A farmland forest net construction method facilitating unmanned aerial vehicle aerial prevention

Abstract

This invention discloses a method for constructing farmland shelterbelts to facilitate drone-based aerial spraying, relating to the fields of agricultural engineering and forestry ecological technology. The method includes the following steps: Drone flight path pre-planning: During the farmland planning stage, based on farmland parameters, drone performance, and spraying requirements, an improved AI algorithm generates multiple drone flight paths prioritizing straight lines and minimizing turns. The output includes drone flight path planning data containing the drone flight path coordinate set, width, and intersection points with field boundaries. Based on the path planning data, the shelterbelt network orientation is adjusted so that over 80% of the paths are parallel or oblique to the shelterbelt network. Movable nodes are set at the optimal intersection points. This invention, during the farmland planning stage, pre-plans drone flight paths based on an improved AI algorithm, using maximizing straight line segment length, minimizing turns, and ensuring complete coverage as evaluation functions, and sets movable nodes at the optimal intersection points to facilitate drone-based aerial spraying.

Description

Technical Field

[0001] This invention relates to the fields of agricultural engineering and forestry ecological technology, specifically to a method for constructing farmland shelterbelts that facilitates drone-based aerial spraying. Background Technology

[0002] Farmland shelterbelts are a core component of modern agricultural ecological barriers, possessing multiple ecological functions such as windbreak and sand fixation, microclimate regulation, and biodiversity protection. Meanwhile, drone-based aerial spraying, with its advantages of high efficiency, precision, and labor savings, has become a core technology for field management, including pesticide spraying and foliar fertilization. The synergistic efficiency of these two approaches directly impacts agricultural production benefits and ecological sustainability; therefore, "constructing farmland shelterbelts that facilitate drone-based aerial spraying" has become an important topic in agricultural engineering. However, there is a fundamental contradiction between the traditional farmland shelterbelt construction and drone-based aerial spraying synergistic models, which urgently needs to be addressed.

[0003] In the traditional model, farmland shelterbelts follow a passive logic of "building the shelterbelt first, then implementing aerial spraying": after farmland reclamation, a fixed shelterbelt (main and secondary forest belts) is designed based on the needs of windbreak and sand fixation, and tall trees (such as metasequoia and poplar) are planted to form a rigid protective system. However, when drones are used for spraying, it is discovered that there are numerous vertical intersections between the fixed shelterbelt and the drone's flight path. Drones must detour or lower their altitude, increasing spraying time by 30%-50%, wasting over 20% of pesticides due to drift, and potentially missing corner areas, affecting the control effect. This "shelter first, then aerial spraying" model essentially pits the "ecological function" of the shelterbelt against the "operational needs" of the drone, failing to resolve the contradiction at the system design level.

[0004] To resolve this conflict, existing improvement plans attempt to "passively modify existing forest networks," such as proposing the concept of "mobile forest networks," which involves transplanting some tall trees into mobile planting boxes, and then moving the trees to create pathways during aerial spraying. However, such plans still have significant drawbacks: First, the modification is delayed, requiring destructive transplanting of existing fixed forest networks, with a single tree transplant costing ≥200 yuan and potentially damaging the root system and affecting survival; second, there is redundancy in nodes, requiring one mobile node every 50 meters to cover all possible drone flight paths, resulting in a large number of nodes (more than 200 for 1,000 mu of farmland), and the movement operation relies on large machinery, which is costly; third, the repositioning accuracy is poor, with the repositioning error after node movement often exceeding ±1 meter, disrupting the continuity of the forest network and weakening its windbreak function; fourth, the adaptation is passive, failing to consider drone performance (such as operating width and turning radius), and the forest network layout is still out of sync with the needs of drone flight paths, making it impossible to achieve efficient operations based on "straight-line aerial spraying."

