Crop growth period big data analysis and fertilizer optimization method and device

Abstract

The invention relates to the technical field of data analysis, in particular to a crop growth period big data analysis and fertilizer optimization method and device.A machine learning algorithm is adopted, a linear regression model corresponding to the crop yield, the growth stage, meteorological data and the soil nutrient content is established, and the crop predicted yield is set; according to the crop growth period big data analysis and fertilizer optimization method and device, a soil nutrient sensor is used for detecting the current soil nutrient content, the current growth stage of crops is determined, the soil nutrient content needed by the current crops in different growth stages is calculated according to a linear regression model, and the crops are precisely fertilized. The crop yield after accurate fertilization through the linear regression model is received, the learning rate is set, the linear regression model is optimized according to the error value of the current crop yield and the predicted yield, the optimized linear regression model can guide investment of resources such as fertilizer more accurately, and fertilizer waste is reduced while the crop yield is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and in particular to a method and device for big data analysis and fertilizer optimization during the growth period of crops. Background Art

[0002] Crops have different requirements for soil nutrients at different growth stages, which are due to differences in their physiological processes, changes in metabolic pathways, and different requirements for environmental adaptability. For example, during the seed germination period, nitrogen is needed to promote protein synthesis, and phosphorus is needed to help root growth. During the vegetative growth period, nitrogen is used for chlorophyll synthesis, and potassium regulates stomata and transport. During the reproductive growth period, phosphorus promotes flower bud differentiation and energy metabolism, and potassium improves fruit quality and expansion.

[0003] In order to achieve scientific planting of crops, crop growers will build a linear regression model of crops based on a large amount of historical data during the long-term process of planting crops. Accurate analysis of crops at different growth stages can create a suitable growth environment for them, such as precisely controlling soil nutrients, water, light and other conditions, thereby effectively promoting the healthy growth of crops and significantly improving the yield and quality of crops.

[0004] In the process of constructing a linear regression model, once the historical data is insufficient or inaccurate, the linear regression model may have flaws and loopholes, and cannot accurately reflect the real laws and characteristics of crop growth, which will cause deviations and errors in simulating and predicting crop growth conditions and providing guidance for agricultural production decisions, seriously affecting the effectiveness of scientific planting and the improvement of agricultural production efficiency. In view of this, we propose a method and device for big data analysis and fertilizer optimization during the crop growth period. Summary of the invention

[0005] The purpose of the present invention is to solve the key problems of the current lag in updating of the linear regression model of crops and the lack of precision in fertilization at different growth stages.

[0006] To achieve the above purpose, the present invention provides a crop growth period big data analysis and fertilizer optimization method that can closely follow the specific growth stage of the crop, use accurate and scientific strategies to carry out fertilization operations, and at the same time, rely on advanced data processing and analysis mechanisms to update the linear regression model, including the following method steps:

[0007] S1. Obtain the nutrient absorption spectrum characteristic electrical signals, the mean value interval of nutrient consumption rate at different growth stages and meteorological data through a web crawler, as well as receive the historical crop yield data, soil nutrients and meteorological data corresponding to the crop growth stage in the planting records, and classify and store them into the crop growth period database according to the historical crop growth stage;

[0008] S2. Use machine learning algorithms to establish a linear regression model corresponding to crop yield, growth stage, meteorological data, and soil nutrient content;

[0009] S3. Set the predicted yield of crops, use soil nutrient sensors to detect the current soil nutrient content, determine the current growth stage of crops, calculate the soil nutrient content required for different growth stages of crops based on the linear regression model, and accurately fertilize crops;

[0010] S4. Receive the current crop yield after precise fertilization, set the learning rate, and optimize the linear regression model according to the error value between the current crop yield and the predicted yield.

[0011] As a further improvement of the technical solution, the crop growth period database simulates a browser through a web crawler, and sends HTTP requests to the academic database and the meteorological data server respectively, requesting to obtain the electrical signals of the absorption spectrum characteristics of the nutrient substances in the soil, the mean interval of the nutrient consumption rate in different growth stages, and the meteorological signals;

[0012] After receiving the request, the corresponding server returns the corresponding web page data to the web crawler according to the requested URL, so that the web crawler obtains the absorption spectrum characteristic electrical signals of various nutrients in the soil, the average range of nutrient consumption rate in different growth stages and meteorological signals.

[0013] Receive the corresponding historical crop yield data, soil nutrient content at different historical growth stages and historical meteorological data in the planting records, and classify the soil data according to different growth stages of crops and store them in the crop growth period database.

