Unmanned seeder task allocation and path planning method based on particle swarm optimization of improved weighting strategy

Abstract

The invention discloses an unmanned seeder task allocation and path planning method based on particle swarm optimization of an improved weighting strategy, and belongs to the field of agricultural machinery seeding task allocation and path planning. A task allocation model based on two factors of time and oil consumption is established, a particle swarm algorithm based on an improved weighting strategy is provided for task allocation, an AB line type operation method of a full-coverage path is specified, operation plots can be completely covered, related requirements of sowing are met, unmanned sowing agricultural machinery can achieve autonomous sowing operation between the plots, and the working efficiency is improved. The invention provides a semicircular-fishtail-shaped operation method for the agricultural machine according to the size of the field edge at the field edge turning position of the agricultural machine, and the problems that in the prior art, when a seeder turns at the corner, time is wasted, and the turning mode cannot be dynamically adjusted according to the size of a land parcel are solved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of task allocation and path planning for agricultural seeding, and particularly relates to a method for task allocation and path planning of unmanned seeding machines based on an improved weighted strategy particle swarm optimization algorithm. BACKGROUND

[0002] With the implementation of agricultural policies and the deepening of rural land circulation, agricultural machinery cooperatives and farm operation modes are rapidly developing in China, and agricultural machinery and land are gradually concentrated. Agricultural automatic navigation technology, as a core technology of smart agriculture, has promoted the development of multi-agricultural machine cooperative operation. This technology can improve operation efficiency, reduce production costs, and improve production efficiency.

[0003] However, traditional multi-agricultural machine cooperative operation technology still has limitations in task allocation, automatic navigation, environmental perception, automatic control, and information sharing. In particular, in task allocation, the control algorithm is prone to local optimization. Therefore, an improved weighted strategy particle swarm optimization algorithm (IPSO) is proposed to optimize the task allocation and path planning of unmanned seeding machines. This algorithm enhances global search ability and convergence accuracy by dynamically adjusting inertia weight, effectively avoids local optimization, and improves the efficiency and adaptability of path planning.

[0004] This task allocation and path planning method based on IPSO not only optimizes task allocation, but also dynamically adjusts the path according to the farmland environment and operation requirements, thereby achieving efficient and precise operation in complex environments.

[0005] In the prior art, a traditional particle swarm algorithm is generally used to solve the problem of agricultural machine task allocation and dynamic path, and the process is as follows: S1, initializing the particle swarm: selecting a suitable coding mode (such as real number coding or binary coding) to represent the task allocation scheme, randomly generating a certain number of particles (individuals) as the initial population, and ensuring that these particles can cover the task allocation solution space; S2, evaluating the fitness: calculating the fitness value of each particle position, and the fitness function can be designed according to the task completion efficiency, path length, energy consumption and other factors. The higher the fitness value, the better the task allocation scheme; S3, for each particle, compare its current fitness value with the individual historical best fitness value. If the current fitness is higher, update the individual best position; S4, find the particle with the highest fitness in the entire particle swarm, and take its position as the global best position; S5, according to the current position, speed, individual best position and global best position of the particle, update the speed and position of the particle. The speed update formula usually includes inertia weight, cognitive factor C1 and social factor C2 to balance the global search and local search ability; S6, repeat S2 to S5 until the maximum iteration number or the fitness value converges; S7, output the particle position corresponding to the global best position, which is the optimal task allocation scheme. This task allocation scheme corresponds to the operation scheme of the agricultural machine, and the probability of the inertia factor in the existing method is fixed in the iteration process, which does not reflect a random process. The inertia weight usually adopts a fixed value or a simple linear decreasing mode, and lacks the ability to adaptively adjust according to the current search state, which may cause the algorithm to easily fall into a stagnant state and a local optimal solution, thereby greatly reducing the feasible solution.

[0006] The AB line path planning commonly used in unmanned seeding machines is a straight line route presetting A and B points, which is simple and easy to implement, but is difficult to cope with complex environments and obstacles, and has poor adaptability. SUMMARY

[0007] The present application provides an unmanned seeding machine task allocation and path planning method based on an improved weighted strategy particle swarm algorithm to solve the problems existing in the prior art.