[0005] In summary, existing technologies either suffer from conflicting drone flight paths due to the "build the forest network first, then apply the drone" approach, or fall into the trap of high costs and low precision due to "post-application modifications," failing to achieve synergy between the "ecological function of the forest network" and the "efficiency of drone flight application" from the outset. Therefore, there is an urgent need for an "integrated farmland-forest network design method guided by drone flight paths," incorporating flight application needs into the farmland planning stage. This method, through a source design approach of "pre-planning drone flight paths → forest network adaptation → dynamic node adjustment," fundamentally resolves the contradiction between the forest network and drone flight application. This patent addresses this industry pain point by proposing a "farmland forest network construction method that facilitates drone flight application," filling the technological gap in "source-based collaborative design." Summary of the Invention

[0006] The technical solution of this invention addresses the problem of overly simplistic solutions in existing technologies by providing a solution that is significantly different from existing technologies. Specifically, the purpose of this invention is to provide a method for constructing farmland shelterbelts that facilitates drone-based aerial spraying, thereby solving the problems of conflicting flight paths between traditional farmland shelterbelts and drone-based aerial spraying, high post-project modification costs, and poor repositioning accuracy.

[0007] To achieve the above objectives, this invention provides the following technical solution: a method for constructing farmland shelterbelt networks that facilitates drone-based aerial spraying. The core of this invention is "integrated farmland-shelter network design guided by drone flight paths." Through source design involving "pre-planning of drone flight paths, shelterbelt network adaptation, and dynamic adjustment of nodes," the synergy between efficient drone-based aerial spraying and the ecological functions of the shelterbelt network is achieved. Specifically, it includes the following steps:

[0008] (1) Pre-planning of UAV flight paths: In the farmland planning stage, based on farmland parameters, UAV performance and flight requirements, multiple UAV flight paths are generated with the goal of prioritizing straight lines and minimizing turns by improving AI algorithms. The output includes UAV flight path coordinate set, width and intersection point with field boundary. The UAV flight path planning is completed in the farmland design stage as a "guiding benchmark" for forest network construction to avoid UAV flight path conflicts in the later stage.

[0009] Implementation details:

[0010] Input parameters: Field shape, e.g., rectangular, aspect ratio ≤ 3:1, single plot area ≤ 50 mu, reduce turning and boundary obstacles; Drone operating width, e.g., 8-12m, such as DJI T40 with a width of 10m, endurance, wind resistance level; Aerial spraying requirements: Insecticides need to cover the entire area, foliar fertilizers can be applied in sections.

[0011] Improved AI algorithm: A weighted comprehensive evaluation function that maximizes straight segment length, minimizes the number of turns, and ensures complete coverage.

[0012] Let cost function be This represents the total length of the straight segments in the drone's flight path for spraying. This represents the total length of the drone's flight path for spraying. The number of turns in the drone's flight path for spraying. This refers to the area of ​​farmland covered by the drone's flight path. This refers to the total area of ​​a single plot of land. The width of the node channel. For the drone's operating swath width, To match the width of the drone flight path to operational requirements.

[0013] The main drone's flight path is set along the long side of the field (spacing = operating width), and the secondary drone's flight path is connected to the short side of the main drone's flight path. This represents the proportion of straight line segments (reflecting the maximization of straight line segment length). The maximum number of turns allowed per plot of land (taken as 3). This is the normalized value for the number of turns (reflecting the minimization of the number of turns). This represents the coverage ratio (reflecting complete coverage; a value of 1 indicates full coverage).

[0014] Among them, the weighting coefficients, Prioritize straight line segments. Focus on reducing turns, Ensure full coverage and drone flight paths, and + + =1.

[0015] Weight of this patent =0.6、 =0.3、 =0.1 Quantified based on "core needs of drone-based aerial spraying": Priority ranking: Straight lines are the core of aerial spraying efficiency (9 / 10 points), followed by the number of turns (6 / 10 points), and coverage and channels are fundamental (3 / 10 points). Therefore, the weight ratio is 6:3:1. AHP calculation: The comparison matrix was constructed using the analytic hierarchy process, and the weights were calculated to be 0.6, 0.3, and 0.1, which met the consistency requirements.

[0016] Output data: Coordinate set of drone flight path for spraying, including the start point, end point, width 10m, and coordinates of potential intersections with the forest network.

[0017] (2) Forest network structure adapted to UAV flight path design: Based on the UAV flight path planning data, adjust the traditional forest network orientation so that more than 80% of the UAV flight paths are parallel or oblique to the forest network. For the UAV flight paths-forest network intersection points that cannot be avoided, set movable nodes. The movable nodes are a combination of locally movable planting boxes and tall trees. Fixed forest belts are retained on both sides of the nodes.

[0018] The design of flight paths for drone-based forest network structures also includes: The forest network alignment is adjusted so that the main forest belt runs along the short side of the field and the secondary forest belt runs along the long side, so that the angle between the forest network and the drone flight path is ≥60°.

