Multi-unmanned aerial vehicle task allocation method based on distance selection algorithm

Abstract

The invention relates to a multi-unmanned aerial vehicle task allocation method based on a distance selection algorithm. The method comprises the steps of constructing an initial task arrangement model, calculating a task point dynamic clustering center, constructing a task point dynamic allocation model of the distance selection algorithm, executing a task route space relation evaluation function, adjusting a task allocation scheme and optimizing flight task allocation. The method is a flight task allocation problem in consideration of task space distribution characteristics, task allocation rationality, economical efficiency and the like. Example results show that the most economical and most reasonable multi-unmanned aerial vehicle task allocation can be realized under the multi-unmanned aerial vehicle and multi-task conditions by using the distance selection algorithm, and the method can provide an effective cooperation mode for civil unmanned aerial vehicles and improve the task execution efficiency and economic benefits at the same time.

Description

Technical Field

[0001] The present invention belongs to the field of unmanned aerial vehicle task allocation, and relates to designing an optimal allocation scheme using constraints such as economic performance and rationality of task allocation, and is specifically a multi-unmanned aerial vehicle task allocation method based on a distance selection algorithm. Background Art

[0002] So far, the task allocation and route generation linkage model has not been formed in the task allocation research of multi-UAV coordinated rapid emergency monitoring in emergency events under complex dynamic environments at home and abroad, and there is no guidance plan for rapid response to emergencies. Task allocation alone is affected by various types of factors, making the modeling and solution of task allocation and route planning problems increasingly difficult. How to use the spatial characteristics of the task points themselves to plan efficient and reasonable task allocation and flight route linkage plans so that the overall task execution efficiency can achieve the best results still requires in-depth research and exploration. The solution to this problem has extremely important practical significance for broadening the application of civil UAVs in military and military application efficiency. Summary of the invention

[0003] A multi-UAV task allocation method based on a distance selection algorithm is characterized by the following implementation steps:

[0004] (1) Construct the initial task arrangement model z, which is mathematically described as follows:

[0005] z=[x,y] (1)

[0006] x=[x1;x2;...xi;...xn]

[0007] y=[y1;y2;...yi;...yn]

[0008] In the above formula, z, x, y represent the initial task arrangement model values ​​to be assigned and their corresponding horizontal coordinate vectors and vertical coordinate vectors, in meters; x1, y1, x2, y2, xi, yi, xn, yn represent the horizontal coordinates and vertical coordinates of the 1st, 2nd, i and nth task points in the planned airspace, in kilometers; “[]”, “,” and “;” represent the symbols of the matrix symbol, column and row spacing, respectively;

[0009] (2) Calculate the dynamic cluster center of the task point:

[0010] Using formulas (2) to (3), we can get the dynamic clustering center of the task points of the k-means clustering algorithm:

[0011] [index_cluster, cluster]=kmeans_func(z, cluster_num) (2)

[0012] meancluster=mean(cluster) (3)

[0013] In formulas (2) to (3), cluster_num, kmeans_func, index_cluster, and cluster are the number of clustered categories, the k-means clustering function, and the categories and cluster center sets corresponding to the task points obtained after clustering; mean and meancluster represent the function for calculating the average value and the abscissa and ordinate of the average cluster center point obtained after calculation, respectively, in kilometers;

[0014] (3) Constructing a dynamic allocation model of task points for distance selection algorithm:

[0015] First, use formulas (4) to (7) to obtain the distance Dy from each task point to the cluster center point:

[0016] X = [meancluster; z] (4)

[0017] D=pdist(X) (5)

[0018] dl=length(X) (6)

[0019] Dy=D(1:dl-1) (7)

[0020] In formulas (4) to (7), X, pdist, and D are the coordinate matrices of the constructed cluster center points and task points, the function for calculating the distance between the coordinate points, and the distance matrix calculated; length and dl are the function for calculating the maximum number of dimensions of the matrix and its calculation results, respectively; “:” and Dy are the colon calculation symbol (used to generate equidistant series, which can be omitted when the step length is 1) and the distance from the cluster center point to each task point calculated;

[0021] Secondly, the task point dynamic allocation model of the distance selection algorithm is obtained by using formulas (8) to (11) to allocate the task point set of each drone:

[0022] [Ds, Di] = sort (Dy, 'option') (8)

[0023] meanfrn=fix(dl / flyn) (9)

