Helicopter aerial pesticide application landing point position planning method and system

Abstract

The application relates to a helicopter aerial pesticide application landing point position planning method and system. The method comprises the following steps: acquiring the vertex position of each forest area; determining the number of turns, the number of pesticide application routes and the direction vector of each forest area by using a vector modeling method according to the vertex position of each forest area; determining four access points of the corresponding forest area according to the parity of the number of pesticide application routes and the direction vector; determining the boundary of each forest area according to the vertex position of each forest area, and constructing a binary classification model of the landing point position according to the forest area boundary; judging whether the landing point position is outside the forest area by using the binary classification model of the landing point position; when the landing point position is outside the forest area, the target landing point position is determined by using an ant colony algorithm and a whale algorithm; and when the landing point position is inside the forest area, the target landing point position is determined by using a whale algorithm. The application can accurately determine the helicopter aerial pesticide application landing point position, and further improve the aerial pesticide application efficiency of the forest area.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of forestry management, in particular to a helicopter aerial pesticide application take-off and landing point position planning method and system. BACKGROUND

[0002] China's artificial forest area is vast, and due to single variety, the disease and pest situation is very serious, with an annual stable occurrence area of 180 million mu. Moreover, forestry diseases and pests spread rapidly. Aerial pesticide application by manned helicopters can achieve an area of more than 1000 mu per operation, and plays an irreplaceable role in artificial forest disease and pest control. In aerial pesticide application, the first step is to select the helicopter take-off and landing point position, and then plan the pesticide application route. However, the existing aircraft take-off and landing point position planning mainly relies on manual experience, and only according to the forest area, the forest take-off and landing point position is roughly planned, which lacks theoretical basis, and the possibility of the aircraft take-off and landing point position being optimal is low. Planning a reasonable aircraft take-off and landing point position can reduce the length of the dispatch route between the aircraft and each forest area, which is of great significance to reducing the energy consumption of the aircraft and improving the operation efficiency. At present, there is no literature on planning the helicopter take-off and landing point position, and there is no tool that can plan the take-off and landing point position, which brings some difficulties to the planning of the helicopter aerial pesticide application take-off and landing point.

[0003] Therefore, there is an urgent need for a helicopter aerial pesticide application take-off and landing point position planning method or system that can accurately determine the helicopter aerial pesticide application take-off and landing point position, thereby improving the efficiency of forest aerial pesticide application. SUMMARY

[0004] The purpose of the present application is to provide a helicopter aerial pesticide application take-off and landing point position planning method and system that can accurately determine the helicopter aerial pesticide application take-off and landing point position, thereby improving the efficiency of forest aerial pesticide application.

[0005] To achieve the above-mentioned purpose, the present application provides the following scheme:

[0006] A helicopter aerial pesticide application take-off and landing point position planning method, comprising:

[0007] obtaining the vertex position of each forest area;

[0008] determining the number of turns, the number of pesticide application routes and the direction vector of each forest area by vector modeling method according to the vertex position of each forest area;

[0009] determining four access points of the corresponding forest area according to the parity of the number of pesticide application routes and the direction vector of each forest area;

[0010] determining the boundary of each forest area according to the vertex position of each forest area, and constructing a binary classification model of the take-off and landing point position according to the forest area boundary; the binary classification model of the take-off and landing point position is used to judge whether the take-off and landing point position is outside the forest area;

[0011] A binary classification model of take-off and landing point locations is used to determine whether the take-off and landing point is outside the forest area;

[0012] When the take-off and landing point is located outside the forest area, the ant colony algorithm and whale algorithm are used to determine the target take-off and landing point location.

[0013] When the take-off and landing point is located within a forest area, the whale algorithm is used to determine the target take-off and landing point location.

[0014] Optionally, the step of determining the number of U-turns, the number of spraying routes, and the direction vector for each forest area using vector modeling based on the vertex position of each forest area specifically includes:

[0015] Based on the vertex location of each forest area, a mathematical model for evaluating pesticide application routes is constructed using vector modeling.

[0016] Based on the mathematical model for evaluating the application route, the number of U-turns and excess coverage are used as evaluation indicators. The NSGA-II and Pareto optimal solutions are adopted to determine the direction vector.

[0017] Optionally, the step of determining the boundary of each forest area based on the vertex position of each forest area, and constructing a binary classification model of the take-off and landing point positions based on the forest area boundaries, specifically includes:

[0018] The boundaries of each forest area are determined based on the apex position of each forest area;

[0019] Based on the boundary of each forest area, a step function is used to determine the corresponding AND classifier for the forest area; the AND classifier is used to determine whether the take-off and landing points are outside the corresponding forest area;

[0020] An OR classifier is constructed based on each AND classifier; the OR classifier takes the output of each AND classifier as input; the OR classifier is a binary classification model for take-off and landing point locations.

[0021] Optionally, when the take-off and landing point is located outside the forest area, the ant colony algorithm and whale algorithm are used to determine the target take-off and landing point location, specifically including:

[0022] An initial whale population is randomly generated within the search area of ​​the take-off and landing points; each individual whale represents a take-off and landing point.

[0023] Initialize the pheromone level and iteration count for the ant colony algorithm;

[0024] Place m ants randomly at random points in n forest areas; the random points are the four entry and exit points of the corresponding forest areas.

[0025] Set the ant index k = 1;

[0026] A movement within the forest area is performed based on the parity of the number of pesticide application routes.

