Intelligent agricultural irrigation method

Abstract

The invention discloses an intelligent agricultural irrigation method which comprises the following steps: S1, carrying out environment monitoring on a target area to obtain environment information of the target area; s2, performing soil detection on the target area to obtain soil information of each sub-planting area in the target area; s3, analyzing crops planted in each sub-planting area in the target area, and determining growth stages of various crops; s4, according to the environment information of the target area, the soil information of each sub-planting area in the target area and the growth stage of each type of crops, calculating irrigation coefficients of each type of crops; and S5, determining irrigation methods of various crops. According to the method, the growth stages of various crops are determined, the growth stages of various crops are obtained by comparing the crop images with the growth crop standard models of various crops, and the growth stages of various crops are obtained according to the crop images in combination with the CNNs, so that the growth stages of various crops can be accurately judged; and determining irrigation methods of various crops.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural irrigation, and in particular to a smart agricultural irrigation method. Background Art

[0002] With the rapid development of artificial intelligence technology, it has greatly promoted the development of social economy; artificial intelligence technology can automate and optimize many complex tasks and processes, thereby significantly improving productivity and work efficiency, and artificial intelligence technology has the ability to process and analyze large amounts of data, discover patterns from them, and help decision makers make clearer decisions.

[0003] As an important part of modern technology, artificial intelligence technology is widely used in various industries and fields, such as industrial production, medical care, education and agriculture. Among them, agriculture, as the foundation of people's survival, has an extremely important impact and status on our society and economy; combining artificial intelligence technology with agriculture can bring many benefits to agriculture, including: improving agricultural productivity, effective agricultural resource management and effective protection of crops.

[0004] The combination of agriculture and artificial intelligence technology creates a new management method for smart agriculture, providing a more scientific and effective agricultural production plan. Smart agriculture uses advanced technology and data analysis to improve agricultural production efficiency and sustainable development. The implementation of smart agriculture includes many aspects, such as crop planting management, growth status detection, and intelligent irrigation.

[0005] Smart irrigation is a key technology in smart agriculture, which aims to improve the efficiency of water resource utilization by accurately controlling and optimizing the amount of irrigation water, while ensuring the growth and yield of crops. Many excellent methods have been proposed in the existing literature on the research of smart irrigation of crops:

[0006] For example, the invention patent application with publication number CN113229123A discloses a method, device, equipment and storage medium for intelligent irrigation of crops; the invention includes: obtaining a type of regional environmental information of the area where the crops are located, and dividing the area where the crops are located according to the type of regional environmental information to obtain different crop irrigation areas; obtaining crop images of the crop irrigation area, and performing image recognition on the crop images to obtain crop attribute information of the crops; obtaining a type of regional environmental information of the crop irrigation area, and determining the water requirement and fertilizer requirement corresponding to the crop irrigation area according to the crop attribute information and the type of regional environmental information; determining a corresponding crop irrigation mode according to the water requirement and the fertilizer requirement, and irrigating the crop irrigation area according to the crop irrigation mode.

[0007] For another example, an invention patent application with publication number CN112400676A discloses an intelligent irrigation system based on big data; the system includes a cloud platform, a user management module, an environmental monitoring module, a water volume prediction module, an equipment monitoring module and an auxiliary material addition module; the invention sets up an environmental monitoring module, which monitors the growth environment of crops and determines whether crops need irrigation based on the monitoring results; and the system sets up a water volume prediction module, which uses historical data and artificial intelligence algorithms to predict the water consumption for crop irrigation.

[0008] Although the invention patent application mentioned above can realize the calculation of the irrigation water volume for crops, it does not combine environmental information, soil information and growth stage data of various crops for accurate judgment during the calculation process of irrigation water volume, which will cause deviations in the judgment of irrigation methods for crops. This will not only fail to achieve good irrigation effects, but also cause waste of water resources. Summary of the invention

[0009] The purpose of this invention is to propose a smart agricultural irrigation method, which combines environmental information, soil information and growth stage data of various crops to make accurate judgments, so as to achieve accurate judgments on crop irrigation methods, which can not only achieve good irrigation effects, but also effectively save water resources.

