional space unmanned aerial vehicle path planning method based on swarm intelligence algorithm, comprising the following steps:

[0006] S1, defining an initial boundary and sampling in the initial boundary to generate an initial population containing multiple candidate paths;

[0007] S2, constructing a target function according to an optimization target; calculating the fitness value of the candidate path based on the target function and the position function, and selecting the candidate path with high fitness value as the global optimal solution;

[0008] S3, setting the maximum iteration number, the maximum iteration amount and the dynamically updated iteration number, and pre-setting a specific iteration number and a target probability;

[0009] Two dynamic parameters are introduced and a boundary adjustment coefficient, an adaptive factor and a dynamic probability varying with the current iteration number are designed;

[0010] S4, judging whether the boundary adjustment coefficient reaches the specific iteration number; if yes, boundary updating and resampling are performed, the objective function is updated, and S7 is entered; if no, S5 is entered;

[0011] S5, switching global search and local search according to the stage where the current iteration number is located, and introducing the adaptive factor to balance the search step length in global search and local search;

[0012] In local search, judging whether the designed dynamic probability reaches the target probability; if yes, the dynamic probability is used as a coefficient to update the position function and the candidate path, and S7 is entered; if no, the objective function and the global optimal solution are updated, and S7 is entered;

[0013] S6, judging whether the current iteration number is equal to the dynamically updated iteration number; if yes, the objective function is updated, the suboptimal solution of the candidate path is selected, and S7 is entered; if no, the convergence curve is recorded, and S7 is entered;

[0014] S7, judging whether the current iteration number reaches the maximum iteration amount; if no, returning to S4 to continue the iteration loop; if yes, S8 is entered;

[0015] S8, taking the corresponding candidate path as the output result.

[0016] As a preferred scheme, the two dynamic parameters are defined as:

[0017]

[0018] Wherein, d1 and d2 are the two dynamic parameters respectively; t is the current iteration number; Max_Iter is the maximum iteration amount; exp(·) is the natural exponential function.

[0019] As a preferred scheme, the adaptive factor a is defined as:

[0020]

[0021] Wherein, ∈ is a minimum normal number.

[0022] As a preferred scheme, the boundary adjustment coefficient CM is defined as:

[0023]

[0024] Wherein, rand() represents a random function between 0 and 1.

[0025] As a preferred solution, in S5, the updated position function is updated with a dynamic probability as a coefficient:

[0026]

[0027] Wherein, X new 2(i,j) is the updated position function; is the dynamic probability; alpha is the adaptive factor; Destination_position(j) is the target position; X(i,j) is the current position.

[0028] As a preferred solution, the iteration number CE NEW is dynamically updated by using the following formula:

[0029]

[0030] Wherein, CE new is the iteration number of dynamic update; CE is the adjustable control parameter; T max is the maximum iteration number.

[0031] As a preferred solution, the steps of switching global search and local search according to the stage where the current iteration number is located in S5 are: judging the size relationship between the current iteration number t and the maximum iteration number T max ; if t<=T max , global search is performed; if t>T max , local search is performed.

[0032] As a preferred solution, the sampling method is optimal Latin hypercube sampling.

[0033] Compared with the prior art, the beneficial effects of the present application are:

[0034] The present application provides a three-dimensional space unmanned aerial vehicle path planning method based on swarm intelligence algorithm, and the innovation of the model is that the trajectory planning problem of the unmanned aerial vehicle is converted into an optimization problem, so that the unmanned aerial vehicle converges quickly and optimizes the multi-objective consideration of the aircraft distance within a limited time, in order to make the model optimal, the algorithm innovation is that a plurality of strategies are fused, the population initialization is optimized, a new adaptive factor is introduced, and the traditional local search strategy is changed, so that the algorithm is more suitable for the unmanned aerial vehicle path planning problem in the three-dimensional environment. In addition, the benchmark function is introduced, and has strong computing power. In the simulation experiment, benchmark function test, path planning test and initialization simulation test are carried out, and the competitiveness and feasibility of the IETO algorithm in solving such high-dimensional complex optimization problems are verified from multiple angles. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0036] Figure 1 is the simulation diagram of the adaptive alpha factor iterative automatic adjustment search strategy in the embodiments of the present application;

[0037] Figure 2 is the contrast diagram of the optimal Latin hypercube sampling in the embodiments of the present application;

[0038] Figure 3 is the contrast planar diagram of the three-dimensional path planning swarm intelligence algorithm in the embodiments of the present application;

[0039] Figure 4 is the contrast overhead view of the three-dimensional path planning swarm intelligence algorithm in the embodiments of the present application;

[0040] Figure 5 is the CEC2017 ten kinds of benchmark function engineering problem diagram in the embodiments of the present application;

[0041] Figure 6 is the algorithm convergence curve of F1, F2 two kinds of benchmark function engineering problems in the embodiments of the present application;

