Field rice water and fertilizer intelligent management and control method fusing environment perception and unmanned aerial vehicle image
By integrating environmental perception and drone images, an intelligent rice management and control model was constructed, which solved the problem of insufficient water and fertilizer management in the promotion of new rice varieties, realized intelligent irrigation and fertilization of rice, and improved water and fertilizer utilization efficiency and crop yield.
Patent Information
- Application Number
- CN202411104451.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-11-29
AI Technical Summary
In the process of promoting new rice varieties, the lack of supporting good varieties and methods has led to insufficient water and fertilizer management, the inability to achieve accurate rice growth information acquisition, inefficient growth monitoring, and data planning that cannot be predicted in advance, and the lack of intelligent control with unified standards.
By integrating environmental perception and drone images, a fusion mechanism model of phenotypic big data and meteorological environment information is constructed, combined with multi-source information parallel detection of remote sensing and environmental sensors, a mechanism model and multi-objective decision-making crop intelligence driven by coordination of spatiotemporal data is established. The management and control model uses deep learning methods to extract rice phenotype information, establish the best irrigation and fertilization model, and connect it to the integrated irrigation and fertilization device to achieve intelligent management and control.
It has realized intelligent control of water and fertilizers in rice production, improved the efficiency of climate resources and water and fertilizer utilization, provided high-throughput rice growth monitoring and intelligent regulation of fertilization strategies, and ensured that the crops are in the optimal growth state.
Smart Images

Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] The present invention relates to the fields of smart agriculture, agricultural aviation and intelligent field management and control, and in particular to a method for intelligent field rice water and fertilizer management by integrating environmental perception and unmanned aerial vehicle images. Background Art
[0002] Research on intelligent management and control methods for field rice is one of the hot topics that has emerged in the context of the current development of agricultural science and technology. In the process of promoting new rice varieties, agricultural production faces a series of challenges due to the lack of supporting good varieties and good methods, among which the lack of water and fertilizer management is a major problem. In order to solve such problems, scholars at home and abroad have carried out a series of research and are committed to building intelligent management and control models to improve rice yields and agricultural sustainability.
[0003] Conventional rice planting management has problems such as untimely acquisition of rice growth information, low efficiency and poor accuracy of rice growth monitoring, inability to predict data in advance and carry out planning and design in advance, and no unified standards for precise intelligent water and fertilizer management of large fields. In order to solve the above problems and shortcomings, the present invention provides a method for intelligent water and fertilizer management of large-scale rice by integrating environmental perception and drone images. Summary of the invention
[0004] 1. Technical issues to be resolved The present invention proposes a method for intelligent control of field rice water and fertilizer by integrating environmental perception and unmanned aerial vehicle images, so as to improve the climate resources and water and fertilizer utilization efficiency of rice production; the intelligent rice control method is aimed at the demand for the lack of supporting improved varieties and good methods in the process of promoting new rice varieties, and can realize intelligent control of water and fertilizer in rice production by constructing a fusion mechanism model of phenotypic big data and meteorological environmental information; at the same time, multi-source information parallel detection methods such as remote sensing and environmental sensors, and rice water and fertilizer diagnosis and regulation technology are studied, and a multi-objective decision-making crop intelligent control model driven by mechanism models and spatiotemporal data coordination is constructed to improve crop yield and water and fertilizer utilization efficiency.
[0005] (II) Technical solution The present invention provides a method for intelligent control of water and fertilizer for rice in a large field by integrating environmental perception and unmanned aerial vehicle images. By deploying meteorological stations and multi-source sensors, the meteorological and environmental information in the field is acquired in real time. At the same time, based on the deep learning method, the meteorological and environmental information is predicted at an early stage, and the best irrigation model for a period of time is established in combination with the elevation difference terrain data of the rice planting land; the overall growth of rice in the field is monitored by the unmanned aerial vehicle platform, and the rice phenotypic big data information is extracted from the visible light images and multispectral images taken by the unmanned aerial vehicle. After feature selection, the vegetation index VI that can reflect the fertilizer deficiency state of rice is established, the field rice is classified and judged by the VI information, and the best fertilization model is established in combination with the expert guidance value; finally, the best irrigation model and the best fertilization model are connected to the irrigation and fertilization integrated device to realize the intelligent control of the large field, as shown in the attached figure. Figure 1 The figure shows the overall technical roadmap of the present invention. In order to realize the contents of the present invention, the following scheme is adopted.