[0019] The unavoidable intersection point is defined as the narrowest point of the forest network that the drone's flight path must cross (the width of the forest belt is ≤20m). Each work area has one node, and the forest belt is broken at the node to form a fixed belt-node-fixed belt structure.

[0020] The forest network layout proactively adapts to the flight paths of drones, reducing intersections and setting up nodes only where avoidance is impossible.

[0021] Implementation details: Prioritize avoidance: Adjust the orientation of the forest network to "the main forest belt along the short side of the field and the secondary forest belt along the long side", so that the angle between the forest network and the drone flight path is ≥60°, and more than 80% of the drone flight paths are parallel to the forest network.

[0022] Node supplement: For unavoidable intersections, the drone flight path must pass through the narrowest part of the forest network, with a forest width of ≤20m. One movable node is set up, and a 5m fixed strip is left on both sides of the node to ensure local wind protection, forming a fixed strip-node-fixed strip structure.

[0023] (3) Construction of movable nodes: A movable node is constructed at the intersection point to achieve "moving away during aerial spraying and fixing at other times". It includes a lightweight planting box and a tall tree with medium crown width and slow growth. The planting box is equipped with a wheel set, a locking device, a traction interface and a root protection device.

[0024] Implementation details: Planting box design: Fiberglass outer frame, e.g., 1.2m×1.2m×1.2m, wall thickness 3mm, strength ≥150MPa; pull-out PP inner box, 0.8m×0.8m×0.6m, filled with coconut coir + water-retaining gel; outer frame equipped with 4 polyurethane universal wheels, 20cm in diameter, load capacity 150kg / wheel; electromagnetic locking device: normally energized and locked, 3A unlocking current when power off; U-shaped traction ring, 20mm in diameter, tensile strength ≥500N; bottom equipped with elastic buffer pad, 5cm thick, made of rubber.

[0025] Tree selection: Pond cypress or dawn redwood, with an initial crown width of 3m. Wrap the root system of each tree with natural materials such as non-woven fabric or straw rope to retain moisture, prevent frost damage, protect against abrasion, and stabilize the root ball. Plant the trees in inner boxes at a depth of 0.3m, and compact the soil around the base. The planting method is not limited to the above; the specific method should be adapted based on the chosen tree species and height.

[0026] (4) Node dynamic adjustment: During aerial spraying, the movable nodes are moved by the power device to form a channel. After aerial spraying, the nodes are reset to their original positions and the accuracy is calibrated to restore the ecological function of the forest network.

[0027] The node movement distance is dynamically calculated based on the path width and the planting box size. ; in, The distance the node moves; The coefficient 2 is a safety margin, which refers to the extra space beyond the width of the passageway to the operating width of the drone. It is set to 2m, based on safety requirements such as airflow disturbance and positioning error during drone flight. The width of the planting box in the movable node refers to the dimension of the movable planting box along the vertical direction of the forest belt, such as 1.2m in the 1.2m × 1.2m planting box of this patent.

[0028] when =10m, When =1.2m, =10+2−1.2=10.8m, rounded to 15m to ensure channel redundancy and avoid movement errors.

[0029] Dynamic node adjustment specifically includes: The power unit is a small electric tractor or a manual pusher, which moves along a magnetic strip track buried in the ground.

[0030] Implementation details:

[0031] Mobile execution: Before aerial spraying, the mobile trigger automatically identifies nodes by reading the coordinate set of the drone's flight path. For example, if "node A is located at X=100m, Y=200m", a command to "move 15m to the outside of the forest belt" is sent. Power is provided by a small electric tractor, which moves along a magnetic track buried in the ground, 5cm wide and 10cm deep. During the movement, the tilt of the trees is monitored by a camera.

[0032] Reset mechanism: After aerial spraying, the tractor moves the node in the opposite direction to its original position, uses a laser rangefinder to calibrate the distance to the fixed belt, and locks it after confirming a seamless connection.

[0033] (5) System feedback optimization: Collect aerial spraying data and optimize the forest network route or number of nodes based on the number of detours and node movement time to achieve the coordinated evolution of farmland-forest network-drone aerial spraying path.

[0034] Implementation details: Data Acquisition: The UAV flight control system records flight time, number of detours, node movement time, and uniformity of pesticide coverage.