[0024] X=X(2:end) (10)

[0025] frwd(i)=X(1+meanfrn*(i-1):meanfrn*i), i=1, 2,..., flyn (11)

[0026] In formulas (8) to (11), sort, option, Ds, and Di are the sorting function, the sorting method, the distance after sorting, and the index corresponding to the original matrix, respectively; flyn, fix, meanfrn, and end are the number of drones, the integer function and the number of task points to be assigned to each drone calculated by it, and the last element of the array, respectively; frwd(i) is the set of task points assigned to the i-th drone and its horizontal and vertical coordinate sets;

[0027] Finally, the routes formed by each UAV mission point set are obtained using formulas (12) to (14):

[0028] Cfp=[xq,yq;Cf;xz,yz] (12)

[0029] [cx, cy]=yiqunhanshu(Cfp) (13)

[0030] frwdn(i)=[cx;cy]′, i=1, 2,..., flyn (14)

[0031] In formulas (12) to (14), xq, yq, xz, yz, Cf, and cfp are the horizontal and vertical coordinates of the starting point of the UAV flight, the horizontal and vertical coordinates of the mission point converted to double precision type, and the horizontal and vertical coordinates of the track points that form the route. yiqunhanshu, cx, and cy are the ant colony sorting functions that ensure the economic efficiency of the UAV flight and the horizontal and vertical coordinates of the track points of the i-th UAV obtained after sorting.

[0032] (4) Evaluation function of the spatial relationship of the mission route:

[0033] Using formulas (15) to (19), we can obtain the set of UAV track points after the route spatial relationship evaluation function is adjusted:

[0034] [fic, fixc]=liniset(fij, fijx) (15)

[0035] fij=[frwdn(i,j),frwdn(i,j+1)] (16)

[0036] fixj=[frwdn(i+1,j), frwdn(i+1,j+1) (17)

[0037] frwdn(i, j+1)=fij (18)

[0038] frwdn(i+1,j+1)=fixj (19)

[0039] In formulas (15) to (19), fij, fijx, linset, fic, and fixc are the horizontal and vertical coordinates of the jth and j+1th track points of the ith UAV, the evaluation function of the route spatial relationship and the function for obtaining the number of statistical elements, and the horizontal and vertical coordinates of the jth and j+1th track points of the ith UAV to be adjusted, respectively. frwdn(i, j) and frwdn(i, j+1) are the horizontal and vertical coordinates of the jth track point of the ith UAV and the horizontal and vertical coordinates of the j+1th track point, respectively. frwdn(i+1, j) and frwdn(i+1, j+1) are the horizontal and vertical coordinates of the jth track point of the ith UAV and the horizontal and vertical coordinates of the j+1th track point, respectively.

[0040] (5) Adjust the task allocation plan and optimize the flight task allocation:

[0041] First, use formulas (20) to (21) to obtain the adjusted UAV track point set:

[0042] [cx1,cy1]=yctsp(frwdn) (20)

[0043] ltn(i)=[cx1 cy1] (21)

[0044] In formulas (20) and (21), yctsp, cx1, and cy1 are the genetic traveling salesman route optimization function and its adjusted horizontal and vertical coordinate sequences, respectively, and ltn is the adjusted route set. This method achieves economic optimization of flight mission allocation and provides an effective method for improving the efficiency of UAV mission completion.

[0045] Compared with the prior art, the present invention has the following advantages:

[0046] a. Achieve optimal task allocation that meets the constraints of UAV flight efficiency and economy. The combination of distance selection algorithm and cluster analysis algorithm provides an effective method to improve the allocation of multiple UAVs and multiple tasks, as well as flight efficiency and economic performance.

[0047] b. Dynamic adjustment of multi-UAV track points using genetic traveling salesman algorithm. The dynamically adjusted track points are generated using the spatial relationship evaluation function of the mission routes of two adjacent UAVs, which realizes the characteristics of rapid adjustment of adjacent UAV track points to complete multi-UAV collaborative emergency monitoring. It allocates tasks and generates and optimizes routes in complex dynamic emergency environments, which plays an important role in quickly and efficiently completing emergency monitoring tasks. DETAILED DESCRIPTION