[0008] To achieve the above purpose, the present application provides an unmanned seeding machine task allocation and path planning method based on an improved weighted strategy particle swarm algorithm, comprising the following steps:

[0009] According to the operation scene of the multiple agricultural machines, the multiple agricultural machine operation task set and the operation task site information are constructed;

[0010] According to the performance parameters and operation parameter information of the agricultural machine operation, a multiple agricultural machine task scheduling evaluation model is constructed;

[0011] A cloud-based ad hoc network multiple agricultural machine task scheduling system is constructed;

[0012] The inertia weight coefficient of the traditional particle swarm algorithm is dynamically adjusted to obtain an improved weighted strategy particle swarm algorithm;

[0013] The uploaded task is distributed by the improved weighted strategy particle swarm algorithm, and the rationality of the distributed task is evaluated by a multi-tractor task scheduling evaluation model to obtain an optimal distribution result.

[0014] According to the distribution result, a path is generated, and the path is adjusted according to the obstacle points by a full coverage path algorithm.

[0015] When the tractor reaches the head of the field, the turning mode is switched based on the size of the head, and a turning path is obtained according to the turning mode.

[0016] Optionally, the set of multi-tractor operation tasks includes operation types, operation areas, and operation plot quantities; and the operation task field information includes field numbers, field endpoints, head endpoints, entrance endpoints, and obstacle endpoints.

[0017] Optionally, the multi-tractor task scheduling evaluation model includes constraint conditions and a cost function.

[0018] The constraint conditions include that the number of tractors does not exceed a threshold value required to complete the task, the number of tasks is greater than the number of tractors, and the tractors can cross-row operation.

[0019] The expression of the cost function is:

[0020]

[0021] wherein c is a function of the coordination cost, δ and μ are weight coefficients, t z is the operation time of the zth tractor, c z is the operation fuel consumption of the zth tractor, and m is the number of tractors.

[0022] The expression of the operation time is:

[0023]

[0024] wherein m is the mth tractor, L d is the row length, v z is the tractor speed of the zth tractor, r is the turning radius of the tractor, and n is the number of operation tasks.

[0025] The expression of the operation fuel consumption is:

[0026]

[0027] wherein m is the mth tractor, L d is the row length, o0 is the fuel consumption per unit time, v zLet be the speed of the z-th agricultural machine, n be the number of tasks the agricultural machine is performing, and r be the turning radius of the agricultural machine.

[0028] Optionally, the method can be implemented using a cloud-based self-organizing network-based multi-agricultural machinery task scheduling system. This system includes: an agricultural machinery management module, a task planning and scheduling module, a communication and self-organizing network management module, and a user interface and control module. The agricultural machinery management module assigns IDs to agricultural machinery and tracks their current location. The task planning and scheduling module performs dynamic task allocation optimization based on the location and status of the agricultural machinery. The communication and self-organizing network management module monitors the status of the agricultural machinery and configures tasks.

[0029] Optionally, the dynamic adjustment of the inertia weight coefficient of the traditional particle swarm optimization algorithm includes linear update of the inertia coefficient and nonlinear update of the inertia coefficient.

[0030] The expression for linearly updating the inertia coefficient is:

[0031]

[0032] Where, ω min With ω max These are the minimum and maximum inertia coefficients, respectively; t is the current time; and T is the maximum iteration time.

[0033] The expression for the nonlinear update of the inertia coefficient is:

[0034]

[0035] Where K is a constant used to adjust the rate at which the inertia factor changes with the number of iterations, t is the current time, and T is the maximum iteration time.

[0036] Optionally, obtaining the optimal allocation result includes:

[0037] Generate initial population particles according to the coding method required by the task;

[0038] The fitness value of each particle is calculated using a multi-agricultural machinery task scheduling evaluation model.

[0039] Update the individual's optimal position based on its fitness value;

[0040] The optimal solution is updated using an elite strategy;

[0041] Find the particle with the highest fitness and use its position as the global best position;

[0042] Update the particle's velocity and position based on its current position, velocity, individual best position, and global best position;

[0043] The weight coefficient is dynamically adjusted, the speed and position are updated, and when the iteration number is reached, the optimal allocation result is obtained.