[0035] Optimization strategy: If the detour rate of a certain drone's flight path is greater than 5%, such as the forest network deviating, adjust the direction of the forest network to avoid deviation; if the node movement time is greater than 10 minutes, such as the tractor malfunctioning, split the single node into two nodes and halve the movement distance.

[0036] Compared with the prior art, the beneficial effects of the present invention are: 1. This patent fundamentally changes the inefficient traditional "flying around" model by implementing a source design of "pre-planning of drone flight paths → adapting drone flight paths to forest networks." Specifically, during the farmland design phase, based on an improved AI algorithm, drone flight paths are pre-planned using the evaluation functions of maximizing straight-line segment length, minimizing turning times, and ensuring no omissions in coverage. The primary drone flight path is prioritized along the long side of the field, while secondary drone flight paths connect to the short side, ensuring that over 80% of the drone flight paths are parallel or oblique to the forest network, thus reducing the number of intersections between drone flight paths and the forest network from the source. For unavoidable intersections, movable nodes are only placed at the narrowest point. During flight, a small electric tractor moves the nodes outward to form a channel, enabling the drone to "cross in a straight line" instead of "flying around," shortening flight time and reducing the rate of flying around.

[0037] 2. This patent abandons the redundant design of "mobile entire forest network" and adopts a "dynamic adjustment of local nodes" strategy, significantly reducing construction and maintenance costs. On one hand, the number of nodes is minimized: nodes are only placed at intersections where the drone's flight path and the forest network cannot avoid them. The planting boxes use lightweight fiberglass frames, allowing for easy movement by hand or a small tractor. On the other hand, it avoids later modification costs: the forest network is constructed simultaneously with farmland construction, and the node planting boxes have pre-embedded wheels and tracks, eliminating the need for destructive transplantation of existing forest networks. Furthermore, the repositioning accuracy is calibrated using a laser rangefinder, ensuring the integrity of the forest network and avoiding secondary repair costs due to repositioning errors.

[0038] 3. This patent, through a mechanism of "dynamic node adjustment + system feedback optimization," meets the needs of aerial spraying while preserving the ecological functions of the forest network, breaking through the functional limitations of traditional "static forest networks." Specifically, fixed forest belts are retained on both sides of the nodes to ensure the preservation of the forest network's windbreak and sand-fixing functions; during aerial spraying, the nodes are moved to form a channel, and after spraying, they are reset by electromagnetic locking devices to ensure seamless connection between the fixed belts and the nodes, restoring the continuity of the forest network. Furthermore, the system dynamically optimizes the forest network's orientation or the number of nodes by collecting aerial spraying data, achieving the synergistic evolution of "farmland-forest network-drone aerial spraying path." This "dynamic balance" model avoids the ecological damage of "felling trees for aerial spraying" and solves the efficiency problem of "fixed forest networks hindering aerial spraying," providing an innovative paradigm for the integration of agricultural ecology and smart agriculture. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating a method for constructing farmland shelterbelts that facilitates drone-based aerial spraying, according to the present invention. Detailed Implementation

[0040] 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.

[0041] Example: 2000 mu (approximately 1,333,340 square meters) (coordinated design of 20 standard farm plots)

[0042] 1. Basic parameters and classification of farmland

[0043] Total size: 2,000 mu (approximately 1,333,340 square meters), divided into 20 standard plots (each plot is 100 mu, approximately 66,667 square meters). The plots are arranged in a rectangular pattern of 2 rows and 10 columns, with a long side of 283 meters, a short side of 236 meters, and a length-to-width ratio of 1.2:1 to 3:1, which meets the requirements for regular plots.

[0044] Boundary conditions: The farmland is surrounded by drainage ditches (1m wide and 0.8m deep) and there are no obstacles such as utility poles; the soil is loam with good aeration, which is suitable for tree growth.

[0045] Drone selection: DJI T40 agricultural drone, with a working width of 10m, a flight time of 35 minutes per flight, a single flight coverage of 50-80 acres, and a wind resistance level of 6.

[0046] 2. Pre-planning of drone flight paths for spraying

[0047] Step 2.1: Input Parameter Definition

[0048] Farmland parameters: Plot shape: rectangular 283m×236m, single plot area 100 mu, boundary coordinates are established with the southwest corner of the farmland as the origin, and the total area is 2830m long (10 plots×283m) and 472m wide (2 plots×236m).