[0027] Based on the probability of selection, the roulette wheel rule is used to determine the next forest area that the ant will visit;

[0028] Update local pheromones;

[0029] Determine if the ant index k ≥ m;

[0030] If k < m, then return to the step of moving within the forest area once based on the parity of the number of spraying routes in the forest area;

[0031] If so, then determine whether all m ants have visited all forest areas;

[0032] If all forest areas have been visited, then update the global pheromones;

[0033] Conversely, return to the step of setting the ant index k=1;

[0034] Determine whether the number of iterations in the ant colony algorithm has reached the maximum number of iterations for the ant colony algorithm;

[0035] If so, the individual whale's fitness value is determined based on the scheduling route at the current take-off and landing point location;

[0036] Determine whether the fitness values ​​of all individual whales have been calculated.

[0037] If the fitness values ​​of all individual whales have been calculated, then determine whether the number of iterations of the whale algorithm has reached the maximum number of iterations of the whale algorithm;

[0038] If so, the corresponding take-off and landing point location will be taken as the optimal take-off and landing point location, and the corresponding scheduling route will be determined.

[0039] If not, the whale population is updated using bubble netting, prey encirclement, and random prey hunting methods, and the process returns to the step of using a binary classification model based on the take-off and landing point location to determine whether the take-off and landing point location is outside the forest area.

[0040] If the fitness values ​​of all individual whales have not been calculated, return to the step of using the binary classification model of take-off and landing location to determine whether the take-off and landing location is outside the forest area;

[0041] Conversely, if the ant colony algorithm fails, proceed to the next iteration until the maximum number of iterations for the ant colony algorithm is reached.

[0042] Optionally, when the take-off and landing point is located within a forest area, the whale algorithm is used to determine the target take-off and landing point location, specifically including:

[0043] An initial whale population is randomly generated within the search area of ​​the take-off and landing points; each individual whale represents a take-off and landing point.

[0044] Set the take-off and landing point location to the preset individual whale fitness value;

[0045] Determine whether the fitness values ​​of all individual whales have been calculated.

[0046] If the fitness values ​​of all individual whales have been calculated, then determine whether the number of iterations of the whale algorithm has reached the maximum number of iterations of the whale algorithm;

[0047] If so, the corresponding take-off and landing point location will be taken as the optimal take-off and landing point location, and the corresponding scheduling route will be determined.

[0048] If not, the whale population is updated using bubble netting, prey encirclement, and random prey hunting methods, and the process returns to the step of using a binary classification model based on the take-off and landing point location to determine whether the take-off and landing point location is outside the forest area.

[0049] If the fitness values ​​of all individual whales have not been calculated, return to the step of using the binary classification model of take-off and landing location to determine whether the take-off and landing location is outside the forest area.

[0050] A helicopter aerial spraying take-off and landing point location planning system, applied to the aforementioned helicopter aerial spraying take-off and landing point location planning method, the system comprising:

[0051] The vertex position acquisition module is used to acquire the vertex position of each forest area;

[0052] The module for determining the number of U-turns, the number of pesticide application routes, and the direction vector is used to determine the number of U-turns, the number of pesticide application routes, and the direction vector for each forest area based on the vertex position of each forest area using vector modeling.

[0053] The entry and exit point determination module is used to determine the four entry and exit points of the corresponding forest area based on the parity of the number of spraying routes and the direction vector of each forest area.

[0054] The binary classification model construction module is used to determine the boundary of each forest area based on the vertex position of each forest area, and to construct a binary classification model of the take-off and landing point positions based on the forest area boundaries; the binary classification model of the take-off and landing point positions is used to determine whether the take-off and landing point positions are outside the forest area;

[0055] The judgment module is used to determine whether the take-off and landing point is outside the forest area using a binary classification model of the take-off and landing point location;

[0056] The first module for determining the target take-off and landing point location is used to determine the target take-off and landing point location using ant colony algorithm and whale algorithm when the take-off and landing point location is outside the forest area;

[0057] The second module for determining the target take-off and landing point location is used to determine the target take-off and landing point location using the whale algorithm when the take-off and landing point location is within a forest area.

[0058] Optionally, the module for determining the number of U-turns, the number of application routes, and the direction vector specifically includes:

[0059] The mathematical model determination unit is used to construct a mathematical model for evaluating pesticide application routes based on the vertex position of each forest area using vector modeling.

[0060] The direction vector determination unit is used to determine the direction vector based on the mathematical model of the drug application route evaluation, using the number of U-turns and excess coverage as evaluation indicators, and employing the optimal solutions of NSGA-II and Pareto.

[0061] Optionally, the binary classification model construction module specifically includes:

[0062] The forest area boundary determination unit is used to determine the boundary of each forest area based on the vertex position of each forest area;

[0063] The AND classifier determination unit is used to determine the corresponding AND classifier for each forest area based on the boundary of each forest area using a step function; the AND classifier is used to determine whether the take-off and landing points are outside the corresponding forest area;

[0064] The binary classification model construction unit is used to construct an OR gate classifier based on each AND sub-classifier; the OR gate classifier takes the output of each AND classifier as input; the OR gate classifier is a binary classification model of the take-off and landing point positions.