[0044] The updating speed and position include:

[0045] V zd = ωV zd + c1r1(p zd -x zd ) + c2r2(p gd -x zd );

[0046] x zd = x zd + v zd ;

[0047] Wherein, ω is a weight coefficient, V zd is the zth particle speed, c1 and c2 are respectively a first learning factor and a second learning factor, P zd and P gd are respectively an individual optimal position and a global optimal position, X zd is a current zth particle position, and r1 and r2 are random numbers between 0 and 1.

[0048] Optionally, the full coverage path algorithm includes:

[0049] A grid map of a farm field is constructed, all boundary and obstacle points are marked as 1, and non-obstacle points in the working area are marked as 0;

[0050] A working starting point is found, the first grid marked as 0 is found from left to right in the last row of the grid map, and the grid is the starting coordinate; the coverage value of the starting point is set to 1;

[0051] The next point is found, and whether the current point is in the dead zone is checked; if so, the A* algorithm is used to escape;

[0052] In the working area, the indexes of n points around the current point are set, and the conditions of the surrounding points are judged: the first kind: the surrounding points are not covered and are not obstacle points, the value of the surrounding points is set to K, indicating that the points are not covered; the second kind: the surrounding points are obstacle points, the value of the surrounding points is set to-K, indicating that the points are obstacles and cannot be passed through; the third kind: the surrounding points are covered points but are not obstacle points, the value of the surrounding points is set to 0, indicating that the points have been passed through; wherein, n and K are positive integers;

[0053] During the search, the processing measures for different situations include: no surrounding point is covered, a new path is determined according to the distance from the current point; there is an obstacle point around, the obstacle point is removed from the path; the surrounding point is covered, whether to enter the dead zone is judged, if yes, the A* algorithm is used to get rid of the trouble;

[0054] The new path is set as the current path, whether the end point or the next point does not exist is judged, if yes, the search is stopped, otherwise the above operation is repeatedly executed.

[0055] Optionally, the turning mode includes fishtail shape and semicircular turning.

[0056] When the turning radius of the agricultural machine is greater than half of the working width of the agricultural machine, the fishtail type turning path is adopted for turning, and the fishtail type turning path is as follows:

[0057] C1=(2+pi)R-W;

[0058] Wherein, R is the turning radius, and W is the width of the agricultural machine.

[0059] The fishtail type turning path is composed of two circular arc segments with a radius of R and a straight line segment, wherein the circular arc segment and the straight line segment are tangent to each other.

[0060] When the turning radius of the agricultural machine is not greater than half of the working width of the agricultural machine, the semicircular turning path is adopted for turning, and the semicircular turning path is as follows:

[0061]

[0062] Wherein, R is the turning radius, and W is the width of the agricultural machine.

[0063] Compared with the prior art, the present application has the following advantages and technical effects:

[0064] The unmanned seeding machine task allocation and path planning method based on the improved weighted strategy particle swarm algorithm, by acquiring parameter information of the agricultural machine and parameter information of the farmland to be worked, setting task parameters according to n agricultural seeding tasks, constructing a multi-agricultural machine working task set and a multi-task scheduling evaluation model of the multi-agricultural machine, according to the task parameters, the task scheduling evaluation model of the multi-agricultural machine and the improved weighted strategy particle swarm algorithm, a full coverage path planning algorithm, and in combination with real-time state information of the agricultural machine received in real time, an optimal allocation scheme of the agricultural machine is acquired. The improved weighted strategy particle swarm algorithm designed in the present application can better balance the global search and local search capabilities. A larger inertia coefficient omega value is conducive to global search, and a smaller inertia coefficient omega value enhances the local search capability. Through dynamic adjustment, the particle swarm algorithm can achieve the best balance between global exploration and local optimization. BRIEF DESCRIPTION OF DRAWINGS

[0065] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this application. The embodiments of this application, and of which specific embodiments will be described, are illustrated in the drawings by specific embodiments of the application. In the drawings:

[0066] Figure 1 A work field block map of an embodiment of the application;

[0067] Figure 2 A grid map of No. 1 field of a work field block of an embodiment of the application;

[0068] Figure 3 A fish tail map of a work turn of an embodiment of the application;

[0069] Figure 4 A semi-circular map of a work turn of an embodiment of the application;

[0070] Figure 5 A ridge method work map of a work mode of an embodiment of the application;

[0071] Figure 6 A task allocation flowchart of an embodiment of the application;

[0072] Figure 7 A selection path map of a particle swarm algorithm with improved weighting strategy of an embodiment of the application. DETAILED DESCRIPTION

[0073] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0074] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown herein.

