CN117480927B - Intelligent fertilization method and device - Google Patents

Intelligent fertilization method and device Download PDF

Info

Publication number
CN117480927B
CN117480927B CN202311534388.9A CN202311534388A CN117480927B CN 117480927 B CN117480927 B CN 117480927B CN 202311534388 A CN202311534388 A CN 202311534388A CN 117480927 B CN117480927 B CN 117480927B
Authority
CN
China
Prior art keywords
data
soil
information
clustering
fertilization
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
Application number
CN202311534388.9A
Other languages
Chinese (zh)
Other versions
CN117480927A (en
Inventor
魏建林
谭德水
崔荣宗
郑福丽
李燕
吴小宾
马垒
王利
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong Academy of Agricultural Sciences
Original Assignee
Shandong Academy of Agricultural Sciences
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shandong Academy of Agricultural Sciences filed Critical Shandong Academy of Agricultural Sciences
Priority to CN202311534388.9A priority Critical patent/CN117480927B/en
Publication of CN117480927A publication Critical patent/CN117480927A/en
Application granted granted Critical
Publication of CN117480927B publication Critical patent/CN117480927B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01C—PLANTING; SOWING; FERTILISING
    • A01C21/00—Methods of fertilising, sowing or planting
    • A01C21/007—Determining fertilization requirements
    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01C—PLANTING; SOWING; FERTILISING
    • A01C15/00—Fertiliser distributors
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/56—Extraction of image or video features relating to colour
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
    • G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/776—Validation; Performance evaluation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30181—Earth observation
    • G06T2207/30188—Vegetation; Agriculture

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Databases & Information Systems (AREA)
  • Soil Sciences (AREA)
  • Environmental Sciences (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Quality & Reliability (AREA)
  • Data Mining & Analysis (AREA)
  • Probability & Statistics with Applications (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides an intelligent fertilization method and device, wherein the method comprises the following steps: acquiring soil information of a region to be fertilized and soil form information in a fertilization process; the soil information includes: image data, soil physical and chemical property data, historical fertilization data, initial crop growth conditions and crop root system distribution conditions; the soil morphology information comprises ground vibration data and surface disturbance data; preprocessing soil information and soil morphology information to obtain preprocessed data; inputting the preprocessing data into a fertilizing position determining model based on a deep learning neural network to obtain a position determining result; and controlling the horizontal distance (based on the position of crops) of the downward movement of the fertilizing drill leg and the vertical distance (i.e. depth) of the downward movement of the fertilizing drill leg of the intelligent fertilizer applicator based on the ground according to the position determination result. According to the invention, the influence of soil information and soil morphology information on the fertilization position is more comprehensively considered, so that the fertilizer utilization rate and the precision degree of a fertilization determination model are improved.