[0042] Figure 7 is the algorithm convergence curve of F3, F4 two kinds of benchmark function engineering problems in the embodiments of the present application;

[0043] Figure 8 is the algorithm convergence curve of F5, F6 two kinds of benchmark function engineering problems in the embodiments of the present application;

[0044] Figure 9 is the algorithm convergence curve of F7, F8 two kinds of benchmark function engineering problems in the embodiments of the present application;

[0045] Figure 10 is the algorithm convergence curve of F9, F10 two kinds of benchmark function engineering problems in the embodiments of the present application;

[0046] Figure 11 is the algorithm contrast box plot in the path planning in the embodiments of the present application. DETAILED DESCRIPTION

[0047] In the following, the present application will be specifically described through exemplary embodiments. However, it should be understood that the elements, structures and features in one embodiment can also be beneficially combined into other embodiments without further description.

[0048] It should be noted that, unless otherwise defined, the technical terms or scientific terms used herein should be understood as the general meaning understood by those skilled in the art to which the present application belongs. The "one", "a" or "the" and the like similar words used in the patent application description and claims of the present application do not represent the quantity limitation, but indicate that there is at least one; The "first", "second" and "third" used herein should not be regarded as a limitation on the order of the components, but only to distinguish different components; "Include" or "contain" and the like similar words indicate that the elements or objects appearing before "include" or "contain" cover the elements or objects listed after "include" or "contain" and their equivalents, but do not exclude other elements or objects with the same function.

[0049] In order to more clearly describe the specific structure of the three-dimensional space unmanned aerial vehicle path planning method based on swarm intelligence algorithm, combined with the accompanying drawings Figures 1-11 The embodiment is described:

[0050] As Figures 1-11 shown, a three-dimensional space unmanned aerial vehicle path planning method based on swarm intelligence algorithm, characterized in that: comprising the following steps:

[0051] S1, define the initial boundary and sample in the initial boundary to generate an initial population containing a plurality of candidate paths;

[0052] The sampling method is optimal Latin hypercube sampling;

[0053] S101, before entering iteration, the total population is assigned by random initialization, and the initial individual number is calculated as follows:

[0054]

[0055] Start its work by randomly generating a set of potential populations; here is an Nxd matrix, for example, i 1.1 represents an individual, each column represents a dimension, and the whole I represents the whole population, which is represented by a specific matrix;

[0056] S102, after initialization sampling, define the upper and lower boundaries of the initialization, increase the constraints, and the formula is as follows:

[0057] i ij = Low j +(Up j -Low j )·rand()

[0058] Where rand() represents falling in the interval [0,1], and rand() is a random function between 0 and 1; Up j and Lowj is the lower and upper bound of the jth dimension;

[0059] S103, complete the constraint condition, set the algorithm to control the parameters, define the i i is a 1xd vector, representing a single random initialization formula as follows:

[0060]

[0061] S104, in order for each element to be calculated by the above equation, and to ensure that the value of the individual in each dimension is within the specified range, in order to simplify the formula, we calculate i i vector to add boundary constraints to speed up the convergence of initialization, the formula is as follows:

[0062]

[0063] S105, we need to extract the sample points from it;

[0064] Split interval: divide each dimension into equal parts, for D dimensions, the range of the ith interval is:

[0065]

[0066] Generate preliminary samples: select a random point in each interval; Specifically, for the x n,d th dimension, the coordinates in the interval can be expressed as:

[0067]

[0068] where x n,d is the sample point coordinates; LB d is the lower bound of dimension d; n is the sample index; u n,d is a random offset in the interval; UB d is the upper bound of dimension d; N is the total number of samples;

[0069] S106, the position of the preliminary sampling point, to dynamically change the conditional coefficient of variation between the maximum sampling point distance and the minimum sampling point distance; The formula is as follows:

[0070] Maximin Distance = min 1≤i,j≤N ‖x i -x j ‖

[0071] where x i is the ith sample point; x j is the jth sample point;

[0072] Coefficient of variation objective function:

[0073]

[0074] wherein is the variance; is the square of the mean;

[0075] After the set condition is reached, the optimal solution is found, and the sampling is completed.

[0076] S2, constructing a target function according to an optimization target; calculating a fitness value of a candidate path based on the target function and a position function, and selecting a candidate path with a high fitness value as a global optimal solution;

[0077] The fitness value is used to evaluate the pros and cons of a path solution, and a higher fitness value represents a shorter path length and a smoother path.