[0075] Embodiment one

[0076] As shown in the drawings, an unmanned seeder task allocation and path planning method based on a particle swarm algorithm with improved weighting strategy is provided in the embodiment, which comprises: Figures 5-6 According to the work scene of the multiple agricultural machines, a multiple agricultural machine work task set and work task site information are constructed;

[0077] According to the performance parameters and work parameter information of the agricultural machine work, a multiple agricultural machine task scheduling evaluation model is constructed;

[0078] A centralized multiple agricultural machine task scheduling system based on cloud self-organizing network is constructed;

[0079] A centralized multiple agricultural machine task scheduling system based on cloud self-organizing network is constructed;

[0080] When the information of the work plot and the information of the working agricultural machine are uploaded to the cloud ad hoc network system, the particle swarm algorithm with improved weighting strategy is used to allocate the new task, and the rationality of the allocated task is evaluated through the multi-agricultural machine task scheduling evaluation model to obtain the optimal allocation result.

[0081] According to the allocation result, a path is generated, and the full coverage path algorithm is used to dynamically adjust the path according to the obstacle points.

[0082] When the agricultural machine reaches the head, the turning mode is switched based on the size of the head, and the forward path is obtained according to the turning mode.

[0083] Specifically, the following steps are included:

[0084] S100, according to the working scene of the multi-agricultural machine, a multi-agricultural machine working task set {S1, S2, S3} is constructed, and a multi-agricultural machine working task site {Site1, Site2, Site3} is constructed.

[0085] S200, according to the performance parameters of the agricultural machine operation and the parameter information of the operation, a multi-agricultural machine task scheduling evaluation model is constructed.

[0086] S300, a centralized multi-agricultural machine task scheduling system based on cloud ad hoc network center is constructed.

[0087] S400, when the information of the work plot and the information of the working agricultural machine, and there is a new task to be completed, the system can upload the information to the cloud ad hoc network system, allocate the new task through the particle swarm algorithm with improved weighting strategy, evaluate the rationality of the allocated task through the multi-agricultural machine task scheduling evaluation model, and obtain the optimal allocation result.

[0088] S500, when the allocation result is obtained, input the system to obtain the path (AB line path); during the running process, the path is adjusted through the full coverage path algorithm according to the obstacle points.

[0089] S600, when the allocation result is obtained, the turning mode can be switched when the agricultural machine reaches the head.

[0090] Further, the step S100 specifically includes:

[0091] S101, the multi-agricultural machine working task set includes: working type type, working area M1, M2, M3, and working plot quantity num.

[0092] S102, the operation task field of the plurality of agricultural machines specifically includes: ① field number: 1, 2, 3, …; ② field end points: (x1, y1), (x2, y2), (x3, y3), (x4, y4); ③ head end points: (x5, y5), (x6, y6), (x7, y7), (x8, y8); ④ entrance end points: (x9, y9), (x 10 , y 10 ), (x 11 , y 11 ), (x 12 , y 12 ); ⑤ obstacle end points: (x 13 , y 13 ), (x 14 , y 14 ), (x 15 , y 15 ), (x 16 , y 16 ); ⑥ field import / export end points: (x 17 , y 17 ), (x 18 , y 18 ).

[0093] Further, step S200 specifically includes:

[0094] S201, define parameters, define symbols, and specify that a plurality of agricultural machines simultaneously start operation, and the last agricultural machine stops operation after completing operation.