[0006] A method for intelligent control of rice water and fertilizer in a field by integrating environmental perception and drone images, the specific steps include: S1: Obtain field meteorological information and crop growth environment information, combine current meteorological environment data and subsequent meteorological forecast data, as well as elevation difference terrain data of rice planting areas, build a field intelligent irrigation control model, and realize early planning of field irrigation; Specifically, in step S1, the field environment information refers to meteorological information such as air temperature, air humidity, light, CO2 concentration, wind speed and wind direction, and the crop growth environment information refers to environmental information such as soil temperature, soil humidity, pH, nitrogen, phosphorus and potassium content; More specifically, field meteorological information is acquired by deploying internet-connected micro-meteorological stations in rice-growing fields. After the meteorological stations upload relevant data to the cloud, users can view them remotely, and subsequent meteorological data are predicted through the LSTM model. Crop growth environment information is acquired by deploying multi-source sensors in more refined rice-growing plots. The sensors process the signals of each plot and send them to the meteorological stations, which are then transmitted to users through the cloud.
[0007] S2: Use drones equipped with visible light and multispectral cameras to shoot above rice fields to obtain orthophotos of the fields. The shooting process covers the entire growth cycle of rice. Specifically, in step S2, the whole growth cycle of rice includes the seedling stage, tillering stage, booting stage, heading stage, milky stage and maturity stage. In each growth cycle, a sufficient amount of rice field images are obtained, and the time nodes of the transition of each period are recorded.
[0008] S3: stitching the acquired images of rice growth stages to obtain complete and clear field images, and then using batch processing scripts to further crop and segment the region of interest ROI; Specifically, in step S3, the image stitching process refers to aggregating the discretized rice images of different regions taken by the drone and processing them using remote sensing software to restore a single small image into a complete rice field image. At the same time, ROI refers to the area to be focused on, and the image of this part of the area can be segmented from the field image using automated tools.
[0009] S4: Enhance the ROI image selected in the above step S3. The image enhancement method is based on a deep learning semantic segmentation network to separate the rice growing part in the field from the background information such as soil, so as to achieve the effect of extracting the rice part.
[0010] S5: Process the image after the rice part is extracted in the above step S4, calculate the crop phenotypic information of the whole growth cycle of rice, verify the significance of each phenotypic parameter, and perform feature selection, as shown in the attached figure. Figure 2 As shown, it is a schematic diagram of obtaining phenotypic information of the UAV and extracting vegetation index according to the present invention.
[0011] Specifically, in terms of crop phenotypic information, the vegetation index VI extracted from drone images can characterize the growth and fertilizer deficiency of rice. The rice growth vegetation index includes the visible light vegetation index and the multispectral vegetation index. The two different data types reflect the differences in the color information of rice appearance and the internal nutritional information of rice. At the same time, due to the large number of vegetation index types and the different crop characteristics represented, in terms of rice growth and fertilizer deficiency discrimination, it is necessary to verify the significant impact of each vegetation index, select the best feature combination, enhance the prediction results and reduce the amount of calculation.
[0012] A method for intelligent control of rice water and fertilizer in a field by integrating environmental perception and drone images, the specific steps also include: S6: According to the vegetation index combination after feature selection described in the above step S5, the growth-VI relationship formula at different stages during the whole growth cycle of rice is established; and the rice fertilizer deficiency information is indirectly represented by the fertilizer deficiency-growth relationship formula using different growth information.
[0013] Specifically, in step S6, the growth potential-VI relationship formula needs to input the characteristic vegetation index of rice seedling stage, tillering stage, booting stage, heading stage, milky stage and mature stage for calculation respectively. Since the fertilizer deficiency judgment criteria of rice in different growth periods are different, it is necessary to add variables representing time nodes to the formula to further calculate the growth potential and fertilizer deficiency degree of rice in each period.