Description

Intelligent fertilization method and device
Technical Field
The invention relates to the technical field of fertilization, in particular to an intelligent fertilization method and device.
Background
The traditional fertilization mode adopts uniform fertilization operation, namely, an average fertilizer amount is applied in a cultivated area. However, the soil fertility of different plots in the same cultivated area is different and even has larger difference, so that the traditional uniform fertilization can cause insufficient fertilization of the plots with lower fertility to influence the growth of crops, and the plots with higher fertility can be excessively fertilized, so that the fertilizer is wasted, the production cost is increased, and even the environmental pollution can be caused. The "4R" nutrient management strategy proposed by the International Plant Nutrient Institute (IPNI) covers all scientific principles related to fertilization and is widely accepted in the industry. The 4R nutrient management is to select the Right fertilizer variety (Right source), use the Right fertilizer amount (RIGHT RATE), apply at the Right time (RIGHTTIME) and place (RIGHTPLACE), and the first letter of the 4 Right English letters is R, so the name is 4R. When in fertilization, the fertilizer is applied to the distribution range of the crop root system which is favorable for absorption according to the crop growth requirement amount and proportion as accurately as possible, so that the utilization rate of the fertilizer is improved.
At present, an accurate variable rate fertilizer applicator is adopted in the agricultural field, real-time positioning information is received through a GPS on a fertilizer application machine, fertilizer demand calculation of a land where the fertilizer applicator is located is combined, and variable rate fertilization is realized by adjusting the rotating speed of a fertilizer discharging shaft. In chinese patent ZL200520106231.7, the upper computer is connected to the GPS module and the Geographic Information System (GIS) module, the controller is connected to the upper computer, and the hydraulic valve is connected to the controller through a relay, and the opening of the hydraulic valve is adjusted by a control signal generated by the relay, so as to drive the rotation speed of the hydraulic motor connected to the hydraulic valve, and control the rotation speed of the fertilizer mechanism connected to the hydraulic motor through a gear. Because the fertilizer discharge amount has a certain corresponding relation with the rotating speed of the fertilizer mechanism, the aim of variable fertilization according to the needs can be achieved by controlling the rotating speed of the fertilizer mechanism. And a rotation speed sensor can be arranged on the motor gear part and used as a feedback quantity to be connected into the controller for correcting the rotation speed of the motor so as to achieve dynamic stability.
Therefore, most of the prior art only considers the control of the fertilizing amount, but the precise control of the fertilizing position is rarely involved, so that the fertilizing effect is poor, and the crop yield cannot be effectively improved.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide an intelligent fertilization method and device.
In order to achieve the above object, the present invention provides the following solutions:
In one aspect, the invention provides an intelligent fertilization method, which is applied to an intelligent fertilizer distributor and comprises the following steps:
Acquiring soil information of a region to be fertilized and soil form information in a fertilization process; the soil information includes: image data, soil physical and chemical property data, historical fertilization data, initial crop growth conditions and crop root system distribution conditions; the soil morphology information comprises ground vibration data and surface disturbance data;
preprocessing the soil information and the soil morphology information to obtain preprocessed data;
inputting the preprocessing data into a fertilizing position determining model based on a deep learning neural network to obtain a position determining result;
And controlling the horizontal distance (based on the position of crops) of the intelligent fertilizer applicator and the vertical distance (based on the ground) of the intelligent fertilizer applicator according to the position determination result. .
Preferably, the construction method of the fertilization position determination model comprises the following steps:
constructing a historical soil information database and a historical soil morphology information database by a method of investigating and collecting materials;
Dividing the historical soil information database and the historical soil morphology information database based on a K-fold cross validation method to generate a training set matrix and a validation set matrix;
Based on the training set matrix and the verification set matrix, determining an optimal smoothing factor of the generalized regression neural network model by utilizing a differential evolution algorithm;
and inputting the optimal smoothing factors into the generalized regression neural network model for training to obtain a fertilization position determination model.
Preferably, the preprocessing is performed on the soil information and the soil morphology information to obtain preprocessed data, including:
normalizing the soil information and the soil morphology information to obtain normalized data;
clustering the normalized data to obtain a plurality of clustering subsets;
and obtaining the preprocessing data according to the clustering subset.
Preferably, clustering the normalized data to obtain a plurality of cluster subsets includes:
Constructing a target clustering function according to the distance from the data point in the normalized data to the clustering center; wherein, the objective clustering function is:
where v i denotes the ith cluster center, n denotes the total number of data points, Representing the membership degree of the data point x j to the ith cluster center, d ij＝||xj-vi||,dij represents the distance from the data point x j to the ith cluster center, and lambda represents a weighting parameter;
Carrying out iterative solution on the target clustering function to obtain a clustering update model;
and clustering the normalized data by using the cluster updating model to obtain a plurality of cluster subsets.
Preferably, the preprocessing of the soil information and the soil morphology information is performed to obtain preprocessed data, and then the method further comprises:
Performing picture enhancement processing on the image data in the preprocessing data to obtain a data set;
Extracting weeds and crops in the data set by adopting an HSV color model, and then marking the crops;
inputting the data set marked with the crops into a crop growth model based on a YOLO network for identification to obtain crop growth information;
the crop growth information is added to the pre-treatment data.
Preferably, after controlling the horizontal distance (based on the position of the crops) of the intelligent fertilizer applicator and the vertical distance (i.e. depth) of the intelligent fertilizer applicator, based on the ground, the intelligent fertilizer applicator further comprises:
after a preset growing period, detecting the crop growth condition and crop yield index of the area to be fertilized to obtain crop biological characters and yield index information;
Training and parameter optimization are carried out on the fertilization position determination model according to the crop biological characters and the yield index information so as to obtain an optimized fertilization position determination model; the optimized fertilization position determination model is used for determining the position of next fertilization.
The invention also provides an intelligent fertilizer apparatus, which is applied to an intelligent fertilizer distributor, and comprises:
The data acquisition module is used for acquiring soil information of a region to be fertilized and soil form information in the fertilization process; the soil information includes: image data, soil physical and chemical property data, historical fertilization data, initial crop growth conditions and crop root system distribution conditions; the soil morphology information comprises ground vibration data and surface disturbance data;
The pretreatment module is used for carrying out pretreatment on the soil information and the soil morphology information to obtain pretreatment data;
the position determining module is used for inputting the preprocessing data into a fertilizing position determining model based on a deep learning neural network to obtain a position determining result;
And the control module is used for controlling the horizontal distance (based on the position of crops) of the intelligent fertilizer applicator for the downward movement of the fertilizing drill legs and the vertical distance (i.e. depth) of the fertilizing drill legs (based on the ground) according to the position determination result.
Preferably, the location determining module specifically includes:
A database construction unit for constructing a historical soil information database and a historical soil morphology information database by a method of investigating and collecting materials;
the dividing unit is used for dividing the historical soil information database and the historical soil morphology information database based on a K-fold cross verification method to generate a training set matrix and a verification set matrix;
The factor determining unit is used for determining the optimal smoothing factor of the generalized regression neural network model by utilizing a differential evolution algorithm based on the training set matrix and the verification set matrix;
And the training unit is used for inputting the optimal smoothing factor into the generalized regression neural network model for training to obtain a fertilization position determination model.
Preferably, the preprocessing module specifically includes:
the normalization unit is used for performing normalization processing on the soil information and the soil morphology information to obtain normalized data;
the clustering unit is used for clustering the normalized data to obtain a plurality of clustering subsets;
and the data acquisition unit is used for acquiring the preprocessing data according to the clustering subset.
Preferably, the normalization unit comprises:
the function construction unit is used for constructing a target clustering function according to the distance from the data point in the normalized data to the clustering center; wherein, the objective clustering function is:
where v i denotes the ith cluster center, n denotes the total number of data points, Representing the membership degree of the data point x j to the ith cluster center, d ij＝||xj-vi||,dij represents the distance from the data point x j to the ith cluster center, and lambda represents a weighting parameter;
The iteration unit is used for carrying out iteration solution on the target clustering function to obtain a clustering update model;
And the clustering unit is used for clustering the normalized data by using the clustering update model to obtain a plurality of clustering subsets.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
The invention provides an intelligent fertilization method and system, wherein the method comprises the following steps: acquiring soil information of a region to be fertilized and soil form information in a fertilization process; the soil information includes: image data, soil physical and chemical property data, historical fertilization data, initial crop growth conditions and crop root system distribution conditions; the soil morphology information comprises ground vibration data and surface disturbance data; preprocessing the soil information and the soil morphology information to obtain preprocessed data; inputting the preprocessing data into a fertilizing position determining model based on a deep learning neural network to obtain a position determining result; and controlling the horizontal distance (based on the position of crops) of the intelligent fertilizer applicator and the vertical distance (based on the ground) of the intelligent fertilizer applicator according to the position determination result. According to the invention, the influence of soil information and soil morphology information on the fertilization position is more comprehensively considered, and the related influence of each information is converted into the optimal fertilization position through a fertilization position determination model based on a deep learning neural network. Thereby improving the fertilizer utilization rate and the precision degree of the fertilizer application determination model.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the following description will be briefly described with reference to specific crop application embodiments and the drawings that are required to be used, it is obvious that the drawings in the following description are only one embodiment of the present invention, and other drawings can be obtained according to these drawings without inventive effort to a person skilled in the art.
Fig. 1 is a flowchart of a method according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of a device structure according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention aims to provide an intelligent fertilization method and system, which more comprehensively consider the influence of soil information and soil morphology information on fertilization positions, and convert the relevant influence of each information into an optimal fertilization position by determining a model based on the fertilization positions of a deep learning neural network. Thereby improving the fertilizer utilization rate and the precision degree of the fertilizer application determination model.
In order that the above objects, features and advantages of the invention will be readily apparent, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings and are applied to corn crops.
Fig. 1 is a flowchart of a method provided by an embodiment of the present invention, and as shown in fig. 1, the present invention provides an intelligent fertilizing method, which is applied to an intelligent fertilizer distributor, and includes:
Step 100: acquiring soil information of a region to be fertilized of a corn planting field and soil form information in a fertilization process; the soil information includes: image data, soil physical and chemical property data, historical fertilization data, initial corn plant growth conditions and corn root system distribution conditions; the soil morphology information comprises ground vibration data and surface disturbance data;
step 200: preprocessing the soil information and the soil morphology information to obtain preprocessed data;
Step 300: inputting the preprocessing data into a fertilizing position determining model based on a deep learning neural network to obtain a position determining result;
step 400: and controlling the horizontal distance (based on the position of crops) of the intelligent fertilizer applicator and the vertical distance (based on the ground) of the intelligent fertilizer applicator according to the position determination result.
Specifically, in this embodiment, image data can be obtained by a binocular camera provided on the intelligent fertilizer applicator; wherein the image data includes a top layer image of the soil and a corn plant image. In the embodiment, through an expert analysis engine, the image data can be used for analysis, so that initial corn plant growth conditions, corn root system distribution conditions, ground vibration data and soil surface disturbance data are obtained.
Optionally, the crop root system distribution comprises analyzing the image data to obtain a schematic diagram of the quantity and the spatial distribution of the root system in the soil.
In addition, the specific steps for identifying the ground vibration data in this embodiment include:
s1: continuously collecting surface layer images of soil according to a preset shooting frequency;
S3: preprocessing the acquired image;