[0048] The technical solution of the present invention is further described below in conjunction with an embodiment. In the embodiment, there are 5 drones flying, the coordinates of the lower left corner of the planned airspace are [0km, 0km, 50m], and the coordinates of the upper right corner are [100km, 100km, 50m]. At (90km, 33km), (35km, 71km), (52km, 62km), (83km, 34km), (64km, 49km), (24km, 48km), (49km, 46km), (92km, 69km), (99km, 56km), (45km, 59km), (19km, 28km), (38km, 14km), (1km, 55km), (71km, 4 1km), (56km, 39km), (97km, 78km), (63km, 23km), (32km, 99km), (3km, 68km), (34km, 30km), (33km, 87km), (55km, 85km), (75km, 43km), (84km, 88km), (53km, 2km), (15km, 47km), (88km, 50km), (66km, 77km), (41km , 22km), (51km, 76km), (39km, 94km), (78km, 11km), (67km, 80km), (65km, 51km), (25km, 6km), (40km, 7km) , (77km, 72km), (13km, 36km), (69km, 90km), (29km, 96km), (14km, 44km), (54km, 61km), (87km, 70km), (47 There are 50 mission points at (km, 60km), (44km, 75km), (58km, 74km), (8km, 63km), (68km, 40km), (81km, 81km), (31km, 4km); the performance parameters of the UAV are: maximum flight time 35 hours, minimum cruising speed 300 km / h, range greater than 3000 km; the coordinates of the starting point and the target point are (0km, 0km, 50m) and (100km, 100km, 50m) respectively;

[0049] (1) Construct the initial task arrangement model z, which is mathematically described as follows:

[0050] z=[x,y] (1)

[0051] x=[90;35;52;83;64;24;49;92;99;45;19;38;1;71;56;97;63;...32;3;34;33;55;75;84;53 ;15;88;66;41;51;39;78;67;65;25;40;77;...13;69;29;14;54;87;47;44;58;8;68;81;31]

[0052] y=[33;71;62;34;49;48;46;69;56;59;28;14;55;41;39;...78;23;99;68;30;87;85;43;88 ;2;47;50;77;22;76;94;11;80;...51;6;7;72;36;90;96;44;61;70;60;75;74;63;40;81;4]

[0053] In the above formula, z, x, y represent the initial task arrangement model values ​​to be assigned and their corresponding horizontal coordinate vectors and vertical coordinate vectors, in meters; x1, y1, x2, y2, xi, yi, xn, yn represent the horizontal coordinates and vertical coordinates of the 1st, 2nd, i and nth task points in the planned airspace, in kilometers; “[]”, “,” and “;” represent the symbols of the matrix symbol, column and row spacing, respectively;

[0054] (2) Calculate the dynamic cluster center of the task point:

[0055] Using formulas (2) to (3), we can get the dynamic clustering center of the task points of the k-means clustering algorithm:

[0056] [index_cluster, cluster]=kmeans_func(z, 2) (2)

[0057] meancluster=mean(cluster) (3)

[0058] In formulas (2) to (3), cluster_num, kmeans_func, index_cluster, and cluster are the number of clustered categories, the k-means clustering function, and the categories and cluster center sets corresponding to the task points obtained after clustering; mean and meancluster represent the function for calculating the average value and the abscissa and ordinate of the average cluster center point obtained after calculation, respectively, in kilometers;

[0059] (3) Constructing a dynamic allocation model of task points for distance selection algorithm:

[0060] First, use formulas (4) to (7) to obtain the distance Dy from each task point to the cluster center point:

[0061] X = [meancluster; z] (4)

[0062] D=pdist(X) (5)

[0063] dl=length(X) (6)

[0064] Dy=D(1:dl-1) (7)

[0065] In formulas (4) to (7), X, pdist, and D are the coordinate matrices of the constructed cluster center points and task points, the function for calculating the distance between the coordinate points, and the distance matrix calculated; length and dl are the function for calculating the maximum number of dimensions of the matrix and its calculation results, respectively; “:” and Dy are the colon calculation symbol (used to generate equidistant series, which can be omitted when the step length is 1) and the distance from the cluster center point to each task point calculated;

[0066] Secondly, the task point dynamic allocation model of the distance selection algorithm is obtained by using formulas (8) to (11) to allocate the task point set of each drone:

[0067] [Ds, Di] = sort (Dy, 'descend') (8)

[0068] meanfrn=fix(dl / flyn) (9)

[0069] X=X(2:end) (10)