[0049] Drone performance: 10m operating swath, 35-minute flight time, wind resistance level 6.

[0050] Aerial spraying requirements: Apply pesticide during the rice heading stage, requiring full coverage. Pesticide type: Imidacloprid suspension concentrate, dosage 15L / mu.

[0051] Step 2.2: Improve the AI ​​algorithm for drone flight path generation

[0052] Evaluation function ;

[0053] This represents the total length of the straight segments in the drone's flight path for spraying. This represents the total length of the drone's flight path for spraying. The number of turns in the drone's flight path for spraying. This refers to the area of ​​farmland covered by the drone's flight path. This represents the total area of ​​a single field.

[0054] Drone flight path layout:

[0055] The main drone flight path is set along the long side of the field at 283m, with a spacing equal to the working width of 10m, for a total of 47 paths: total width 472m / 10m=47.2, taking 47 paths to cover the entire area.

[0056] Secondary drone flight paths: The shortest path connecting the main paths, with a length equal to 236m of the short side of the field, taken as 50m to avoid excessively long turns, totaling 46 paths: 46 secondary paths are set between the 47 main paths.

[0057] Output: Coordinate set of drone flight paths. The path coordinate set includes the starting / ending points of 47 main paths, the connection points of 46 secondary paths, and potential intersections with the forest network. Through GIS overlay analysis, a total of 4 unavoidable intersections were identified, located at the four corners of the field cluster.

[0058] Step 2.3: Verification of Drone Flight Path

[0059] Calculate the evaluation function: =47×283+46×50=13,301+2,300=15,601m; =15,601m (no redundant turns); =0 (line priority); =2000 / 2000=1; Therefore =0.6×(15,601 / 15,601)+0.3×(1−0 / 3)+0.1×1=0.6+0.3+0.1=1.0 (Full marks, optimal path).

[0060] 3. Forest network structure adapted to UAV flight path design

[0061] Step 3.1: Adjustment of the forest network orientation

[0062] Main forest belt: set along the short side (236m) of the field, a total of 12: 10 fields + two side boundaries, spacing = the long side of the field 283m, that is, X=0,283,566,...,2830m, a total of 12.

[0063] Sub-forest belts: set along the long side (2830m) of the field, a total of 4: 2 field rows + two side boundaries, spacing = 236m of the short side of the field, i.e. Y = 0, 236, 472m, a total of 4.

[0064] Angle between the forest network and the drone flight path: The angle between the main forest belt and the secondary drone flight path is 90°-25°=65°, which is oblique and reduces vertical intersection. The target is that more than 80% of the drone flight paths are parallel to the forest network. In this embodiment, the main drone flight path is parallel to the secondary forest belt and there is no intersection.

[0065] Step 3.2: Setting up movable nodes

[0066] Node locations: 4 unavoidable intersections (the four corners of the field cluster, with coordinates (0,0), (2830,0), (0,472), (2830,472) respectively), each node is located at the narrowest point where the path crosses the forest network;

[0067] Node structure: The forest belt is broken at each node, with one movable node and a 5m fixed belt on each side, forming a "5m fixed belt-node-5m fixed belt" structure. The total forest belt width = 5 + 1.2 + 5 = 11.2m.

[0068] 4. Construction of movable nodes

[0069] Step 4.1: Planting Box Design

[0070] Outer frame: made of fiberglass (1.2m×1.2m×1.2m, wall thickness 3mm, strength ≥150MPa), with 4 polyurethane casters at the bottom, 20cm in diameter, with a load capacity of 150kg / wheel, and an electromagnetic locking device: normally locked when powered on, unlocking current 3A when powered off, with a U-shaped traction ring on the side, 20mm in diameter, tensile strength ≥500N, and a 5cm thick rubber buffer pad at the bottom.

[0071] Inner box: a pull-out PP plastic box (0.8m×0.8m×0.6m), filled with coconut coir + water-retaining gel matrix, with drainage holes at the bottom of the inner box.

[0072] Step 4.2: Tree Planting

[0073] Variety selection: Pond cypress, initial crown width 3m, slow growth rate, annual height increase ≤0.5m, tolerant of short-term movement;

[0074] Planting method: Wrap the roots of each pond cypress tree with non-woven fabric, plant it in the inner box at a depth of 0.3m, cover the base with soil and compact it to ensure stability, one tree per node, for a total of 4 trees per 4 nodes.