[0095] Performance parameters of the agricultural machines: m represents the number of agricultural machines, the set {a1, a2, … a m} represents m agricultural machines, the set {v1, v2, … v m} represents the speed of m agricultural machines, and l represents the number of rows of operation, and Ti represents the i-th operation task. The performance parameters of the i-th agricultural machine can be represented as a = {v i , d i , w, t i}(i = 1, 2, m), wherein v i represents the average speed (km / h) of the i-th agricultural machine, d i represents the operation width (m) of the i-th agricultural machine, and t i represents the operation completion time (h) of the i-th agricultural machine.

[0096] S202, the multi-agricultural machine task scheduling evaluation model includes a cost function and a constraint condition,

[0097] The constraint condition includes that the number of agricultural machines does not exceed the threshold value of the required task completion amount, the number of tasks is greater than the number of agricultural machines, and the agricultural machines can cross-row operation;

[0098] The cost function is:

[0099]

[0100] where c is a function of the coordination cost, and δ and μ are weight coefficients, t z is the operation time of the zth agricultural machine, and c z is the fuel consumption of the zth agricultural machine. When δ = 1 and μ = 0, it represents the time of a single factor; when δ = 0 and μ = 1, it represents the fuel consumption of a single factor.

[0101] S203, calculating the operation time of a single agricultural machine, which is composed of two parts:

[0102] The operation time of the agricultural machine is composed of two parts: the straight-line time of the agricultural machine and the turning time of the agricultural machine at the head.

[0103]

[0104] where m is the mth agricultural machine, L d is the ridge length, v z is the speed of the zth agricultural machine, and r is the turning radius of the agricultural machine.

[0105] where,

[0106]

[0107] S204, calculating the fuel consumption c z of a single agricultural machine;

[0108] The fuel consumption of the agricultural machine is composed of two parts: the fuel consumption of straight-line operation and the fuel consumption of turning. Since the fuel consumption is related to time, it is not difficult to assume that the fuel consumption per unit time is o0, and the following formula is obtained:

[0109]

[0110] where m is the mth agricultural machine, L d is the ridge length, v z is the speed of the zth agricultural machine, and r is the turning radius of the agricultural machine.

[0111] where,

[0112]

[0113] Further, the constraint conditions include:

[0114] (1) Limitation of the number of agricultural machines: the number of agricultural machines cannot exceed the maximum required by the task.

[0115] (2) Rationality of task allocation: the amount of tasks cannot be unevenly distributed.

[0116] (3) Operation flow synchronization: all agricultural machines need to start from the garage at the same time, and return to the garage after completing the task, i.e., as the task is completed.

[0117] Further, the particle swarm algorithm based on the improved weighting strategy in step S400 comprises:

[0118] S401, generating an initial population with a number of N according to the encoding mode, selecting a suitable encoding mode (such as real number encoding or binary encoding) to represent the task allocation scheme, randomly generating a certain number of particles (individuals) as the initial population, and ensuring that these particles can cover the space of task allocation solutions, wherein N is a positive integer;

[0119] S402, calculating the fitness value of each particle's position, and the fitness function can be designed according to the task completion efficiency, path length, energy consumption and other factors. The higher the fitness value, the better the task allocation scheme;

[0120] S403, for each particle, comparing its current fitness value with the individual historical best fitness value. If the current fitness is higher, update the individual best position;

[0121] S404, using the elite strategy to update the optimal solution, selecting the optimal individual, and updating the offspring population;

[0122] S405, for each particle, comparing its current fitness value with the individual historical best fitness value. If the current fitness is higher, update the individual best position;

[0123] S406, find the particle with the highest fitness value in the entire particle swarm, and take its position as the global best position;

[0124] S407, updating the speed and position of the particle according to the current position, speed, individual best position and global best position of the particle;

[0125] S408, dynamically adjusting the weight coefficient ω to update the position and speed;

[0126] S409, judging whether the iteration number is reached. If the iteration number is not reached, return to step S402, if the iteration number is reached, select the individual with the highest fitness value of the current population as the solution of the task allocation, and output the optimal result, i.e., the optimal task allocation scheme;

[0127] S410, inputting the output result into the system to generate three points on the path of an AB line, and planning the following points according to the full coverage path method;

[0128] S411, when reaching the end of the road, a different turning method will be selected.