[0010] In order to achieve the above object, the present invention adopts the following technical solutions:

[0011] A smart agricultural irrigation method comprises the following steps:

[0012] S1: Perform environmental detection on the target area to obtain environmental information of the target area;

[0013] S2: Conduct soil monitoring in the target area to obtain soil information of each sub-planting area in the target area;

[0014] S3: Analyze the crops planted in each sub-planting area within the target area to determine the growth stage of each crop;

[0015] S4: Calculate the irrigation coefficient of each crop according to the environmental information of the target area, the soil information of each sub-planting area in the target area and the growth stage of each crop;

[0016] S5: Determine irrigation methods for each crop.

[0017] Preferably, the environment of the target area is detected by a preset agricultural meteorological monitoring instrument to obtain the environmental information of the target area.

[0018] Preferably, obtaining soil information includes the following steps:

[0019] Divide the target area according to the crop types to obtain sub-planting areas within the target area;

[0020] The soil nutrient monitoring device preset in each sub-planting area is used to detect the soil in each sub-planting area in the target area, so as to obtain the soil information of each sub-planting area in the target area.

[0021] Preferably, the growth stages of various crops are obtained by comparing and analyzing the crop images with the crop standard models of various crops at various growth stages, including the following steps:

[0022] By using high-definition cameras preset in each sub-planting area to collect images of crops planted in each sub-planting area in the target area, crop images corresponding to each type of crop are obtained;

[0023] Use contour scanners to perform all-round scanning of crop images corresponding to various crops, and establish crop contour models corresponding to various crops;

[0024] Extracting the corresponding crop standard model of each growth stage of the target crop from the crop standard model database according to the crop type;

[0025] Compare and analyze the crop contour models corresponding to each type of crop with the corresponding crop standard models of each growth stage of the target crop, and calculate the similarity;

[0026] According to the maximum similarity, the growth stage matching the crop contour model corresponding to each type of crop is screened out from the crop standard models of each growth stage of the target crop to obtain the growth stage of each type of crop.

[0027] Preferably, using convolutional neural networks (CNNs) to obtain the growth stages of various crops based on crop images includes the following steps:

[0028] Collect historical crop images of various crops at various growth stages, set labels according to the growth stages of target crops corresponding to each historical crop image, and form a historical data set;

[0029] Divide the historical data set into training set, validation set and test set according to the preset ratio;

[0030] Construct the loss function and optimizer of the crop growth stage recognition model based on convolutional neural networks (CNNs);

[0031] The training set is input into the crop growth stage recognition model for model training;

[0032] The loss value is calculated based on the loss function, and the optimizer updates the model parameters according to the loss value and network gradient information;

[0033] If the loss value is less than the preset threshold, the model training ends, and the current crop growth stage recognition model is the trained crop growth stage recognition model. Otherwise, return to ④ and continue to use the training set for model training.

[0034] Input the validation set into the trained crop growth stage recognition model, evaluate the generalization ability of the model by observing the performance on the validation set, and tune the hyperparameters and structure of the model;

[0035] The test set is input into the optimized crop growth stage recognition model to evaluate the model performance;

[0036] After obtaining the pre-trained crop growth stage recognition model, crop images corresponding to various crops collected by a high-definition camera are input into the pre-trained crop growth stage recognition model to obtain the growth stages of various crops.

[0037] Preferably, calculating the irrigation coefficient of each crop includes calculating the soil influencing factor IFs, which has the following steps:

[0038] According to the growth stage of each crop, the standard soil air content Ai', standard soil water content Mi' and standard soil temperature Ti' corresponding to the target crop i (i = 1, 2, ..., k, k is the number of crop types, that is, the number of sub-planting areas in the target area) are extracted from the soil information database;

[0039] Obtain the soil air content Ai, soil water content Mi and soil temperature Ti of the sub-planting area of ​​the target crop i located in the target area;

[0040] The soil impact factor IFs is calculated using the following formula:

[0041]

[0042] Among them, α, β, and χ are the influence weights of soil air content, soil moisture content, and soil temperature, respectively.

[0043] Preferably, calculating the irrigation coefficient of each crop also includes calculating the environmental impact factor IFe, which has the following steps:

[0044] The environmental impact factor IFe is calculated using the following formula:

[0045]

[0046] Among them, t, H, and L are air temperature, air humidity, and light intensity, respectively; δ and ε are the influence weights of light intensity and air humidity, respectively.

[0047] Preferably, calculating the irrigation coefficient of each crop also includes calculating the irrigation coefficient ICi of the target crop i:

[0048] ICi=IFs*ω 1 +IFs*ω 2

[0049] Among them, ω 1 ,ω 2 They are the influence weights of soil influencing factor IFs and environmental influencing factor IFe respectively.