[0065] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0066] This invention provides a method and system for planning helicopter aerial spraying take-off and landing points. Based on the vertex position of each forest area, the boundary of each forest area is determined. A binary classification model for take-off and landing point locations is constructed based on the forest area boundaries. Specifically, a combinatorial logic classification method is used to control the feasible region of helicopter take-off and landing points. When the take-off and landing point is outside the forest area, an ant colony algorithm and a whale algorithm are used to determine the target take-off and landing point location; when the take-off and landing point is inside the forest area, the whale algorithm is used to determine the target take-off and landing point location. In other words, the algorithm of this invention adopts a nested fusion approach. The outer algorithm uses the whale algorithm to search for take-off and landing point locations, while the inner algorithm uses a modified ant colony algorithm to calculate the fitness of the whale algorithm. This invention reduces the number of times a helicopter needs to turn around in a single forest area, reducing the difficulty of pilot operation; it can shorten the length of the scheduling route between forest areas, saving non-spraying operation time, and effectively reducing aviation fuel consumption, thus saving economic costs for helicopter aerial spraying operations and improving the efficiency of aerial spraying in forest areas. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 This is a schematic diagram of a helicopter aerial spraying take-off and landing point location planning method provided by the present invention.

[0069] Figure 2 This is a schematic diagram of the overall process of a helicopter aerial spraying take-off and landing point location planning method provided by the present invention.

[0070] Figure 3 Planning the mission map for the take-off and landing point locations;

[0071] Figure 4 A schematic diagram of a binary classification model (combinatorial logic classification) for take-off and landing point locations;

[0072] Figure 5 A schematic diagram of the forest area's entrance and exit points;

[0073] Figure 6 This is a graph showing the search strategy of the ant colony algorithm.

[0074] Figure 7 This is a diagram showing the planning results for the optimal take-off and landing point locations.

[0075] Figure 8 This is a convergence curve diagram of the present invention.

[0076] Figure 9 This involves planning take-off and landing points based on human experience and scheduling routes planned using a double-nested genetic algorithm. Detailed Implementation

[0077] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] The purpose of this invention is to provide a method and system for planning the take-off and landing points of helicopter aerial spraying, which can accurately determine the take-off and landing points of helicopter aerial spraying, thereby improving the efficiency of aerial spraying in forest areas.

[0079] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0080] Figure 1 This is a schematic diagram of a helicopter aerial spraying take-off and landing point location planning method provided by the present invention, as shown below. Figure 1 As shown, the present invention provides a method for planning the take-off and landing point locations for helicopter aerial spraying, comprising:

[0081] S101, obtain the vertex position of each forest area.

[0082] S102. Based on the vertex position of each forest area, the vector modeling method is used to determine the number of U-turns, the number of spraying routes, and the direction vector for each forest area.

[0083] S102 specifically includes:

[0084] Based on the vertex location of each forest area, a mathematical model for evaluating pesticide application routes is constructed using vector modeling.

[0085] Based on the mathematical model for evaluating pesticide application routes, the number of U-turns and redundant coverage are used as evaluation indicators. The NSGA-II and Pareto optimal solutions are employed to determine the direction vector. According to the Pareto optimal solution, application schemes with fewer U-turns are prioritized, followed by those with lower redundant coverage.

[0086] S103, determine the four entry and exit points for each forest area based on the parity of the number of spraying routes and the direction vector. For example... Figure 4 As shown, in counter-clockwise order, the first two points are exactly on the same application route.

[0087] S104. Based on the vertex position of each forest area, determine the boundary of each forest area, and construct a binary classification model of the take-off and landing point positions based on the forest area boundaries; the binary classification model of the take-off and landing point positions is used to determine whether the take-off and landing point positions are outside the forest area.

[0088] S104 specifically includes:

[0089] Determine the boundary (boundary equation) of each forest area based on the vertex location of each forest area.

[0090] Based on the boundary of each forest area, a step function is used to determine the corresponding AND classifier for the forest area; the AND classifier is used to determine whether the take-off and landing points are outside the corresponding forest area.

[0091] An OR classifier is constructed based on each AND classifier; the OR classifier takes the output of each AND classifier as input; the OR classifier is a binary classification model for take-off and landing point locations.

[0092] This method employs combinational logic classification to determine whether the takeoff and landing point is located within the forest area. Based on the coordinates of each vertex in the forest area, the boundary equations for each area are calculated, ensuring that takeoff and landing points biased towards the forest polygon side are positive when substituted into these equations. A step function is then used as the input to an AND gate. For each polygonal forest area, an AND sub-classifier is constructed. The outputs of the AND gates corresponding to all forest areas are then used as the input to an OR gate. Based on the OR gate output, it can be determined whether the takeoff and landing point is within the forest area; flag = 1 indicates within the forest area (including the forest boundary), and flag = 0 indicates outside the forest area.

[0093] S105 uses a binary classification model of take-off and landing point locations to determine whether the take-off and landing point locations are outside the forest area.

[0094] S106. When the take-off and landing point is outside the forest area, the ant colony algorithm and whale algorithm are used to determine the target take-off and landing point.

[0095] S107. When the take-off and landing point is located within a forest area, the whale algorithm is used to determine the target take-off and landing point location.

