CN116738704B - Digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles - Google Patents
Digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles Download PDFInfo
- Publication number
- CN116738704B CN116738704B CN202310671870.0A CN202310671870A CN116738704B CN 116738704 B CN116738704 B CN 116738704B CN 202310671870 A CN202310671870 A CN 202310671870A CN 116738704 B CN116738704 B CN 116738704B
- Authority
- CN
- China
- Prior art keywords
- fertilizer
- particles
- coordinate system
- disc
- fertilizer particles
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000003337 fertilizer Substances 0.000 title claims abstract description 315
- 239000002245 particle Substances 0.000 title claims abstract description 180
- 238000000034 method Methods 0.000 title claims abstract description 51
- 230000007480 spreading Effects 0.000 title claims abstract description 44
- 238000003892 spreading Methods 0.000 title claims abstract description 44
- 230000033001 locomotion Effects 0.000 claims abstract description 41
- 239000011159 matrix material Substances 0.000 claims abstract description 31
- 230000009466 transformation Effects 0.000 claims abstract description 29
- 238000009826 distribution Methods 0.000 claims abstract description 22
- 238000001514 detection method Methods 0.000 claims abstract description 16
- 230000004720 fertilization Effects 0.000 claims abstract description 12
- 239000008187 granular material Substances 0.000 claims description 74
- 230000008569 process Effects 0.000 claims description 28
- 238000012549 training Methods 0.000 claims description 15
- 230000005484 gravity Effects 0.000 claims description 12
- 239000013598 vector Substances 0.000 claims description 11
- 230000001133 acceleration Effects 0.000 claims description 8
- 238000013519 translation Methods 0.000 claims description 8
- 230000003068 static effect Effects 0.000 claims description 7
- 238000003708 edge detection Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000005520 cutting process Methods 0.000 claims description 5
- 238000002474 experimental method Methods 0.000 claims description 4
- 238000004519 manufacturing process Methods 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 230000026058 directional locomotion Effects 0.000 claims description 3
- 230000005764 inhibitory process Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000010998 test method Methods 0.000 claims description 3
- 230000001788 irregular Effects 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000003491 array Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000005251 gamma ray Effects 0.000 description 1
- 230000012010 growth Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008635 plant growth Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
- G06F17/13—Differential equations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Geometry (AREA)
- Evolutionary Computation (AREA)
- Operations Research (AREA)
- Computer Hardware Design (AREA)
- Computing Systems (AREA)
- Fertilizing (AREA)
Abstract
The invention provides a digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles, which comprises the following steps: s1, establishing a coordinate system; s2, calculating a transformation matrix; s3, establishing a movement model of fertilizer particles; s4, verifying a movement track model of the fertilizer particles; s5, constructing different DEM models to obtain the distribution of fertilizer particles; s6, detecting and analyzing uniformity of fertilizer particles. According to the invention, the running track of fertilizer particles is analyzed according to a field digital elevation model and by calculating a transformation matrix, meanwhile, the running track model of the fertilizer particles is verified, irregular fields encountered by a fertilizer spreader are simulated by building different DEM models, and then, the detection and uniformity analysis are carried out on fertilizer particle images under different field DEM models based on an image processing technology, so that the problem of uniformity of distribution of the fertilizer particles in the irregular fields is investigated, and reasonable and effective precise fertilization is realized.
Description
Technical Field
The invention belongs to the technical field of intelligent monitoring, and particularly relates to a digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles.
Background
The fertilizer particles are necessary agricultural investment for plant growth and crop yield optimization, and reasonable fertilization has positive significance for the formation of grain crop yield and quality, can effectively reduce pollution of water resources, improve soil structure, and are necessary measures for sustainable development of precise agriculture in China. Meanwhile, satellite navigation and automation control have been largely used in the field of precise variable fertilization. For example, high-end modern centrifugal fertilizer spreaders use Global Positioning Systems (GPS) and Geographic Information Systems (GIS) and variable fertilizer application rate technology (VRT), and by adjusting parameters of the fertilizer spreader in real time to adapt to the surrounding environment, and simultaneously combining with a fertilizer application prescription map, the optimal fertilizer application rate is achieved, and uniform fertilizer granule application is achieved. However, the field digital elevation model and the change thereof represent that the centrifugal fertilizer spreader cannot handle the disturbance of the fertilization process, the application errors also occur due to some disturbance, such as the guiding errors of the fertilizer spreader, geometric singularities of land plots, application errors of a fertilizer spreading operation platform and the like, so that when the fertilizer spreader works on the uneven ground DEM, fertilizer particles deviate from the existing running track in the transition process from one area to the other area, the spreading mode of the fertilizer particles possibly covers different areas, thereby affecting the uniformity of the distribution of the fertilizer particles in the fertilizer spreading area, and further leading to the reduction of the utilization rate of the fertilizer, the waste of resources, the increase of economic cost and a certain influence on the growth of grain crops. At present, few researches relate to the influence of uneven ground on a fertilizer spreading mode of a fertilizer spreader, so that a model is needed to combine the digital elevation of the field with the movement of the fertilizer spreader, calculate the running track of fertilizer particles in the fertilizer spreading process of the fertilizer spreader, and survey the uniformity of the fertilizer particles by adopting a novel method for detecting the uniformity of the fertilizer particles after the fertilizer is spread, thereby realizing reasonable and effective precise fertilization according to different digital elevation models of the field, improving the fertilization rate and improving the yield and quality of crops.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles, researches the influence of uneven ground on the fertilizer spreading mode of a fertilizer spreader, and realizes high-efficiency and accurate uniformity detection of fertilizer particles.
The present invention achieves the above technical object by the following means.
A digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles comprises the following steps:
step 1: establishing a coordinate system;
step 2: calculating a transformation matrix;
step 3: establishing a motion trail model of fertilizer particles;
step 4: verifying a movement track model of fertilizer particles;
step 5: constructing different DEM models to simulate the distribution of fertilizer particles;
step 6: and detecting and analyzing uniformity of the fertilizer particles.
Further, the step 1 specifically includes:
by R f ＝(0 f X f Y f Z f ) Representing the field coordinate system, origin 0 f Coordinate axis X f Coordinate axis Y f Selected on a given DEM, on a horizontal plane, coordinate axis Z f Perpendicular to the horizontal plane;
by R f′ ＝(0 f′ X f′ Y f′ Z f′ ) Representing a constant and field coordinate system R f Aligned mobile coordinate system with origin 0 f′ The method comprises the steps of connecting a middle point field on an orthogonal projection field of a connecting line between a left fertilizer spreading disc and a right fertilizer spreading disc of a fertilizer spreader;
by R t ＝(0 t X t Y t Z t ) Represents the coordinate system of the fertilizer distributor, and has an origin 0 t And 0 to f′ Overlapping;
by R d ＝(0 d X d Y d Z d ) Representing the disk coordinate system with origin 0 d Is positioned at the center of the fertilizer spreading disc, Y d The axis is the advancing direction of the fertilizer distributor, Z d Through the origin 0 d Vertical fertilizer spreading disc plane;
by R v ＝(0 v X v Y v Z v ) Representing the disk blade coordinate system with origin 0 v Is positioned at the center of the fertilizer spreading disc.