[0050] Preferably, the irrigation method corresponding to each type of crop is obtained according to the irrigation coefficient of each type of crop, including the following steps:

[0051] The corresponding irrigation method is extracted from the irrigation method database according to the growth stage of the target crop i and the irrigation coefficient ICi.

[0052] Preferably, convolutional neural networks (CNNs) are used to obtain irrigation methods corresponding to various crops according to environmental information of the target area, soil information of each sub-planting area within the target area, and growth stages of various crops, including the following steps:

[0053] Collect various environmental information, soil information and growth stage data of various crops, set labels according to corresponding irrigation methods, and form historical data sets;

[0054] Divide the historical data set into training set, validation set and test set according to the preset ratio;

[0055] Construct the loss function and optimizer of the crop irrigation method recognition model based on convolutional neural networks (CNNs);

[0056] The training set is input into the crop irrigation method recognition model for model training;

[0057] The loss value is calculated based on the loss function, and the optimizer updates the model parameters according to the loss value and network gradient information;

[0058] If the loss value is less than the preset threshold, the model training ends, and the current crop irrigation method recognition model is the trained crop irrigation method recognition model. Otherwise, return to step 4 and continue to use the training set for model training.

[0059] Input the validation set into the trained crop irrigation method recognition model, evaluate the generalization ability of the model by observing the performance on the validation set, and tune the hyperparameters and structure of the model;

[0060] The test set is input into the tuned crop irrigation method identification model to evaluate the model performance;

[0061] After obtaining the pre-trained crop irrigation method recognition model, the collected environmental information of the target area, the soil information of the sub-planting area of ​​the target crop i within the target area, and the growth stage of the target crop i are input into the pre-trained crop irrigation method recognition model to obtain the irrigation method corresponding to the target crop i.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] 1. In the process of determining the growth stage of various crops, the present invention obtains the growth stage of various crops by comparing and analyzing the crop images with the crop standard models of various crops at various growth stages. In combination with the use of convolutional neural networks (CNNs) to obtain the growth stage of various crops based on the crop images, the growth stage of various crops can be accurately determined, providing a guarantee for the subsequent determination of the irrigation methods for various crops.

[0064] 2. In the process of determining the irrigation method for each crop, the present invention obtains the corresponding irrigation method for each crop according to the irrigation coefficient of each crop, and on the other hand, uses convolutional neural networks (CNNs) to obtain the corresponding irrigation method for each crop according to the environmental information of the target area, the soil information of each sub-planting area in the target area, and the growth stage of each crop, thereby accurately determining the irrigation method for each crop and achieving a good irrigation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 This is a schematic diagram of the process of a smart agricultural irrigation method proposed in the present invention. DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0067] Reference Figure 1 , a smart agricultural irrigation method, comprising:

[0068] S1: Perform environmental detection on the target area to obtain environmental information of the target area;

[0069] S2: Conduct soil monitoring in the target area to obtain soil information of each sub-planting area in the target area;

[0070] S3: Analyze the crops planted in each sub-planting area within the target area to determine the growth stage of each crop;

[0071] S4: Calculate the irrigation coefficient of each crop according to the environmental information of the target area, the soil information of each sub-planting area in the target area and the growth stage of each crop;

[0072] S5: Determine irrigation methods for each crop.

[0073] The above-mentioned environmental monitoring of the target area is performed to obtain the environmental information of the target area. The specific operations are as follows:

[0074] The environment of the target area is detected by a preset agricultural meteorological monitoring instrument to obtain the environmental information of the target area;

[0075] Among them, environmental information includes air temperature, air humidity and light intensity.

[0076] The above-mentioned soil monitoring of the target area is performed to obtain the soil information of each sub-planting area within the target area. The specific operations are as follows:

[0077] Divide the target area according to the crop types to obtain sub-planting areas within the target area;

[0078] The soil information of each sub-planting area in the target area is obtained by performing soil testing on each sub-planting area in the target area through a soil nutrient monitor preset in each sub-planting area;

[0079] Among them, soil information includes soil air content, soil moisture content and soil temperature.