[0014] As a further improvement of the present technical solution, the steps of establishing a linear regression model using a machine learning algorithm in S2 are as follows:

[0015] Receive historical crop yield data stored in the crop growth period database, soil nutrient content and meteorological data corresponding to different crop growth stages;

[0016] Due to the characteristic basis of crop growth stage requirements, the interrelationship of physiological processes and the synergistic mechanism of environmental factors, there is a linear relationship between crop yield data and soil nutrient content and meteorological data corresponding to different crop growth stages. Therefore, the relationship between them is expressed by the following linear regression model:

[0017] y1=β1x1+β2x2+…+β i x i ;

[0018] y2=β1x1+β2x2+…+β i x i ; ...;

[0020] y i =β1x1+β2x2+…+β i x i ;

[0021] Among them, β1-β i is the regression coefficient, which means x1-x i The impact on production, x1-x i is y1-y i Soil nutrient content and meteorological data corresponding to the growth stage, such as nitrogen content x1, phosphorus content x2, average temperature x3, precipitation x4, potassium content x i , y1-y i It is the historical yield corresponding to different growth stages of crops.

[0022] As a further improvement of this technical solution, different regression coefficients in the linear regression model are calculated by the least squares method:

[0023]

[0024] Among them, β i is the regression coefficient, n is the number of samples of multiple crop yields, soil nutrient content at different growth stages, and meteorological data, x i is the soil nutrient content and meteorological data of the i-th sample, y i is the crop yield of the i-th sample;

[0025] is the average value of multiple soil nutrient content sample data:

[0026] is the average value of multiple crop yield sample data

[0027] The relationship between crop yield and soil nutrient content and meteorological data was determined by calculating the regression coefficient.

[0028] As a further improvement of the technical solution, the step of using the soil nutrient sensor to detect the current soil nutrient content in S3 is as follows:

[0029] The output end of the soil nutrient sensor emits light with a specific wavelength range to the soil around the crops. When the light with a specific wavelength range irradiates the soil, the nutrients in the soil will absorb part of the light energy, causing the light intensity through the soil to change;

[0030] The receiving end of the soil nutrient sensor receives the light that passes through the soil and converts the light transmission signal into an electrical signal;

[0031] Receive the absorption spectrum characteristic electrical signals of various nutrients in the soil obtained by the web crawler;

[0032] The electrical signals are acquired and converted to obtain the content of various nutrients in the soil;

[0033] The calculation formula is as follows:

[0034]

[0035] Among them C meas is the current content of nutrients in the soil, I meas The electrical signal converted by the soil nutrient sensor, I std is the electrical signal obtained by web crawler, C std For I std The electrical signal corresponds to the known nutrient content.

[0036] As a further improvement of the technical solution, the step of determining the current crop growth stage in S3 is as follows:

[0037] Receive the soil nutrient content difference between two adjacent time points of the current crop, and calculate the nutrient consumption rate of the crop according to the change of the nutrient difference;

[0038]

[0039] Among them, v is the current rate of nutrient consumption of crops, C1-C2 is the difference in soil nutrient content between two adjacent time points, and v=Δt is the difference between adjacent time points;

[0040] Receive the mean range of nutrient consumption rates of crops at different growth stages obtained in S1:

[0041] Seedling stage mean interval: [X A ,X AI ];

[0042] Jointing stage mean range: [X AI ,X B ];

[0043] …;

[0044] Maturity mean range: [X Z ,X ZI ]

[0045] Receive the current nutrient consumption rate of crops: v;

[0046] If X A <v<X AI , it means that the current crops are in the seedling stage. If X A <v<X AI , XAI <v<X B This indicates that the current crops are in the jointing stage.

[0047] As a further improvement of the technical solution, the step of calculating the soil nutrient content required for different growth stages of the current crop according to the linear regression model in S3 is as follows:

[0048] Receive current crop growth stage, soil nutrient content and crop yield forecast;

[0049] Call out the linear regression model in S2 that matches the current growth stage;

[0050] y i预 =β1x 1需 +β2x 2需 +…+β i x i需 ;

[0051] Calculate the nutrients required for the current growth stage through the linear regression model;

[0052]

[0053] Among them, y i预 To predict crop yield, β i is the correlation coefficient corresponding to the i-th crop yield, x 1需 Current soil nutrient content and meteorological data;

[0054] Calculate the difference between the current soil nutrients of crops and the required nutrients, and output control signals to accurately fertilize crops;

[0055] x i调 =x i需 -x i实 ;

[0056] Among them, x i调 To adjust the parameters of soil nutrients, x i需 is the soil nutrients required by current crops, x i实 The actual soil nutrients of the current crops detected.

[0057] As a further improvement of the technical solution, the steps of optimizing the linear regression model in S4 are as follows:

[0058] Receive the actual yield Y of m farmlands after precise fertilization 实际 , the predicted crop yield y set in S3 i设 ;

[0059] Calculate the mean square error between the actual output and the predicted output

[0060] Among them, Y 实际 -Y i设 is the error value between the predicted yield and the actual yield of each farmland, and MSE is the average difference between the predicted yield and the actual yield of all farmlands. The smaller the MSE value, the higher the accuracy of the established linear regression model.