Further, the specific process of the step 2 is as follows:
step 2.1: first, a secondary field coordinate system R is established f To a mobile coordinate system R f′ Is then constructed from the translation transformation matrix of the moving coordinate system R f′ Coordinate system R of fertilizer distributor t Is a rotation transformation matrix of (a);
step 2.2: establishing a coordinate system R of a secondary fertilizer distributor t To the disc coordinate system R d Is a translation transformation matrix of (a);
step 2.3: establishing a slave disc coordinate system R d To the disc blade coordinate system R v Is a transformation matrix of (a);
step 2.4: establishing a site coordinate system R f And a disc blade coordinate system R v Global transformation matrix between.
Further, the specific process of the step 3 is as follows:
step 3.1: analyzing the movement of fertilizer particles on a fertilizer spreading disc, and establishing a mechanical equation of the movement of the fertilizer particles from the center point of the disc to the edge of the disc blade and a speed equation of the particles on the blade;
step 3.2: when the fertilizer particles run along the track, an equation of the track running of the fertilizer particles influenced by resistance and gravity is established;
step 3.3: the position of the fertilizer granules is represented by a discretized time series.
Further, the specific process of the step 3.1 is as follows:
and (3) taking the center of the fertilizer spreading disc as an origin of coordinates, and deducing a dynamic motion vector equation from a dropping point to a blade edge point of the fertilizer particles through Newton's second law:
wherein,is the friction of the fertilizer particles relative to the disc blades; />The force applied by the disc blade to the fertilizer spreading particles; />Centrifugal force of the fertilizer particles along the blades; />Is the gravity of the fertilizer spreading particles; />And->Orthogonal reaction forces of the disc and the blade, respectively; />Is the coriolis force;
the movement of fertilizer particles on the disc blades is:
wherein the initial condition ζv (0) =ζv is applied by considering the constant angular velocity of the rotation of the fertilizer spreading disc d And
time t of particle leaving disc blade e Can be calculated by numerical value, t e Instead of t in the equation, the radial component vr of the fertilizer granule velocity at the blade edge exit point is derived:
wherein the velocity v of the fertilizer granules at the outlet point e Expressed as:v t is the tangential velocity of the fertilizer particles.
Further, the specific process of the step 3.2 is as follows:
to rotate the disk at the center 0 d As the origin, Z d Is the rotation axis X d And the direction of movement of the fertilizer distributor is taken as a pointing axis, a reference coordinate system of the ballistic flight of fertilizer particles is established, the coordinate of the fertilizer particles at the outlet is used as the initial position of the ballistic movement of the particles, and the initial velocity component (v X ,v Y ,v Z ) The method comprises the following steps:
wherein, gamma tr Is fertilizer particles relative to X in a reference coordinate system d Is a horizontal track angle of (2); v e Indicating the speed of the fertilizer granules at the exit point;
the mechanical equation of the influence of the resistance coefficient and the gravity on the running condition of the fertilizer particles is as follows:
decomposing equations to describe the disk blade coordinate system R v Lower axis X v ,Y v ,Z v Directional movement and defines the landing position of the fertilizer granule:
wherein F is g Is the gravity of fertilizer particles; f (f) d Resistance to fertilizer granules: c (C) d Is the air resistance coefficient; v is the movement speed of the fertilizer particles, A is the projection area of the fertilizer particles; boundary conditions are considered to define the coordinates (X g ,Y g ,Z g ) And the velocity component (v) X ,v Y ,v Z ) Are all zero;instead of representing fertilizer particles in a disc blade coordinate system R v Along coordinate axis X as reference frame v ,Y v ,Z v Acceleration in the direction.
Further, the specific process of the step 3.3 is as follows:
recording the position and speed of the fertilizer granules in each time step in units of a certain time step, and representing the position of the fertilizer granules as a time sequence as follows:
{(t 1 ,X 1 ,Y 1 ,Z 1 ),(t 2 ,X 2 ,Y 2 ,Z 2 ),...,(t n ,X n ,Y n ,Z n )}；
wherein t is n Time representing the nth time step, (X) n ,Y n ,Z n ) Indicating the position of the fertilizer granule within the nth time step;
the position and speed of the fertilizer granules in each time step were numerically calculated using the implicit euler method:
position: s (t+Δt) =s (t) +v (t+Δt) Δt
Speed of: v (t+Δt) =v (t) +a (t+Δt) Δt
Where a (t+Δt) is the acceleration calculated by newton's second law; Δt represents the increment of the unit time step t.
Further, the specific process of the step 4 is as follows:
spreading the 25 virtual collection disc arrays perpendicular to the direction of movement of the fertilizer spreader, the arrays of collection discs being arranged on a static spread-out map obtained from the left and right discs, calculating the number and mass of particles per virtual tray by discretized time representation of the fertilizer particle trajectory model and the positions of the fertilizer particles, using the mass of particles collected in each virtual tray to define a lateral distribution of the applied mass; verifying the accuracy of the movement track model of the fertilizer particles by using the determination coefficient R:
wherein m is meas (y) A method for producing the sameIs the mean and standard deviation of the measurements, m mod (y) and->Is the average value and the standard deviation in the virtual tray; t is the number of collection trays; />Is m meas (y) standard deviation; />Is m mod Standard deviation of (y); i is the serial number of the collection disc.
Further, the specific process of the step 5 is as follows:
in the transition process of the fertilizer spreader from one area to another area, representing a non-flat DEM model encountered by the fertilizer spreader, simulating different DEM models encountered in the working process of the fertilizer spreader (7) by changing the longitudinal inclination angle or the transverse inclination angle of the fertilizer spreading disc, and obtaining fertilizer particle distribution areas under various DEM models;
the distribution of fertilizer particles is obtained by adopting a static test method, a coordinate system of a fertilizer distributor disc is obtained according to a DEM model, a projection point of the center initial position of the fertilizer distributor disc (3) on the ground is taken as a coordinate origin, and 0 is built by vertical projection d X d Y d Two-dimensional rectangular coordinate system, wherein Y d The axial direction is the advancing direction of the fertilizer spreader, the size of the fertilizer granule collecting area is 1540cm multiplied by 830cm rectangular area, the size of the collecting small grid is 50cm multiplied by 50cm, Y d The distance between each row of small grids in the axial direction is 80cm, 7 rows are put together, X d The distance between each row of small nets in the axial direction is 80cm, 11 rows are all arranged, all the collecting small nets on the ground form a collecting matrix of 7X 11, and the fertilizer distributor is positioned at X d The central position of the collecting grids in the axial direction and the distance from the collecting grids in the first row is 150cm, the grids in each row and each column are marked, and each small area is marked with a digital label n ij (i=1, 2, …,7,j =1, 2, …, 11) to obtain different fertilizer granule (5) distributions by varying the inclination of the fertilizer spreading disc to the ground.