[0078] S3, setting a maximum iteration number T max , a maximum iteration amount Max_Iter and a dynamically updated iteration number CE new , a preset specific iteration number and a target probability;

[0079] Two dynamic parameters (d1, d2) are introduced, and a boundary adjustment coefficient CM, an adaptability factor a and a dynamic probability that change with the current iteration number t are designed;

[0080]

[0081] wherein d1 and d2 are two dynamic parameters respectively; t is the current iteration number; Max_Iter is the maximum iteration amount; exp(·) is the natural exponential function; after introducing the two dynamic parameters d1 and d2, the scalar function can form an asymmetric state space within the value range through dynamic jitter, thereby improving the search strategy;

[0082] The initial update position is defined as follows, which creates conditions for subsequent local search:

[0083]

[0084] S4, judging whether the boundary adjustment coefficient CM reaches the specific iteration number; if yes, boundary updating and resampling are performed, the target function is updated, S7 is entered to iteratively update the position of each individual and record the convergence curve; if not, S5 is entered.

[0085] The boundary adjustment coefficient CM is:

[0086]

[0087] Wherein, ∈ is a minimum normal number; rand() represents a random function between 0 and 1; CM is obtained according to the current iteration number t in each iteration, which is a parameter with randomness and decaying from small to large; it is determined by CM whether to search outward or to search in a small range around the current optimal solution, so as to avoid the population into a dead cycle;

[0088] S5, switching global search and local search according to the stage where the current iteration number t is located, and introducing adaptive factor α in global search and local search to balance the search step length; judging the size relationship between the current iteration number t and the maximum iteration number T max ; if t≤T max , global search is performed; if T max >T, local search is performed;

[0089] The calculation formula of adaptive factor α is:

[0090]

[0091] Wherein, ∈ is a minimum normal number; the probability calculated by α is used to perform local search on the current solution, and a balance is obtained between global search and convergence speed;

[0092] In local search, it is judged whether the designed dynamic probability reaches the target probability (for example, the target probability can be set to 30%, 70%); if yes, the dynamic probability is used as a coefficient to update the position function X(i,j) and the candidate path, and S7 is entered; if no, the objective function and the global optimal solution are updated, and S6 is entered;

[0093] Wherein, the updated position function of global search is:

[0094] X new 1(i,j)=X(i,j)+(-1) round(rand()) ×3×(rand()×α×(Destination_position(j)-X(i,j)))

[0095] Wherein, the updated position function with dynamic probability as a coefficient is:

[0096]

[0097] Wherein, X new 1(i,j) is the updated position function; X new 2(i,j) is the updated position function; is the dynamic probability; α is the adaptive factor; Destination_position(j) is the target position, the position of the currently known global optimal solution; X(i,j) is the current position, the starting point of calculation;

[0098] In this step, the local search is triggered in a probabilistic manner, the local perturbation step is generated, and the new position is tried and evaluated, and if it is satisfied, the better solution is accepted;

[0099] S6, judge whether the current iteration number t is equal to the dynamically updated iteration number CE new ; if yes, update the objective function, select the suboptimal solution of the candidate path, and enter S7; if no, record the convergence curve (the convergence curve is used to analyze the algorithm performance), and enter S7;

[0100] The dynamically updated iteration number CE NEW is generated by the following formula:

[0101]

[0102] Wherein, CE new is the dynamically updated iteration number; CE is an adjustable control parameter; T max is the maximum iteration number; this design makes the update of the suboptimal solution triggered at certain specific iteration points, and the time of the next triggering is controlled by the dynamic formula, avoiding premature convergence or search stagnation caused by fixed period, and is used to guide the subsequent search direction;

[0103] S7, judge whether the current iteration number t reaches the maximum iteration amount Max_Iter; if no, return to S4 to continue the iteration loop; if yes, enter S8;

[0104] S8, take the corresponding candidate path as the output result, and obtain the planned optimal path.

[0105] In order to test the ability of the algorithm, a series of experiments are carried out to test and verify the reliability and performance of the algorithm; in order to keep the integrity and accuracy of the research results, a total of 10 benchmark functions ( Figure 5 ) will be used; these results will provide a clear and accurate view of the ability and robustness of the algorithm;

[0106] In order to keep the consistency of the comparison experiment, all algorithms use the same basic parameter set, including the total size n = 100, the maximum iteration number T = 100 and the running number m = 30; the images for qualitative analysis include the average convergence curve and the box plot; the performance indicators for quantitative analysis include the best value (BV), the average value (MV), the standard deviation (STD), the root mean square error (RMSE) and the Friedman mean rank (FMR); we prove that the algorithm can achieve the best result in many cases, and the benchmark test result is ( Figures 6-10 );