s4: and acquiring pixel skeleton line segments of the edges of the objects in the preprocessed images, acquiring target points on the pixel skeleton line segments in the continuously shot images in the same way, calculating vibration data of the target points, and generating displacement variation. The method can realize vibration identification and detection through the edge information of the image based on the video image processing technology.
Further, the surface disturbance data of the soil in this embodiment may be obtained by a trained soil disturbance detection model, which includes the geometric dimensions of the macroscopic disturbance profile of the soil.
As an alternative embodiment, the crop growth conditions in this example include plant height, leaf number, leaf thickness, color, fruit yield and fruit quality.
Further, in the embodiment, image data in the fertilization process can be acquired in real time, and the fertilization position is determined in the real-time image, so that the fertilization position determination process is not interrupted in the fertilization process.
Preferably, the construction method of the fertilization position determination model comprises the following steps:
constructing a corn planting historical soil information database and a historical soil morphology information database by a method of investigating and collecting data;
Dividing the historical soil information database and the historical soil morphology information database based on a K-fold cross validation method to generate a training set matrix and a validation set matrix;
Based on the training set matrix and the verification set matrix, determining an optimal smoothing factor of the generalized regression neural network model by utilizing a differential evolution algorithm;
and inputting the optimal smoothing factors into the generalized regression neural network model for training to obtain a fertilization position determination model.
Specifically, in the embodiment, the constructed sample data set is divided by adopting the K-fold cross validation method, so that limited sample data can be fully utilized, the problem that the generalized regression neural network is seriously dependent on sample division can be solved, and the training precision is improved.
Further, in this embodiment, each influence factor is scored by an expert model, and corresponding weights are configured according to the scoring condition, so that the integration of each influence factor is realized, and the comprehensive influence factor is obtained.
Furthermore, the optimal fertilization position can be accurately predicted in the image acquired in real time by combining the influence factors and the trained fertilization position determination model.
Preferably, the preprocessing is performed on the soil information and the soil morphology information to obtain preprocessed data, including:
normalizing the soil information and the soil morphology information to obtain normalized data;
clustering the normalized data to obtain a plurality of clustering subsets;
and obtaining the preprocessing data according to the clustering subset.
Preferably, clustering the normalized data to obtain a plurality of cluster subsets includes:
Constructing a target clustering function according to the distance from the data point in the normalized data to the clustering center; wherein, the objective clustering function is:
where v i denotes the ith cluster center, n denotes the total number of data points, Representing the membership degree of the data point x j to the ith cluster center, d ij＝||xj-vi||,dij represents the distance from the data point x j to the ith cluster center, and lambda represents a weighting parameter;
Carrying out iterative solution on the target clustering function to obtain a clustering update model;
and clustering the normalized data by using the cluster updating model to obtain a plurality of cluster subsets.
Optionally, the data preprocessing in the embodiment includes clustering the data, and because the original nutrient content, the pH, the salt content, the volume weight, the texture, the fertilizing amount and the corresponding corn yield of the planted field soil have complex nonlinear relations, the data is difficult to describe by using a single function model, so that the original data sample is firstly divided into different training subsets by adopting a clustering algorithm, the central number of the radial basis function neural network can be effectively reduced, and the training of the network is more efficient and accurate.
Preferably, the preprocessing of the soil information and the soil morphology information is performed to obtain preprocessed data, and then the method further comprises:
Performing picture enhancement processing on the image data in the preprocessing data to obtain a data set;
extracting weeds and corn plants in the data set by adopting an HSV color model, and then marking the corn plants;
inputting the data set marked with the corn plants into a corn growth model based on a YOLO network for identification to obtain corn growth information;
the maize growth information is added to the pre-processing data.