Further, the specific process of the step 6 is as follows:
step 6.1: image acquisition and preprocessing;
installing a CCD camera on a fixed support, and acquiring an original image of fertilizer particles in a fertilizer particle collecting area from a vertical position until the fertilizer particle image of the DEM model area is acquired; then cutting the acquired images according to different DEM models, naming the images by corresponding digital labels, and classifying the images; then, marking the obtained original image by using a labelme tool, making a json file, setting a training set, and carrying out data enhancement on the training set, wherein the data enhancement comprises rotation and horizontal overturning;
step 6.2: training a model;
training the images processed in the step 6.1 by using a Mask RCNN network structure, inputting fertilizer particle images into a backbone network to obtain a group of feature images, and then up-sampling and splicing the feature images with different sizes to form a feature pyramid, and sending the feature pyramid into a regional proposal network; the position regression and classification of the fertilizer particles are preliminarily predicted by utilizing RPN, and a series of proposal areas are obtained; extracting a group of feature vectors from the proposed area on the corresponding feature map by RoIALign, classifying the RoIALign according to the feature vectors and performing frame regression; screening by using a confidence coefficient threshold value and non-maximum value inhibition, setting a confidence coefficient threshold value in a non-maximum value algorithm, only reserving detection results with the confidence coefficient larger than the threshold value, and discarding other detection results to obtain a final prediction frame and classification; cutting the feature map according to the final pre-selected frame, and generating a prediction mask by utilizing a semantic segmentation network;
obtaining an RoI region by using a prediction frame, carrying out edge detection on the RoI to obtain an edge detection dividing line of particles, dividing the whole RoI region into a plurality of mutually separated mask regions by using the edge dividing line, intersecting a mask predicted by a model with a candidate mask, indicating that the candidate mask belongs to a part of an object when the proportion of the intersecting part to the whole candidate mask exceeds a threshold value, selecting all the candidate masks with the proportion larger than the threshold value, and combining the selected candidate masks to obtain a final mask;
step 6.3: evaluating a model;
evaluation of the detection model was evaluated using the cross-over ratio IoU:
wherein P represents the proportion of the correct sample among all the fertilizer granule samples detected, T P Indicating the proportion of correct sample of detected fertilizer particles, F P Indicating the number of false samples of detected fertilizer particles, R is the recall, F N A negative number of samples representing classification errors; by F 1 As a comprehensive evaluation index; the intersection ratio IoU is used for evaluating the accuracy of a prediction frame or a prediction mask, A gt Represent the true value region, A pre Representing a prediction region;
step 6.4: detecting and evaluating the overall uniformity of the fertilizer particles 5;
the average transverse coefficient is used as a uniformity experiment index and is deduced by the following equation:
wherein,is the average transverse coefficient; m is m ij (i=1, 2, …,7,j =1, 2, …, 11) expressed as the mass of the detected fertilizer particles of the ith row and jth column corresponding to the fertilizer particle image in the small grid; n is a constant number of cells per row; m is M Total (S) Collecting the total mass of all fertilizer granule images in the area for wrapping the fertilizer granules; n (N) T Is the total number of small grids in the same area;
in a flat field, the average transverse coefficient fertilization error of the fertilizer distributor is within +/-5 percent, and the fertilization is considered to be uniform; outside this level, an average lateral coefficient higher than 105% indicates over-fertilization, while an average lateral coefficient lower than 95% indicates under-fertilization.
The invention has the following beneficial effects:
the invention designs a model combining the field elevation and the movement of a fertilizer distributor, the model can establish a centrifugal fertilizer granule running track model through simple matrix transformation, verify the model, build different DEM models to obtain the distribution of fertilizer granules, grid-divide a collected fertilizer granule area, obtain an image of the distribution of the fertilizer granules by using a CCD camera, detect the number and the quality of fertilizer granules in each grid-divided small area by combining the existing machine vision algorithm, detect the uniformity of the fertilizer granules by an average transverse coefficient, correct a control system of a fertilizer spreading disc according to the detection result of the uniformity of the fertilizer, change the transverse inclination and the longitudinal inclination angle of the disc, change the flight track of the fertilizer granules to adapt to different DEM models, uniformly distribute the fertilizer on undulating lands, solve the problem of fertilizer application uniformity of irregular fields, thereby realizing reasonable, effective and intelligent precise fertilizer application, and improving the utilization rate of the fertilizer and the yield and the quality of crops. Compared with the traditional detection method, the whole detection process is more intelligent, saves time and economic cost, can more accurately and rapidly realize the detection of the uniformity of fertilizer particles, and has stronger popularization value.
Drawings
FIG. 1 is a flow chart of a method for modeling and detecting uniformity of a digital elevation model track of centrifugal fertilizer granules;
FIG. 2 is a schematic diagram of the trajectory of the fertilizer distributor under DEM;
fig. 3 is a rear view of the dual disc fertilizer applicator;
FIG. 4 is a top view of a disk blade;
FIG. 5 is a mechanical vector diagram of the movement of fertilizer granules on a disc;
FIG. 6 is a graph of fertilizer granule outlet angle and trajectory angle;
FIG. 7 is a schematic diagram of disc tilt control;
FIG. 8 is a schematic view of a fertilizer granule collection area;
FIG. 9 is a schematic representation of an improved Mask-RCNN model for fertilizer granule testing.
In the figure: 1-a digital elevation model; 2-a track line; 3-a fertilizer spreading disc; 4-leaves; 5-fertilizer granules; 6-small grids; 7-fertilizer distributor.
Detailed Description
The invention will be further described with reference to the drawings and the specific embodiments, but the scope of the invention is not limited thereto.
The digital elevation model track modeling and uniformity detecting method of the centrifugal fertilizer spreading particles is shown in fig. 1, and comprises the following steps:
step 1: establishing a coordinate system;
by R f ＝(0 f X f Y f Z f ) Representing the field coordinate system, which is considered as a rigid body since the spreader 7 is connected to the spreader 7 by means of the back coupling, referring to fig. 2, when the spreader 7 moves along the trajectory line 2 at a given DEM, the field coordinate system follows the right-hand cartesian coordinate system principle, in which the origin 0 f Coordinate axis X f Coordinate axis Y f Selected on a given digital elevation model 1, they are required to be on a horizontal plane, coordinate axis Z f Perpendicular to the horizontal plane;
referring to FIG. 2, R is employed f′ ＝(0 f′ X f′ Y f′ Z f′ ) Representing a constant and field coordinate system R f Aligned mobile coordinate system with origin 0 f′ About the fertilizer distributor 7On the orthogonal projection field of the point field in the connecting line between the discs, its axis X f′ 、Y f′ 、Z f′ Is through a field coordinate system R f Through the process ofTranslating the transformed;
referring to FIG. 2, R is employed t ＝(0 t X t Y t Z t ) Represents the coordinate system of the fertilizer distributor, and has an origin 0 t And 0 to f′ Overlap with its axis X t 、Y t 、Z t By translating the coordinate system R f′ Through the process ofRotating and transforming;
referring to FIGS. 2 and 3, R is employed d ＝(0 d X d Y d Z d ) Representing the disk coordinate system with origin 0 d Is positioned at the center of the disc, Y d The axis is the advancing direction of the fertilizer distributor 7, Z d Through the origin 0 d Vertical to the plane of the fertilizer spreading disc 3;
referring to FIG. 4, R is employed v ＝(0 v X v Y v Z v ) Representing the coordinate system of the disc blade 4, the origin 0 thereof v Is positioned at the center of the disc and has an axis X v 、Y v 、Z v Is through a disc coordinate system R d Through matrix transformation.