[0061] Definition of loss function: Take mean square error MSE as loss function L(β), where β = (β1, β2, …, β i ) T is the regression coefficient vector;

[0062] Calculate the loss function with respect to each regression coefficient β j , β j =β1-β i The partial derivative of

[0063] For the linear regression model Among them, x1, x2, …, x i is the soil nutrient content, The partial derivatives of other coefficients can be calculated similarly;

[0064] Assume that the learning rate η is a positive number used to control the step size of each update. According to the gradient descent algorithm, the regression coefficient is updated

[0065] As a further improvement of this technical solution, the steps of setting the learning rate are as follows:

[0066] Set the initial learning rate η = 0.001, and adjust the learning rate by Δη = 0.001;

[0067] receiving an adjusted performance indicator MSE;

[0068] If MSE new <MSE init , the linear regression model in the feature analysis module can converge faster, continue to increase the learning rate and repeat the above process;

[0069] If MSE new Change MSE init If no change occurs, it means that the learning rate is too large and it is not possible to increase crop yields by providing soil nutrients. At this time, the linear regression model is no longer adjusted.

[0070] The device of the big data analysis and fertilizer optimization method during the growth period of crops comprises a fertilization structure, wherein the fertilization structure comprises a fertilization aircraft, a fertilizer storage box for storing mixed fertilizers is installed in the middle of the fertilization aircraft, two sealing doors are installed on the top of the fertilizer storage box, a feeder is installed on the side of the fertilization aircraft close to the fertilizer storage box, a feeding end of the feeder is through-connected with the fertilizer storage box, a discharging end of the feeder is through-connected with a connecting pipe, and the outer side wall of the connecting pipe is through-connected with three feeding nozzles.

[0071] When precise fertilization is required, the staff first injects the mixed fertilizer into the fertilizer storage box, and then seals the fertilizer storage box through two sealing doors. The fertilization aircraft flies above the crops that need to be fertilized, and then starts the feeder to extract the fertilizer inside the fertilizer storage box, and then sprays the fertilizer on the crops through the connecting pipe and the feeder nozzle.

[0072] Compared with the prior art, the present invention has the following beneficial effects:

[0073] 1. In the crop growth period big data analysis and fertilizer optimization method and device, a linear regression model corresponding to crop yield, growth stage, meteorological data and soil nutrient content is established through a machine learning algorithm, and a linear regression model including crop growth stages is established through a machine learning algorithm, so that the yield can be predicted according to the characteristics of different growth stages;

[0074] For example, during the seed germination period, the model will comprehensively consider the impact of nitrogen and phosphorus content in the soil and meteorological data such as temperature and humidity at the time on the seed germination rate, which plays an important role in laying the foundation for subsequent yield estimates.

[0075] The soil nutrient sensor can detect the current soil nutrient content and determine the current growth stage of crops. By detecting the soil nutrient content in real time and determining the growth stage of crops, the best time for fertilization can be seized.

[0076] For example, when it is detected that the nitrogen content in the soil decreases rapidly during the nutritional growth stage of crops, it means that the crops' absorption of nitrogen is accelerated. At this time, nitrogen fertilizer should be added in time to ensure that crops are supplied with nutrients when they need them most, avoiding problems such as yellowing leaves and slow growth due to late fertilization, so that the growth process of crops is not affected.

[0077] 2. In the method and device for big data analysis and fertilizer optimization during the crop growth period, the soil nutrients required for the crops in the current growth stage are calculated according to the set predicted crop yield, and then the fertilization structure is used to accurately fertilize the crops. The soil nutrients required for the current growth stage are calculated according to the set predicted crop yield, which can ensure that the fertilizer supply is highly matched with the actual needs of the crops. During the fertilization process of the fertilization structure, the fertilization aircraft flies above the crops that need to be fertilized, and then the feeder is started to extract the fertilizer in the fertilizer storage box, and then the fertilizer is sprayed on the crops through the connecting pipe and the feeder nozzle. The fertilization aircraft is not restricted by the terrain and can quickly cross vast farmlands or complex planting areas to complete large-scale fertilization tasks in a short time.

[0078] 3. In the crop growth period big data analysis and fertilizer optimization method and device, the crop yield after accurate fertilization through the linear regression model is received, the learning rate is set, and the linear regression model is optimized according to the error value between the current crop yield and the predicted yield. The optimized linear regression model can more accurately guide the input of resources such as fertilizers, while ensuring the crop yield and reducing fertilizer waste;

[0079] For example, through continuous optimization of the model, growers can accurately purchase and use fertilizers based on more accurate fertilization recommendations, avoiding capital occupation and waste caused by excessive fertilizer reserves. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 It is a flow chart of the working principle of the present invention;

[0081] Figure 2 It is a three-dimensional diagram of the fertilization structure of the present invention;

[0082] Figure 3 It is a schematic three-dimensional diagram of the feed nozzle of the present invention.