Further, a YOLOv algorithm cuts a data set marked with corn into pictures with 416 x 416, a feature map with a 32-time downsampling size of 13 x 13 is obtained after convolution calculation and pooling, a feature map with a 16-time downsampling size of 26 x 26 is obtained by transversely connecting a result of 2-time downsampling of the feature map with a previous layer output result of 32-time downsampling, and a feature map with a 8-time downsampling size of 52 x 52 is obtained by transversely connecting a result of 2-time upsampling of the feature map of 16-time downsampling with a previous two layers output result of 32-time downsampling. And obtaining three feature graphs with different sizes of 13, 26 and 52 after feature extraction, and finally obtaining three ruler vectors with different sizes after full connection and 1*1 convolution calculation. And identifying the feature frames through the size vectors, so as to obtain the maize growth information.
Preferably, after controlling the horizontal distance (based on the position of the crops) of the intelligent fertilizer applicator and the vertical distance (i.e. depth) of the intelligent fertilizer applicator, based on the ground, the intelligent fertilizer applicator further comprises:
After a preset growing period, detecting indexes such as plant height, spike height, stem thickness, leaf number, leaf thickness, leaf area, spike length, spike thickness, spike grain number, hundred grain weight and the like of the corn in the area to be fertilized to obtain biological character and yield index information of the corn;
Training and parameter optimization are carried out on the fertilization position determination model according to the biological characters and the yield index information of the corn so as to obtain an optimized fertilization position determination model; the optimized fertilization position determination model is used for determining the position of next fertilization.
Optionally, in this embodiment, by detecting biological properties and yield indexes of soil and corn after fertilization, and adjusting parameters of the fertilization judgment model according to the detection result, the accuracy of the fertilization position determination model can be improved.
Corresponding to the method, the invention also provides an intelligent fertilizer apparatus, as shown in fig. 2, applied to an intelligent fertilizer apparatus, comprising:
The data acquisition module is used for acquiring soil information of a region to be fertilized and soil form information in the fertilization process; the soil information includes: image data, soil analysis data, historical fertilization data, initial corn growth conditions and corn root system distribution conditions; the soil morphology information comprises ground vibration data and surface disturbance data;
The pretreatment module is used for carrying out pretreatment on the soil information and the soil morphology information to obtain pretreatment data;
the position determining module is used for inputting the preprocessing data into a fertilizing position determining model based on a deep learning neural network to obtain a position determining result;
And the control module is used for controlling the horizontal distance (based on the position of crops) of the intelligent fertilizer applicator for the downward movement of the fertilizing drill legs and the vertical distance (i.e. depth) of the fertilizing drill legs (based on the ground) according to the position determination result.
Preferably, the location determining module specifically includes:
A database construction unit for constructing a historical soil information database and a historical soil morphology information database by a method of investigating and collecting materials;
the dividing unit is used for dividing the historical soil information database and the historical soil morphology information database based on a K-fold cross verification method to generate a training set matrix and a verification set matrix;
The factor determining unit is used for determining the optimal smoothing factor of the generalized regression neural network model by utilizing a differential evolution algorithm based on the training set matrix and the verification set matrix;
And the training unit is used for inputting the optimal smoothing factor into the generalized regression neural network model for training to obtain a fertilization position determination model.
Preferably, the preprocessing module specifically includes:
the normalization unit is used for performing normalization processing on the soil information and the soil morphology information to obtain normalized data;
the clustering unit is used for clustering the normalized data to obtain a plurality of clustering subsets;
and the data acquisition unit is used for acquiring the preprocessing data according to the clustering subset.
Preferably, the normalization unit comprises:
the function construction unit is used for constructing a target clustering function according to the distance from the data point in the normalized data to the clustering center; wherein, the objective clustering function is:
where v i denotes the ith cluster center, n denotes the total number of data points, Representing the membership degree of the data point x j to the ith cluster center, d ij＝||xj-vi||,dij represents the distance from the data point x j to the ith cluster center, and lambda represents a weighting parameter;
The iteration unit is used for carrying out iteration solution on the target clustering function to obtain a clustering update model;
And the clustering unit is used for clustering the normalized data by using the clustering update model to obtain a plurality of clustering subsets.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present invention and the core ideas thereof; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.