[0096] The whale algorithm is the main algorithm for searching take-off and landing point locations, and it is used throughout the entire fusion algorithm. In the two-dimensional solution space, whales update their positions using three methods: bubble netting, surrounding prey, and random prey search. Surrounding prey, where all whales move closer to the current optimal whale, guarantees algorithm convergence; bubble netting and random prey search reflect the whales' exploratory nature, ensuring the algorithm escapes local optima. The whale algorithm's position update methods ensure its strong global search capability. The fitness of the whale algorithm is designed as the reciprocal of the optimal scheduling route between global regions. Using vector modeling to evaluate the application route for each forest area, without considering the length of turning routes, each forest area will have 4 possible entry and exit points. Once the entry point of a forest area is determined, the departure point is also determined based on the parity of the number of routes, resulting in 16 scheduling routes between every two forest areas. Solving this multi-forest area scheduling route problem is more complex than the Traveling Salesman Problem. Two whale fitness calculation methods are designed based on the whale (take-off and landing point) positions in the forest areas. When whales are in forest areas, they are assigned a relatively low fitness value. When whales are not in forest areas, an improved ant colony algorithm is used to calculate their fitness. Compared to calculating the fitness of each whale using the ant colony algorithm, this method not only meets the practical requirement of not being able to set up take-off and landing points in forest areas, but also improves the algorithm's processing speed. This is because manually assigning a whale's fitness value requires only one line of computer instructions, while the improved ant colony algorithm requires approximately 200 lines of instructions.

[0097] To determine whether a whale is within a forest area, this invention draws inspiration from combinational logic circuits and proposes a combinational logic classification method, as shown in steps S5 and S7. Combinational logic circuits have two important logic elements: AND gates and OR gates. The logic of an AND gate is that if all inputs are 1, the output is 1; otherwise, the output is 0. The logic of an OR gate is that if one input is 1, the output is 1; otherwise, the output is 0. This invention calculates the linear equations of each boundary line of the forest area based on the coordinates of the vertices of each boundary, and corrects the coefficients of the linear equations based on any point within the forest area, ensuring that the landing and takeoff points biased towards the forest area are all positive. Thus, a step function can be used as the input to the AND gate. When a landing and takeoff point falls within a certain forest area, the corresponding AND gate outputs 1; otherwise, it outputs 0. By setting one AND gate for each forest area, it is possible to determine whether the landing and takeoff point is within the current forest area. The outputs of all AND gates are used as inputs to OR gates, with the OR gate outputting a flag: flag = 1 indicates the location is within the forest area (including forest boundaries), and flag = 0 indicates the location is outside the forest area. This combinational logic classification method, compared to a three-layer neural network, eliminates the need to set weights for each layer, making it easy to program and computationally efficient. Compared to deep learning neural networks, it also eliminates the need for manual dataset labeling and training, and deep learning neural networks struggle to achieve 100% classification accuracy. However, the combinational logic classification method only requires randomly sampling one point within each forest area to determine the boundary equation coefficients, achieving 100% classification accuracy.

[0098] For whales located outside the forest area, an improved ant colony algorithm is used to calculate their fitness, i.e., to solve for global inter-regional scheduling routes. Steps S8-S20 explain the search principle of the improved ant colony algorithm. Ant movement is designed as lateral movement within the forest area and longitudinal movement between forest areas. Movement within the forest area is determined by the parity of the route, a deterministic movement. Longitudinal movement between forest areas is determined by a roulette wheel method based on pheromone concentration, an uncertain movement, and also a heuristic movement. The improvement to the ant colony algorithm perfectly suits the needs of solving inter-regional scheduling routes. With the powerful path search capability of the ant colony algorithm, inter-regional scheduling routes can be solved very quickly. Compared to using a double-nested genetic algorithm or a step-by-step two-stage genetic algorithm, this algorithm has a faster solution speed; compared to the case where there are only two entry and exit points within the forest area, this algorithm can solve for shorter inter-regional scheduling routes. The improved ant colony algorithm quickly and efficiently solves for whale fitness, providing more possibilities for the integration of multiple intelligent algorithms.

[0099] Figure 2 This is a schematic diagram of the overall process of a helicopter aerial spraying take-off and landing point location planning method provided by the present invention, as shown below. Figure 2 As shown, the present invention specifically includes the following steps:

[0100] S1. The algorithm begins; input the vertex positions of each forest area.

[0101] S2. Use vector modeling to determine the number of U-turns, number of pesticide application routes, and direction vectors for each forest area;

[0102] S3. Determine the four entry and exit points for each forest area based on the direction vector and the parity of the number of spraying routes, sort them counterclockwise, and ensure that the first two points are located on the same spraying route.

[0103] S4. Set the parameters for the whale algorithm and ant colony algorithm. The parameters for the whale algorithm include: number of whales and number of iterations. The parameters for the ant colony algorithm include: number of ants, number of iterations, pheromone evaporation coefficient, pheromone intensity, and heuristic factor. The number of ants is set to 4 times the number of forest areas, and the number of whales is set to 6 times the number of forest areas.

[0104] S5. Based on the forest area boundary, construct a binary classification model of the take-off and landing point locations using the combinatorial logic classification method;

[0105] S6. Randomly generate an initial whale population within the solution space (search area for take-off and landing points), with each individual whale representing a take-off and landing point location;

[0106] S7. Use combinational logic classification to determine whether the whale (take-off and landing point location) is outside the forest area. If it is, the improved ant colony algorithm will be used to calculate the fitness of the whale algorithm, and S8 will be executed. Otherwise, a relatively small fitness value will be manually assigned, and S21 will be executed.

[0107] S8. Initialize the pheromone of the ant colony algorithm and set the pheromone of the blocks to the initial value.

[0108] S9, the number of iterations for the ant colony algorithm is Nc = 1;

[0109] S10, the number of iterations for the ant colony algorithm is Nc = Nc + 1;

[0110] S11. Randomly place m ants at random points in n forest areas. These points are the four entry and exit points of the forest area.

