Wetland operation path planning method for combine harvester based on particle swarm optimization

Abstract

The invention relates to the technical field of operation path planning, in particular to a combined harvester wetland operation path planning method based on particle swarm optimization, and the method comprises the following steps: S1, building a wetland passage probability model, and outputting the passage probability value of each position point; s2, constructing a rasterized passable probability matrix based on the pass probability values output in S1; s3, designing a wetland fitness weight function taking a path point passing probability accumulated value as a core; s4, executing a particle swarm optimization algorithm, and outputting a global optimal path point sequence; and S5, performing arc interpolation processing on the global optimal path point sequence output in the S4 to generate an operation path. According to the method, through fusion of passage probability modeling and arc interpolation optimization, a continuous and feasible wetland operation path is generated on the premise of meeting the turning constraint of the combine harvester, so that the path trafficability and the operation execution stability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of operation path planning, and in particular to a wetland operation path planning method for a combine harvester based on particle swarm optimization. Background Art

[0002] In the scenario of automated agricultural machinery operations, wetland areas have soft soil and poor trafficability, which can easily cause combine harvesters to get stuck, slip, or fail to complete the planned trajectory. This places higher demands on path planning algorithms.

[0003] Existing path planning methods are mostly based on grid graph search or static cost maps. While these methods can provide the shortest path, they fail to fully consider dynamic characteristics such as soil moisture and compaction, making the planned paths difficult to implement in real-world operations. Furthermore, combine harvesters are constrained by a minimum turning radius during steering, and sharp angles or twists in the path often exceed the machine's controllable range, increasing the risk of operation failure. Therefore, a particle swarm optimization-based path planning method for combine harvesters operating in wetlands is urgently needed to address these issues. Summary of the Invention

[0004] Based on the above objectives, the present invention provides a combine harvester wetland operation path planning method based on particle swarm optimization to solve the problems of low passage probability, long path and high operation cost in complex wetland environment.

[0005] The path planning method for combine harvester wetland operation based on particle swarm optimization includes the following steps:

[0006] S1: Obtain the soil moisture distribution map and soil compaction distribution map of the operation area, establish a wetland pass probability model that integrates the soil moisture threshold and compaction threshold, and output the pass probability value of each location point;

[0007] S2: Rasterize the work area into a uniform cell grid and construct a rasterized passability probability matrix based on the passability probability value output by S1;

[0008] S3: Based on the gridded passability probability matrix generated in S2, a wetland fitness weight function is designed with the cumulative value of the passability probability of the path points as the core;

[0009] S4: Initialize the particle swarm position representation path sequence, take the wetland fitness weight function defined in S3 as the optimization target, execute the particle swarm optimization algorithm with dynamic inertia factor, and output the global optimal path point sequence;

[0010] S5: Based on the minimum turning radius constraint of the combine harvester, circular interpolation is performed on the global optimal path point sequence output by S4 to generate a smooth and executable operation path.

[0011] Optionally, the S1 specifically includes:

[0012] S11: The conductivity sensor installed at the front of the combine harvester chassis measures the soil electrical conductivity (EC) value in real time, and the soil density value is collected synchronously with the gamma-ray soil density meter.

[0013] S12: Substitute the electrical conductivity EC value collected in S11 into the following formula to convert it into a soil moisture value H, and convert the collected soil density value into a soil compaction value C. Then, use the Kriging interpolation method to interpolate the H and C values ​​to generate a soil moisture distribution map and a soil compaction distribution map covering the entire operation area;

[0014] S13: Set a safety humidity threshold and a minimum compaction threshold, and build a dual-threshold judgment model to output a pass probability value.

[0015] Optionally, the dual-threshold decision model is expressed as:

[0016] When H≤H safe And C≥C min When P=1;

[0017] When H>H safe or C<C min hour, Where P is the pass probability, ranging from [0, 1]; λ is the humidity penalty coefficient; H safe is the safe humidity threshold, C min is the minimum compaction threshold.

[0018] Optionally, the S2 specifically includes:

[0019] S21: Establish a plane rectangular coordinate system for the working area and divide it along the X-axis and Y-axis directions of the coordinate system to generate square grid units:

[0020] S22: Call the pass probability value calculated in S1, assign the pass probability value corresponding to the coordinates of the center point of each grid cell to the corresponding cell, and establish an initial mapping relationship;

[0021] S23: soil type mutation detection is performed on each grid cell. When the difference in the passage probability values ​​of two or more sampling points within the cell is greater than the set value, correction is performed;

[0022] S24: Construct a two-dimensional matrix with the grid row number and column number as the row index and column index respectively, so as to obtain the gridded passable probability matrix.