[0080] The above analysis is performed on the crops planted in each sub-planting area within the target area to determine the growth stage of each crop. The specific operations are as follows:

[0081] (a) By comparing and analyzing crop images with standard crop models at different growth stages of various crops, the growth stages of various crops are obtained:

[0082] ① The crops planted in each sub-planting area in the target area are collected by using high-definition cameras preset in each sub-planting area to obtain crop images corresponding to each type of crop;

[0083] ②Use contour scanner to perform all-round scanning of crop images corresponding to various crops, and establish crop contour models corresponding to various crops;

[0084] ③ Extract the corresponding crop standard model of each growth stage of the target crop from the crop standard model database according to the crop type;

[0085] ④ Compare and analyze the crop contour models corresponding to each type of crop with the corresponding crop standard models of each growth stage of the target crop, and calculate the similarity;

[0086] ⑤ According to the maximum similarity, the growth stage matching the crop contour model corresponding to each type of crop is selected from the crop standard models of each growth stage of the target crop to obtain the growth stage of each type of crop.

[0087] It should be noted that the similarity calculation can be feature matching, structural similarity, etc.

[0088] (b) Using convolutional neural networks (CNNs) to obtain the growth stages of various crops based on crop images:

[0089] ① Collect historical crop images of various crops at various growth stages, set labels according to the growth stages of target crops corresponding to each historical crop image, and form a historical data set;

[0090] ② Divide the historical data set into training set, validation set and test set according to the preset ratio;

[0091] ③ Construct the loss function and optimizer of the crop growth stage recognition model based on convolutional neural networks (CNNs);

[0092] ④ Input the training set into the crop growth stage recognition model for model training;

[0093] ⑤ Calculate the loss value based on the loss function, and the optimizer updates the model parameters according to the loss value and network gradient information;

[0094] ⑥If the loss value is less than the preset threshold, the model training ends, and the current crop growth stage recognition model is the trained crop growth stage recognition model. Otherwise, return to ④ and continue to use the training set for model training;

[0095] ⑦ Input the validation set into the trained crop growth stage recognition model, evaluate the generalization ability of the model by observing the performance on the validation set, and tune the hyperparameters and structure of the model;

[0096] ⑧ Input the test set into the optimized crop growth stage recognition model to evaluate the model performance;

[0097] ⑨ After obtaining the pre-trained crop growth stage recognition model, the crop images corresponding to various crops collected by the high-definition camera are input into the pre-trained crop growth stage recognition model to obtain the growth stages of various crops.

[0098] (c) Make a comprehensive judgment on the growth stages of various crops obtained by methods (a) and (b), and finally determine the growth stages of various crops.

[0099] The above calculation of irrigation coefficients for various crops is based on the environmental information of the target area, the soil information of each sub-planting area within the target area, and the growth stage of various crops. The specific operations are as follows:

[0100] (a) Calculate soil impact factors IFs;

[0101] ① According to the growth stage of each crop, the standard soil air content Ai', standard soil water content Mi' and standard soil temperature Ti' corresponding to the target crop i (i = 1, 2, ..., k, k is the number of crop types, that is, the number of sub-planting areas in the target area) are extracted from the soil information database;

[0102] ② Obtain the soil air content Ai, soil water content Mi and soil temperature Ti of the sub-planting area of ​​target crop i within the target area;

[0103] ③ Use the following formula to calculate soil impact factor IFs:

[0104]

[0105] Among them, α, β, and χ are the influence weights of soil air content, soil moisture content, and soil temperature, respectively;

[0106] (b) Calculate the environmental impact factor IFe:

[0107] The environmental impact factor IFe is calculated using the following formula:

[0108]

[0109] Among them, t, H, and L are air temperature, air humidity, and light intensity, respectively; δ and ε are the influence weights of light intensity and air humidity, respectively;

[0110] (c) Calculate the irrigation coefficient ICi of target crop i:

[0111] ICi=IFs*ω 1 +IFs*ω 2

[0112] Among them, ω 1 ,ω 2 They are the influence weights of soil influencing factor IFs and environmental influencing factor IFe respectively.