[0111] S12, Set the ant index k = 1;

[0112] S13. Based on the parity of the number of forest route lines, perform one movement within the forest area, that is, move laterally within the forest area once, without updating the pheromone.

[0113] S14. Based on the selection probability The calculation is performed according to the following formula, and the roulette wheel rule is used to select the next forest area that the ant will visit;

[0114]

[0115] In the formula, s,w=1,2,3,4,i,j=1,2,...n. n is Let S represent point S in the i-th forest area, and n jw Let w represent point w in the i-th forest area.

[0116] S15. Ant colony algorithm updates local pheromones;

[0117] S16. Determine if the ant index k ≥ m; if so, proceed to S17, otherwise return to S13.

[0118] S17. Determine whether all m ants have visited all forest areas. If yes, proceed to S18. Otherwise, return to S12.

[0119] S18. Update global pheromones using the ant colony algorithm. Update global pheromones according to the following formula.

[0120]

[0121]

[0122] In the formula, s,w=1,2,3,4,i,j=1,2,...n. Represents path n is n jw The increase in pheromones. This indicates that the k-th ant is on path n. is n jw The pheromones left behind. t represents the current time, which can be understood as the current iteration number, and n represents the total number of forest areas. This indicates that at iteration number t, all ants have traversed all forest areas and the subsequent path n. is n jw The pheromone. k represents the ant's number, ρ represents the pheromone evaporation coefficient, a constant between [0,1], typically taken as 0.5. L k This represents the sum of the path lengths of all forest intervals traversed by the k-th ant.

[0123] S19. Determine if Nc ≥ Ncmax. If yes, execute S20; otherwise, return to S10.

[0124] S20. Output the optimal scheduling route at the current take-off and landing point, and use the reciprocal of the length of the scheduling route as the individual whale's fitness value.

[0125] S21. Assign smaller fitness values ​​to take-off and landing points within the forest area;

[0126] S22. Determine whether the fitness values ​​of all individual whales have been calculated. If yes, proceed to S23; otherwise, proceed to S7.

[0127] S23. Determine whether the number of iterations of the whale algorithm has reached the maximum number of iterations Tmax. If yes, execute S33; otherwise, execute S24.

[0128] S24. Calculate the values ​​of a, A, and C in the whale algorithm;

[0129]

[0130]

[0131]

[0132] in The integer is a random number in the range [0,1]. t represents the current iteration number, T max This indicates the maximum number of iterations for the whale algorithm. It is a real number that gradually decreases from 2 to 0, decreasing with the number of iterations. and It is a coefficient vector.

[0133] S25. Generate a random number p; p is a random number between [0,1].

[0134] S27. Determine if p < 0.5. If not, proceed to S28; otherwise, proceed to S29.

[0135] S28. The whale algorithm updates the position according to the bubble net method, using the following formula;

[0136]

[0137]

[0138] b = 1;

[0139]

[0140] In the formula, This indicates the location of the whale with the best current fitness. This represents the position matrix of the i-th whale. This represents the position matrix of the i-th whale in the next iteration. This represents the distance from the i-th whale to its prey (the best solution obtained so far). b is a constant that determines the shape of the bubble net spiral. is a real number that varies between [-2, -1], and l is a random number between [-2, 1].

[0141] S29. Determine if |A| < 1. If yes, execute S30; otherwise, execute S31.

[0142] S30. The whale algorithm updates the position according to the prey encirclement method, using the following formula;

[0143]

[0144]

[0145] In the formula, This indicates the location of the whale with the best current fitness. This represents the position matrix of the i-th whale. This represents the position matrix of the i-th whale in the next iteration. Let represent the distance from the i-th whale to the optimal whale in the encircling pattern, where b is a constant that determines the shape of the bubble net spiral. is a real number that varies between [-2, -1], and l is a random number between [-2, 1].

[0146] S31. The whale algorithm updates the position by randomly searching for prey, according to the following formula;

[0147]

[0148]

[0149] In the formula, This refers to a random whale within the current whale population. This represents the position matrix of the i-th whale. This represents the position matrix of the i-th whale in the next iteration. and It is a coefficient vector.

[0150] S32. Obtain the updated whale population. Return to execute S7.

[0151] S33: Output the optimal take-off and landing point locations and their corresponding optimal scheduling routes.

[0152] When the take-off and landing point is outside the forest area, the ant colony algorithm and whale algorithm are used to determine the target take-off and landing point location. According to the corresponding steps (S8-S20) above, the movement method where ants leave pheromones with each step is abandoned, and their movement is modified into two forms: lateral movement within the forest area and longitudinal movement between forest areas. When ants move laterally within the forest area, the movement is based on the parity of the number of flight paths within the forest area; that is, when an ant enters one entry / exit point in a forest area, the other entry / exit point is randomly determined. When ants move longitudinally between forest areas, the selection probability is calculated, and a roulette wheel rule is used to determine the entry point to the next forest area.