Claims (3)

1. An intelligent fertilization method is applied to an intelligent fertilizer distributor and is characterized by comprising the following steps:
Acquiring soil information of a region to be fertilized and soil form information in a fertilization process; the soil information includes: image data, soil physical and chemical property data, historical fertilization data, initial crop growth conditions and crop root system distribution conditions; the soil morphology information comprises ground vibration data and surface disturbance data;
preprocessing the soil information and the soil morphology information to obtain preprocessed data;
inputting the preprocessing data into a fertilizing position determining model based on a deep learning neural network to obtain a position determining result;
Controlling the horizontal distance of the falling of the fertilizing drill leg and the vertical distance of the falling of the fertilizing drill leg of the intelligent fertilizer applicator according to the position determining result; the horizontal distance is based on the crop position; the vertical distance is depth based on the ground;
the construction method of the fertilization position determination model comprises the following steps:
constructing a historical soil information database and a historical soil morphology information database by a method of investigating and collecting materials;
Dividing the historical soil information database and the historical soil morphology information database based on a K-fold cross validation method to generate a training set matrix and a validation set matrix;
Based on the training set matrix and the verification set matrix, determining an optimal smoothing factor of the generalized regression neural network model by utilizing a differential evolution algorithm;
inputting the optimal smoothing factors into the generalized regression neural network model for training to obtain a fertilization position determination model;
preprocessing the soil information and the soil morphology information to obtain preprocessed data, wherein the preprocessing data comprises:
normalizing the soil information and the soil morphology information to obtain normalized data;
clustering the normalized data to obtain a plurality of clustering subsets;
Obtaining the preprocessing data according to the clustering subset;
clustering the normalized data to obtain a plurality of cluster subsets, including:
Constructing a target clustering function according to the distance from the data point in the normalized data to the clustering center; wherein, the objective clustering function is: ；
Wherein, Represents the ith cluster center, n represents the total number of data points,/>Representing data points/>Membership degree of i-th cluster center,/>，/>Representing the distance of data point x j to the ith cluster center,/>Representing the weighting parameters;
Carrying out iterative solution on the target clustering function to obtain a clustering update model;
Clustering the normalized data by using the cluster updating model to obtain a plurality of cluster subsets;
After controlling the horizontal distance of the falling of the fertilizing drill leg and the vertical distance of the falling of the fertilizing drill leg of the intelligent fertilizer applicator according to the position determination result, the intelligent fertilizer applicator further comprises:
after a preset growing period, detecting the crop growth condition and crop yield index of the area to be fertilized to obtain crop biological characters and yield index information;
Training and parameter optimization are carried out on the fertilization position determination model according to the crop biological characters and the yield index information so as to obtain an optimized fertilization position determination model; the optimized fertilization position determination model is used for determining the position of next fertilization.
2. The intelligent fertilization method according to claim 1, wherein the preprocessing of the soil information and the soil morphology information to obtain preprocessed data further comprises:
Performing picture enhancement processing on the image data in the preprocessing data to obtain a data set;
Extracting weeds and crops in the data set by adopting an HSV color model, and then marking the crops;
inputting the data set marked with the crops into a crop growth model based on a YOLO network for identification to obtain crop growth information;
the crop growth information is added to the pre-treatment data.
3. An intelligent fertilizer injection unit is applied to intelligent fertilizer distributor, its characterized in that includes:
The data acquisition module is used for acquiring soil information of a region to be fertilized and soil form information in the fertilization process; the soil information includes: image data, soil physical and chemical property data, historical fertilization data, initial crop growth conditions and crop root system distribution conditions; the soil morphology information comprises ground vibration data and surface disturbance data;
The pretreatment module is used for carrying out pretreatment on the soil information and the soil morphology information to obtain pretreatment data;
the position determining module is used for inputting the preprocessing data into a fertilizing position determining model based on a deep learning neural network to obtain a position determining result;
The control module is used for controlling the horizontal distance of the downward movement of the fertilizing drill leg and the vertical distance of the downward movement of the fertilizing drill leg of the intelligent fertilizer applicator according to the position determination result; the horizontal distance is based on the crop position; the vertical distance is depth based on the ground; after controlling the horizontal distance of the falling of the fertilizing drill leg and the vertical distance of the falling of the fertilizing drill leg of the intelligent fertilizer applicator according to the position determination result, the intelligent fertilizer applicator further comprises:
after a preset growing period, detecting the crop growth condition and crop yield index of the area to be fertilized to obtain crop biological characters and yield index information;
Training and parameter optimization are carried out on the fertilization position determination model according to the crop biological characters and the yield index information so as to obtain an optimized fertilization position determination model; the optimized fertilizing position determining model is used for determining the position of next fertilizing;
the position determining module specifically comprises:
A database construction unit for constructing a historical soil information database and a historical soil morphology information database by a method of investigating and collecting materials;
the dividing unit is used for dividing the historical soil information database and the historical soil morphology information database based on a K-fold cross verification method to generate a training set matrix and a verification set matrix;
The factor determining unit is used for determining the optimal smoothing factor of the generalized regression neural network model by utilizing a differential evolution algorithm based on the training set matrix and the verification set matrix;
The training unit is used for inputting the optimal smoothing factor into the generalized regression neural network model for training to obtain a fertilization position determination model;
The preprocessing module specifically comprises:
the normalization unit is used for performing normalization processing on the soil information and the soil morphology information to obtain normalized data;
the clustering unit is used for clustering the normalized data to obtain a plurality of clustering subsets;
The data acquisition unit is used for acquiring the preprocessing data according to the clustering subset;
the normalization unit includes:
the function construction unit is used for constructing a target clustering function according to the distance from the data point in the normalized data to the clustering center; wherein, the objective clustering function is: ；
Wherein, Represents the ith cluster center, n represents the total number of data points,/>Representing data points/>Membership degree of i-th cluster center,/>，/>Representing the distance of data point x j to the ith cluster center,/>Representing the weighting parameters;
The iteration unit is used for carrying out iteration solution on the target clustering function to obtain a clustering update model;
And the clustering unit is used for clustering the normalized data by using the clustering update model to obtain a plurality of clustering subsets.
CN202311534388.9A 2023-11-17 2023-11-17 Intelligent fertilization method and device Active CN117480927B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311534388.9A CN117480927B (en) 2023-11-17 2023-11-17 Intelligent fertilization method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311534388.9A CN117480927B (en) 2023-11-17 2023-11-17 Intelligent fertilization method and device