[0014] S7: According to the growth potential-VI relationship formula and the fertilizer deficiency-growth potential relationship formula described in the above step S6, a fusion calculation is performed to obtain a field intelligent fertilization control model, as shown in the attached figure. Figure 3 As shown, it is a schematic diagram of calculating the rice growth distribution and making fertilization decisions according to the present invention.
[0015] Specifically, in step S7, based on the calculated rice growth information, the standardized fertilizer deficiency difference values of rice in different areas in the field are determined, and a field fertilizer deficiency distribution map is made. Combined with the field management plan recommended by experts, the intelligent fertilization control model is output, and the field fertilization decision prescription map is obtained after calculation, so that the field fertilization strategy can be corrected in time to keep the rice in the best growth state during its growth and development.
[0016] S8: Using integrated water and fertilizer equipment, remote control can be performed to achieve intelligent and automated irrigation and fertilization operations for field crops, as shown in the attached Figure 4 As shown, it is a schematic diagram of an intelligent management and control system for the entire growth cycle of field rice that integrates environmental perception and drone images according to the present invention; Specifically, in step S8, the water-fertilizer integrated equipment includes a water source, a fertilizer storage tank, an electromagnetic control valve, a high-pressure water pump, and integrally arranged water pipes and wires. The start and stop control of each part of the equipment can be remotely controlled through the network to control the precise irrigation and fertilization operations of a specific community.
[0017] More specifically, the execution basis of the integrated water and fertilizer equipment comes from the field intelligent irrigation and fertilization control model described in the above steps S1 and S7, and the cloud platform is used as a transfer station to deploy the intelligent control model. After obtaining relevant data information and performing analysis and calculation, control decisions are made to enable the integrated water and fertilizer equipment to perform corresponding work, so as to achieve the optimal state of the rice planting area.
[0018] (III) Beneficial effects The present invention provides a method for intelligent control of field rice water and fertilizer by integrating environmental perception and drone images. By integrating and calculating big data information such as meteorology, environment and phenotype, an intelligent irrigation and fertilization control model for field rice is established. At the same time, the intelligent control model is combined with a water and fertilizer integrated machine to realize remote automatic and intelligent management measures such as irrigation and fertilization of field rice, thus achieving more refined and accurate crop growth management.
[0019] The present invention also provides a high-throughput rice growth monitoring method, which uses drones to capture images of rice fields and mine their phenotypic information, establishes a rice growth evaluation method based on a combination of multiple vegetation index features, and realizes rapid identification, calculation and analysis of the growth of field rice under different management scenarios.
[0020] The present invention also provides an intelligent control method for rice field fertilization strategy. According to the relationship between growth potential and vegetation index, the fertilizer deficiency of rice is indirectly characterized by the vegetation index, and the distribution of fertilizer deficiency in the field is judged. Finally, the method is corrected in combination with the management plan recommended by experts to make a prescription map for field fertilization. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 The overall technical roadmap of the method for intelligent control of field rice water and fertilizer by integrating environmental perception and drone images of the present invention; Figure 2 Schematic diagram of drone phenotypic information acquisition and vegetation index extraction; Figure 3 This is a schematic diagram of calculating rice growth distribution and making fertilization decisions based on a combination of characteristic vegetation indices; Figure 4 This is a schematic diagram of an intelligent management and control system for the entire growth cycle of field rice that integrates environmental perception and drone images according to the present invention. DETAILED DESCRIPTION
[0022] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.
[0023] The purpose of this invention is to propose an intelligent management and control method for field rice water and fertilizer by integrating environmental perception and drone images.
[0024] In order to make the above-mentioned objects, features and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to pictures and examples.
[0025] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.
[0026] like Figure 1 As shown, the present invention relates to a method for intelligent management and control of rice water and fertilizer in a field by integrating environmental perception and drone images, and the specific steps are as follows: S1: Obtain field weather information and deploy a micro-weather station in the rice planting area. The micro-weather station includes a main control system, a temperature observation instrument, a humidity observation instrument, a wind speed and direction observation instrument, a CO2 concentration observation instrument, and a solar radiation observation instrument, which can record weather information such as air temperature, air humidity, wind speed and direction, CO2 concentration, and light intensity. At the same time, the micro-weather station records and transmits data through the cloud through the main control system, can upload data in real time and store it for a long time, and users can view it remotely through the cloud platform on the web page.
[0027] To obtain the field crop growth environment information, multi-source sensors are deployed in more detailed rice planting areas. The sensors can obtain and process the growth environment information of each planting area, such as soil temperature, soil moisture, EC value, pH value, nitrogen content, phosphorus content and potassium content. At the same time, one end of the sensor is deployed in the field, and the other end is connected to the micro-meteorological station, and the crop growth environment information and field meteorological information are sent to the user end through the cloud for viewing.
[0028] The above-mentioned field meteorological information and environmental information are predicted early with the help of the time series prediction model LSTM. The prediction time span is 30 days. The intelligent irrigation control model for rice fields is mainly constructed based on the temperature, soil temperature and elevation difference terrain data of the rice planting area. The temperature mainly determines the irrigation frequency, the soil temperature mainly determines the water layer height, and the elevation difference data determines the total amount of single irrigation. The specific mathematical formula of the intelligent irrigation control model is as follows:
[0029] Where N represents the total number of irrigation times required in the next 30 days; T>35 Indicates the number of days when the temperature is above 35 degrees Celsius, at which it is recommended to irrigate every 3 days; N 26<T<35 Indicates the number of days with air temperature between 26 and 35 degrees Celsius, at which it is recommended to irrigate every 4 days; N T<26 Indicates the number of days the air temperature is below 26 degrees Celsius, at which temperature irrigation is recommended every 6 days.
[0030]
[0031] In the formula, I represents the single irrigation depth in the next 30 days; I ori Indicates the basic water layer height in the field; N (Tair-Tsoil)<0 Indicates the number of days when the air temperature is lower than the soil temperature. When the air temperature is lower than the soil temperature, it means that there is a large water shortage in the field. The direct sunlight causes the temperature to rise, and irrigation is carried out with deep water layer; N <0(Tair-Tsoil)<5 Indicates the number of days when the air temperature is slightly higher than the soil temperature. When the difference between the air temperature and the soil temperature is not large, irrigation is carried out with the normal water layer; N (Tair-Tsoil)>5 It indicates the number of days when the air temperature is much higher than the soil temperature. If the soil temperature is significantly lower than the air temperature, it means that there is sufficient water in the field and irrigation can be carried out with shallow water.
[0032]
[0033] Where W represents the total irrigation water consumption in a single irrigation; I / N represents the average water layer depth in a single irrigation; S field represents the total area of field rice cultivation; DSM h and DSM lThey represent the high point coordinate distance and the low point coordinate distance of the field respectively. The total water consumption for a single time can be calculated based on the geometric relationship.
[0034] S2: Use drone equipment equipped with visible light and multispectral cameras to shoot above the rice field. The shooting process covers the entire growth cycle of rice, including the seedling stage, tillering stage, booting stage, heading stage, milk stage and maturity stage of rice growth. During the rice growth cycle, use drones to obtain rice field images every 3 days. During each operation, the drone flight altitude is set to 30m, the flight speed is set to 2.5m / s, the gimbal angle is set to -90°, and the shooting is set to be performed at equal time intervals. The heading overlap rate and the lateral overlap rate are both set to 80%, and the time nodes of the transition of each period need to be recorded.
[0035] S3: The original image obtained by taking pictures with a drone is relatively discrete, with low recognizability and is not conducive to processing and calculation, and needs to be stitched. The image stitching software used in the present invention is DJI Terra software, which can automatically identify the content information of the image taken by the drone, and stitch the discretized rice images of each area to restore the complete original rice field image. After obtaining the rice field image, the area to be focused on can be segmented again with the help of automated tools to achieve accurate processing.
[0036] S4: In order to achieve accurate extraction of crop phenotypic information in step S5 and reduce the interference of irrelevant information, the field rice image is segmented using the semantic segmentation network SegFormer based on deep learning. Since the rice part is significantly different from the background part, it will have a greater impact on the final result when calculating the vegetation index. By masking the image, the rice growth part and the soil background part are separated into the foreground and background, and only the effective image information of the rice is retained.
[0037] S5: mining and calculating the crop phenotypic information of the whole growth cycle of rice. By calculating the pixel values of the rice part in the image, the crop phenotypic information described in the present invention can be obtained to characterize the growth and fertilizer deficiency of rice. The vegetation index VI can be used to characterize the image level. There are many types of vegetation indices. The present invention mainly uses the following vegetation indices: VI Type VI Calculation formula Visible light R 1.4RG Visible light G JZ 2G-RB Visible light ExGR ExG-ExR Visible light NGRDI (GR)(G+R) Visible light NGBDI (GB)(G+B) Visible light RGRI R / G Visible light GRRI G / R Visible light VDVI (2G-RB)(2G+R+B) Visible light VEG <![CDATA[G / (R a *B (1-a) ),a=0.667]]> Multispectral NDVI (NIR-R) / (NIR+R) Multispectral GNDVI (NIR-G) / (NIR+G) Multispectral LCI (NIR-RE) / (NIR+R) Multispectral NDRE (NIR-RE) / (NIR+RE) Multispectral OSAVI (NIR-R) / (NIR+R+0.16) R, G, and B correspond to the red, green, and blue color components of the image, respectively. NIR is the near-infrared band reflectivity, and RE is the red edge band reflectivity. In order to reduce the dimension of feature data and the amount of calculation, it is necessary to perform feature selection on each vegetation index. The feature selection method described in the present invention is based on the Pearson correlation coefficient method. First, for each vegetation index, the Pearson correlation coefficient r between it and the rice fertility deficiency state is calculated; the feature threshold is set to 0.5, and the vegetation index with a correlation coefficient r>0.5 is selected for combination; the above vegetation index combination is exhaustively searched to optimize the feature data subset, and the correlation coefficient between the feature combination and the rice fertility deficiency state is calculated again to determine the optimal subset with a higher correlation coefficient and a smaller number of features.
[0038] S6: The growth of rice can be characterized by the vegetation index. According to the combination of vegetation indices after feature selection, the growth-VI characterization formula of rice throughout its growth cycle is established. The vegetation index at the visible light image level reflects the color and texture characteristics of the rice part, which can be used to preliminarily judge the growth characteristics of rice; the vegetation index at the multi-spectral level reflects the intrinsic characteristics of the rice part through spectral reflectance, which can be used to judge the absorption of water and nutrients by rice. An expert rice planting management area is set up in the field as the best control for rice growth, and other areas are indirectly characterized. The specific rice growth-VI characterization formula is as follows:
[0039] Where G represents the standardized index of rice growth, which is obtained by averaging the vegetation index combination after feature screening and the standard index value; VI i Represents a subset of features in the vegetation index combination; VI i-std Indicates that the expert planting management area is VI i The corresponding standardized vegetation index value; T represents the rice growth time node.
[0040] The above rice characterization formula uses the vegetation index value of the expert planting management area as the standard to calculate which part of the rice growth in other areas reaches the standard value. At the same time, based on the above rice growth-VI characterization formula, it can indirectly reflect the fertilizer deficiency status of different areas in the rice field. If the growth is too different from the standard value, it means that there is a large fertilizer deficiency, otherwise it means that there is a small fertilizer deficiency. The specific rice fertilizer deficiency-growth characterization formula is as follows:
[0041] Where N x represents the fertilizer deficiency index of rice field area, which is calculated by subtracting the standard fertilizer requirement index of the expert management area from the fertilizer requirement index of growth trend inversion; α is the empirical factor, which represents the weighted coefficient of fertilizer requirement in different growth cycles of rice; T all Indicates the total growth period of this rice variety, T / T all Indicates which growth cycle the rice is in; N stdIt represents the standard fertilizer requirement index of the expert management area.
[0042] S7: Calculate the growth index value of each area in the rice field by the rice growth-VI characterization formula described in step S6 above, and make a growth distribution map according to the specific value; then calculate the fertilizer deficiency index value of each area in the rice field by the rice fertilizer deficiency-growth characterization formula described in step S6 above. The data of the two are integrated to construct an intelligent fertilization control model, and its specific mathematical formula is as follows:
[0043] In the formula, F represents the amount of additional fertilizer applied to the fertilizer-deficient area; ρ represents the average fertilizer concentration per plant; R c Indicates the number of rice plants in the area to be top-dressed; C std , C x are the rice growth coverage of the overall area and the rice growth coverage of the area to be top-dressed; β is the empirical factor for field rice fertilization. The intelligent fertilization control model calculates the specific amount of fertilizer for each area that needs additional fertilization, and makes a fertilization decision prescription map.
[0044] S8: Use water-fertilizer integrated equipment to remotely control the irrigation and fertilization of rice in the field. The water-fertilizer integrated equipment consists of a water source, a fertilizer storage tank, an electromagnetic control valve, a high-pressure water pump, integrally arranged water pipes and wires, and an integrated control system. The conventional workflow of the equipment is: first, use a mobile phone to remotely connect to the integrated control system of the equipment. If only irrigation operations are required, open the water source, high-pressure water pump, and electromagnetic control valves of each rice planting area, and set the irrigation amount. The water flow is transported to the field along the pipeline to complete the irrigation of the field; if fertilization operations are required, open the water source and high-pressure water pump, and add the required total fertilizer amount to the fertilizer storage tube. Subsequently, the electromagnetic valves of each planting area are opened in turn and fertilization is carried out according to the proportional duration to achieve remote automated management of rice fields.
[0045] The method for intelligent control of field rice water and fertilizer by integrating environmental perception and drone images described in the present invention can deploy the intelligent irrigation control model and intelligent fertilization control model in the above steps S1 and S7 in the water-fertilizer integrated equipment control system. By acquiring data through field monitoring in the field, the relevant big data information is input into the intelligent control model for calculation, the irrigation and fertilization decision-making amount is obtained, and the water-fertilizer integrated equipment is automatically called for irrigation and fertilization operations, so as to realize the automation and intelligence of field control.
[0046] The contents not described in detail in the specification of the present invention belong to the prior art known to the professional and technical personnel in the field. Although the illustrative specific embodiments of the present invention are described above to facilitate the understanding of the present invention by the technical personnel in the field, it should be clear that the present invention is not limited to the scope of the specific embodiments. For ordinary technicians in the field of this technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the attached claims, these changes are obvious, and all inventions and creations using the concept of the present invention are protected.
Claims
1. A method for intelligent control of field rice water and fertilizer by integrating environmental perception and drone images. The method acquires field meteorological and environmental information in real time by deploying meteorological stations and multi-source sensors. At the same time, based on deep learning methods, early prediction of meteorological and environmental information is performed. Combined with the elevation difference terrain data of the rice planting area, an optimal irrigation model for a period of time is established. The overall growth of field rice is monitored by a drone platform. The rice phenotypic big data information is extracted from the visible light images and multispectral images taken by the drone. After feature selection, a vegetation index VI that can reflect the fertilizer deficiency state of rice is established. The field rice is classified and judged by the VI information, and the optimal fertilization model is established in combination with the expert guidance value. Finally, the optimal irrigation model and the optimal fertilization model are connected to the irrigation and fertilization integrated device to realize intelligent control of the field. The method is completed by the following steps: S1 obtains field meteorological information and crop growth environment information, specifically meteorological information such as temperature, humidity, light, CO2 concentration, wind speed and direction, and environmental information such as soil temperature, soil humidity, pH, nitrogen, phosphorus and potassium content, and predicts the time series information. It combines the current meteorological and environmental data and the predicted value data, as well as the elevation difference terrain data of the rice planting area, to build a field intelligent irrigation control model to achieve early planning of field irrigation; S2 uses drones equipped with visible light and multi-spectral cameras to shoot above rice fields to obtain orthophotos of the fields. The shooting process covers the entire growth cycle of rice. S3 splices several images of rice growth stages to obtain complete and clear field images, and then uses batch processing scripts to further crop and segment the region of interest ROI; S4 performs enhancement processing on the ROI image selected in the above step S3, wherein the image enhancement method is based on a deep learning semantic segmentation network, which separates the rice growing part in the field from the background information such as soil, so as to achieve the effect of extracting the rice part; S5 processes the image after the rice part is extracted in the above step S4, mines the crop phenotypic information of the whole growth cycle of rice, verifies the significance of each phenotypic parameter, and performs feature selection; S6: establishing a growth potential-VI relationship formula at different stages of the whole growth cycle of rice according to the vegetation index combination after feature selection described in the above step S5; and indirectly representing the rice fertilizer deficiency information with the fertilizer deficiency-growth potential relationship formula using different growth potential information; S7 performs a fusion calculation based on the growth potential-VI relationship formula and the fertilizer deficiency-growth potential relationship formula described in the above step S6 to obtain a field intelligent fertilization control model; S8 uses integrated water and fertilizer equipment for remote control to achieve intelligent and automated irrigation and fertilization operations for field crops.
2. According to claim 1, a method for intelligent management and control of field rice water and fertilizer by integrating environmental perception and drone images, characterized in that: Acquiring field meteorological information and crop growth environment information as described in S1: S1.1 Field environmental information refers to meteorological information such as air temperature, air humidity, light, CO2 concentration, wind speed and wind direction, while crop growth environmental information refers to environmental information such as soil temperature, soil humidity, pH, nitrogen, phosphorus and potassium content; S1.2 Field meteorological information is obtained by deploying an internet-connected micro-meteorological station in the rice-growing field. The weather station uploads the relevant data to the cloud, and users can view it remotely. Crop growth environment information is obtained by deploying multi-source sensors in more detailed rice-growing plots. The sensors process the signals of each plot and send them to the weather station, which are then transmitted to users through the cloud. S1.3 uses the time series prediction model LSTM to make early predictions on the above-mentioned field meteorological information and environmental information, and constructs a field rice intelligent irrigation control model based on the air temperature, soil temperature and elevation difference terrain data of the rice planting area; the air temperature determines the irrigation frequency, the soil temperature determines the water layer height, and the elevation difference data determines the total amount of single irrigation. The mathematical formula of the intelligent irrigation control model of the present invention is expressed as follows: The predicted irrigation times N is:
3. Single irrigation depth I is:
4. The average total water consumption for a single irrigation is:
5. According to claim 1, a method for intelligent management and control of field rice water and fertilizer by integrating environmental perception and drone images, characterized in that: The orthophoto images of the field described in S2 cover the entire growth cycle of rice: S2.1 The whole growth cycle of rice includes seedling stage, tillering stage, booting stage, heading stage, milky stage and maturity stage. In each growth cycle, sufficient images of rice fields are obtained and the time nodes of each stage transition are recorded; S2.2 During the rice growth cycle, the image of the rice field is acquired using a drone every 3 days, and the time nodes of the transition between each period are recorded.
6. According to claim 1, a method for intelligent management and control of field rice water and fertilizer by integrating environmental perception and drone images, characterized in that: As described in S3, several images of rice growth stages are spliced, and the region of interest ROI is further cropped and segmented: S3.1 Image stitching processing refers to the process of aggregating the discrete rice images of different areas taken by drones and processing them using remote sensing software to restore a single small image into a complete rice field image; S3.2 The ROI area refers to the rice field area of focus, and the automated processing tool is used to segment the area of focus from the field image.
7. According to claim 1, a method for intelligent management and control of field rice water and fertilizer by integrating environmental perception and drone images, characterized in that: Separation of rice growing in the field from background information such as soil as described in S4: S4.1 Since the rice part is significantly different from the background part, it will have a great impact on the final result when calculating the vegetation index. The semantic segmentation network SegFormer based on deep learning is used to segment the field rice image; S4.2 performs mask processing on the image, focusing on separating the foreground and background of the rice growing part from the soil background part, and only retains the effective image information of the rice.
8. The method for intelligent management and control of field rice water and fertilizer by integrating environmental perception and drone images according to claim 1 is characterized in that: As described in S5, mining the crop phenotypic information of the entire growth cycle of rice, verifying the significance of each phenotypic parameter, and performing feature selection: S5.1 mining and calculating the crop phenotypic information of the whole growth cycle of rice. By calculating the pixel values of the rice part in the image, the crop phenotypic information described in the present invention can be obtained to characterize the growth and fertilizer deficiency of rice, which can be characterized by the vegetation index VI at the image level; S5.2 performs feature selection on each vegetation index. The feature selection method described in the present invention is based on the Pearson correlation coefficient method. Vegetation indices with a Pearson correlation coefficient r>0.5 are selected for combination, and an exhaustive search is performed on the vegetation index combination to optimize the feature data subset. The correlation coefficient between the feature combination and the fertilizer deficiency state of rice is calculated again to determine the optimal subset with a higher correlation coefficient and a smaller number of features.
9. The method for intelligent management and control of field rice water and fertilizer by integrating environmental perception and drone images according to claim 1 is characterized in that: The growth potential-VI relationship formula at different stages of the rice growth cycle described in S6 is established, and different growth potential information is used to indirectly represent the rice fertilizer deficiency information: S6.1 The growth of rice can be characterized by the vegetation index. The vegetation index at the visible light image level reflects the color and texture characteristics of the rice part, which can be used to preliminarily judge the growth characteristics of rice; the vegetation index at the multispectral level reflects the intrinsic characteristics of the rice part through spectral reflectance, which can be used to judge the absorption of water and nutrients by rice. The combination of the two establishes a rice growth-VI relationship formula. The mathematical formula of the rice growth-VI relationship described in the present invention is expressed as follows: 10.S6.2 Based on the rice growth potential-VI characterization formula, the fertilizer deficiency status of different areas in the rice field can be indirectly reflected. If the growth potential differs too much from the standard value, it means that the fertilizer deficiency is greater, and vice versa, it means that the fertilizer deficiency is less. The mathematical formula of the relationship between rice fertilizer deficiency and growth potential described in the present invention is expressed as follows:
11. The method for intelligent management and control of rice water and fertilizer in a field by integrating environmental perception and drone images according to claim 1, characterized in that: The field intelligent fertilization control model described in S7 is: The growth index values of each area in the rice field are calculated by the rice growth-VI characterization formula, and the fertilizer deficiency index values of each area in the rice field are calculated by the rice fertilizer deficiency-growth characterization formula. The two data are integrated, and the intelligent fertilization control model is constructed in combination with the expert advice and management standards. The mathematical formula of the rice intelligent fertilization control model of the present invention is expressed as follows:
12. The method for intelligent management and control of field rice water and fertilizer by integrating environmental perception and drone images according to claim 1, characterized in that: The use of integrated water and fertilizer equipment as described in S8 for remote control can achieve intelligent and automated irrigation and fertilization operations for field crops: S8.1 The integrated water and fertilizer equipment includes water source, fertilizer storage tank, electromagnetic control valve, high-pressure water pump and integrally arranged water pipes and wires. The start and stop control of each part of the equipment can be remotely controlled through the network to control the precise irrigation and fertilization operations of a specific plot; S8.2 The intelligent irrigation control model and intelligent fertilization control model for rice fields are deployed in the water-fertilizer integrated equipment control system. The cloud platform serves as a transfer station. Data is obtained through on-site monitoring of the field, and relevant big data information is input into the intelligent control model for calculation to obtain irrigation and fertilization decision-making quantities. The water-fertilizer integrated equipment is automatically called to perform irrigation and fertilization operations, thereby realizing automation and intelligence of field control.
Citation Information
Cited By
Precise topdressing method of agricultural unmanned aerial vehicle for rice
CN120814398A
A method for precise topdressing of rice using agricultural drones
CN120814398B
Fruit planting water and fertilizer management and control system and method based on deep learning
CN120827043A
Crop growth prediction method and device, equipment and storage medium
CN121838124A