Step 2: calculating a transformation matrix;
step 2.1: establishing a secondary field coordinate system R f Coordinate system R of fertilizer distributor t Is a transformation matrix of (a);
first, a secondary field coordinate system R is established f To a mobile coordinate system R f′ Is a translation transformation matrix of:
wherein,representing the secondary field coordinate system R f To a mobile coordinate system R f′ Is a translation transformation matrix of (a); the projection of the digital elevation position of the fertilizer distributor 7 in the field can be defined by R when the fertilizer distributor 7 moves along the track line 2 f A set of coordinate points (d) x ，d y ，d z ) A representation; tra XYZ Representation for the slave field coordinate system R f To a mobile coordinate system R f′ Is a mobile transformation tag of (2);
then, a slave movement coordinate system R is established f′ Coordinate system R of fertilizer distributor t Is a rotation transformation matrix of (a):
wherein,representing the slave movement of the coordinate system R f′ Coordinate system R of fertilizer distributor t Is a rotation transformation matrix of (a); />Representing a moving coordinate system R f′ X in (2) f′ Shaft rotation theta p Is a mark of (2); />Representing a moving coordinate system R f′ Y in (3) f′ Shaft rotation theta r Is a mark of (2); />Representing a moving coordinate system R f′ Z in (a) f′ Shaft rotation theta y Is a mark of (2); pitch angle theta p Roll angle theta r Yaw angle theta y Respectively about the vector axis X f′ ，Y f′ ，Z f′ Euler angle of clockwise rotation.
Step 2.2: establishing a coordinate system R of a secondary fertilizer distributor t To the disc coordinate system R d Is a translation transformation matrix of:
wherein,representing the coordinate system R of the fertilizer distributor t To the disc coordinate system R d Is a translation transformation matrix of (a); reference figures 3,e and h respectively show the disk center to fertilizer distributor coordinate system R t Origin 0 t E > 0 represents a counterclockwise rotating right disk and e < 0 represents a clockwise rotating left disk.
Step 2.3: establishing a slave disc coordinate system R d To the disc blade coordinate system R v Is a transformation matrix of (a):
wherein,representing the slave disc coordinate system R d To the disc blade coordinate system R v Referring to FIG. 4, by a circular disk coordinate system R d Z of (2) d Shaft rotation angle gamma l (outlet angle of fertilizer granules 5, i.e. projection of blade 4 on disk and disk coordinate system R v Y of (2) d Angle between axes) and translating the length +.>The translational transformation matrix (distance between the centre of the disc and the edge of the blade 4).
Step 2.4: establishing a site coordinate system R f And a disc blade coordinate system R v Global transformation matrix between:
wherein,representing the secondary field coordinate system R f To the disc blade coordinate system R v Is a global transformation matrix of (a); x, y, z are the field coordinate system R f X, Y, Z is the disk blade coordinate system R v And the coordinates of the outlet point of the lower fertilizer granule 5.
Step 3: establishing a motion trail model of the fertilizer particles 5;
step 3.1: referring to fig. 5, in fig. 5, (X, Y, Z) is the exit point coordinates of the edge of the blade 4,for the distance of the drop point of the fertilizer granule 5 to the center of the disc, ζ v For the distance of the fertilizer granule 5 from the centre of the disc when moving on the blade, < >>Is the distance between the edge of the blade 4 and the center of the disc;
analyzing the movement of the fertilizer particles 5 on the fertilizer spreading disc 3, and establishing a mechanical equation of the movement of the fertilizer particles 5 from the center point of the disc to the edge of the disc blade 4 and a speed equation of the particles on the blade 4:
assuming that the fertilizer particles 5 (hereinafter referred to as particles) are uniformly spherical, and meanwhile, no interaction force exists between the particles, the particles impact the disc at the falling point without rebound, the radial speed of the particles at the point is zero, the particles pass through the disc in a pure rolling way along the disc blade 4, the center of the disc is taken as the origin of coordinates, and the motion vector equation of the fertilizer particles 5 from the falling point to the edge point of the blade 4 is deduced through Newton's second law:
wherein,is the friction of the fertilizer particles 5 with respect to the disc blades 4; />Is the force exerted by the disc blades on the particles; />Centrifugal force of the fertilizer particles 5 along the blades 4; />Is the gravity of the particles; />And->The orthogonal reaction forces of the disc and the blade 4, respectively; />Is the coriolis force; the particles move laterally along the blade 4 about the centre of rotation 0 v Rotating blade coordinate system R v Defined as follows; /> F fd Mu.m.g, mu is the coefficient of friction between the particles and the disc, omega d Is the rotation speed of the disc, m is the mass of particles, and ζ v (t) is the movement time t of the particles on the blade 4 to the center 0 of the disc d Is a distance of (2); /> Is the angular acceleration in the direction of the spin of the particles, canDefined by the relationship between the rotational and translational speeds of a particle of diameter d: /> Indicating the angular velocity in the direction of the spin of the particles at a certain moment;
the movement of the fertilizer granules 5 on the disc blades 4 is:
wherein, referring to FIG. 6, the initial conditions in the equation are applied by considering the constant angular velocity of the disk rotationAnd->Time t of particle leaving disc blade e Can be calculated by numerical value, t e Instead of t in the equation, the radial component vr of the velocity of the fertilizer granules 5 at the exit point of the edge of the blade 4 is derived:
wherein the velocity v of the fertilizer granules 5 at the outlet point e Expressed as:v t is the tangential velocity of the fertilizer particles 5.