[0083] The meaning of each number in the figure is:

[0084] 100. Fertilization structure; 101. Fertilization aircraft; 102. Fertilizer storage box; 103. Sealed door; 104. Feeder; 105. Connecting pipe; 106. Feeding nozzle; DETAILED DESCRIPTION

[0085] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0086] The method for big data analysis and fertilizer optimization during the crop growth period includes the following steps:

[0087] S1. Obtain the nutrient absorption spectrum characteristic electrical signals, the mean value interval of nutrient consumption rate at different growth stages and meteorological data through a web crawler, as well as receive the historical crop yield data, soil nutrients and meteorological data corresponding to the crop growth stage in the planting records, and classify and store them into the crop growth period database according to the historical crop growth stage;

[0088] S2. Use machine learning algorithms to establish a linear regression model corresponding to crop yield, growth stage, meteorological data, and soil nutrient content;

[0089] Based on this linear regression model, the type and amount of fertilizer required can be accurately calculated according to the current growth stage of crops, the existing nutrient content of the soil, and weather forecast information. Compared with traditional fertilization based on experience, it can avoid insufficient fertilization that leads to limited crop growth or excessive fertilization that causes waste of resources and environmental pollution. For example, in vegetable planting, the model can accurately determine the amount and time of nitrogen fertilizer application based on the current available nitrogen content in the soil, the growth stage of the vegetables, and future weather conditions, ensuring that the vegetables obtain sufficient nutrients while reducing the risk of fertilizer loss, improving fertilizer utilization, and reducing production costs;

[0090] S3. Set the predicted yield of crops, use soil nutrient sensors to detect the current soil nutrient content, determine the current growth stage of crops, calculate the soil nutrient content required for different growth stages of crops based on the linear regression model, and accurately fertilize crops;

[0091] Because precision fertilization can meet the nutrient needs of crops at each growth stage, it avoids the restriction of growth and development due to nutrient deficiency or imbalance, thus effectively increasing crop yields;

[0092] S4, receiving the current crop yield after precise fertilization, setting a learning rate, and optimizing a linear regression model according to the error value between the current crop yield and the predicted yield;

[0093] Existing technologies may have certain limitations in scientific planting, such as inaccurate simulation of crop growth, which leads to extensive planting management. This method can further explore the growth laws of crops by continuously optimizing the linear regression model, providing stronger support for scientific planting.

[0094] For example, the model can analyze in detail the impact of the interaction of various environmental factors and agricultural operations at different growth stages on yield. Based on this, growers can adopt more refined planting patterns, such as precise intercropping patterns and reasonable planting density adjustments, to improve land utilization and crop yields and enhance the effectiveness of scientific planting.

[0095] The crop growth period database simulates a browser through a web crawler, sending HTTP requests to the academic database and the meteorological data server, respectively, requesting to obtain the electrical signals of the absorption spectrum characteristics of nutrients in the soil, the mean range of nutrient consumption rates at different growth stages, and meteorological signals;

[0096] After receiving the request, the corresponding server returns the corresponding web page data to the web crawler according to the requested URL, so that the web crawler obtains the absorption spectrum characteristic electrical signals of various nutrients in the soil, the average range of nutrient consumption rate in different growth stages and meteorological signals.

[0097] Receive the corresponding historical crop yield data, soil nutrient content at different historical growth stages and historical meteorological data in the planting records, and classify the soil data according to different growth stages of crops and store them in the crop growth period database.

[0098] Crops have different requirements and absorption of soil nutrients at different growth stages;

[0099] For example, during the seed germination stage, the demand for nitrogen, phosphorus, potassium and other macronutrients in the soil may be low, but the soil permeability and humidity requirements are high; in the flowering and fruiting stage, the demand for phosphorus and potassium will increase sharply to promote flower bud differentiation and fruit development. Storing soil data by growth stage can better match the characteristics of each stage and facilitate subsequent analysis of the impact of soil nutrient changes at different stages on crop yields.

[0100] Historical meteorological data also have different effects at different growth stages;

[0101] For example, during the seedling stage of crops, moderate rainfall and suitable temperature are conducive to the healthy growth of seedlings; while during the mature stage, excessive rainfall may lead to the breeding of pests and diseases or fruit rot. Therefore, it is necessary to analyze the impact of meteorological data in combination with the growth stage of crops. Classified storage helps to study these impacts in a targeted manner.

[0102] As planting records accumulate, the amount of data will become larger and larger. Classifying soil data by growth stage and storing it in the crop growth period database can make data management more orderly.