[0153] This invention plans helicopter take-off and landing locations, specifically for manned large agricultural aircraft. The algorithm employs a nested fusion approach: the outer algorithm uses a whale algorithm to search for take-off and landing locations, while the inner algorithm uses an improved ant colony algorithm to calculate the fitness of the whale algorithm. Combinatorial logic classification controls the feasible region of helicopter take-off and landing locations. The whale algorithm updates locations in three ways: bubble netting, prey encirclement, and random search, demonstrating strong global search capabilities. The inner improved ant colony algorithm calculates the fitness of the whale algorithm by modifying ant movement to include both lateral movement within the forest area and longitudinal movement between forest areas. Only one algorithm is needed to solve the scheduling path between forest areas with four entry and exit points, resulting in fast computation and strong global search capabilities. The use of combinatorial logic classification cleverly limits the feasible region of take-off and landing locations. By directly assigning smaller fitness values ​​to take-off and landing points located within the forest area, the search speed is accelerated without affecting the search behavior of the whale algorithm. This invention can reduce the number of times a helicopter needs to turn around in a single forest area, reducing the difficulty of pilot operation; it can shorten the length of the dispatch route between forest areas, saving non-pesticide operation time, and effectively reducing the consumption of aviation fuel, saving the economic cost of helicopter aerial pesticide application operations, and improving the efficiency of aerial pesticide application in forest areas.

[0154] helicopter from Figure 3 In the multi-forest region shown, the whale algorithm selects a take-off and landing point location from the forest region. Figure 4 Under the constraints of the combinational logic classification method shown, the take-off and landing points are controlled outside the forest area, entering... Figure 5 The text describes a forest area with four entry and exit points. After completing the pesticide application task in that area, the user exits from another of the four entry points. Figure 6 The ant search strategy shown involves flying to the next forest area, continuing until all forest areas have been traversed, before returning to the starting and landing point. The whale algorithm continuously searches for the starting and landing point location, ultimately obtaining... Figure 7 The optimal take-off and landing points and their corresponding optimal scheduling routes are shown in the figure. The convergence curve of the multi-intelligent algorithm fusion is as follows: Figure 8 As shown.

[0155] Input the coordinates of each forest region vertex; the vertices of the forest region polygon are sorted counter-clockwise. Figure 2 Taking the take-off and landing point location planning task as an example, the coordinates of each vertex in the forest area are obtained, as shown in Table 1:

[0156] Table 1

[0157] Vertex Abscissa x Ordinate y A1 50.15 521.187 A2 217.466 320.607 A3 469.379 356.731 A4 378.12 524.039 B1 410.916 308.725 B2 466.052 228.873 B3 632.409 300.169 B4 640.965 405.687 C1 660.809 470.567 C2 703.903 372.337 C3 850.932 380.575 C4 827.483 518.097 D1 246.658 206.666 D2 253.418 123.012 D3 330.875 66.538 D4 426.922 67.665 D5 477.622 159.769 D6 361.858 229.903

[0158] by Figure 3 Taking the planning task as an example, the coordinates of the four possible entry and exit points for each forest area are shown in Table 2, as follows:

[0159] Table 2

[0160]

[0161]

[0162] Structure such as Figure 4 The binary classification model shown calculates the coefficients of the linear equations for each forest area based on the two-point formula. Within each forest area, one interior point is randomly selected as the basis for coefficient matrix correction. The interior points within the four randomly selected forest areas are shown in Table 3. Table 3 is as follows:

[0163] Table 3

[0164]

[0165] According to the principle that the value of a point biased towards the forest area should be greater than 0 when substituted into the linear equation, the coefficients are corrected. For example, if an interior point in A is substituted into the boundary equation 1 of forest area A, and the value is negative, then the coefficients of the corresponding boundary equation 1 are inverted. The equation coefficient matrices before and after correction are shown in Table 4.

[0166] Table 4

[0167]

[0168] The step function outputs 1 if the input is greater than or equal to 0, and 0 otherwise. The AND gate outputs 1 if all inputs are 1, and 0 otherwise. The OR gate outputs 1 if the input contains 1, and 0 otherwise. Clearly, the number of AND gates is the same as the number of forest areas, with one AND gate corresponding to each forest area. The number of AND gate inputs is equal to the number of corresponding forest area boundaries. Given a random helicopter take-off and landing point (x, y), if the point is within forest area A, then all AND gate inputs for forest area A are 1, and the AND gate output is 1. According to OR gate logic, the OR gate output is also 1, indicating that point A is within a forest area. If the take-off and landing point is not within any forest area, then all AND gate outputs are 0, and according to OR gate logic, the OR gate output is also 0. Therefore, using the combinational logic classification method of this invention, it is easy to determine whether a point is within a forest area, and the classification accuracy is 100%.

[0169] Planning according to the algorithm provided by this invention yields the following results: Figure 7 The optimal takeoff and landing point locations and their corresponding scheduling routes are shown. The coordinates of the takeoff and landing point locations are (498.2748954, 492.5671026), and the optimal scheduling route length is 835.4922368m. The iterative convergence curve of this invention is attached. Figure 8 As shown, the algorithm converges relatively quickly. Figure 9As shown, by planning takeoff and landing point locations using the method provided by this invention, a shorter scheduling route distance can be achieved, reducing the scheduling route length by 15.52%. This is of great significance for saving fuel consumption and shortening operation time for helicopters.