Publications (2)

Publication Number Publication Date
CN117480927A CN117480927A (en) 2024-02-02
CN117480927B true CN117480927B (en) 2024-06-11

Family

ID=89668940

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311534388.9A Active CN117480927B (en) 2023-11-17 2023-11-17 Intelligent fertilization method and device

Country Status (1)

Country Link
CN (1) CN117480927B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105929733A (en) * 2016-04-27 2016-09-07 山东省农业科学院玉米研究所 Intelligent maize sowing and fertilizing control system and control method
CN106508204A (en) * 2016-04-27 2017-03-22 山东省农业科学院玉米研究所 Corn precise localization synchronal sowing control system and method
CN108684278A (en) * 2017-04-06 2018-10-23 点豆（山东）网络技术有限公司 Intelligent fertilizer distributing methods, devices and systems
CN115063252A (en) * 2022-06-08 2022-09-16 山东省农业科学院 Crop fertilizer accurate application method and system based on neural network
CN116258060A (en) * 2022-12-30 2023-06-13 武汉大学 Soil testing formula fertilization method based on machine learning
CN116918546A (en) * 2023-07-11 2023-10-24 吉林隆源农业服务有限公司 Corn high-yield planting and fertilizing method
CN117055420A (en) * 2023-09-12 2023-11-14 河北省农林科学院粮油作物研究所 Farmland soil variable fertilization control system and method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP7195176B2 (en) * 2018-02-27 2022-12-23 株式会社トプコン Fertilization design device, agricultural equipment, fertilization design method, and fertilization design program

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105929733A (en) * 2016-04-27 2016-09-07 山东省农业科学院玉米研究所 Intelligent maize sowing and fertilizing control system and control method
CN106508204A (en) * 2016-04-27 2017-03-22 山东省农业科学院玉米研究所 Corn precise localization synchronal sowing control system and method
CN108684278A (en) * 2017-04-06 2018-10-23 点豆（山东）网络技术有限公司 Intelligent fertilizer distributing methods, devices and systems
CN115063252A (en) * 2022-06-08 2022-09-16 山东省农业科学院 Crop fertilizer accurate application method and system based on neural network
CN116258060A (en) * 2022-12-30 2023-06-13 武汉大学 Soil testing formula fertilization method based on machine learning
CN116918546A (en) * 2023-07-11 2023-10-24 吉林隆源农业服务有限公司 Corn high-yield planting and fertilizing method
CN117055420A (en) * 2023-09-12 2023-11-14 河北省农林科学院粮油作物研究所 Farmland soil variable fertilization control system and method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于ComGIS的精准农业变量施肥处方推荐系统研究;侯顺艳;李志远;王秀;;河北农业大学学报;20051130(第06期);第101-103页 *

Also Published As

Publication number Publication date
CN117480927A (en) 2024-02-02

Similar Documents

Publication Publication Date Title
CN100416590C (en) Method for automatically identifying field weeds in crop seeding-stage using site and grain characteristic
CN111104858A (en) Large-scale crop phenological extraction method based on morphological model method
CN108710766B (en) Greenhouse plant liquid manure machine fertilizer regulation parameter calculation method based on growth model
CN111598045B (en) Remote sensing farmland change detection method based on object spectrum and mixed spectrum
CN114442705B (en) Intelligent agricultural system based on Internet of things and control method
CN113034301A (en) Crop growth management system and method
CN111967441A (en) Crop disease analysis method based on deep learning
CN114828618A (en) Computer-implemented method for providing test design and test instruction data for comparative testing of yield, profitability, efficacy, or vegetation index for at least two products or the same product at different application schedules
CN117322214A (en) Crop fertilizer accurate application method and system based on neural network
CN116076340A (en) Precise decision making and precise irrigation control method based on mobile irrigation system
CN114782840A (en) Real-time wheat phenological period classification method based on unmanned aerial vehicle RGB images
CN116128672A (en) Model-data combined driving intelligent greenhouse fertilizer preparation method and system
CN117480927B (en) Intelligent fertilization method and device
CN116843493B (en) Ecological digital monitoring management system is planted in tea-oil camellia woods based on thing networking
Arora et al. Automated dosing system in hydroponics with machine learning
CN116595333B (en) Soil-climate intelligent rice target yield and nitrogen fertilizer consumption determination method
CN114514914A (en) Intelligent sensing fertilization and pesticide spraying method and device
CN116894514A (en) Crop yield prediction method and system based on soil quality index
CN116452358A (en) Intelligent agriculture management system based on Internet of things
CN115392016A (en) Silage corn growth and development prediction method based on remote sensing data assimilation
CN115690609A (en) Method for monitoring growth vigor of crops in growth period
DE202022102591U1 (en) System for monitoring plant health in precision agriculture using image processing and convolutional neural network
CN112950700A (en) Plant leaf stomatal conductance measuring method based on deep learning and image processing
CN115144342B (en) Unmanned aerial vehicle multispectral citrus blade nitrogen content estimation method based on semi-supervised learning
CN115796558B (en) Tobacco planting area intelligent management system based on deep neural network

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