[0023] Optionally, the S3 specifically includes:

[0024] S31: The path to be evaluated is represented as an ordered coordinate sequence R; at the same time, according to the grid division rule of S21, each coordinate point is mapped to the corresponding grid cell index to generate a path grid index sequence G;

[0025] S32: extracting the pass probability value corresponding to each index of the G sequence from the rasterized pass probability matrix, and calculating the total pass probability value ProbSum of the path;

[0026] S33: Based on the path coordinate sequence R, the path length L is calculated according to the actual travel distance of the harvester;

[0027] S34: setting the length penalty coefficient α according to the maximum diagonal length of the field;

[0028] S35: Define a wetland fitness weight function, the expression is: Fitness = ProbSum-α·L, where Fitness is the path fitness value.

[0029] Optionally, the expression of the length penalty coefficient α is: Among them, L max The maximum diagonal length of the field.

[0030] Optionally, the S4 specifically includes:

[0031] S41: Based on the starting and ending positions and the passable grid set, multiple initial paths are generated by encoding in grid index order to form a particle swarm;

[0032] S42: calling the wetland fitness weight function to calculate the path fitness value of each particle and eliminating particles that do not meet the minimum turning radius constraint;

[0033] S43: Compare the current fitness with the historical best value, and update the optimal path of the individual particle and the global optimal path of the group respectively;

[0034] S44: Linearly decrease the inertia factor according to the iteration progress and update the particle speed and position;

[0035] S45: When the global optimal fitness value does not improve within the set number of iterations or reaches the maximum number of iterations, the current global optimal path point sequence is output.

[0036] Optionally, the S42 specifically includes:

[0037] S421: Set the minimum turning radius R of the combine harvester in wetland operation min As a curvature constraint benchmark;

[0038] S422: For each turning structure in the particle path, extract any three consecutive points to form two vectors;

[0039] S423: Calculate the angle between two vector segments;

[0040] S424: Construct an equivalent arc model based on the obtained included angle and calculate the actual turning radius R of this section q ;

[0041] S425: For all turning radii R in the particle path q Perform traversal judgment, if there is any turning radius that does not meet R q <R min ; then the particle path is eliminated.

[0042] Optionally, the S44 specifically includes:

[0043] S441: Initialize the inertia factor decrement parameter and set the initial inertia factor ω before the particle swarm optimization algorithm starts max and the final inertia factor ω min , and define the maximum number of iterations T max ;

[0044] S442: At the tth iteration, calculate the current inertia factor ω (t) , the expression is:

[0045] S443: Using the current inertia factor ω (t) , combining the individual historical optimal path with the current global optimal path, update the velocity vector of particle p in the t+1 round and position vector

[0046] S444: Update the path position vector Decodes into a sequence of waypoints.

[0047] Optionally, the S5 specifically includes:

[0048] S51: For all turning segments to be interpolated in the path point sequence, the midpoints thereof are used as the interpolation starting points, and a tangent direction vector determined by the inbound segment and the outbound segment is constructed;

[0049] S52: Calculate the vector that bisects the angle between two tangent lines As the center direction of the circle; at the same time, select the arc radius as the minimum turning radius R min , from the midpoint along Direction positioning arc center (x c ,y c );

[0050] S53: After determining the starting point, end point and center of the arc, perform equal-angle stepping along the arc trajectory according to the center angle and the arc direction, and interpolate several transition points;

[0051] S54: The generated arc point sequence is used to replace the sharp-angled broken line segments in the original path, and is continuously spliced ​​with the previous and next path segments to form a final smooth operation path point sequence.

[0052] Beneficial effects of the present invention:

[0053] The present invention achieves a quantitative assessment of the passability and energy efficiency of wetland operation paths by constructing a passability probability model that integrates soil moisture and compaction parameters, combining gridded area processing with fitness function design. On this basis, a particle swarm optimization algorithm with a dynamic inertia factor is adopted, which can quickly converge to a global optimal path point sequence with high passability probability, short path length and low operation cost in complex wetland environments.