[0113] The above-mentioned irrigation methods for various crops (including irrigation time, irrigation water volume and irrigation mode, etc.) are specifically operated as follows:

[0114] (a) According to the irrigation coefficient of each crop, the corresponding irrigation method of each crop is obtained:

[0115] Extract the corresponding irrigation method from the irrigation method database according to the growth stage of the target crop i and the irrigation coefficient ICi;

[0116] (b) Using convolutional neural networks (CNNs), we can obtain the irrigation methods corresponding to each crop based on the environmental information of the target area, the soil information of each sub-planting area within the target area, and the growth stage of each crop:

[0117] ① Collect various environmental information, soil information and growth stage data of various crops, set labels according to the corresponding irrigation methods, and form a historical data set;

[0118] ② Divide the historical data set into training set, validation set and test set according to the preset ratio;

[0119] ③ Construct the loss function and optimizer of the crop irrigation method recognition model based on convolutional neural networks (CNNs);

[0120] ④ Input the training set into the crop irrigation method recognition model for model training;

[0121] ⑤ Calculate the loss value based on the loss function, and the optimizer updates the model parameters according to the loss value and network gradient information;

[0122] ⑥If the loss value is less than the preset threshold, the model training ends, and the current crop irrigation method recognition model is the trained crop irrigation method recognition model. Otherwise, return to ④ and continue to use the training set for model training;

[0123] ⑦ Input the validation set into the trained crop irrigation method recognition model, evaluate the generalization ability of the model by observing the performance on the validation set, and tune the hyperparameters and structure of the model;

[0124] ⑧ Input the test set into the optimized crop irrigation method recognition model to evaluate the model performance;

[0125] ⑨ After obtaining the pre-trained crop irrigation method recognition model, the collected environmental information of the target area, the soil information of the sub-planting area of ​​the target crop i located in the target area, and the growth stage of the target crop i are input into the pre-trained crop irrigation method recognition model to obtain the irrigation method corresponding to the target crop i;

[0126] (c) Comprehensively judge the irrigation methods corresponding to the target crop i obtained by methods (a) and (b), and finally determine the irrigation method corresponding to the target crop i.

[0127] The present invention is conceived to:

[0128] In the process of determining the growth stage of various crops, the growth stage of various crops is obtained by comparing and analyzing the crop images with the crop standard models of various crops at various growth stages. Combined with the use of convolutional neural networks (CNNs) to obtain the growth stage of various crops based on crop images, the growth stage of various crops can be accurately determined, providing a guarantee for the subsequent determination of irrigation methods for various crops.

[0129] In the process of determining the irrigation methods for various crops, on the one hand, the corresponding irrigation methods for various crops are obtained according to the irrigation coefficients of various crops. On the other hand, convolutional neural networks (CNNs) are used to obtain the corresponding irrigation methods for various crops according to the environmental information of the target area, the soil information of each sub-planting area in the target area, and the growth stage of various crops. This makes it possible to accurately determine the irrigation methods for various crops and achieve good irrigation effects.

[0130] The above calculations, identifications and model building are all based on the rich resources and computing power of CS5202LS. Specifically:

[0131] CS5202LS hardware:

[0132] (1) Single board, single board complete machine

[0133] CS5202LS has a variety of product forms:

[0134] Single board mode: separate board

[0135] ·Original customized case: using a case specially customized by Giskeda

[0136] Developer customized case: using developer's own customized case and chassis

[0137] (2) Various combination modes of multi-board complete machine

[0138] In order to meet different application requirements on site, CS5202LS can expand various IO boards based on the main control board and combine them into various multi-board complete machine forms.

[0139] CS5202LS software:

[0140] (1) Software level description:

[0141] On the CS5202LS hardware platform, Giskaida provides an application development environment and development kit for industrial applications, which are comprehensive in functions, reasonable in structure, highly extensible, streamlined and efficient. Developers can develop final products with different functions according to different application requirements.

[0142] Hardware board resource layer

[0143] On a 140mm x 100mm (8-layer) single board, the LoongArch 2K1000LA CPU, DDR and various IO interface circuits are integrated to implement all IO drivers at the operating system level.

[0144] Software basic resource layer

[0145] Through interaction with IO drivers (including serial port RS232 / RS485, network port, CAN, 4G / 5G, WIFI, PCIe, etc.), protocol analysis (300+ protocol libraries), MQTT service, real-time database, time series database (historical database), relational database (historical database), audio and video push-pull streaming / Onvif / GB28181 and other functions are realized.

[0146] Application function extension layer

[0147] By interacting with protocol parsers, real-time databases, time series databases, and historical databases, the expansion of various data processing application functions is achieved.

[0148] Application service layer

[0149] Dynamically scalable lightweight C+HTTP application server, without the need for large, complex and unstable middleware, to achieve efficient application services;

[0150] This service mechanism can directly interact with the HMI / SCADA of the C / S architecture of the application display layer; it can also provide services for device configuration, B / S HMI interaction, and external interface calls through the Nginx server of the publishing layer;

[0151] The streaming media server directly interacts with the push-pull stream / pan-tilt control components of the software basic resource layer, and provides services for B / SHMI interaction and external interface calls through the Nginx server of the publishing layer.