Step 3.2: when the fertilizer particles 5 run along the track, an equation of the resistance and gravity affecting the track running of the fertilizer particles 5 is established:
origin 0 d Is positioned at the center of the rotary disc 3, and the rotary shaft is Z d The directional axis is X d And the direction of movement of the fertilizer distributor 7 is a reference coordinate system of the ballistic flight of the particles, and the coordinates (X, Y, Z) of the particles at the outlet are used as initial positions of the ballistic movement of the particles; in a selected reference frame, the initial velocity component (v X ,v Y ,v Z ) The method comprises the following steps:
wherein, gamma tr Is the particle relative to X in the above reference coordinate system d Is a horizontal track angle of (2); gamma ray tr ＝γ out -γ l ，γ out For horizontal outlet angle, gamma out ＝arctg(v r /v e )；
The mechanical equation of the influence of the drag coefficient and gravity on the running condition of the fertilizer granules 5 is:
decomposing the above equation, describing the disk blade coordinate system R v Lower axis X v ,Y v ,Z v Directional movement and defines the landing position of the particles:
wherein the particles move in the air, and the particles self gravity F g ：F g =mg, g is the gravitational acceleration, the resistance force f the particle receives d ：ρ is the air density, C d Is the air resistance coefficient, v is the particle movement speed,a is the projected area of fertilizer granule 5, +.>Finally, boundary conditions are considered to define the coordinates (X g ,Y g ,Z g ) And the velocity component (v) X ,v Y ,v Z ) Are all zero;instead of representing particles in a disc blade coordinate system R v Along coordinate axis X as reference frame v ,Y v ,Z v Acceleration in the direction;
step 3.3: the position of the fertilizer granule 5 is represented by a discretized time sequence:
the position of the particle can be represented by a discretized time series, typically in units of time steps, the position and velocity of the particle within each time step being recorded, the position of the particle being represented by the position and velocity equation of the particle having been calculated as a time series as follows:
{(t 1 ,X 1 ,Y 1 ,Z 1 ),(t 2 ,X 2 ,Y 2 ,Z 2 ),...,(t n ,X n ,Y n ,Z n )}；
wherein t is n Time representing the nth time step, (X) n ,Y n ,Z n ) Indicating the position of the particle within the nth time step;
the position and velocity of the particles within each time step may be numerically calculated using the implicit euler method:
position: s (t+Δt) =s (t) +v (t+Δt) Δt
Speed of: v (t+Δt) =v (t) +a (t+Δt) Δt
Wherein a (t+Δt) is the acceleration calculated by newton's second law, and is solved by an iterative method; Δt represents the increment of the unit time step t.
Step 4: verifying a movement track model of the fertilizer particles 5:
verifying a movement trace model of the fertilizer granules 5 by using a simulated static virtual collection disc method, scattering 25 'virtual' collection disc arrays perpendicular to the moving direction of the fertilizer distributor 7, the shape and size of which are the same as those of the collection discs used in defining the transverse distribution measurement, the arrays of the collection discs being arranged on static development figures obtained from the left and right discs, calculating the quantity and quality of the granules of each virtual tray in discrete time representation by the fertilizer granule trace model and the positions of the fertilizer granules 5, and defining the transverse distribution of the applied quality by using the quality of the granules collected in each virtual tray; verifying the accuracy of the movement track model of the fertilizer granules 5 by using the determination coefficient R:
wherein m is meas (y) A method for producing the sameIs the mean and standard deviation of the measurements, m mod (y) and->Is the average value and the standard deviation in the virtual tray; t is the number of collection trays; />Is m meas (y) standard deviation; />Is m mod Standard deviation of (y); i is the serial number of the collection disc.
Step 5: different digital elevation model 1 (DEM) models were built to simulate the distribution of fertilizer particles 5:
referring to fig. 7 and 8, in the transition process of the fertilizer spreader 7 from one area to another, a non-flat ground digital elevation model 1 encountered by the fertilizer spreader 7 is represented, different digital elevation models 1 encountered in the working process of the fertilizer spreader 7 are simulated by changing the longitudinal inclination angle or the transverse inclination angle of the fertilizer spreading disc 3, and the distribution areas of fertilizer particles 5 under various DEM models are obtained;
referring to fig. 7 and 8, when a static test method is adopted to obtain the distribution of fertilizer particles 5, a coordinate system of a fertilizer distributor disc is obtained according to a field digital elevation model 1, a projection point of the initial position of the center of the fertilizer distributor disc 3 on the ground is taken as a coordinate origin, and a vertical projection is established to form 0 d X d Y d Two-dimensional rectangular coordinate system, wherein Y d The axial direction is the advancing direction of the fertilizer spreader 7, the size of the fertilizer granule collecting area is 1540cm multiplied by 830cm rectangular area, the size of the collecting small grid 6 is 50cm multiplied by 50cm, Y d The distance between each row of small grids 6 in the axial direction is 80cm, 7 rows are put together, X d The distance between each row of small grids 6 in the axial direction is 80cm, 11 rows are all arranged, and all the collecting small grids 6 on the ground form a 7X 11 collecting matrix, wherein the fertilizer spreading device is positioned at the X d The central position of the collecting grids in the axial direction and the distance from the collecting grids 6 in the first row is 150cm, the grids 6 in each row and each column are marked, and each small area is marked with a digital label n ij (i=1, 2, …,7,j =1, 2, …, 11) to obtain different fertilizer granule 5 distributions by varying the inclination of the fertilizer spreading disc 3 to the ground.
Step 6: detecting and analyzing uniformity of the fertilizer particles 5;
step 6.1: image acquisition and preprocessing;
installing a CCD camera on a fixed support, and acquiring original images of fertilizer particles 5 in a fertilizer particle collecting area 5 from a vertical position, so as to ensure that each image comprises a small grid 6 which is divided in a gridding way; after the images of the fertilizer particles 5 in the first grid area are acquired, moving a fixed bracket provided with a CCD camera to the next grid area, and acquiring the images of the fertilizer particles 5 in the next grid area until the images of the fertilizer particles 5 in all the areas of the digital elevation model 1 are acquired;
after the distribution images of fertilizer particles 5 under different analog digital elevation models 1 are obtained through a CCD camera, the obtained images are cut according to different digital elevation models 1, so that only grid-sized areas are ensured to be contained, and then the corresponding digital labels are used for naming and classifying;
then, marking the obtained original image by using a labelme tool to manufacture a json file, wherein 0 represents a background, 1 represents fertilizer particles 5, 400 different images are used for a training set, 77 images are contained in a verification set, 77 images represent the total number of small grids 6 in the collecting area of the fertilizer particles 5 in each experiment, the small grids are grouped into a group, and each image corresponds to a small grid n ij A label; to obtain more data, the training set is data enhanced, including rotation and horizontal flipping.
Step 6.2: training a model;
training the images processed in the step 6.1 by using a Mask RCNN network structure, inputting fertilizer particle images into a backbone network to obtain a group of feature images, then up-sampling and splicing the feature images with different sizes to form a feature pyramid, and sending the feature pyramid into a Regional Proposal Network (RPN); the position regression and classification of the fertilizer particles are preliminarily predicted by utilizing RPN, and a series of proposal areas are obtained; extracting a group of feature vectors from the proposed area on the corresponding feature map by RoIALign, classifying the RoIALign according to the feature vectors and performing frame regression; screening by using a confidence coefficient threshold value and non-maximum value inhibition, setting a confidence coefficient threshold value in a non-maximum value algorithm, only reserving detection results with the confidence coefficient larger than the threshold value, and discarding other detection results to obtain a final prediction frame and classification; and cutting the feature map according to the final pre-selected frame, and generating a prediction mask by utilizing a semantic segmentation network.