[0129] The method for updating the position and velocity of the particle is:

[0130] Velocity update: V zd = ωV zd + c1r1(p zd - x zd ) + c2r2(p gd - x zd ) (6) ;

[0131] Position update: x zd = x zd + v zd (7) ;

[0132] where ω is a weight coefficient, V zd is the velocity of the zth particle, c1 and c2 are learning factors, P zd and P gd are the individual optimal position and the global optimal position, X zd is the current zth particle position, and r1 and r2 are random numbers between 0 and 1.

[0133] Preferably, the method for searching the full-coverage path comprises the following steps: S01, constructing a grid map of a farm field plot, the field map being as shown in Figure 1 , and the grid map being as shown in Figure 2 , marking all points outside the boundary and the obstacle points as 1, and marking the non-obstacle points in the working area as 0. S02, searching for a working starting point, which is achieved by traversing from left to right on the last row of the grid map, searching for the first grid marked as 0, and then setting the starting coordinates; setting the coverage value of the starting point as 1. S03, starting to search for the next point, checking whether the current point is in the dead zone, and if so, using the A* algorithm to escape. S04, in the working area, setting the indices of the eight surrounding points of the current point, and judging the conditions of the surrounding points: the first condition is that the surrounding points are not covered and are not obstacle points, the value of the surrounding points is set as 10, indicating that the surrounding points are not covered; the second condition is that the surrounding points are obstacle points, the value of the surrounding points is set as -10, indicating that the surrounding points are obstacles and cannot be passed through; the third condition is that the surrounding points are covered points but are not obstacle points, the value of the surrounding points is set as 0, indicating that the surrounding points have been passed through. When searching, the measures for different conditions are as follows: the surrounding points are not covered, a new path is determined according to the distance from the current point; the surrounding points are obstacle points, the obstacle points are removed from the path; the surrounding points are covered, it is judged whether the surrounding points enter the dead zone, and if so, the A* algorithm is used to escape. S05, setting the new path as the current path, judging whether the current path is the terminal point or whether there is a next point, and if so, the searching is stopped, otherwise, the step 4 is executed.

[0134] Further, the S408 further comprises: nonlinearly updating the inertia coefficient and linearly updating the inertia coefficient:

[0135] (1) Linear update:

[0136]

[0137] where ωminand ωmaxare the minimum and maximum inertia coefficients, respectively, t is the current time, and T is the maximum iteration time. min max where ωminand ωmaxare the minimum and maximum inertia coefficients, respectively, t is the current time, and T is the maximum iteration time.

[0138] (2) Nonlinear update:

[0139]

[0140] where K is a constant used to adjust the rate of change of the inertia factor with the iteration number, t is the current time, and T is the maximum iteration time.

[0141] Further, the cloud-based self-organizing network centralized multi- agricultural machine task scheduling system of step S300 includes an agricultural machine end, a server end, and a client end. The agricultural machine itself carries a GPS system and a set of self-organizing network communication systems, which are used to realize real-time positioning of the agricultural machine and interactive communication with other agricultural machines. The server end part has a storage function, which is used to store the historical trajectory of the agricultural machine, the real-time state information of the agricultural machine, and the experimental plot information. The client end is used to issue instructions and tasks to the agricultural machine.

[0142] Further, the path mode of step S500 includes:

[0143] Fish tail turning mode: when the turning radius of the agricultural machine is greater than half of the working width of the agricultural machine, a fish tail turning path is used for turning, and the fish tail turning path is as follows:

[0144] C1=(2+π)R-W (10)

[0145] where R is the turning radius, and W is the width of the agricultural machine.

[0146] The fish tail turning path is composed of two circular arc segments with a radius of R and a straight line segment, and the circular arc segment and the straight line segment are tangent to each other.

[0147] When the minimum turning radius of the agricultural machine is Rmin, the working width of the agricultural machine is W, the reverse distance of the agricultural machine is x, the turning path length of the agricultural machine is S, the centers of the circular arc turning of the agricultural machine are P1 and P2, and the boundary of the turning land is perpendicular to the working path, the agricultural machine driving direction and turning distance are as shown in Figure 3 .

[0148] Half-circle turning mode: when the turning radius of the agricultural machine is not greater than half of the working width of the agricultural machine, a half-circle turning path is used for turning, and the half-circle turning path is as follows:​

[0149]

[0150] wherein R is the turning radius, and W is the width of the agricultural machine.

[0151] Suppose the working width of the agricultural machine is marked as W, and accordingly, the turning radius R of the agricultural machine can be set as W / 2. If the head boundary is perpendicular to the working path and serves as the turning area, the driving direction of the agricultural machine and its turning trajectory are as shown in Figure 4 .

[0152] Further, the path mode of step S500 further includes a furrowing method:

[0153] The furrowing method refers to a working method of opening new furrows in the process of tillage. This method is convenient for subsequent agricultural operations such as irrigation and drainage, helps to improve the soil environment, and promotes the healthy growth of crops. The path diagram of the furrowing method is as shown in Figure 7 .

[0154] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An unmanned seeder task allocation and path planning method based on improved weighted strategy particle swarm optimization algorithm, characterized in that, The method comprises the following steps: According to the working scene of multiple agricultural machines, the working task set and the working task site information of multiple agricultural machines are constructed; According to the performance parameters and working parameter information of agricultural machine operation, a multi-agricultural machine task scheduling evaluation model is constructed; The inertia weight coefficient of the traditional particle swarm algorithm is dynamically adjusted to obtain an improved weighted strategy particle swarm algorithm; According to the working task set and the working task site information of multiple agricultural machines and the multi-agricultural machine task scheduling evaluation model, the uploaded tasks are distributed by the improved weighted strategy particle swarm algorithm, and the rationality of the distributed tasks is evaluated by the multi-agricultural machine task scheduling evaluation model to obtain the optimal distribution result; According to the distribution result, a path is generated, and the path is adjusted according to the obstacle points by a full coverage path algorithm; When the agricultural machine reaches the head of the field, the turning mode is switched based on the size of the head, and the turning path is obtained according to the turning mode.

2. The unmanned seeder task allocation and path planning method based on improved weighted strategy-based particle swarm optimization algorithm according to claim 1, characterized in that, The working task set of multiple agricultural machines includes the working type, the working area, and the number of working plots; the working task site information includes the site number, the site endpoint, the endpoint of the head, the entry endpoint, and the endpoint of the obstacle.

3. The unmanned seeder task allocation and path planning method based on improved weighted strategy-based particle swarm optimization algorithm according to claim 1, characterized in that, The multi-agricultural machine task scheduling evaluation model includes constraint conditions and a cost function: The constraint conditions include that the number of agricultural machines does not exceed the threshold value of the required task completion amount, the number of tasks is greater than the number of agricultural machines, and the agricultural machines can cross-row work; The expression of the cost function is: Wherein, c is the function of the synergy cost, δ and μ are weight coefficients, t z is the operation time of the zth agricultural machine, c z is the operation oil consumption of the zth agricultural machine, and m is the number of agricultural machines. The expression of the working time is: Wherein, m is the mth agricultural machine, L d is the ridge length, v z is the speed of the zth agricultural machine, r is the turning radius of the agricultural machine, and n is the number of tasks of the agricultural machine operation; The expression of the working oil consumption is: wherein m is the mth agricultural machine, L d is the ridge length, o0is the oil consumption per unit time, v z is the speed of the zth agricultural machine, n is the number of tasks of the agricultural machine operation, r is the turning radius of the agricultural machine, 4. The unmanned seeder task allocation and path planning method based on improved weighted strategy-based particle swarm optimization algorithm according to claim 1, characterized in that, The method is implemented by a multi-agricultural machine task scheduling system based on cloud ad hoc network, and the system comprises an agricultural machine management module, a task planning and scheduling module, a communication and ad hoc network management module, and a user interface and control module; wherein the agricultural machine management module assigns an ID to the agricultural machine and tracks the current position of the agricultural machine; the task planning and scheduling module performs dynamic task distribution optimization according to the agricultural machine position and state factors; and the communication and ad hoc network management module is used for monitoring the state of the agricultural machine and configuring tasks.