[0103] The steps in S2 to establish a linear regression model using machine learning algorithms are as follows:

[0104] Receive historical crop yield data stored in the crop growth period database, soil nutrient content and meteorological data corresponding to different crop growth stages;

[0105] Due to the characteristic basis of crop growth stage requirements, the interrelationship of physiological processes and the synergistic mechanism of environmental factors, there is a linear relationship between crop yield data and soil nutrient content and meteorological data corresponding to different crop growth stages. Therefore, the relationship between them is expressed by the following linear regression model:

[0106] y1=β1x1+β2x2+…+β i x i ;

[0107] y2=β1x1+β2x2+…+β i x i ;

[0108] …;

[0109] y i =β1x1+β2x2+…+β i x i ;

[0110] Among them, β1-β i is the regression coefficient, which means x1-x i The impact on production, x1-x i is y1-y i Soil nutrient content and meteorological data corresponding to the growth stage, such as nitrogen content x1, phosphorus content x2, average temperature x3, precipitation x4, potassium content x i , y1-y i The historical yields of crops at different growth stages;

[0111] The regression coefficient reflects the direction and degree of the impact of soil nutrient content on yield. At a certain growth stage, if the regression coefficient corresponding to a certain soil nutrient is positive and has a large value, it means that increasing the soil nutrient content at this stage will increase the yield, that is, the demand of crops for this soil nutrient at this stage may be more significant;

[0112] At the same time, by observing the changes in regression coefficients at different growth stages, we can analyze the dynamic characteristics of the demand for soil nutrients by crops throughout the growth cycle;

[0113] For example, if the regression coefficient of nitrogen content is larger at the seedling stage, and the regression coefficient of potassium content is larger at the maturity stage, it means that the crop has a higher demand for nitrogen at the seedling stage and a higher demand for potassium at the maturity stage.

[0114] Calculate different regression coefficients in the linear regression model using the least squares method:

[0115]

[0116] Among them, β iis the regression coefficient, n is the number of samples of multiple crop yields, soil nutrient content at different growth stages, and meteorological data, x i is the soil nutrient content and meteorological data of the i-th sample, y i is the crop yield of the i-th sample;

[0117] is the average value of multiple soil nutrient content sample data:

[0118] is the average value of multiple crop yield sample data

[0119] The relationship between crop yield and soil nutrient content and meteorological data was determined by calculating the regression coefficient.

[0120] The steps of using soil nutrient sensor in S3 to detect the current soil nutrient content are as follows:

[0121] The output end of the soil nutrient sensor emits light with a specific wavelength range to the soil around the crops. When the light with a specific wavelength range irradiates the soil, the nutrients in the soil will absorb part of the light energy, causing the light intensity through the soil to change;

[0122] For example, particles and nutrients in the soil scatter incident light. The intensity and angular distribution of scattered light are related to the physical properties and nutrient content of the soil. Generally speaking, the smaller the soil particles and the higher the nutrient content, the stronger the intensity of scattered light, and the angular distribution of scattered light will also change. By collecting and analyzing the information of scattered light, relevant data on soil nutrients can be obtained.

[0123] The receiving end of the soil nutrient sensor receives the light that passes through the soil and converts the light transmission signal into an electrical signal;

[0124] Receive the absorption spectrum characteristic electrical signals of various nutrients in the soil obtained by the web crawler;

[0125] The electrical signals are acquired and converted to obtain the content of various nutrients in the soil;

[0126] The calculation formula is as follows:

[0127]

[0128] Among them C meas is the current content of nutrients in the soil, I meas The electrical signal converted by the soil nutrient sensor, I std is the electrical signal obtained by web crawler, C std For I std The electrical signal corresponds to the known nutrient content.

[0129] The steps in S3 to determine the current crop growth stage are as follows:

[0130] Receive the soil nutrient content difference between two adjacent time points of the current crop, and calculate the nutrient consumption rate of the crop according to the change of the nutrient difference;

[0131]

[0132] Among them, v is the current rate of nutrient consumption of crops, C1-C2 is the difference in soil nutrient content between two adjacent time points, and v=Δt is the difference between adjacent time points;

[0133] Receive the mean range of nutrient consumption rates of crops at different growth stages obtained in S1:

[0134] Seedling stage mean interval: [X A ,X AI ];

[0135] Jointing stage mean range: [X AI ,X B ];

[0136] …;

[0137] Maturity mean range: [X Z ,X ZI ]

[0138] Receive the current nutrient consumption rate of crops: v;

[0139] If X A <v<X AI , it means that the current crops are in the seedling stage. If X A <v<X AI , X AI <v<X B This indicates that the current crops are in the jointing stage.