[0107] In order to measure the efficiency of the proposed optimization algorithm, an extensive comparative study of other optimization algorithm choices was conducted using 4 classic benchmark functions as the test basis; the results of the algorithm were juxtaposed with those of four other optimization algorithms, namely the Fishing Optimization Algorithm (CFOA), the Genetic Algorithm (GA), the Whale Optimization Algorithm (WOA), and the Exponential Triangular Optimization (ETO). These algorithms were selected based on their popularity and novelty, and they all serve similar benchmark functions, ensuring the reliability and superiority of IETO (the present invention). In these experiments, parameters such as the number of search agents (N), the number of iterations, and the number of dimensions (Dim) were kept at the previous values; each algorithm was executed independently 30 times in a single function with parameters set to N=30 and a maximum number of iterations of 1000, and the actual simulation was as follows Figure 3 、 Figure 4 As shown, the optimization results were not just random; the results of these runs were analyzed to determine the best fit, average and standard deviation (Std) of the fit values ​​obtained in these runs; in addition, a ranking test was performed to rank the best fit values ​​produced by the algorithms and finally the average values ​​were ranked to calculate the IETO ranking and compare the algorithms. The test results are shown in Figure 11 shown.

[0108] Test results show that the proposed algorithm significantly improves upon the original model in all aspects. Its convergence speed is 7.2% faster than that of mainstream algorithms, and its function test is 6.8% higher than the best performance. It is superior to other UAV path planning algorithms.

[0109] Parts not described in detail in the above embodiments are prior art.

[0110] It should be noted that although the present invention has been described with reference to the above embodiments, the present invention may also have other various embodiments. Without departing from the spirit and scope of the present invention, it is obvious that those skilled in the art may make various corresponding changes and modifications to the present invention, and such changes and modifications shall fall within the scope of protection of the appended claims and their equivalents.

Claims

1. A three-dimensional UAV path planning method based on swarm intelligence algorithm, characterized by: The following steps are involved: S1. Define the initial boundary and sample within the initial boundary to generate an initial population containing multiple candidate paths; S2. Construct an objective function based on the optimization goal; Calculate the fitness value of the candidate path based on the objective function and the position function, and select the candidate path with the highest fitness value as the global optimal solution; S3, setting the maximum number of iterations, the maximum iteration amount and the number of iterations to be dynamically updated, and presetting a specific number of iterations and target probability; Two dynamic parameters are introduced and the boundary adjustment coefficient, adaptability factor and dynamic probability that change with the current iteration number are designed; S4, determining whether the boundary adjustment coefficient reaches the specific number of iterations; if so, performing boundary update and resampling, updating the objective function, and proceeding to S7; If not, go to S5; S5. Switching between global search and local search according to the stage of the current iteration number, introducing the adaptive factor to balance the search step in both global search and local search; In the local search, determine whether the designed dynamic probability reaches the target probability; if so, use the dynamic probability as a coefficient to update the position function and candidate paths, and enter S7; if not, update the target function and the global optimal solution, and enter S6; S6. Determine whether the current number of iterations is equal to the dynamically updated number of iterations; if so, update the objective function, select the suboptimal solution of the candidate path, and proceed to S7; If not, record the convergence curve and go to S7; S7, determine whether the current number of iterations reaches the maximum number of iterations; if not, return to S4 to continue the iteration loop; if so, enter S8; S8. Take the corresponding candidate path as the output result.

2. The three-dimensional UAV path planning method based on swarm intelligence algorithm according to claim 1 is characterized by: The two dynamic parameters are defined as: Where d1 and d2 are two dynamic parameters; t is the current number of iterations; Max_Iter is the maximum number of iterations; and exp(·) is the natural exponential function.

3. The three-dimensional UAV path planning method based on swarm intelligence algorithm according to claim 2 is characterized by: The adaptability factor α is defined as: Among them, ∈ is a minimal positive constant.

4. The three-dimensional UAV path planning method based on swarm intelligence algorithm according to claim 2, characterized in that: The boundary adjustment coefficient CM is defined as: Among them, rand() represents a random function between 0 and 1.

5. The three-dimensional UAV path planning method based on swarm intelligence algorithm according to claim 3 is characterized by: In S5, the updated position function with dynamic probability as the coefficient is: Among them, X new2 (i, j) is the updated position function; is the dynamic probability; α is the adaptability factor; Destination_position(j) is the target position; X(i,j) is the current position.

6. The three-dimensional UAV path planning method based on swarm intelligence algorithm according to claim 1, characterized in that: Dynamically updated number of iterations CE NEW It is generated using the following formula: Among them, CE new is the number of iterations of dynamic update; CE is the adjustable control parameter; T max is the maximum number of iterations.

7. The three-dimensional UAV path planning method based on swarm intelligence algorithm according to claim 1, characterized in that: The steps of switching between global search and local search according to the stage of the current number of iterations in S5 are: judging the current number of iterations t and the maximum number of iterations T max The size relationship; if t≤T max , then perform a global search; if t>T max , then perform a local search.

8. The three-dimensional space UAV path planning method based on swarm intelligence algorithm according to claim 1, characterized in that: The sampling method is optimal Latin hypercube sampling.

Images