[0075] 5. Dynamic node adjustment

[0076] Step 5.1: Moving before aerial spraying

[0077] Triggering condition: The system reads the coordinate set of the drone's flight path, automatically identifies 4 nodes (such as (0,0), (2830,0), etc.), and sends the command "Move 15m outward from the forest belt ( =10+2-1.2=10.8m, rounded to 15m)” command, forming a 15m wide passage.

[0078] Execution process:

[0079] Power source: Small electric tractor, with magnetic strip tracks buried in the ground;

[0080] Speed: 0.3m / s, single node movement time = 15m / 0.3m / s = 50 seconds, total movement time of 4 nodes = 4 × 50 seconds = 200 seconds ≈ 3.3 minutes, including positioning time, total 4 minutes;

[0081] Monitoring: Using cameras to monitor the tilt of trees in real time.

[0082] Step 5.2: Reset after aerial spraying

[0083] Reset operation: After the aerial spraying is completed, the tractor moves in the opposite direction along the magnetic strip track, and the node returns to its original position. The distance between the node and the fixed belt is calibrated with a laser rangefinder to ensure seamless connection with the fixed belt.

[0084] Locking Confirmation: Power on the electromagnetic locking device to check for seamless connection between the node and the fixing strip, thereby restoring the continuity of the forest network.

[0085] 6. System feedback optimization

[0086] Step 6.1: Data Acquisition

[0087] The drone flight control system recorded: total flight time ≈ 20 hours, 2000 mu, 34 sorties × 35 minutes = 1190 minutes ≈ 19.8 hours, including 4 minutes of node movement, 0% detour rate, 4 minutes of node movement time, and pesticide coverage uniformity of 0.93.

[0088] Forest network monitoring: Trees in the fixed belts were undamaged, and the root systems of trees at the nodes were not loose.

[0089] Step 6.2: Optimization and Adjustment

[0090] If the rate of bypassing a certain area is greater than 5%, such as the forest network shifting, the spacing of the main forest belt will be adjusted to 290m (originally 283m).

[0091] If the node movement takes more than 10 minutes, such as due to a tractor failure, the single node will be split into two nodes.

[0092] This implementation case, through the integrated design of "path pre-planning - forest network adaptation - node dynamic adjustment" for 2,000 mu of farmland (20 plots × 100 mu), achieved significant results such as drone-based straight-line flight spraying (0% detour rate), reduced node movement costs, and improved pesticide utilization rate, verifying the practicality and universality of the solution in large-scale farmland (100 mu per plot).

[0093] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing farmland shelterbelts that facilitates drone-based aerial spraying, characterized in that, Includes the following steps: (1) Pre-planning of UAV flight path: In the farmland planning stage, based on farmland parameters, UAV performance and flight requirements, multiple flight paths are generated by improving AI algorithms with the evaluation functions of maximizing straight segment length, minimizing turning number and covering without omission. The output includes planning data containing path coordinate set, width and intersection point with field boundary. (2) Forest network structure adapted to UAV flight path design: Based on the path planning data, adjust the direction of the forest network so that more than 80% of the path is parallel or oblique to the forest network, and set movable nodes at the optimal intersection point to form a fixed belt-node-fixed belt structure; (3) Construction of movable nodes: Construct lightweight planting boxes and tall trees with medium crown width / slow growth at the intersection point; (4) Dynamic adjustment of nodes: During aerial spraying, the nodes are moved by the power unit according to the path instructions to form the drone aerial spraying channel. The moving distance = the working width of the path + 2m safety margin - the width of the forest belt. After aerial spraying, the laser rangefinder is used to calibrate and reset to restore the ecological function of the forest network. (5) System feedback optimization: Collect aerial spraying data, and optimize path planning and node parameters in a coordinated manner based on the bypass rate, node movement time and pesticide coverage, so as to achieve the coordinated evolution of farmland-forest network-path-node.