[0070] frwd(i)=X(1+meanfrn*(i-1):meanfrn*i), i=1, 2,..., 5 (11)

[0071] In formulas (8) to (11), sort, option, Ds, and Di are the sorting function, the sorting method, the distance after sorting, and the index corresponding to the original matrix, respectively; flyn, fix, meanfrn, and end are the number of drones, the integer function and the number of task points to be assigned to each drone calculated by it, and the last element of the array, respectively; frwd(i) is the set of task points assigned to the i-th drone and its horizontal and vertical coordinate sets;

[0072] Finally, the routes formed by each UAV mission point set are obtained using formulas (12) to (14):

[0073] Cfp=[1,1;Cf;100,100] (12)

[0074] [cx, cy]=yiqunhanshu(Cfp) (13)

[0075] frwdn(i)=[cx;cy]′,i=1,2,...,5 (14)

[0076] In formulas (12) to (14), xq, yq, xz, yz, Cf, and cfp are the horizontal and vertical coordinates of the starting point of the UAV flight, the horizontal and vertical coordinates of the mission point converted to double precision type, and the horizontal and vertical coordinates of the track points that form the route. yiqunhanshu, cx, and cy are the ant colony sorting functions that ensure the economic efficiency of the UAV flight and the horizontal and vertical coordinates of the track points of the i-th UAV obtained after sorting.

[0077] (4) Evaluation function of the spatial relationship of the mission route:

[0078] Using formulas (15) to (19), we can obtain the set of UAV track points after the route spatial relationship evaluation function is adjusted:

[0079] [fic, fixc]=liniset(fij, fijx) (15)

[0080] fij=[frwdn(i,j),frwdn(i,j+1)] (16)

[0081] fixj=[frwdn(i+1,j), frwdn(i+1,j+1)] (17)

[0082] frwdn(i, j+1)=fij (18)

[0083] frwdn(i+1,j+1)=fixj (19)

[0084] In formulas (15) to (19), fij, fijx, linset, fic, and fixc are the horizontal and vertical coordinates of the jth and j+1th track points of the ith UAV, the evaluation function of the route spatial relationship and the function for obtaining the number of statistical elements, and the horizontal and vertical coordinates of the jth and j+1th track points of the ith UAV to be adjusted, respectively. frwdn(i, j) and frwdn(i, j+1) are the horizontal and vertical coordinates of the jth track point of the ith UAV and the horizontal and vertical coordinates of the j+1th track point, respectively. frwdn(i+1, j) and frwdn(i+1, j+1) are the horizontal and vertical coordinates of the jth track point of the ith UAV and the horizontal and vertical coordinates of the j+1th track point, respectively.

[0085] (5) Adjust the task allocation plan and optimize the flight task allocation:

[0086] First, use formulas (20) to (21) to obtain the adjusted UAV track point set:

[0087] [cx1,cy1)=yctsp(frwdn) (20)

[0088] ltn(i)=[cx1 cy1] (21)

[0089] In formulas (20) to (21), yctsp, cx1, cy1 are the genetic traveling salesman route optimization function and its adjusted horizontal and vertical coordinate sequences, respectively, and ltn is the adjusted route set.

[0090] The experimental results are as follows: the maximum range of the assigned drone is 395.98 kilometers, the planning time is 6.5 seconds, and an average of 10 tasks are assigned to the drone. The distance selection algorithm used in this method can also obtain the flight route with the shortest range, which can achieve the rapid and efficient completion of drone collaborative flight missions and provide technical support for emergency response.