In order to modify the output Mask, the Mask RCNN network architecture is modified by making the edge of the output Mask be the edge of the actual fertilizer particle, referring to fig. 9, the RoI area is obtained by using the prediction frame, the edge detection is performed on the RoI, the edge detection dividing line of the particle is obtained, the whole RoI area is divided into a plurality of mutually separated Mask areas by using the edge dividing line, the intersection is performed between the Mask predicted by the model and the candidate Mask, when the proportion of the intersection part to the whole candidate Mask exceeds the threshold value, the candidate Mask is represented as a part of the object, all the candidate masks with the proportion larger than the threshold value are selected, and the selected candidate masks are combined to obtain the final Mask, so that the boundary of the fertilizer particle can be more accurately represented, and a detection model capable of accurately detecting the fertilizer particle is obtained.
In order to obtain a model capable of accurately detecting the fertilizer particles 5, the invention adopts the pre-trained DenseNet as a backbone network in the training process, and enhances the output of the DenseNet network by utilizing a channel attention mechanism, so that the edge information of the fertilizer particles 5 more accords with the outline of an object; on the output prediction mask, generating a mask conforming to the boundary of the fertilizer particles by using the prediction mask of the edge segmentation combined network, so as to ensure that the size of the fertilizer particles is accurate; the invention also performs fine tuning on the ROI feature extractor and the mask generator, adopts a random gradient descent (SGD) optimizer for training, has a learning rate of 0.001, a momentum of 0.9, a weight attenuation of 0.0001, 250 steps per epoch, a training batch size of 1, 50 epochs in total, a 512x512 pixel size for an input image, a size of [16,32,64,128,256] for different feature layers Anchor of PRN (region proposed network), a sliding window stride of [4,8,16,32,64], and an Anchor setting ratio of [0.5,1,2] for each layer.
Step 6.3: evaluating a model;
evaluation of the detection model was evaluated using the cross-over ratio (intersection over union, ioU):
wherein P represents the proportion of the correct sample among all the detected fertilizer granule 5 samples, T P Indicating the proportion of correct sample of detected fertilizer particles 5, F P Indicating the number of erroneous samples of the detected fertilizer particles 5, R is the recall, indicating the proportion of the correct samples of the detected fertilizer particles 5 to all the correct samples, F N A negative number of samples representing classification errors; by F 1 As a comprehensive evaluation index; the intersection ratio IoU is used for evaluating the accuracy of a prediction frame or a prediction mask and is expressed by the ratio of the intersection and the union of a real region and a prediction region, A gt Represent the true value region, A pre Representing the prediction area.
Step 6.4: detecting and evaluating the overall uniformity of the fertilizer particles 5;
the average transverse coefficient is used as a uniformity experiment index and is deduced by the following equation:
wherein,is the average transverse coefficient; m is m ij (i=1, 2, …,7,j =1, 2, …, 11) expressed as the mass of the detected fertilizer granules 5 in the ith row j column of the grid 6 corresponding to the fertilizer granule 5 image; n is a constant number of cells 6 per row; m is M Total (S) Collecting the total mass of all fertilizer granule 5 images in the area for wrapping the fertilizer granule 5; n (N) T Is the total number of small grids 6 in the same area;
in a flat field, the average transverse coefficient fertilization error of the fertilizer distributor 7 is within +/-5%, so that fertilization can be considered to be uniform; beyond this level, an average transverse coefficient calculated by the formula of greater than 105% indicates over-fertilization, while an average transverse coefficient of less than 95% indicates under-fertilization.
The examples are preferred embodiments of the present invention, but the present invention is not limited to the above-described embodiments, and any obvious modifications, substitutions or variations that can be made by one skilled in the art without departing from the spirit of the present invention are within the scope of the present invention.
Claims (4)
1. A method for modeling and detecting uniformity of a digital elevation model track of centrifugal fertilizer spreading particles is characterized by comprising the following steps:
step 1: establishing a coordinate system;
step 2: calculating a transformation matrix;
step 3: establishing a motion trail model of the fertilizer particles (5);
step 4: verifying a motion trail model of the fertilizer particles (5);
step 5: constructing different DEM models to simulate the distribution of fertilizer particles (5);
step 6: detecting and analyzing uniformity of the fertilizer particles (5);
the step 1 specifically comprises the following steps:
by R f ＝(0 f X f Y f Z f ) Representing the field coordinate system, origin 0 f Coordinate axis X f Coordinate axis Y f Selected on a given DEM, on a horizontal plane, coordinate axis Z f Perpendicular to the horizontal plane;
by R f′ ＝(0 f′ X f′ Y f′ Z f′ ) Representing a constant and field coordinate system R f Aligned mobile coordinate system with origin 0 f′ The method comprises the steps that the orthogonal projection fields of the dot fields in the connecting line between the left fertilizer spreading disc (3) and the right fertilizer spreading disc (7) are arranged;
by R t ＝(0 t X t Y t Z t ) Represents the coordinate system of the fertilizer distributor, and has an origin 0 t And 0 to f′ Overlapping;
by R d ＝(0 d X d Y d Z d ) Representing the disk coordinate system with origin 0 d Is positioned at the center of the fertilizer spreading disc (3), Y d The axis is the advancing direction of the fertilizer distributor (7), Z d Through the origin 0 d Perpendicular to the plane of the fertilizer spreading disc (3);
by R v ＝(0 v X v Y v Z v ) Representing the disk blade coordinate system with origin 0 v Is positioned at the center of the fertilizer spreading disc (3);
the specific process of the step 2 is as follows:
step 2.1: first, a secondary field coordinate system R is established f To a mobile coordinate system R f′ Is then constructed from the translation transformation matrix of the moving coordinate system R f′ Coordinate system R of fertilizer distributor t Is a rotation transformation matrix of (a);
step 2.2: establishing a coordinate system R of a secondary fertilizer distributor t To the disc coordinate system R d Is a translation transformation matrix of (a);
step 2.3: establishing a slave disc coordinate system R d To the disc blade coordinate system R v Is a transformation matrix of (a);
step 2.4: establishing a site coordinate system R f And a disc blade coordinate system R v A global transformation matrix between;
the specific process of the step 3 is as follows:
step 3.1: analyzing the movement of the fertilizer particles (5) on the fertilizer spreading disc (3), and establishing a mechanical equation of the movement of the fertilizer particles (5) from the center point of the disc to the edge of the disc blade (4) and a speed equation of the fertilizer particles (5) on the blade (4);
step 3.2: when the fertilizer particles (5) run along the track, an equation of the resistance and gravity affecting the track running of the fertilizer particles (5) is established;
step 3.3: representing the position of the fertilizer granules (5) by a discretized time sequence;
the specific process of the step 4 is as follows:
spreading an array of 25 virtual collection trays perpendicular to the direction of movement of the fertilizer spreader (7), the array of collection trays being arranged on a static spread-out map obtained from the left and right discs, calculating the number of particles and mass of each virtual tray by discretized time representation of the fertilizer particle (5) trajectory model and the position of the fertilizer particles (5), defining the lateral distribution of the applied mass using the mass of the fertilizer particles (5) collected in each virtual tray; verifying the accuracy of the motion trail model of the fertilizer particles (5) by using the decision coefficient R:
wherein m is meas (y) A method for producing the sameIs the mean and standard deviation of the measurements, m mod (y) and->Is the average value and the standard deviation in the virtual tray; t is the number of collection trays; />Is m meas (y) standard deviation; />Is m mod Standard deviation of (y); i is the serial number of the collection disc;
the specific process of the step 5 is as follows:
when the fertilizer distributor (7) transits from one area to another, a non-flat DEM model encountered by the fertilizer distributor (7) is represented, different DEM models encountered in the working process of the fertilizer distributor (7) are simulated by changing the longitudinal inclination angle or the transverse inclination angle of the fertilizer distributor disc (3), and the distribution areas of fertilizer particles (5) under various DEM models are obtained;
the distribution of fertilizer particles (5) is obtained by adopting a static test method, a coordinate system of a fertilizer distributor disc is obtained according to a DEM model, and a projection point of the center initial position of the fertilizer distributor disc (3) on the ground is taken as a coordinate origin, and 0 is built by projection in the vertical direction d X d Y d Two-dimensional rectangular coordinate system, wherein Y d The axial direction is the advancing direction of the fertilizer distributor (7), and the fertilizer granule (5) collecting areaA rectangular area with the size of 1540cm multiplied by 830cm, a collecting small grid (6) with the size of 50cm multiplied by 50cm and Y d The distance between each row of small grids (6) in the axial direction is 80cm, 7 rows are arranged in total, X d The distance between each row of small grids (6) in the axial direction is 80cm, 11 rows are all arranged, all the small grids (6) on the ground form a 7X 11 collecting matrix, and the fertilizer distributor is positioned at the X d The central position of the collecting grids in the axial direction and the distance from the collecting grids (6) in the first row is 150cm, the grids (6) in each row and each column are marked, and each small area is marked with a digital label n ij (i=1, 2, …,7,j =1, 2, …, 11) marking, obtaining different fertilizer granule (5) distributions by varying the inclination of the fertilizer spreading disc (3) to the ground;
the specific process of the step 6 is as follows:
step 6.1: image acquisition and preprocessing;
installing a CCD camera on a fixed support, and acquiring an original image of fertilizer particles (5) in a fertilizer particle collecting area from a vertical position until the fertilizer particles (5) images of all DEM model areas are acquired; then cutting the acquired images according to different DEM models, naming the images by corresponding digital labels, and classifying the images; then, marking the obtained original image by using a labelme tool, making a json file, setting a training set, and carrying out data enhancement on the training set, wherein the data enhancement comprises rotation and horizontal overturning;
step 6.2: training a model;
training the images processed in the step 6.1 by using a Mask RCNN network structure, inputting fertilizer particle images into a backbone network to obtain a group of feature images, and then up-sampling and splicing the feature images with different sizes to form a feature pyramid, and sending the feature pyramid into a regional proposal network; the position regression and classification of the fertilizer particles are preliminarily predicted by utilizing RPN, and a series of proposal areas are obtained; extracting a group of feature vectors from the proposed area on the corresponding feature map by RoIALign, classifying the RoIALign according to the feature vectors and performing frame regression; screening by using a confidence coefficient threshold value and non-maximum value inhibition, setting a confidence coefficient threshold value in a non-maximum value algorithm, only reserving detection results with the confidence coefficient larger than the threshold value, and discarding other detection results to obtain a final prediction frame and classification; cutting the feature map according to the final pre-selected frame, and generating a prediction mask by utilizing a semantic segmentation network;
obtaining an RoI region by using a prediction frame, carrying out edge detection on the RoI to obtain an edge detection dividing line of particles, dividing the whole RoI region into a plurality of mutually separated mask regions by using the edge dividing line, intersecting a mask predicted by a model with a candidate mask, indicating that the candidate mask belongs to a part of an object when the proportion of the intersecting part to the whole candidate mask exceeds a threshold value, selecting all the candidate masks with the proportion larger than the threshold value, and combining the selected candidate masks to obtain a final mask;
step 6.3: evaluating a model;
evaluation of the detection model was evaluated using the cross-over ratio IoU:
wherein P represents the proportion of the correct sample among all the detected fertilizer granule (5) samples, T P Indicating the proportion of correct sample of detected fertilizer particles (5), F P Indicating the number of false samples of the detected fertilizer granule (5), R is the recall, F N A negative number of samples representing classification errors; by F 1 As a comprehensive evaluation index; the intersection ratio IoU is used for evaluating the accuracy of a prediction frame or a prediction mask, A gt Represent the true value region, A pre Representing a prediction region;
step 6.4: detecting and evaluating the overall uniformity of the fertilizer particles 5;
the average transverse coefficient is used as a uniformity experiment index and is deduced by the following equation:
wherein,is the average transverse coefficient; m is m ij (i=1, 2, …,7,j =1, 2, …, 11) expressed as the mass of the detected fertilizer granules (5) of the ith row and jth column corresponding to the fertilizer granule image in the small grid (6); n is a constant number of cells (6) per row; m is M Total (S) Collecting the total mass of all fertilizer granule (5) images in the area for the coated fertilizer granules; n (N) T Is the total number of small grids (6) in the same area;
in a flat field, the average transverse coefficient fertilization error of the fertilizer distributor (7) is within +/-5 percent, and the fertilization is considered to be uniform; outside this level, an average lateral coefficient higher than 105% indicates over-fertilization, while an average lateral coefficient lower than 95% indicates under-fertilization.
2. The method for modeling and detecting uniformity of a digital elevation model track of centrifugal fertilizer granules according to claim 1, wherein the specific process of step 3.1 is as follows:
the center of the fertilizer spreading disc (3) is taken as the origin of coordinates, and a dynamic motion vector equation from the dropping point of the fertilizer particles (5) to the edge point of the blade (4) is deduced through Newton's second law:
wherein,is the friction force of fertilizer particles (5) relative to the disc blades (4); />Is the force exerted by the disc blades (4) on the fertilizer particles (5); />Centrifugal force of the fertilizer particles (5) along the blades (4); />Is the gravity of the fertilizer particles (5); />And->Orthogonal reaction forces of the disc and the blade (4), respectively; />Is the coriolis force;
the movement of the fertilizer particles (5) on the disc blades (4) is as follows:
wherein the initial conditions are applied by taking into account the constant angular velocity of rotation of the fertilizer spreading disc (3)And->Time t of leaving the disc blade of fertilizer particles (5) e By numerical valuesCalculation with t e Instead of t in the equation, the radial component v of the velocity of the fertilizer particles (5) at the exit point of the blade (4) edge is derived r ：
Wherein the velocity v of the fertilizer particles (5) at the exit point e Expressed as:v t is the tangential velocity of the fertilizer particles (5).