2. The method for constructing farmland shelterbelts for drone-based aerial spraying according to claim 1, characterized in that: The pre-planning of the drone flight path for spraying in step (1) specifically includes: The farmland parameters include the shape of the plots, the area of ​​each plot, and the coordinates of boundary obstacles; The performance characteristics of the drone include operating width, flight time, and wind resistance level; The evaluation function of the improved AI algorithm is a weighted comprehensive evaluation function, with the objectives of maximizing the length of straight segments, minimizing the number of turns, and ensuring complete coverage. The specific expression is as follows: ； in, A comprehensive score for the drone's flight path. The higher the value, the better the drone's flight path for spraying. This represents the total length of the straight segments in the drone's flight path for spraying. This represents the total length of the drone's flight path for spraying. The percentage of straight line segments; The number of turns in the drone's flight path for spraying. The maximum number of turns allowed per plot of land. This is the normalized value for the number of turns; This refers to the area of ​​farmland covered by the drone's flight path. This refers to the total area of ​​a single plot of land. This refers to the coverage ratio; The width of the node channel; For the drone's operating swath width; Match the width of the drone flight path to operational requirements; Among them, the weighting coefficients, Prioritize straight line segments. Focus on reducing turns, Ensure full coverage and drone flight paths, and + + =1; The improved AI algorithm uses this evaluation function. The cost function for the drone flight path is defined as follows: the main drone flight path is set along the long side of the field, and the secondary drone flight path is the short side drone flight path that connects to the main drone flight path.

3. The method for constructing farmland shelterbelts for drone-based aerial spraying as described in claim 1, characterized in that: The synchronous calculation of the optimal location of the movable node in step (1) involves overlaying path and forest network data in GIS, identifying the narrowest point of the path-forest network intersection, and setting the node coordinates with the goal of minimizing the node movement distance and maximizing the channel width.

4. The method for constructing farmland shelterbelts for drone-based aerial spraying according to claim 1, characterized in that: The node movement distance mentioned in step (4) is dynamically calculated based on the path operation width: ； in, The distance the node moves; Coefficient 2 is a safety margin, referring to the extra space beyond the width of the passageway when it exceeds the operating width of the drone; The width of the planting box in the movable node refers to the dimension of the movable planting box along the vertical direction of the forest belt.

5. A method for constructing farmland shelterbelts for drone-based aerial spraying according to claim 4, characterized in that: The collaborative optimization path and node parameters described in step (5) adopt a feedback rule: The aerial spraying data collection includes the drone's flight time, number of detours, node movement time, and uniformity of pesticide coverage. If the detour rate of a certain route is greater than 5%, adjust the direction of the forest network or increase the number of nodes; If the node movement takes longer than 10 minutes, optimize the power unit or shorten the movement distance. The planting box has polyurethane casters with a diameter of 15-20cm and a load-bearing capacity of ≥100kg / wheel; the electromagnetic lock is normally energized and locked, and unlocked when the power is off, with an unlocking current of ≤5A.

6. The method for constructing farmland shelterbelts for drone-based aerial spraying according to claim 1, characterized in that: The dynamic adjustment of nodes in step (4) specifically includes: The power unit is a small electric tractor or a manual pusher, which moves along a magnetic strip track buried in the ground; The movement trigger automatically identifies nodes by reading the coordinate set of the drone's flight path and sends commands; The reset mechanism is calibrated using a laser rangefinder to ensure seamless connection between the node and the fixed strip.

7. The method for constructing farmland shelterbelts for drone-based aerial spraying according to claim 1, characterized in that: The construction of the movable node in step (3) specifically includes: The planting box is a nested structure of a fiberglass outer frame and a pull-out inner box. The outer frame has dimensions of 1.2m × 1.2m × 1.2m, and a single box weighs ≤ 50kg. It is equipped with 4 directional wheels and an electromagnetic lock at the bottom, and a U-shaped traction ring on the side. The inner box is filled with coconut coir and a water-retaining gel matrix, and the outer frame is equipped with an elastic cushioning pad. The tall trees mentioned are pond cypress, Metasequoia glyptostroboides, dawn redwood, or bald cypress, with an initial crown width of 3-4m.

8. A method for constructing farmland shelterbelts for drone-based aerial spraying according to claim 1, characterized in that: The forest network structure adaptation UAV flight path design mentioned in step (2) also includes: The alignment of the forest network is adjusted so that the main forest belt runs along the short side of the field and the secondary forest belt runs along the long side, so that the angle between the forest network and the drone flight path is ≥60°. The unavoidable intersection point is defined as the narrowest point in the forest network that the drone's flight path must cross. Each work area has one node, and the forest belt is broken at the node to form a fixed belt-node-fixed belt structure.

Images