5. The unmanned seeder task allocation and path planning method based on improved weighted strategy-based particle swarm optimization algorithm as claimed in claim 1, wherein, The dynamic adjustment of the inertia weight coefficient of the traditional particle swarm algorithm includes linear updating of the inertia coefficient and nonlinear updating of the inertia coefficient; The expression of the linear updating of the inertia coefficient is: where ω min and ω max are the minimum and maximum inertia coefficients, respectively, t is the current time, and T is the maximum iteration time. The expression of the nonlinear updating of the inertia coefficient is: Wherein, K is a constant for adjusting the rate of change of the inertia factor with the number of iterations, t is the current time, and T is the maximum iteration time.

6. The unmanned seeder task allocation and path planning method based on improved weighted strategy-based particle swarm optimization algorithm as claimed in claim 1 wherein, The optimal distribution result includes: An initial population particle is generated according to the task requirement coding mode; The fitness value of each particle is calculated by using the multi-agricultural machine task scheduling evaluation model; The individual best position is updated according to the fitness value; The optimal solution is updated by using the elite strategy; The particle with the highest fitness value is found, and its position is taken as the global best position; The speed and position of the particle are updated according to the current position, speed, individual best position, and global best position of the particle; The weight coefficient is dynamically adjusted, the speed and position are updated, and when the number of iterations is reached, the optimal distribution result is obtained; The updating of the speed and position includes: V zd = ωV zd + c1r1(p zd - x zd ) + c2r2(p gd - x zd ); x zd = x zd + v zd ; where ω is a weight coefficient, V zd is the zth particle velocity, c1 and c2 are the first and second learning factors, respectively, P zd , P gd are the individual optimal position and the global optimal position, respectively, X zd is the current zth particle position, and r1 and r2 are random numbers between 0 and 1.

7. The unmanned seeder task allocation and path planning method based on improved weighted strategy-based particle swarm optimization algorithm according to claim 1, characterized in that, The full coverage path algorithm includes: A grid map of the farm plot is constructed, all boundary and obstacle points outside the boundary are marked as 1, and non-obstacle points in the working area are marked as 0. Finding the starting point of the operation, traversing from left to right in the last row of the grid map, finding the first grid marked as 0, which is the starting coordinate; the cover value of the starting point is set to 1; Start searching for the next point, check if the current point is in the dead zone flag, if so, use A* algorithm to escape; In the operation area, set the index of the surrounding n points at the current point, judge the situation of the surrounding points: the first kind: the surrounding point has not been covered, and is not an obstacle point, the value of the surrounding point is set to K, indicating that it has not been covered; the second kind: is an obstacle point, the value of the surrounding point is set to-K, indicating that it is an obstacle and cannot be passed; the third kind: is a covered point, but is not an obstacle point, the value of the surrounding point is set to 0, indicating that the point has been passed; wherein, n and K are positive integers; The treatment measures for different situations during search include: the surrounding points have not been covered, determine the new path according to the distance from the current point; there are obstacle points in the surrounding, remove the obstacle points from the path; the surrounding points have been covered, judge whether to enter the dead zone, if so, use A* algorithm to escape; Set the new path as the current path, judge whether it is the terminal point or there is no next point, if so, stop searching, otherwise, repeat the above operation.

8. The unmanned seeder task allocation and path planning method based on improved weighted strategy-based particle swarm optimization algorithm according to claim 1, characterized in that, The turning mode includes fishtail shape and semicircular turning; When the turning radius of the agricultural machine is greater than half of the width of the agricultural machine, the fishtail turning path is adopted for turning, and the fishtail turning path is as follows: C1=(2+π)R-W; Wherein, R is the turning radius, and W is the width of the agricultural machine; The fishtail turning path is composed of two circular arc segments with a radius of R and a straight line segment, wherein the circular arc segment and the straight line segment are tangent to each other; When the turning radius of the agricultural machine is not greater than half of the width of the agricultural machine, the semicircular turning path is adopted for turning, and the semicircular turning path is as follows: Wherein, R is the turning radius, and W is the width of the agricultural machine.

Images