[0140] The steps in S3 to calculate the soil nutrient content required for different growth stages of current crops based on the linear regression model are as follows:

[0141] Receive current crop growth stage, soil nutrient content and crop yield forecast;

[0142] Call out the linear regression model in S2 that matches the current growth stage;

[0143] y i预 =β1x 1需 +β2x 2需 +…+β i x i需 ;

[0144] Calculate the nutrients required for the current growth stage through the linear regression model;

[0145]

[0146] Among them, y i预 To predict crop yield, β i is the correlation coefficient corresponding to the i-th crop yield, x 1需 Current soil nutrient content and meteorological data;

[0147] Calculate the difference between the current soil nutrients of crops and the required nutrients, and output control signals to accurately fertilize crops;

[0148] x i调 =x i需 -x i实 ;

[0149] Among them, x i调 To adjust the parameters of soil nutrients, x i需 is the soil nutrients required by current crops, x i实 To test the actual soil nutrients of current crops;

[0150] The calculation example is as follows:

[0151] The linear regression model constructed by the feature analysis module 200 for analyzing the corn flaring period is:

[0152] y=0.5x1+0.3x2+0.3x3;

[0153] Among them, y is the set corn yield, x1 is the soil available nitrogen content, x2 is the available phosphorus content, and x3 is the available potassium content;

[0154] The soil nutrient content of the corn during the flaring period detected by the crop growth period database module 100 is:

[0155] The content of available nitrogen is: x1 = 12 mg / kg;

[0156] The available phosphorus content is: x2 = 8 mg / kg;

[0157] The content of fast-acting potassium is: x2 = 6 mg / kg;

[0158] The predicted value of corn yield is: y = 100;

[0159] Calculate the amount of available nitrogen required for corn soil during the flaring period:

[0160]

[0161] Calculating Available Nitrogen Difference

[0162] x i调 =191.6-12=179.6;

[0163] This means that the soil's available nitrogen content is far lower than what is required for corn growth during the flaring period, and appropriate nitrogen fertilizer needs to be supplemented. The same method can be used to calculate the difference in other nutrients such as available phosphorus, thereby comprehensively guiding fertilization operations.

[0164] The steps for optimizing the linear regression model in S4 are as follows:

[0165] Receive the actual yield Y of m farmlands after precise fertilization 实际 , the predicted crop yield y set in S3 i设 ;

[0166] Calculate the mean square error between the actual output and the predicted output

[0167] Among them, Y 实际 -Y i设 is the error value between the predicted yield and the actual yield of each farmland, and MSE is the average difference between the predicted yield and the actual yield of all farmlands. The smaller the MSE value, the higher the accuracy of the established linear regression model.

[0168] Definition of loss function: Take mean square error MSE as loss function L(β), where β = (β1, β2, …, β i ) T is the regression coefficient vector;

[0169] Calculate the loss function with respect to each regression coefficient β j , β j =β1-β i The partial derivative of

[0170] For example, for a linear regression model Among them, x1, x2, …, x i is the soil nutrient content, The partial derivatives of other coefficients can be calculated similarly;

[0171] Assume that the learning rate η is a positive number used to control the step size of each update. According to the gradient descent algorithm, the regression coefficient is updated

[0172] By continuously iterating this process, the loss function MSE gradually decreases until it is impossible to increase crop yields by providing soil nutrients.

[0173] The steps for setting the learning rate are as follows:

[0174] Set the initial learning rate η = 0.001, and adjust the learning rate by Δη = 0.001;

[0175] receiving an adjusted performance indicator MSE;

[0176] If MSE new <MSE init , the linear regression model in the feature analysis module 200 can converge faster, continue to increase the learning rate and repeat the above process;

[0177] If MSE new Change MSE init If no change occurs, it means that the learning rate is too large and it is not possible to increase crop yields by providing soil nutrients. At this time, the linear regression model is no longer adjusted.

[0178] Step S1 obtains the electrical signal of the absorption spectrum characteristics of the nutrient substance, the nutrient consumption rate, meteorological data, historical yield data, and soil nutrients corresponding to the growth stage, and the data provides information support for building a linear regression model in step S2;

[0179] For example, the electrical signals of the absorption spectrum characteristics of nutrient substances help the model to more accurately understand the nutrient absorption mechanism of crops under different spectral conditions, while the data on the mean range of nutrient consumption rates at different growth stages enable the model to better analyze the changing patterns of nutrient demand at each stage. Compared with a single data source or simple data collection method, the collaborative collection of multi-source data can build a model that better fits the actual growth situation of crops.