[0152] Application publishing layer

[0153] The Nginx server provides various services for the application presentation layer and external interfaces through the reverse proxy mechanism.

[0154] Application presentation layer

[0155] Through C / S and B / S methods, HMI human-computer interaction and external interface are provided.

[0156] (2) Main application functions:

[0157] a. Industrial control full function combination

[0158] It has functions such as PLC, gateway, video surveillance, network security, and host computer configuration software. It can be separated or combined to flexibly meet different application scenarios.

[0159] b. Multi-protocol support

[0160] The protocol converter supports more than 300 industrial protocol libraries, covering protocol data analysis of industry protocols such as electricity, water conservancy, petroleum and petrochemicals, rail transportation, PLC, CNC, Internet of Things, robot control, and video surveillance.

[0161] c. Unified data processing

[0162] Integration and linkage of electrical quantity IO data (AIAODIDO), protocol data and video streaming media, unified processing of network port, serial port, wired and wireless, and 4G / 5G.

[0163] d. Unified data model

[0164] The protocol data, protocol model and protocol extension framework based on the real-time database provide great convenience for the development of new protocols.

[0165] e. Support relational database and time series database

[0166] It supports reading and writing data to relational databases (Sqlite, MySQL) and time series databases (InfluxDB), and fully supports the changing upper-level data processing and analysis requirements.

[0167] f. SCADA function

[0168] Built-in high-performance, multi-functional C / S+B / S architecture lightweight SCADA system (embedded full-function configuration software), with simple architecture, high efficiency and integrated design.

[0169] g. HMI function

[0170] It can realize touch screen human-computer interaction (C / S, B / S) through HDMI / USB, and network human-computer interaction (B / S) through browser.

[0171] h. Lightweight WEB server

[0172] There is no need to configure heavy middleware such as TOMCAT, and the built-in lightweight HTTP / MQTT service is simple, efficient, secure and reliable.

[0173] i. Dual-machine redundancy

[0174] Supports dual-network / dual-node active-standby redundancy mechanism to ensure high performance and high reliability.

[0175] j. Various secondary development kits

[0176] Provides APIs and development environments for a variety of secondary development tools, including API interfaces in C / C++, QSA, JAVA, JS, and Python, to achieve diversity and convenience in value-added applications.

[0177] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A smart agricultural irrigation method, characterized in that: The following steps are involved: S1: Perform environmental detection on the target area to obtain environmental information of the target area; S2: Conduct soil monitoring in the target area to obtain soil information of each sub-planting area in the target area; S3: Analyze the crops planted in each sub-planting area within the target area to determine the growth stage of each crop; S4: Calculate the irrigation coefficient of each crop according to the environmental information of the target area, the soil information of each sub-planting area in the target area and the growth stage of each crop; S5: Determine irrigation methods for each crop.

2. A smart agricultural irrigation method according to claim 1, characterized in that: The environmental information of the target area is obtained by conducting environmental detection on the target area through the preset agricultural meteorological monitoring instrument.

3. The smart agricultural irrigation method according to claim 1, characterized in that: The acquisition of soil information includes the following steps: Divide the target area according to the crop types to obtain sub-planting areas within the target area; The soil nutrient monitoring device preset in each sub-planting area is used to detect the soil in each sub-planting area in the target area, so as to obtain the soil information of each sub-planting area in the target area.

4. The smart agricultural irrigation method according to claim 1, characterized in that: By comparing and analyzing the crop images with the crop standard models of each growth stage of each crop, the growth stages of each crop are obtained, including the following steps: By using high-definition cameras preset in each sub-planting area to collect images of crops planted in each sub-planting area in the target area, crop images corresponding to each type of crop are obtained; Use contour scanners to perform all-round scanning of crop images corresponding to various crops, and establish crop contour models corresponding to various crops; Extracting the corresponding crop standard model of each growth stage of the target crop from the crop standard model database according to the crop type; Compare and analyze the crop contour models corresponding to each type of crop with the corresponding crop standard models of each growth stage of the target crop, and calculate the similarity; According to the maximum similarity, the growth stage matching the crop contour model corresponding to each type of crop is screened out from the crop standard models of each growth stage of the target crop to obtain the growth stage of each type of crop.