[0054] The present invention introduces an arc interpolation processing mechanism to replace the turning segments in the path that do not meet the turning constraints of the combine harvester with continuously executable arc segments, so that the generated path maintains good smoothness while meeting the mechanical feasibility, thereby significantly improving the execution stability and operation reliability of the wetland operation path in actual fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0056] Figure 1 This is a schematic diagram of a wetland operation path planning method for a combine harvester according to an embodiment of the present invention;

[0057] Figure 2 Schematic diagram of the process of executing the particle swarm optimization algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. It is also noted that, to provide a more detailed description, the following embodiments are best and preferred embodiments, and those skilled in the art may employ alternative methods for implementing certain known technologies. Furthermore, the accompanying drawings are intended only to provide a more detailed description of the embodiments and are not intended to limit the present invention.

[0059] It should be noted that references in the specification to "one embodiment," "an embodiment," "an exemplary embodiment," "some embodiments," etc. indicate that the described embodiments may include specific features, structures, or characteristics, but not every embodiment necessarily includes such specific features, structures, or characteristics. In addition, when specific features, structures, or characteristics are described in conjunction with an embodiment, it is within the knowledge of persons skilled in the relevant art to implement such features, structures, or characteristics in conjunction with other embodiments (whether or not explicitly described).

[0060] In general, terms can be understood, at least in part, from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending at least in part on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0061] like Figure 1-Figure 2 As shown in FIG, the path planning method for combine harvester wetland operation based on particle swarm optimization includes the following steps:

[0062] S1: Obtain the soil moisture distribution map and soil compaction distribution map of the operation area, establish a wetland pass probability model that integrates the soil moisture threshold and compaction threshold, and output the pass probability value of each location point;

[0063] S2: Rasterize the work area into a uniform cell grid and construct a rasterized passability probability matrix based on the passability probability value output by S1;

[0064] S3: Based on the gridded passability probability matrix generated in S2, a wetland fitness weight function is designed with the cumulative value of the passability probability of the path points as the core;

[0065] S4: Initialize the particle swarm position representation path sequence, take the wetland fitness weight function defined in S3 as the optimization target, execute the particle swarm optimization algorithm with dynamic inertia factor, and output the global optimal path point sequence;

[0066] S5: Based on the minimum turning radius constraint of the combine harvester, circular interpolation is performed on the global optimal path point sequence output by S4 to generate a smooth and executable operation path.

[0067] S1 specifically includes:

[0068] S11: The soil electrical conductivity (EC) value was measured in real time using a conductivity sensor (Veris 3150) installed on the front of the combine harvester chassis. A gamma-ray soil density meter (Troxler 3440) was used to simultaneously collect soil density values. The measurement depth was set to 10-15 cm.

[0069] S12: Substitute the electrical conductivity EC value collected in S11 into the following formula to convert it into a soil moisture value H, and convert the collected soil density value into a soil compaction value C. Then, use the Kriging interpolation method to interpolate the H and C values ​​to generate a soil moisture distribution map and a soil compaction distribution map covering the entire operation area;

[0070] The conversion formula for soil moisture value H is: H = 0.64 × EC + 3.2, where EC is the soil electrical conductivity;

[0071] The conversion formula of soil compaction C is: C = ρ × g × h, where ρ is the soil density and g is the acceleration of gravity, which is 9.8 m / s 2 ; h is the thickness of the measurement layer, set to 0.12m;

[0072] S13: Set a safety humidity threshold and a minimum compaction threshold, and build a dual-threshold judgment model to output a pass probability value.

[0073] The expression of the double threshold decision model is:

[0074] When H≤H safe And C≥C min When P=1;

[0075] When H>H safe or C<C min hour, Where P is the pass probability, ranging from [0, 1]; λ is the humidity penalty coefficient, with a value of 0.1; H safe The safe humidity threshold is set to 35%, C min is the minimum compaction threshold, which is set to 80 kPa. The above steps introduce a quantitative humidity and compaction evaluation mechanism into S1 and construct a segmented passage probability function under dual thresholds, which can significantly improve the operation safety and path planning scientificity of the combine harvester in the wetland environment, avoid the risk of getting stuck or mechanical damage, and thus improve operation continuity and field passage efficiency.

[0076] S2 specifically includes:

[0077] S21: Using the coordinates of the field boundary vertices obtained by the combine harvester's BeiDou positioning system (model: Hemisphere V320) as a reference, establish a rectangular coordinate system for the work area and divide it along the X and Y axes to generate square grid cells of 1.2m × 1.2m:

[0078] S22: Call the pass probability value calculated in S1, assign the pass probability value corresponding to the coordinates of the center point of each grid cell to the corresponding cell, and establish an initial mapping relationship;

[0079] The center point coordinates are calculated using the following formula:

[0080] x c =x min +(j-0.5)×D;

[0081] y c =y min +(i-0.5)×D; where (x min ,y min ) is the coordinate of the lower left corner of the field; i is the grid row number (increasing along the Y axis); j is the grid column number (increasing along the X axis); P c is the passing probability value corresponding to the center point;

[0082] S23: soil type mutation detection is performed on each grid cell. When the difference in the passage probability values ​​of two or more sampling points within the cell is greater than the set value, correction is performed;

[0083] The steps for mutation detection and correction in S23 are as follows:

[0084] When the difference in the probability of passage between two or more sampling points in a cell satisfies: ΔP=|P a -P b |>0.3, the unit is considered to cross the boundary of soil property mutation;

[0085] A conservative decision strategy is adopted at this time:

[0086] If the above conditions are met, then P g =min(P 单元内采样点 );

[0087] If not satisfied, then P g =P c , where P g is the corrected grid passage probability value; P a ,P b is the probability of passage between any two sampling points in the grid; ΔP is the difference in the probability of passage.

[0088] S24: Using the grid row number and column number as the row index and column index respectively, a two-dimensional matrix is ​​constructed to obtain a gridded passability probability matrix. The above steps introduce a fixed-distance grid division scheme based on the width of the combine harvester track, and combine soil mutation detection with a conservative probability correction mechanism to construct a spatial passability probability matrix with actual passability, providing accurate and discrete probability input data support for the particle swarm path optimization algorithm.

[0089] S3 specifically includes:

[0090] S31: The path to be evaluated is represented as an ordered coordinate sequence R, which is represented as: R = {(x1, y1), (x2, y2), ..., (x k ,y k )}; At the same time, according to the grid division rule of S21, each coordinate point is mapped to the corresponding grid cell index to generate a path grid index sequence G, which is expressed as: G = {(i1, j1), (i2, j2), ..., (i k ,j k )};

[0091] S32: Extract the pass probability value corresponding to each index of the G sequence from the rasterized pass probability matrix and calculate the total pass probability value ProbSum of the path. The formula is: Among them, M(i q ,j q ) is the passing probability value of the qth grid of the path;

[0092] S33: Based on the path coordinate sequence R, the path length L is calculated according to the actual travel distance of the harvester. The formula is:

[0093] Among them, (x q ,y q ) is the qth coordinate point on the path;

[0094] S34: setting the length penalty coefficient α according to the maximum diagonal length of the field;

[0095] S35: Define the wetland fitness weight function, expressed as: Fitness = ProbSum-α·L, where Fitness is the path fitness value; by constructing a fitness function that integrates the sum of the pass probability and the path length penalty term, and introducing a dynamic penalty coefficient associated with the field size, the dual constraints of wetland safety and operation energy consumption can be achieved in the path optimization process, effectively improving the global rationality and practical reliability of path selection.

[0096] The expression of the length penalty coefficient α is: Among them, L max The maximum diagonal length of the field.

[0097] The design of the length penalty coefficient is based on the following three considerations:

[0098] The need for normalization of plot size: The geometric size of the plot directly affects the absolute value of the path length. Directly introducing the path length L into the path fitness function will lead to inconsistent penalty intensity between different plots, affecting the fairness of optimization. Therefore, the maximum diagonal length L of the plot is introduced. max As a normalization factor, the following ratio-type dynamic coefficient is constructed: Where γ is the empirical weight parameter;

[0099] Empirical calibration of fuel consumption parameters: A CLAAS LEXION 8900 combine harvester was field-tested in wetland operations. Its fuel consumption per unit distance traveled was converted into an operating cost penalty weight. The empirical calibration value was γ = 0.8. This value indicates that under average wetland resistance, for every additional meter of path length, the unit operating penalty is 0.8, which meets the economic constraints of field operations.

[0100] Match the dimension of the weighted result of the probability of passage: the cumulative value of the probability of passage ProbSum∈[0, k], and the path length L∈[0, L max ], by setting α to 0.8 / L max , which can ensure that α·L and ProbSum are in the same order of magnitude, avoid a certain item in the fitness function from dominating the result, and maintain the optimization balance;

[0101] The penalty coefficient defined in this way brings the following specific technical effects during the path optimization process:

[0102] Implement cross-plot adaptive optimization strategy: By using L max As a scale benchmark, the fitness function can keep the influence of the penalty term relatively constant under different plot sizes, avoiding excessive tolerance of long paths in large plots or excessive compression of paths in small plots, which leads to a decrease in traversability.

[0103] Improving the balance between path planning energy efficiency and traffic rationality: Among candidate paths with similar traffic probabilities, the algorithm prefers to select the shorter path, reducing operation time and fuel consumption. When paths with high traffic probabilities are longer, a reasonable increase in length can still be tolerated to ensure wetland navigation safety, thus achieving a dynamic balance between safety and efficiency.

[0104] Enhance the continuity of the gradient distribution of the fitness function in the optimization algorithm: By smoothly defining the α function, the fitness value change is made more sensitive to path fine-tuning, which helps the particle swarm algorithm search converge faster and avoid falling into the local optimal path.

[0105] S4 specifically includes:

[0106] S41: Based on the starting and ending positions and the passable grid set, multiple initial paths are generated by encoding in grid index order to form a particle swarm;

[0107] S42: calling the wetland fitness weight function to calculate the path fitness value of each particle and eliminating particles that do not meet the minimum turning radius constraint;

[0108] S43: Compare the current fitness with the historical best value, and update the optimal path of the individual particle and the global optimal path of the group respectively;

[0109] S44: Linearly decrease the inertia factor according to the iterative progress, update the particle speed and position, and achieve global exploration first and then local convergence;

[0110] S45: When the global optimal fitness value does not improve within the set number of iterations or reaches the maximum number of iterations, the current global optimal path point sequence is output; the above steps introduce a particle initialization mechanism based on path structure encoding and a speed update strategy with a dynamic inertia factor, which can quickly expand the path diversity in the global search phase, and gradually enhance the local aggregation ability of particles in the later stage of iteration, ultimately achieving efficient path search for wetland high-probability passage areas, and improving the robustness and engineering feasibility of path planning results.

[0111] S42 specifically includes:

[0112] S421: Set the minimum turning radius R of the combine harvester in wetland operation min As a curvature constraint benchmark, it is used to screen whether there are any sharp turns that cannot be executed in the particle path. The value is set according to the model parameters: R min =3.5m, where R min The minimum permissible turning radius for the combine harvester;

[0113] S422: For each turning structure in the particle path, extract any three consecutive points to form two vectors for turning angle analysis, which are defined as: Where: (x q-1 ,y q-1 )、(x q ,y q ) and (x q+1 ,y q+1 ) are the center coordinates of three consecutive grids in the path Represents the two-dimensional vector direction of the incoming segment and the outgoing segment;

[0114] S423: Calculate the angle between the two vector segments to reflect the degree of path bending. The angle calculation formula is: Among them, θ q is the bending angle of the qth turning point, in radians; is the vector dot product; is the vector modulus (i.e. the straight line length of each path);

[0115] S424: Construct an equivalent arc model based on the obtained included angle and calculate the actual turning radius R of this section q , used to determine whether it meets the execution requirements, the formula is: Among them, R q is the estimated turning radius corresponding to the qth turning segment of the path; d is the Euclidean distance between adjacent path points;

[0116] S425: For all turning radii R in the particle path q Perform traversal judgment, if there is any turning radius that does not meet R q <R min ; then the particle path is eliminated; otherwise, the particle is retained for subsequent fitness calculations.

[0117] S44 specifically includes:

[0118] S441: Initialize the inertia factor decrement parameter and set the initial inertia factor ω before the particle swarm optimization algorithm starts max and the final inertia factor ω min , and define the maximum number of iterations T max , used to control the linear decrease of the inertia factor during the entire optimization process;

[0119] S442: At the tth iteration, calculate the current inertia factor ω (t) , the expression is: Among them, ω (t) is the inertia factor used in the tth iteration; ω max is the initial maximum inertia factor, which is usually in the range of 0.9; ω min is the minimum inertia factor, and its value range is usually 0.4; T max is the maximum number of iterations; t is the current iteration round (starting from 0 and increasing);

[0120] S443: Using the current inertia factor ω (t) , combining the individual historical optimal path with the current global optimal path, update the velocity vector of particle p in the t+1 round and position vector The speed update includes three components: inertia, individual cognition and group coordination.

[0121] The speed update formula is:

[0122] The position update formula is: in, is the velocity vector of particle p in round t; is the position vector of particle p in round t; is the historical optimal path position vector of the particle individual; is the global optimal path position vector of the current iteration round; c1, c2 are learning factors, generally set to 2.0; r1, r2 are random numbers in the interval [0, 1], used to enhance the diversity of particle behavior;

[0123] S444: Update the path position vector Decoded into a path point sequence for use in the next round of fitness calculation.

[0124] S5 specifically includes:

[0125] S51: For all turning segments that need to be interpolated in the path point sequence, the middle point (x q ,y q ) is the starting point of interpolation, and the tangent direction vector determined by the inbound and outbound segments is constructed. The expression is:

[0126] Among them, (x q-1 ,y q-1 )、(x q ,y q ) and (x q+1 ,y q+1 ) represent three consecutive path points respectively; are the standardized entry and exit direction vectors respectively;

[0127] S52: Calculate the vector that bisects the angle between two tangent lines As the center direction of the circle; at the same time, select the arc radius as the minimum turning radius R min , from the midpoint (x q ,y q )along Direction positioning arc center (x c ,y c ), the calculation formula is: in, R represents the direction of the bisecting vector; min is the minimum turning radius of the combine harvester; (x c ,y c ) is the coordinate of the arc center;

[0128] S53: After determining the arc start point, end point and center, perform equal angle stepping along the arc trajectory according to the center angle and arc direction (clockwise or counterclockwise), and interpolate several transition points; specifically, the interpolation point coordinates are: (x i ,y i )=(xc +R min cosθ i ,y c +R min sinθ i ), where θ i is the polar angle of the i-th interpolation point; (x i ,y i ) is the coordinate of the interpolated arc point;

[0129] S54: The generated arc point sequence replaces the sharp-angled broken line segments in the original path, and is continuously spliced ​​with the previous and next path segments to form a smooth operation path point sequence that ultimately meets the curvature constraint and is continuously feasible. By introducing an arc interpolation processing mechanism based on the circle center construction method at the path turning segment, the original broken line path can be converted into a continuously differentiable, curvature-controlled arc transition trajectory, ensuring that the turning radius of the generated path at any turning point is not less than the minimum execution radius of the combine harvester, thereby significantly improving the feasibility and execution stability of the path in actual wetland terrain, and avoiding equipment yaw, burial or structural stress concentration problems caused by excessive path curvature.

[0130] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0131] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A path planning method for combine harvester wetland operation based on particle swarm optimization, characterized in that: The following steps are involved: S1: Obtain the soil moisture distribution map and soil compaction distribution map of the operation area, establish a wetland pass probability model that integrates the soil moisture threshold and compaction threshold, and output the pass probability value of each location point; S2: Rasterize the work area into a uniform cell grid and construct a rasterized passability probability matrix based on the passability probability value output by S1; S3: Based on the gridded passability probability matrix generated in S2, a wetland fitness weight function is designed with the cumulative value of the passability probability of the path points as the core; S4: Initialize the particle swarm position representation path sequence, take the wetland fitness weight function defined in S3 as the optimization target, execute the particle swarm optimization algorithm with dynamic inertia factor, and output the global optimal path point sequence; S5: Based on the minimum turning radius constraint of the combine harvester, circular interpolation is performed on the global optimal path point sequence output by S4 to generate a smooth and executable operation path.

2. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 1, characterized in that: Said S1 specifically includes: S11: The conductivity sensor installed at the front of the combine harvester chassis measures the soil electrical conductivity (EC) value in real time, and the soil density value is collected synchronously with the gamma-ray soil density meter. S12: Substitute the electrical conductivity EC value collected in S11 into the following formula to convert it into a soil moisture value H, and convert the collected soil density value into a soil compaction value C. Then, use the Kriging interpolation method to interpolate the H and C values ​​to generate a soil moisture distribution map and a soil compaction distribution map covering the entire operation area; S13: Set a safety humidity threshold and a minimum compaction threshold, and build a dual-threshold judgment model to output a pass probability value.

3. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 2, characterized in that: The expression of the dual-threshold decision model is: When H≤H safe And C≥C min When P=1; When H>H safe or C<C min hour, Where P is the pass probability, ranging from [0, 1]; λ is the humidity penalty coefficient; H safe is the safe humidity threshold, C min is the minimum compaction threshold.

4. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 1, characterized in that: The S2 specifically includes: S21: Establish a plane rectangular coordinate system for the working area and divide it along the X-axis and Y-axis directions of the coordinate system to generate square grid units: S22: Call the pass probability value calculated in S1, assign the pass probability value corresponding to the coordinates of the center point of each grid cell to the corresponding cell, and establish an initial mapping relationship; S23: soil type mutation detection is performed on each grid cell. When the difference in the passage probability values ​​of two or more sampling points within the cell is greater than the set value, correction is performed; S24: Construct a two-dimensional matrix with the grid row number and column number as the row index and column index respectively, so as to obtain the gridded passable probability matrix.

5. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 1, characterized in that: The S3 specifically includes: S31: The path to be evaluated is represented as an ordered coordinate sequence R; at the same time, according to the grid division rule of S21, each coordinate point is mapped to the corresponding grid cell index to generate a path grid index sequence G; S32: extracting the pass probability value corresponding to each index of the G sequence from the rasterized pass probability matrix, and calculating the total pass probability value ProbSum of the path; S33: Based on the path coordinate sequence R, the path length L is calculated according to the actual travel distance of the harvester; S34: setting the length penalty coefficient α according to the maximum diagonal length of the field; S35: Define a wetland fitness weight function, the expression is: Fitness = ProbSum-α·L, where Fitness is the path fitness value.

6. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 1, characterized in that: The expression of the length penalty coefficient α is: Among them, L max The maximum diagonal length of the field.

7. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 1, characterized in that: The S4 specifically includes: S41: Based on the starting and ending positions and the passable grid set, multiple initial paths are generated by encoding in grid index order to form a particle swarm; S42: calling the wetland fitness weight function to calculate the path fitness value of each particle and eliminating particles that do not meet the minimum turning radius constraint; S43: Compare the current fitness with the historical best value, and update the optimal path of the individual particle and the global optimal path of the group respectively; S44: Linearly decrease the inertia factor according to the iteration progress and update the particle speed and position; S45: When the global optimal fitness value does not improve within the set number of iterations or reaches the maximum number of iterations, the current global optimal path point sequence is output.

8. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 7, characterized in that: The S42 specifically includes: S421: Set the minimum turning radius R of the combine harvester in wetland operation min As a curvature constraint benchmark; S422: For each turning structure in the particle path, extract any three consecutive points to form two vectors; S423: Calculate the angle between two vector segments; S424: Construct an equivalent arc model based on the obtained included angle and calculate the actual turning radius R of this section q ; S425: For all turning radii R in the particle path q Perform traversal judgment, if there is any turning radius that does not meet R q <R min ; then the particle path is eliminated.

9. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 8, characterized in that: The S44 specifically includes: S441: Initialize the inertia factor decrement parameter and set the initial inertia factor ω before the particle swarm optimization algorithm starts max and the final inertia factor ω min , and define the maximum number of iterations T max ; S442: At the tth iteration, calculate the current inertia factor ω (t) , the expression is: S443: Using the current inertia factor ω (t) , combining the individual historical optimal path with the current global optimal path, update the velocity vector of particle p in the t+1 round and position vector S444: Update the path position vector Decodes into a sequence of waypoints.

10. The particle swarm optimization-based path planning method for combine harvester wetland operation according to claim 1, characterized in that: The S5 specifically includes: S51: For all turning segments to be interpolated in the path point sequence, the midpoints thereof are used as the interpolation starting points, and a tangent direction vector determined by the inbound segment and the outbound segment is constructed; S52: Calculate the vector that bisects the angle between two tangent lines As the center direction of the circle; at the same time, select the arc radius as the minimum turning radius R min , from the midpoint along Direction positioning arc center (x c ,y c ); S53: After determining the starting point, end point and center of the arc, perform equal-angle stepping along the arc trajectory according to the center angle and the arc direction, and interpolate several transition points; S54: The generated arc point sequence is used to replace the sharp-angled broken line segments in the original path, and is continuously spliced ​​with the previous and next path segments to form a final smooth operation path point sequence.

Images