[0180] The device of the big data analysis and fertilizer optimization method during the growth period of crops comprises a fertilization structure 100, and the fertilization structure 100 comprises a fertilization aircraft 101. A fertilizer storage box 102 for storing mixed fertilizers is installed in the middle of the fertilization aircraft 101, and two sealing doors 103 are installed on the top of the fertilizer storage box 102. A feeder 104 is installed on the side of the fertilization aircraft 101 close to the fertilizer storage box 102, and the feeding end of the feeder 104 is through-connected with the fertilizer storage box 102, and the discharging end of the feeder 104 is through-connected with a connecting pipe 105, and the outer wall of the connecting pipe 105 is through-connected with three feeding nozzles 106.

[0181] When precise fertilization is required, the staff first injects the mixed fertilizer into the fertilizer storage box 102, and then seals the fertilizer storage box 102 through two sealing doors 103. The fertilizing aircraft 101 flies above the crops that need to be fertilized, and then the feeder 104 is started to extract the fertilizer in the fertilizer storage box 102, and then the fertilizer is sprayed on the crops through the connecting pipe 105 and the feeder nozzle 106.

[0182] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. Big data analysis and fertilizer optimization method for crop growth period, characterized by: The method comprises the following steps: S1. Obtain the nutrient absorption spectrum characteristic electrical signals, the mean value interval of nutrient consumption rate at different growth stages and meteorological data through a web crawler, as well as receive the historical crop yield data, soil nutrients and meteorological data corresponding to the crop growth stage in the planting records, and classify and store them into the crop growth period database according to the historical crop growth stage; S2. Use machine learning algorithms to establish a linear regression model corresponding to crop yield, growth stage, meteorological data, and soil nutrient content; S3. Set the predicted yield of crops, use soil nutrient sensors to detect the current soil nutrient content, determine the current growth stage of crops, calculate the soil nutrient content required for different growth stages of crops based on the linear regression model, and accurately fertilize crops; S4. Receive the current crop yield after precise fertilization, set the learning rate, and optimize the linear regression model according to the error value between the current crop yield and the predicted yield.

2. The method for crop growth period big data analysis and fertilizer optimization according to claim 1, characterized in that: The crop growth period database simulates a browser through a web crawler, and sends HTTP requests to the academic database and the meteorological data server respectively, requesting to obtain the electrical signals of the absorption spectrum characteristics of the nutrient substances in the soil, the mean interval of the nutrient consumption rate in different growth stages, and the meteorological signals; After receiving the request, the corresponding server returns the corresponding web page data to the web crawler according to the requested URL, so that the web crawler obtains the absorption spectrum characteristic electrical signals of various nutrients in the soil, the mean range of nutrient consumption rate at different growth stages and meteorological signals; Receive the corresponding historical crop yield data, soil nutrient content at different historical growth stages and historical meteorological data in the planting records, and classify the soil data according to different growth stages of crops and store them in the crop growth period database.

3. The method for crop growth period big data analysis and fertilizer optimization according to claim 2, characterized in that: The steps of using the machine learning algorithm to establish a linear regression model in S2 are as follows: Receive historical crop yield data stored in the crop growth period database, soil nutrient content and meteorological data corresponding to different crop growth stages; Due to the characteristic basis of crop growth stage requirements, the interrelationship of physiological processes and the synergistic mechanism of environmental factors, there is a linear relationship between crop yield data and soil nutrient content and meteorological data corresponding to different crop growth stages. Therefore, the relationship between them is expressed by the following linear regression model: y1=β1x1+β2x2+…+β i x i ; y2=β1x1+β2x2+…+β i x i ; …； y i =β1x1+β2x2+…+β i x i ; Among them, β1-β i is the regression coefficient, which means x1-x i The impact on production, x1-x i is y1-y i Soil nutrient content and meteorological data corresponding to the growth stage, such as nitrogen content x1, phosphorus content x2, average temperature x3, precipitation x4, potassium content x i , y1-y i It is the historical yield corresponding to different growth stages of crops.

4. The method for crop growth period big data analysis and fertilizer optimization according to claim 3, characterized in that: Calculate different regression coefficients in the linear regression model using the least squares method: Among them, β i is the regression coefficient, n is the number of samples of multiple crop yields, soil nutrient content at different growth stages, and meteorological data, x i is the soil nutrient content and meteorological data of the i-th sample, y i is the crop yield of the i-th sample; is the average value of multiple soil nutrient content sample data: is the average value of multiple crop yield sample data The relationship between crop yield and soil nutrient content and meteorological data was determined by calculating the regression coefficient.

5. The method for crop growth period big data analysis and fertilizer optimization according to claim 1, characterized in that: The steps of using the soil nutrient sensor to detect the current soil nutrient content in S3 are as follows: The output end of the soil nutrient sensor emits light with a specific wavelength range to the soil around the crops. When the light with a specific wavelength range irradiates the soil, the nutrients in the soil will absorb part of the light energy, causing the light intensity through the soil to change; The receiving end of the soil nutrient sensor receives the light that passes through the soil and converts the light transmission signal into an electrical signal; Receive the absorption spectrum characteristic electrical signals of various nutrients in the soil obtained by the web crawler; The electrical signals are acquired and converted to obtain the content of various nutrients in the soil; The calculation formula is as follows: Among them C meas is the current content of nutrients in the soil, I meas The electrical signal converted by the soil nutrient sensor, I std is the electrical signal obtained by web crawler, C std For I std The electrical signal corresponds to the known nutrient content.

6. The method for crop growth period big data analysis and fertilizer optimization according to claim 5, characterized in that: The steps of determining the current crop growth stage in S3 are as follows: Receive the soil nutrient content difference between two adjacent time points of the current crop, and calculate the nutrient consumption rate of the crop according to the change of the nutrient difference; Among them, v is the current rate of nutrient consumption of crops, C1-C2 is the difference in soil nutrient content between two adjacent time points, and v=Δt is the difference between adjacent time points; Receive the mean range of nutrient consumption rates of crops at different growth stages obtained in S1: Seedling stage mean interval: [X A ,X AI ]; Jointing stage mean range: [X AI ,X B ]; …； Maturity mean range: [X Z ,X ZI ] Receive the current nutrient consumption rate of crops: v; If X A <v<X AI , it means that the current crops are in the seedling stage. If X A <v<X AI , X AI <v<X B This indicates that the current crops are in the jointing stage.

7. The method for crop growth period big data analysis and fertilizer optimization according to claim 6, characterized in that: The steps of calculating the soil nutrient content required for different growth stages of the current crops according to the linear regression model in S3 are as follows: Receive current crop growth stage, soil nutrient content and crop yield forecast; Call out the linear regression model in S2 that matches the current growth stage; y i预 =β1x 1需 +β2x 2需 +…+b i x i需 ; Calculate the nutrients required for the current growth stage through the linear regression model; Among them, y i预 To predict crop yield, β i is the correlation coefficient corresponding to the i-th crop yield, x 1需 Current soil nutrient content and meteorological data; Calculate the difference between the current soil nutrients of crops and the required nutrients, and output control signals to accurately fertilize crops; x i调 =x i需 -x i实 ; Among them, x i调 To adjust the parameters of soil nutrients, x i需 is the soil nutrients required by current crops, x i实 The actual soil nutrients of the current crops detected.

8. The method for crop growth period big data analysis and fertilizer optimization according to claim 1, characterized in that: The steps of optimizing the linear regression model in S4 are as follows: Receive the actual yield Y of m farmlands after precise fertilization 实际 , the predicted crop yield y set in S3 i设 ; Calculate the mean square error between the actual output and the predicted output Among them, Y 实际 -Y i设 is the error value between the predicted yield and the actual yield of each farmland, and MSE is the average difference between the predicted yield and the actual yield of all farmlands. The smaller the MSE value, the higher the accuracy of the established linear regression model. Definition of loss function: Take mean square error MSE as loss function L(β), where β = (β1, β2, …, β i ) T is the regression coefficient vector; Calculate the loss function with respect to each regression coefficient β j , β j =β1-β i The partial derivative of For the linear regression model where x1, x2, ..., x i is the soil nutrient content, The partial derivatives of other coefficients are calculated similarly; Assume that the learning rate η is a positive number used to control the step size of each update. According to the gradient descent algorithm, the regression coefficient is updated 9. The method for crop growth period big data analysis and fertilizer optimization according to claim 8, characterized in that: The steps of setting the learning rate are as follows: Set the initial learning rate η = 0.001, and adjust the learning rate by Δη = 0.001; receiving an adjusted performance indicator MSE; If MSE new <MSE init , the linear regression model in the feature analysis module (200) can converge faster, continue to increase the learning rate and repeat the above process; If MSE new Change MSE init If no change occurs, it means that the learning rate is too large and it is not possible to increase crop yields by providing soil nutrients. In this case, the linear regression model is no longer adjusted.

10. A method for big data analysis and fertilizer optimization during the growth period of crops, applied to the big data analysis and fertilizer optimization device for the growth period of crops as claimed in any one of claims 1 to 9, characterized in that: The invention comprises a fertilizing structure (100), wherein the fertilizing structure (100) comprises a fertilizing aircraft (101), a fertilizer storage box (102) for storing mixed fertilizer is installed in the middle of the fertilizing aircraft (101), two sealing doors (103) are installed on the top of the fertilizer storage box (102), a feeder (104) is installed on one side of the fertilizing aircraft (101) close to the fertilizer storage box (102), a feeding end of the feeder (104) is through-connected with the fertilizer storage box (102), a discharging end of the feeder (104) is through-connected with a connecting pipe (105), and the outer wall of the connecting pipe (105) is through-connected with three feeding nozzles (106).

Images