5. The smart agricultural irrigation method according to claim 1, characterized in that: Using convolutional neural networks (CNNs) to obtain the growth stages of various crops based on crop images includes the following steps: Collect historical crop images of various crops at various growth stages, set labels according to the growth stages of target crops corresponding to each historical crop image, and form a historical data set; Divide the historical data set into training set, validation set and test set according to the preset ratio; Construct the loss function and optimizer of the crop growth stage recognition model based on convolutional neural networks (CNNs); The training set is input into the crop growth stage recognition model for model training; The loss value is calculated based on the loss function, and the optimizer updates the model parameters according to the loss value and network gradient information; If the loss value is less than the preset threshold, the model training ends, and the current crop growth stage recognition model is the trained crop growth stage recognition model. Otherwise, return to step 4 and continue to use the training set for model training. Input the validation set into the trained crop growth stage recognition model, evaluate the generalization ability of the model by observing the performance on the validation set, and tune the hyperparameters and structure of the model; The test set is input into the optimized crop growth stage recognition model to evaluate the model performance; After obtaining the pre-trained crop growth stage recognition model, crop images corresponding to various crops collected by a high-definition camera are input into the pre-trained crop growth stage recognition model to obtain the growth stages of various crops.

6. The smart agricultural irrigation method according to claim 1, characterized in that: Calculation of irrigation coefficients for various crops includes calculation of soil influencing factors IFs, which has the following steps: According to the growth stage of each crop, the standard soil air content Ai', standard soil water content Mi' and standard soil temperature Ti' corresponding to the target crop i (i = 1, 2, ..., k, k is the number of crop types, that is, the number of sub-planting areas in the target area) are extracted from the soil information database; Obtain the soil air content Ai, soil water content Mi and soil temperature Ti of the sub-planting area of ​​the target crop i located in the target area; The soil impact factor IFs is calculated using the following formula: Among them, α, β, and χ are the influence weights of soil air content, soil moisture content, and soil temperature, respectively.

7. A smart agricultural irrigation method according to claim 6, characterized in that: Calculating the irrigation coefficient of each crop also includes calculating the environmental impact factor IFe, which has the following steps: The environmental impact factor IFe is calculated using the following formula: Among them, t, H, and L are air temperature, air humidity, and light intensity, respectively; δ and ε are the influence weights of light intensity and air humidity, respectively.

8. The smart agricultural irrigation method according to claim 7, characterized in that: Calculating the irrigation coefficient of each crop also includes calculating the irrigation coefficient ICi of target crop i: ICi=IFs*ω1+IFs*ω2 Among them, ω1 and ω2 are the influence weights of soil influencing factor IFs and environmental influencing factor IFe respectively.

9. A smart agricultural irrigation method according to claim 8, characterized in that: According to the irrigation coefficients of various crops, the irrigation methods corresponding to various crops are obtained, including the following steps: The corresponding irrigation method is extracted from the irrigation method database according to the growth stage of the target crop i and the irrigation coefficient ICi.

10. The smart agricultural irrigation method according to claim 8, characterized in that: Convolutional neural networks (CNNs) are used to obtain irrigation methods corresponding to various crops based on the environmental information of the target area, the soil information of each sub-planting area within the target area, and the growth stage of various crops, including the following steps: Collect various environmental information, soil information and growth stage data of various crops, set labels according to corresponding irrigation methods, and form historical data sets; Divide the historical data set into training set, validation set and test set according to the preset ratio; Construct the loss function and optimizer of the crop irrigation method recognition model based on convolutional neural networks (CNNs); The training set is input into the crop irrigation method recognition model for model training; The loss value is calculated based on the loss function, and the optimizer updates the model parameters according to the loss value and network gradient information; If the loss value is less than the preset threshold, the model training ends, and the current crop irrigation method recognition model is the trained crop irrigation method recognition model. Otherwise, return to step 4 and continue to use the training set for model training. Input the validation set into the trained crop irrigation method recognition model, evaluate the generalization ability of the model by observing the performance on the validation set, and tune the hyperparameters and structure of the model; The test set is input into the tuned crop irrigation method identification model to evaluate the model performance; After obtaining the pre-trained crop irrigation method recognition model, the collected environmental information of the target area, the soil information of the sub-planting area of ​​the target crop i within the target area, and the growth stage of the target crop i are input into the pre-trained crop irrigation method recognition model to obtain the irrigation method corresponding to the target crop i.

Images