Claims

1. A multi-UAV task allocation method based on a distance selection algorithm, which features The following steps are implemented: (1) Construct the initial task arrangement model z, which is mathematically described as follows: z=[x,y] (1) x=[x1;x2;...xi;...xn] y=[y1;y2;...yi;...yn] In the above formula, z, x, y represent the initial task arrangement model values ​​to be assigned and their corresponding horizontal and vertical coordinate vectors, in meters; x1, y1, x2, y2, xi, yi, xn, yn represent the horizontal and vertical coordinates of the 1st, 2nd, i and nth task points in the planned airspace, in kilometers; "[]", "," and ";" represent the symbols of the matrix symbol, column and row spacing, respectively; (2) Calculate the dynamic cluster center of the task point: Using formulas (2) to (3), we can get the dynamic clustering center of the task points of the k-means clustering algorithm: [index_cluster, cluster]=kmeans_func(z, cluster_num) (2) meancluster=mean(cluster) (3) In formulas (2) to (3), cluster_hum, kmeans_func, index_cluster, and cluster are the number of clustering categories, the k-means clustering function, and the categories and the set of horizontal and vertical coordinates of the cluster centers corresponding to the task points obtained after clustering; mean and meancluster are the function for calculating the average value, and the horizontal and vertical coordinates of the average cluster center point obtained after calculation, respectively, in kilometers; (3) Constructing a dynamic allocation model of task points for distance selection algorithm: First, use formulas (4) to (7) to obtain the distance Dy from each task point to the cluster center point: X = [meancluster; z] (4) D=pdist(X) (5) dl=length(X) (6) Dy=D(1:dl-1) (7) In formulas (4) to (7), X, pdist, and D are the coordinate matrices of the constructed cluster center points and task points, the function for calculating the distance between the coordinate points, and the distance matrix calculated. Length and dl are the functions for calculating the maximum number of dimensions of the matrix and their calculation results. ":" and Dy are the colon calculation symbols (used to generate equidistant series, which can be omitted when the step length is 1) and the distances from the cluster center points to each task point calculated. Secondly, the task point dynamic allocation model of the distance selection algorithm is obtained by using formulas (8) to (11) to allocate the task point set of each drone: [Ds, Di]=sort(Dy,′option′) (8) meanfrn=fix(dl / flyn) (9) X=X(2:end) (10) frwd(i)=X(1+meanfrn*(i-1);meanfrn*i), i=1, 2,..., flyn (11) In formulas (8) to (11), sort, option, Ds, and Di are the sorting function, the sorting method, the distance after sorting, and the index corresponding to the original matrix, respectively; flyn, fix, meanfrn, and end are the number of drones, the integer function and the number of task points to be assigned to each drone calculated by it, and the last element of the array, respectively; frwd(i) is the set of task points assigned to the i-th drone and its horizontal and vertical coordinate sets; Finally, the routes formed by each UAV mission point set are obtained using formulas (12) to (14): Cfp=[xq,yq;Cf;xz,yz] (12) [cx, cy]=yiqunhanshu(Cfp) (13) frwdn(i)=[cx;cy]′,i=1,2,…,flyn (14) In formulas (12) to (14), xq, yq, xz, yz, Cf, and cfp are the horizontal and vertical coordinates of the starting point of the UAV flight, the horizontal and vertical coordinates of the mission point converted to double precision type, and the horizontal and vertical coordinates of the track points that form the route. yiqunhanshu, cx, and cy are the ant colony sorting functions that ensure the economic efficiency of the UAV flight and the horizontal and vertical coordinates of the track points of the i-th UAV obtained after sorting. (4) Evaluation function of the spatial relationship of the mission route: Using formulas (15) to (19), we can obtain the set of UAV track points after the route spatial relationship evaluation function is adjusted: [fic, fixc]=liniset(fij, fijx) (15) fij=[frwdn(i,j),frwdn(i,j+1)] (16) fixj=[frwdn(i+1,j), frwdn(i+1,j+1)] (17) frwdn(i, j+1)=fij (18) frwdn(i+1,j+1)=fixj (19) In formulas (15) to (19), fij, fijx, linset, fic, and fixc are the horizontal and vertical coordinates of the jth and j+1th track points of the ith UAV, the route spatial relationship evaluation function and the function for obtaining the number of statistical elements, and the horizontal and vertical coordinates of the jth and j+1th track points of the ith UAV to be adjusted, respectively. frwdn(i, j) and frwdn(i, j+1) are the horizontal and vertical coordinates of the jth track point of the ith UAV and the horizontal and vertical coordinates of the j+1th track point, respectively. frwdn(i+1, j) and frwdn(i+1, j+1) are the horizontal and vertical coordinates of the jth track point of the ith UAV and the horizontal and vertical coordinates of the j+1th track point, respectively. (5) Adjust the task allocation plan and optimize the flight task allocation: First, use formulas (20) to (21) to obtain the adjusted UAV track point set: [cx1,cy1]=yctsp(frwdn) (20) ltn(i)=[cx1 cy1] (21) In formulas (20) to (21), yctsp, cx1, cy1 are the genetic traveling salesman route optimization function and its adjusted horizontal and vertical coordinate sequences, respectively, and ltn is the adjusted route set.