[0170] This invention discloses a method for planning helicopter take-off and landing points for aerial pesticide application based on the fusion of multiple intelligent algorithms. It plans helicopter take-off and landing points, specifically for manned large agricultural aircraft. The algorithm employs a nested fusion approach: the outer algorithm uses a whale algorithm to search for take-off and landing points, while the inner algorithm uses an improved ant colony algorithm to calculate the fitness of the whale algorithm. Combinatorial logic classification is used to control the feasible region of the helicopter take-off and landing points. The whale algorithm updates positions in three ways: bubble netting, prey encirclement, and random search, demonstrating strong global search capabilities. The inner improved ant colony algorithm calculates the fitness of the whale algorithm by modifying ant movement to include both lateral movement within the forest area and longitudinal movement between forest areas. Only one algorithm is needed to solve the scheduling path between forest areas with four entry and exit points, resulting in fast computation speed and strong global search capabilities. The use of combinatorial logic classification cleverly controls the feasible region of the take-off and landing points. By directly assigning smaller fitness values ​​to take-off and landing points located within the forest area, the search speed is accelerated without affecting the search behavior of the whale algorithm.

[0171] This invention can reduce the number of times a helicopter needs to turn around in a single forest area, reducing the difficulty of pilot operation; it can shorten the length of the dispatch route between forest areas, saving non-pesticide operation time, and effectively reducing the consumption of aviation fuel, saving the economic cost of helicopter aerial pesticide application operations, and improving the efficiency of aerial pesticide application in forest areas.

[0172] The present invention provides a helicopter aerial spraying take-off and landing point location planning system, which is applied to the aforementioned helicopter aerial spraying take-off and landing point location planning method. The system includes:

[0173] The vertex location acquisition module is used to obtain the vertex location of each forest area.

[0174] The module for determining the number of U-turns, the number of pesticide application routes, and the direction vector is used to determine the number of U-turns, the number of pesticide application routes, and the direction vector for each forest area based on the vertex position of each forest area using vector modeling.

[0175] The entry and exit point determination module is used to determine the four entry and exit points of the corresponding forest area based on the parity of the number of spraying routes and the direction vector of each forest area.

[0176] The binary classification model construction module is used to determine the boundary of each forest area based on the vertex position of each forest area, and to construct a binary classification model of the take-off and landing point positions based on the forest area boundaries; the binary classification model of the take-off and landing point positions is used to determine whether the take-off and landing point positions are outside the forest area.

[0177] The judgment module is used to determine whether the take-off and landing point is outside the forest area using a binary classification model of the take-off and landing point location.

[0178] The first module for determining the target take-off and landing point location is used to determine the target take-off and landing point location using ant colony algorithm and whale algorithm when the take-off and landing point location is outside the forest area.

[0179] The second module for determining the target take-off and landing point location is used to determine the target take-off and landing point location using the whale algorithm when the take-off and landing point location is within a forest area.

[0180] The module for determining the number of U-turns, the number of application routes, and the direction vector specifically includes:

[0181] The mathematical model determination unit is used to construct a mathematical model for evaluating pesticide application routes based on the vertex location of each forest area using vector modeling.

[0182] The direction vector determination unit is used to determine the direction vector based on the mathematical model of the drug application route evaluation, using the number of U-turns and excess coverage as evaluation indicators, and employing the optimal solutions of NSGA-II and Pareto.

[0183] The binary classification model construction module specifically includes:

[0184] The forest area boundary determination unit is used to determine the boundary of each forest area based on the vertex position of each forest area.

[0185] The AND classifier determination unit is used to determine the corresponding AND classifier for each forest area by using a step function based on the boundary of each forest area; the AND classifier is used to determine whether the take-off and landing points are outside the corresponding forest area.

[0186] The binary classification model construction unit is used to construct an OR gate classifier based on each AND sub-classifier; the OR gate classifier takes the output of each AND classifier as input; the OR gate classifier is a binary classification model of the take-off and landing point positions.

[0187] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0188] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for planning the take-off and landing points for helicopter aerial pesticide application, characterized in that, include: Obtain the vertex position of each forest area; Based on the vertex position of each forest area, the vector modeling method is used to determine the number of U-turns, the number of spraying routes, and the direction vector for each forest area; The four entry and exit points for each forest area are determined based on the parity of the number of spraying routes and the direction vector. Based on the vertex position of each forest area, the boundary of each forest area is determined, and a binary classification model of the take-off and landing point positions is constructed based on the forest area boundaries; the binary classification model of the take-off and landing point positions is used to determine whether the take-off and landing point positions are outside the forest area; A binary classification model of take-off and landing point locations is used to determine whether the take-off and landing point is outside the forest area; When the take-off and landing point is located outside the forest area, the ant colony algorithm and whale algorithm are used to determine the target take-off and landing point location. When the take-off and landing point is located within a forest area, the whale algorithm is used to determine the target take-off and landing point location; The process of determining the boundary of each forest area based on its vertex position and constructing a binary classification model for take-off and landing point locations based on the forest area boundaries specifically includes: The boundaries of each forest area are determined based on the apex position of each forest area; Based on the boundary of each forest area, a step function is used to determine the corresponding AND classifier for the forest area; the AND classifier is used to determine whether the take-off and landing points are outside the corresponding forest area; An OR gate classifier is constructed based on each AND gate classifier; the OR gate classifier takes the output of each AND gate classifier as input; the OR gate classifier is a binary classification model for take-off and landing point locations; When the take-off and landing point is located outside the forest area, the ant colony algorithm and whale algorithm are used to determine the target take-off and landing point location, specifically including: An initial whale population is randomly generated within the search area of ​​the take-off and landing points; each individual whale represents a take-off and landing point. Initialize the pheromone level and iteration count for the ant colony algorithm; Place m ants randomly at random points in n forest areas; the random points are the four entry and exit points of the corresponding forest areas. Set the ant index k=1; A movement within the forest area is performed based on the parity of the number of pesticide application routes. Based on the probability of selection, the roulette wheel rule is used to determine the next forest area that the ant will visit; Update local pheromones; Determine if the ant index k ≥ m; If k < m, then return to the step of moving within the forest area once based on the parity of the number of spraying routes in the forest area; If so, then determine whether all m ants have visited all forest areas; If all forest areas have been visited, then update the global pheromones; Conversely, return to the step of setting the ant index k=1; Determine whether the number of iterations in the ant colony algorithm has reached the maximum number of iterations for the ant colony algorithm; If so, the individual whale's fitness value is determined based on the scheduling route at the current take-off and landing point location; Determine whether the fitness values ​​of all individual whales have been calculated. If the fitness values ​​of all individual whales have been calculated, then determine whether the number of iterations of the whale algorithm has reached the maximum number of iterations of the whale algorithm; If so, the corresponding take-off and landing point location will be taken as the optimal take-off and landing point location, and the corresponding scheduling route will be determined. If not, the whale population is updated using bubble netting, prey encirclement, and random prey hunting methods, and the process returns to the step of using a binary classification model based on the take-off and landing point location to determine whether the take-off and landing point location is outside the forest area. If the fitness values ​​of all individual whales have not been calculated, return to the step of using the binary classification model of take-off and landing location to determine whether the take-off and landing location is outside the forest area; Conversely, if the ant colony algorithm fails, proceed to the next iteration until the maximum number of iterations for the ant colony algorithm is reached.

2. The method for planning the take-off and landing point location of helicopter aerial spraying according to claim 1, characterized in that, The method of determining the number of U-turns, the number of spraying routes, and the direction vector for each forest area based on its vertex position using vector modeling includes: Based on the vertex location of each forest area, a mathematical model for evaluating pesticide application routes is constructed using vector modeling. Based on the mathematical model for evaluating the application route, the number of U-turns and excess coverage are used as evaluation indicators. The NSGA-II and Pareto optimal solutions are adopted to determine the direction vector.

3. The method for planning the take-off and landing point location of helicopter aerial spraying according to claim 1, characterized in that, When the take-off and landing point is located within a forest area, the whale algorithm is used to determine the target take-off and landing point location, specifically including: An initial whale population is randomly generated within the search area of ​​the take-off and landing points; each individual whale represents a take-off and landing point. Set the take-off and landing point location to the preset individual whale fitness value; Determine whether the fitness values ​​of all individual whales have been calculated. If the fitness values ​​of all individual whales have been calculated, then determine whether the number of iterations of the whale algorithm has reached the maximum number of iterations of the whale algorithm; If so, the corresponding take-off and landing point location will be taken as the optimal take-off and landing point location, and the corresponding scheduling route will be determined. If not, the whale population is updated using bubble netting, prey encirclement, and random prey hunting methods, and the process returns to the step of using a binary classification model based on the take-off and landing point location to determine whether the take-off and landing point location is outside the forest area. If the fitness values ​​of all individual whales have not been calculated, return to the step of using the binary classification model of take-off and landing location to determine whether the take-off and landing location is outside the forest area.

4. A helicopter aerial spraying take-off and landing point location planning system, applied to the helicopter aerial spraying take-off and landing point location planning method according to any one of claims 1-3, characterized in that, The system includes: The vertex position acquisition module is used to acquire the vertex position of each forest area; The module for determining the number of U-turns, the number of pesticide application routes, and the direction vector is used to determine the number of U-turns, the number of pesticide application routes, and the direction vector for each forest area based on the vertex position of each forest area using vector modeling. The entry and exit point determination module is used to determine the four entry and exit points of the corresponding forest area based on the parity of the number of spraying routes and the direction vector of each forest area. The binary classification model construction module is used to determine the boundary of each forest area based on the vertex position of each forest area, and to construct a binary classification model of the take-off and landing point positions based on the forest area boundaries; the binary classification model of the take-off and landing point positions is used to determine whether the take-off and landing point positions are outside the forest area; The judgment module is used to determine whether the take-off and landing point is outside the forest area using a binary classification model of the take-off and landing point location; The first module for determining the target take-off and landing point location is used to determine the target take-off and landing point location using ant colony algorithm and whale algorithm when the take-off and landing point location is outside the forest area; The second module for determining the target take-off and landing point location is used to determine the target take-off and landing point location using the whale algorithm when the take-off and landing point location is within a forest area. The binary classification model construction module specifically includes: The forest area boundary determination unit is used to determine the boundary of each forest area based on the vertex position of each forest area; The AND classifier determination unit is used to determine the corresponding AND classifier for each forest area based on the boundary of each forest area using a step function; the AND classifier is used to determine whether the take-off and landing points are outside the corresponding forest area; The binary classification model construction unit is used to construct an OR gate classifier based on each AND sub-classifier; the OR gate classifier takes the output of each AND classifier as input; the OR gate classifier is a binary classification model of the take-off and landing point positions.

5. A helicopter aerial spraying take-off and landing point location planning system according to claim 4, characterized in that, The module for determining the number of U-turns, the number of application routes, and the direction vector specifically includes: The mathematical model determination unit is used to construct a mathematical model for evaluating pesticide application routes based on the vertex position of each forest area using vector modeling. The direction vector determination unit is used to determine the direction vector based on the mathematical model of the drug application route evaluation, using the number of U-turns and excess coverage as evaluation indicators, and employing the optimal solutions of NSGA-II and Pareto.

Images