3. The method for modeling and detecting uniformity of a digital elevation model track of centrifugal fertilizer granules according to claim 1, wherein the specific process of step 3.2 is as follows:
with 0 at the centre of the rotating disc (3) d As the origin, Z d Is the rotation axis X d And the direction of movement of the fertilizer distributor (7) is taken as a pointing axis, a reference coordinate system of ballistic flight of the fertilizer particles (5) is established, the coordinates of the fertilizer particles (5) at the outlet are used as initial positions of the ballistic movement of the particles, and the initial velocity component (v X ,v Y ,v Z ) The method comprises the following steps:
wherein, gamma tr Is fertilizer particles (5) relative to X in a reference coordinate system d Is a horizontal track angle of (2); v e Indicating the speed of the fertilizer granules (5) at the exit point;
the mechanical equation of the influence of the resistance coefficient and the gravity on the running condition of the fertilizer particles (5) is as follows:
decomposing equations to describe the disk blade coordinate system R v Lower axis X v ,Y v ,Z v Directional movement and defines the landing position of the fertilizer granule (5):
wherein F is g Is the self gravity of the fertilizer particles (5); f (f) d Resistance to the fertilizer granules (5): c (C) d Is the air resistance coefficient; v is the movement speed of the fertilizer particles (5), and A is the projection area of the fertilizer particles (5); defining coordinates (X) of where fertilizer granules (5) fall in the field elevation model field in consideration of boundary conditions g ,Y g ,Z g ) And the velocity component (v) of the fertilizer granules (5) X ,v Y ,v Z ) Are all zero;instead of representing fertilizer granules (5) in a disc blade coordinate system R v Along coordinate axis X as reference frame v ,Y v ,Z v Acceleration in the direction.
4. The method for modeling and detecting uniformity of a digital elevation model track of centrifugal fertilizer granules according to claim 1, wherein the specific process of step 3.3 is as follows:
recording the position and speed of the fertilizer granule (5) in each time step in units of a certain time step, and representing the position of the fertilizer granule (5) as a time sequence as follows:
{(t 1 ,X 1 ,Y 1 ,Z 1 ),(t 2 ,X 2 ,Y 2 ,Z 2 ),...,(t n ,X n ,Y n ,Z n )}；
wherein t is n Time representing the nth time step, (X) n ,Y n ,Z n ) Representing the position of the fertilizer granule (5) within the nth time step;
the position and the speed of the fertilizer granules (5) in each time step are numerically calculated by implicit Euler method:
position: s (t+Δt) =s (t) +v (t+Δt) Δt
Speed of: v (t+Δt) =v (t) +a (t+Δt) Δt
Where a (t+Δt) is the acceleration calculated by newton's second law; Δt represents the increment of the unit time step t.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310671870.0A CN116738704B (en) | 2023-06-08 | 2023-06-08 | Digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310671870.0A CN116738704B (en) | 2023-06-08 | 2023-06-08 | Digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116738704A CN116738704A (en) | 2023-09-12 |
| CN116738704B true CN116738704B (en) | 2023-12-12 |
Family
ID=87912722
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310671870.0A Active CN116738704B (en) | 2023-06-08 | 2023-06-08 | Digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116738704B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118037814A (en) * | 2024-01-03 | 2024-05-14 | 南京农业大学 | Automatic detection system and method for granular fertilizer deposition distribution mode |
| CN118211766B (en) * | 2024-05-17 | 2024-07-12 | 四川王家渡食品股份有限公司 | Low-temperature luncheon meat production monitoring method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112176136A (en) * | 2020-09-24 | 2021-01-05 | 中南大学 | Method and system for modeling movement locus of furnace charge on U-shaped chute of blast furnace |
| CN114863106A (en) * | 2022-05-09 | 2022-08-05 | 南京农业大学 | Centrifugal fertilizer spreading particle distribution detection method |
-
2023
- 2023-06-08 CN CN202310671870.0A patent/CN116738704B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112176136A (en) * | 2020-09-24 | 2021-01-05 | 中南大学 | Method and system for modeling movement locus of furnace charge on U-shaped chute of blast furnace |
| CN114863106A (en) * | 2022-05-09 | 2022-08-05 | 南京农业大学 | Centrifugal fertilizer spreading particle distribution detection method |
Non-Patent Citations (1)
| Title |
|---|
| 施印炎.中国博士学位论文全文数据库.2020,正文第19-107页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116738704A (en) | 2023-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116738704B (en) | Digital elevation model track modeling and uniformity detecting method for centrifugal fertilizer spreading particles | |
| Yinyan et al. | Numerical simulation of spreading performance and distribution pattern of centrifugal variable-rate fertilizer applicator based on DEM software | |
| Wu et al. | Simulation and parameter optimisation of a centrifugal rice seeding spreader for a UAV | |
| Villette et al. | Hybrid centrifugal spreading model to study the fertiliser spatial distribution and its assessment using the transverse coefficient of variation | |
| CN105203023B (en) | A kind of one-stop scaling method of vehicle-mounted three-dimensional laser scanning system placement parameter | |
| CN104694680B (en) | Control method for radial distribution of blast furnace burden layer structure | |
| Sharipov et al. | Variable rate application accuracy of a centrifugal disc spreader using ISO 11783 communication data and granule motion modeling | |
| CN107860316B (en) | Corn kernel three-dimensional parameter measuring device and its measurement method | |
| CN113012150A (en) | Feature-fused high-density rice field unmanned aerial vehicle image rice ear counting method | |
| CN104166127B (en) | A kind of multi-beam interleaved projections and the high ski-runnings assuming parabola Hough transform jump formula objective accumulation detection method more | |
| CN105950807A (en) | Multi-information fusion modeling method for shapes of burden surfaces in burden distribution process of blast furnace | |
| Landl et al. | Measuring root system traits of wheat in 2D images to parameterize 3D root architecture models | |
| Abbou-Ou-Cherif et al. | On-the-field simulation of fertilizer spreading: Part 3–Control of disk inclination for uniform application on undulating fields | |
| CN114694047A (en) | Corn sowing quality evaluation method and device | |
| CN117132105A (en) | Collision risk assessment method of satellite failure explosion fragments on large-scale constellation | |
| Fan et al. | Airflow characteristics of rotorcraft plant protection UAV operating in rice fields | |
| García-Ramos et al. | Application of a 3-D laser scanner for characterising centrifugal fertiliser spreaders | |
| CN107329190B (en) | Imaging test method for fixed star sensitivity of static meteorological satellite | |
| Van Liedekerke et al. | Recent results of experimentation and DEM modeling of centrifugal fertilizer spreading | |
| CN112781890A (en) | Fire-fighting aircraft drop test ground adhesion density measurement method | |
| Piron et al. | On-the-field simulation of fertilizer spreading: Part 2–Uniformity investigation | |
| Cool et al. | Comparing different methods of using collecting trays to determine the spatial distribution of fertiliser particles | |
| CN111421548B (en) | Mobile robot positioning method and system | |
| CN110148218A (en) | A kind of method of high-volume airborne lidar point cloud data global optimization | |
| Chok | Improving granular fertiliser aerial application for hill country farming: a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Agricultural Engineering at Massey University, Palmerston North, New Zealand |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |