A water-fertilizer integrated drip irrigation control system

By combining the particle swarm algorithm in the integrated water-fertilizer droplet irrigation control system, the water-fertilizer dosage is optimized, and the problem of insufficient accumulation of medicinal ingredients in the existing technology is solved, and the economic value and healthy growth of medicinal crops are improved.

CN118822775BActive Publication Date: 2024-11-29ZHEJIANG YUNZHOU BIG DATA TECH CO LTD
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202411296366.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-18
Publication Date
2024-11-29
Estimated Expiration
2044-09-18

AI Technical Summary

Technical Problem

The existing water and fertilizer drop sprinkler irrigation technology has shortcomings in enhancing the medicinal value of medicinal crops, and has failed to effectively stimulate the accumulation of medicinal ingredients through the stimulation of moisture stress and fertilizer components combined with environmental factors.

Method used

A water and fertilizer integrated drop sprinkler irrigation control system was designed. Through the data acquisition module, graph construction module, graph update module, clustering module and planning module, combined with particle swarm algorithm, the relationship between water and fertilizer and medicinal components is captured, and nutritional elements are optimized and The dosage of water ensures maximum accumulation of medicinal ingredients.

Benefits of technology

Through optimized water and fertilizer dosage, the accumulation of medicinal ingredients is maximized, the economic value of medicinal crops is improved, and the healthy growth of medicinal crops is ensured.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118822775B_ABST
    Figure CN118822775B_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of water and fertilizer regulation, and discloses a water and fertilizer integrated drip irrigation regulation system, comprising: a data acquisition module, used for collecting environment, soil, water and fertilizer data and plant images of M mu experimental land at t times with the same interval within a first preset time period T; a graph construction module, used for mapping the environment, soil, water and fertilizer and plants of each mu of experimental land into nodes; a graph update module, used for updating the feature vector of each node according to the collected pth group of data at the pth time, and re-determining the weight of the edge based on the updated plant node feature; a clustering module, used for clustering the environment, soil and water and fertilizer nodes corresponding to each plant node at t times; and a planning module, used for determining the water and fertilizer allocation at the target time according to the environment, soil data and plant images of the target area in combination with a preset prediction model, and planning a drip irrigation scheme.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of water and fertilizer regulation, and more specifically, to a water and fertilizer integrated drip irrigation regulation system. Background Art

[0002] At present, the cultivation and management of medicinal crops mainly rely on the integrated water and fertilizer drip irrigation technology. This technology combines irrigation and fertilization to deliver water and fertilizer to the root system of crops in a precise manner, ensuring that the crops receive balanced water and nutrients during their growth.

[0003] However, although the existing water and fertilizer drip irrigation technology plays an important role in the basic growth management of medicinal crops, it is insufficient in improving the medicinal value of medicinal crops. Some existing technologies only focus on providing a balanced supply of water and fertilizer, but ignore the comprehensive regulation of water stress, fertilizer component stimulation and environmental factors to stimulate the accumulation of medicinal components of medicinal crops. Summary of the invention

[0004] The invention provides a water-fertilizer integrated drip irrigation control system to solve the technical problems raised in the background technology.

[0005] The present invention provides a water-fertilizer integrated drip irrigation control system, comprising:

[0006] The data collection module (101) is used to collect environment, soil, water and fertilizer data and plant images of an M mu experimental field at t equal intervals within a first preset time period T; wherein the cultivation strategy for the M mu experimental field includes a first strategy and a second strategy;

[0007] The first strategy is to determine the dosage of each nutrient element by the first random parameter based on the standard cultivation scheme during the growth and maturity period of M acres of plants;

[0008] The second strategy is to determine the water dosage by the second random parameter based on the standard cultivation scheme during the growth and maturity period of M acres of plants;

[0009] The first random parameter and the second random parameter linearly expand their value ranges as the planting time of the plants increases; until the number of dead plants in the M mu of plants reaches the death threshold, the value ranges of the first random parameter and the second random parameter are restored to the initial values ​​and recorded as the fth moment; the value ranges of the first random parameter and the second random parameter at the futh moment are obtained and set as the first death range and the second death range;

[0010] The graph construction module (102) is used to map the environment, soil, water, fertilizer and plants of each acre of experimental land into nodes, and construct the initial feature vectors of the environment, soil and water and fertilizer nodes based on the data collected at the first moment, specifically including:

[0011] The average height and average number of leaves per acre of plants are obtained by identifying plant images;

[0012] The health score of plants per acre was obtained based on the average height and average number of leaves per acre. The calculation formula is as follows:

[0013] ;in, represents the average number of leaves of the x-th acre of plants at the u-th moment, represents the average height of the x-th acre of plants at the u-th moment, represents the standard number of leaves of the plant at the uth moment, represents the standard height of the plant at the uth moment, and are all exponential parameters, is the proportionality coefficient;

[0014] By analyzing the green band, near infrared band and blue light band of the plant image, the green band reflectance, near infrared band reflectance and blue light band reflectance of each acre of plant were obtained respectively;

[0015] The concentration of medicinal ingredients per acre of plants is obtained based on the green band reflectance, near infrared band reflectance and blue light band reflectance of plants per acre. The concentration of medicinal ingredients indicates the concentration of medicinal ingredients of medicinal plants. The calculation formula is as follows:

[0016] ;in, It represents the green band reflectance of the leaves of the x-th acre plant at the u-th moment, represents the standard value of the green band reflectance of the leaves at the uth moment, represents the near-infrared reflectivity of the leaves of the x-mu plant at the u-th moment, represents the standard value of the near-infrared reflectivity of the leaf at the uth moment, represents the reflectance of the blue light band of the leaves of the x-th acre plant at the u-th moment, represents the standard value of the blue light band reflectivity of the blade at the u-th moment, represents the sine function, and All represent proportionality coefficients;

[0017] The average time for plant nodes to absorb water and fertilizer is determined by fluorescent dye tracing method as u moments, and the initial feature vector of the plant node is constructed based on the health score and medicinal ingredient concentration obtained by judging the plant image at the uth moment; and the edge is constructed according to the entity relationship, including:

[0018] Build edges based on entity relationships, extract entity relationships among environment, soil, water and fertilizer, and plants, and build edges between environment, soil, and water and fertilizer nodes and their corresponding plant nodes based on entity relationships; build edges between plant nodes based on the similarity of their feature vectors;

[0019] A graph updating module (103) is used to update the feature vector of each node according to the collected p+1th group of data at the pth time;

[0020] The clustering module (104) is used to cluster the environment, soil and water and fertilizer nodes corresponding to each plant node at time t to obtain a clustering vector, which specifically includes:

[0021] Calculate the comprehensive characteristic relationship between the environment, soil and water and fertilizer nodes at the i-th moment; Comprehensive characteristic relationship The calculation formula is as follows:

[0022] ;in, , and Represent the characteristic vectors of water, fertilizer, soil and environment respectively, , and Represent the weight vectors of the characteristic vectors of water, fertilizer, soil and environment, respectively, represents the bias parameter of the comprehensive feature relationship, represents the comprehensive feature relationship vector, express Activation function, Represents element-wise multiplication;

[0023] Calculate the characteristic relationship between water and fertilizer and soil, soil and environment, and environment and water and fertilizer at the i-th moment;

[0024] ;in, represents the characteristic relationship between water, fertilizer and soil at the i-th moment, The bias parameter representing the characteristic relationship between water, fertilizer and soil, Represents the characteristic relationship vector between water, fertilizer and soil, Represents a splicing operation;

[0025] ;in, represents the characteristic relationship between soil and environment at the i-th moment, Bias parameters representing the characteristic relationship between soil and environment, Represents the characteristic relationship vector between soil and environment;

[0026] ;in, represents the characteristic relationship between the environment and water and fertilizer at the i-th moment, The bias parameter representing the characteristic relationship between the environment and water and fertilizer, Represents the characteristic relationship vector between environment and water and fertilizer;

[0027] According to the characteristic relationship between water fertilizer and soil, soil and environment, environment and water fertilizer and the comprehensive characteristic relationship between environment, soil and water fertilizer nodes at the i-th moment, the clustering vector is determined. The calculation formula is as follows:

[0028] ;in, Represents the clustering vector at the i-th moment;

[0029] Collect the clustering vector of the ath plant node and the feature vector of the ath plant node to generate a training sample set of a preset prediction model; collect the similarity of the clustering vectors between the ath and bth plant nodes and the similarity of the feature vectors between the ath and bth nodes to generate a training sample set of a preset verification model;

[0030] The planning module (105) is used to determine the water and fertilizer allocation at the target time according to the environment, soil data and plant images of the target area in combination with a preset prediction model, and plan a drip irrigation plan, specifically including:

[0031] Randomly generate an initial particle swarm with R individuals that meet the constraints, and initialize the current iteration number Time to 1;

[0032] The particle group is encoded, and each element of the code is expressed as the amount of water and various nutrient elements;

[0033] The constraint condition is that the allocation of each nutrient element in the individual solution is within the first death range, and the allocation of water is within the second death range;

[0034] The prediction model is set as a fitness function, the fitness function value represents the concentration of the medicinal ingredient, and the fitness values ​​of X individuals are obtained;

[0035] Setting the update factor ; ;in, represents the initial attack frequency, used for exploration; Indicates the final attack frequency, used for convergence; represents the attack weight; initial cruising breadth, for exploration; Indicates the final cruise width, used for convergence; represents the cruise weight; Time represents the current number of iterations; Tmax represents the maximum number of iterations;

[0036] Update individuals based on update factors;

[0037] Determine Time≥Tmax; if Time≥Tmax, determine the water and fertilizer dosage for the target time, otherwise the cycle updates the individuals based on the update factor.

[0038] Furthermore, at the pth time, the feature vector of each node is updated according to the collected p+1th group of data, including:

[0039] According to the newly collected data of environment, soil, water and fertilizer at the p+1th moment, the feature vectors constructed with the newly collected data are used to replace the feature vectors of the environment, soil and water and fertilizer nodes at the pth moment.

[0040] Further, generating a training sample set of a preset prediction model and generating a training sample set of a preset verification model include:

[0041] The training of the preset prediction model includes: pairing the clustering vector of the a-th plant node at the i-th moment with the feature vector of the a-th plant node at the i+u-th moment as a set of training samples, repeating the process to obtain a prediction model training sample set; training the prediction model using the prediction model training sample set;

[0042] The training of the preset verification model includes: pairing the similarity of the clustering vectors of the a-th plant node and the b-th plant node at the i-th moment with the similarity of the feature vectors of the i-th plant node and the j-th plant node at the i+u-th moment, and if the similarity of the clustering vectors of the a-th plant node and the b-th plant node at the i-th moment is greater than a first similarity threshold, taking them as a group of training samples, repeating the process to obtain a verification model training sample set; using the verification model training sample set to train the verification model;

[0043] During the training process, the mean square error is used as the loss function for the preset prediction model and the preset verification model to optimize the prediction model;

[0044] A second preset time period G is set; the environment, soil, water and fertilizer data and plant images of the P mu experimental field are collected at Q equally spaced moments; based on the environment, soil and water and fertilizer data collected at the i-th moment in the second preset time period G, the health score and medicinal ingredient concentration of the plant at the i+u-th moment are predicted using the prediction model; the verification model and the prediction model are used to verify each other.

[0045] Furthermore, the validation model and the prediction model are used to verify each other, including:

[0046] Based on the clustering vector of plant node a and the clustering vector of plant node b at the i-th moment in the second preset time period G, the prediction feature vectors of node a and node b at the i+u-th moment are obtained respectively through the prediction model;

[0047] Based on the clustering vector of plant node a and the clustering vector of plant node b at the i-th moment in the second preset time period G, it is determined that the similarity between the clustering vector of plant node a and the clustering vector of plant node b is greater than a first similarity threshold;

[0048] If the similarity between the clustering vector of plant node a and the clustering vector of plant node b is greater than the first similarity threshold, the verification values ​​of node a and node b at the i+uth time are obtained respectively through the verification model;

[0049] Determine whether the difference between the similarity of the predicted feature vectors of node a and node b at the i+uth moment and the verification value is greater than a second similarity threshold;

[0050] If the difference between the similarity of the predicted feature vectors of node i and node j at the i+uth moment and the verification value is greater than the second similarity threshold, the prediction model training is completed;

[0051] If the difference between the similarity of the predicted feature vectors of node i and node j at the i+uth moment and the verification value is less than the second similarity threshold, the prediction model and the verification model continue to be trained.

[0052] Furthermore, the individual is updated based on the update factor, including:

[0053] Select the individuals whose health scores are lower than the health threshold among the R individuals as the parent generation update;

[0054] Calculate the attack vector of the i-th individual , the formula is: ;in, represents the current global optimal individual, represents the i-th individual;

[0055] Set the hyperplane of the i-th individual, which is a vector perpendicular to the attack vector of the i-th individual ;

[0056] Calculate the cruise vector of the i-th individual , the formula is: ;in, represents the projection of the attack vector on the hyperplane;

[0057] The update formula of the individual is as follows:

[0058] ;

[0059] ;

[0060] in, represents the update vector of the ith individual, represents the updated value of the i-th individual, and Represents a random number, represents the modulus length of the attack vector, Indicates the modulus of the cruise vector.

[0061] Furthermore, based on the environment, soil data and plant images of the target area, combined with the preset model, the water and fertilizer dosage at the target time is determined, and the drip irrigation plan is planned, including:

[0062] Collect the number of droppers and nozzles and the maximum delivery volume per unit time of the droppers and nozzles;

[0063] According to the duration of the target time and the water and fertilizer allocation at the target time, the actual delivery volume per unit time is determined;

[0064] According to the actual delivery volume per unit time, the drip irrigation scheme is obtained based on the drip irrigation priority strategy;

[0065] Among them, the drip irrigation priority strategy means that drip irrigation is used first to meet the water and fertilizer allocation requirements at the target time; when the drip irrigation delivery per unit time cannot fully meet the water and fertilizer allocation requirements at the target time, the sprinkler irrigation system is used to supplement it.

[0066] The beneficial effects of the present invention are as follows: by capturing the relationship between water, fertilizer and medicinal ingredients during the growth process of medicinal plants, combining with the improved particle swarm algorithm to obtain the water and fertilizer dosage that maximizes the medicinal ingredients, reasonable stimulation is given to the medicinal crops while ensuring the survival of the medicinal crops, thereby allowing the medicinal crops to accumulate more medicinal ingredients and improving the economic value of the medicinal crops. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a module diagram of a water-fertilizer integrated drip irrigation control system of the present invention.

[0068] In the figure: 101, data collection module; 102, graph construction module; 103, graph update module; 104, clustering module; 105, planning module. DETAILED DESCRIPTION

[0069] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0070] like Figure 1 As shown, a water-fertilizer integrated drip irrigation control system comprises:

[0071] The data collection module (101) is used to collect environment, soil, water and fertilizer data and plant images of an M mu experimental field at t equal intervals within a first preset time period T; wherein the cultivation strategy for the M mu experimental field includes a first strategy and a second strategy;

[0072] The first strategy is to determine the dosage of each nutrient element by the first random parameter based on the standard cultivation scheme during the growth and maturity period of M acres of plants;

[0073] The second strategy is to determine the water dosage by the second random parameter based on the standard cultivation scheme during the growth and maturity period of M acres of plants;

[0074] The first random parameter and the second random parameter linearly expand their value ranges as the planting time of the plants increases; until the number of dead plants in the M mu of plants reaches the death threshold, the value ranges of the first random parameter and the second random parameter are restored to the initial values ​​and recorded as the fth moment; the value ranges of the first random parameter and the second random parameter at the futh moment are obtained and set as the first death range and the second death range;

[0075] The graph construction module (102) is used to map the environment, soil, water, fertilizer and plants of each acre of experimental land into nodes, and construct the initial feature vectors of the environment, soil and water and fertilizer nodes based on the data collected at the first moment, specifically including:

[0076] The average height and average number of leaves per acre of plants are obtained by identifying plant images;

[0077] The health score of plants per acre was obtained based on the average height and average number of leaves per acre. The calculation formula is as follows:

[0078] ;in, represents the average number of leaves of the x-th acre of plants at the u-th moment, represents the average height of the x-th acre of plants at the u-th moment, represents the standard number of leaves of the plant at the uth moment, represents the standard height of the plant at the uth moment, and are all exponential parameters, is the proportionality coefficient;

[0079] By analyzing the green band, near infrared band and blue light band of the plant image, the green band reflectance, near infrared band reflectance and blue light band reflectance of each acre of plant were obtained respectively;

[0080] The concentration of medicinal ingredients per acre of plants is obtained based on the green band reflectance, near infrared band reflectance and blue light band reflectance of plants per acre. The concentration of medicinal ingredients indicates the concentration of medicinal ingredients of medicinal plants. The calculation formula is as follows:

[0081] ;in, It represents the green band reflectance of the leaves of the x-th acre plant at the u-th moment, represents the standard value of the green band reflectance of the leaves at the uth moment, represents the near-infrared reflectivity of the leaves of the x-mu plant at the u-th moment, represents the standard value of the near-infrared reflectivity of the leaf at the uth moment, represents the reflectance of the blue light band of the leaves of the x-th acre plant at the u-th moment, represents the standard value of the blue light band reflectivity of the blade at the u-th moment, represents the sine function, and All represent proportionality coefficients;

[0082] The average time for plant nodes to absorb water and fertilizer is determined by fluorescent dye tracing method as u moments, and the initial feature vector of the plant node is constructed based on the health score and medicinal ingredient concentration obtained by judging the plant image at the uth moment; and the edge is constructed according to the entity relationship, including:

[0083] Build edges based on entity relationships, extract entity relationships among environment, soil, water and fertilizer, and plants, and build edges between environment, soil, and water and fertilizer nodes and their corresponding plant nodes based on entity relationships; build edges between plant nodes based on the similarity of their feature vectors;

[0084] A graph updating module (103) is used to update the feature vector of each node according to the collected p+1th group of data at the pth time;

[0085] The clustering module (104) is used to cluster the environment, soil and water and fertilizer nodes corresponding to each plant node at time t to obtain a clustering vector, which specifically includes:

[0086] Calculate the comprehensive characteristic relationship between the environment, soil and water and fertilizer nodes at the i-th moment; Comprehensive characteristic relationship The calculation formula is as follows:

[0087] ;in, , and Represent the characteristic vectors of water, fertilizer, soil and environment respectively, , and Represent the weight vectors of the characteristic vectors of water, fertilizer, soil and environment, respectively, represents the bias parameter of the comprehensive feature relationship, represents the comprehensive feature relationship vector, express Activation function, Represents element-wise multiplication;

[0088] Calculate the characteristic relationship between water and fertilizer and soil, soil and environment, and environment and water and fertilizer at the i-th moment;

[0089] ;in, represents the characteristic relationship between water, fertilizer and soil at the i-th moment, The bias parameter representing the characteristic relationship between water, fertilizer and soil, Represents the characteristic relationship vector between water, fertilizer and soil, Represents a splicing operation;

[0090] ;in, represents the characteristic relationship between soil and environment at the i-th moment, Bias parameters representing the characteristic relationship between soil and environment, Represents the characteristic relationship vector between soil and environment;

[0091] ;in, represents the characteristic relationship between the environment and water and fertilizer at the i-th moment, The bias parameter representing the characteristic relationship between the environment and water and fertilizer, Represents the characteristic relationship vector between environment and water and fertilizer;

[0092] According to the characteristic relationship between water fertilizer and soil, soil and environment, environment and water fertilizer and the comprehensive characteristic relationship between environment, soil and water fertilizer nodes at the i-th moment, the clustering vector is determined. The calculation formula is as follows:

[0093] ;in, Represents the clustering vector at the i-th moment;

[0094] Collect the clustering vector of the ath plant node and the feature vector of the ath plant node to generate a training sample set of a preset prediction model; collect the similarity of the clustering vectors between the ath and bth plant nodes and the similarity of the feature vectors between the ath and bth nodes to generate a training sample set of a preset verification model;

[0095] The planning module (105) is used to determine the water and fertilizer allocation at the target time according to the environment, soil data and plant images of the target area in combination with a preset prediction model, and plan a drip irrigation plan, specifically including:

[0096] Randomly generate an initial particle swarm with R individuals that meet the constraints, and initialize the current iteration number Time to 1;

[0097] The particle group is encoded, and each element of the code is expressed as the amount of water and various nutrient elements;

[0098] The constraint condition is that the allocation of each nutrient element in the individual solution is within the first death range, and the allocation of water is within the second death range;

[0099] The prediction model is set as a fitness function, the fitness function value represents the concentration of the medicinal ingredient, and the fitness values ​​of X individuals are obtained;

[0100] Setting the update factor ; ;in, represents the initial attack frequency, used for exploration; Indicates the final attack frequency, used for convergence; represents the attack weight; initial cruising breadth, for exploration; Indicates the final cruise width, used for convergence; represents the cruise weight; Time represents the current number of iterations; Tmax represents the maximum number of iterations;

[0101] Update individuals based on update factors;

[0102] Determine Time≥Tmax; if Time≥Tmax, determine the water and fertilizer dosage for the target time, otherwise the cycle updates the individuals based on the update factor.

[0103] Environmental data include: ambient temperature, ambient humidity, ambient light intensity and ambient carbon dioxide concentration;

[0104] Soil data include: soil temperature, soil moisture, soil pH value and the content of nitrogen, phosphorus, potassium, calcium, magnesium and sulfur in the soil;

[0105] Water and fertilizer data include: water dosage and the dosage of nitrogen, phosphorus, potassium, calcium, magnesium and sulfur elements;

[0106] It should be noted that the average time for plant nodes to absorb water and fertilizer is determined by the fluorescent dye tracing method; for example, at the first moment in the first preset time period T, rhodamine B fluorescent dye is added to the water and fertilizer. The fluorescent dye is harmless to plants at low concentrations, and the distribution of the dye inside the plant is detected by a fluorescent microscope, so as to determine that the water and fertilizer are absorbed by the plant at the uth moment on average.

[0107] The standard cultivation plan refers to the standard water and fertilizer cultivation plan, which can be collected from the Internet or from planting companies or farmers;

[0108] The entire growth cycle of the plant is divided into four stages: seedling stage, growth stage, maturity stage and picking stage; water and fertilizer management is carried out according to different growth needs; among them, the seedling stage is the initial growth stage of the plant, and the main focus is on root development and seedling health; during the growth stage, the plant grows rapidly, leaves and branches begin to develop, and the water and fertilizer requirements increase; during the maturity stage, the plant growth gradually stabilizes, begins to accumulate nutrients and medicinal ingredients, and is ready to enter the picking stage; during the picking stage, the plant growth is completed and enters the harvesting stage;

[0109] Among them, the first strategy is to adjust the supply of nutrient elements. During the growth and maturity period of the plant, the first random parameter is added on the basis of the standard cultivation plan to adjust the dosage of each nutrient element. The supply of nutrient elements will be more diversified, and data on the impact of changes in fertilizer supply on plant growth and accumulation of medicinal ingredients can be obtained;

[0110] Among them, the second strategy is to adjust the water supply. During the growth and maturity period of the plant, the water dosage is adjusted by adding a second random parameter on the basis of the standard cultivation plan to test the effect of different water conditions on plant growth. By adjusting the water supply, data on the effect of changes in water supply on plant growth and accumulation of medicinal ingredients are obtained.

[0111] For example, the initial value range of the first random parameter is set to , the initial value range of the second random parameter is , the first random parameter and the second random parameter are expressed with three decimal places as the precision; in the standard cultivation scheme, the water dosage in the water-fertilizer dosage at the initial moment is 2, and the dosages of nitrogen, phosphorus and potassium are 0.4, 0.5 and 0.2 respectively; at the second moment, the random value of the first random parameter for nitrogen dosage is 0.961, the random value of the first random parameter for phosphorus dosage is 0.983, the random value of the first random parameter for potassium dosage is 1.004, and the random value of the second random parameter for water is 0.957; the water dosage in the water-fertilizer dosage at the second moment is 1.914, and the dosages of nitrogen, phosphorus and potassium are 0.384, 0.492 and 0.201 respectively;

[0112] As the planting time of the plant increases, the linear expansion range is set to 1.1, and the initial value range of the first random parameter at the second moment is , the initial value range of the second random parameter is ;

[0113] When the number of dead plants in M ​​mu of plants reaches a preset death threshold, mark the current time as the fth moment, record the parameters at the futh moment, and record the value ranges of the first random parameter and the second random parameter at the futh moment; the value ranges are defined as the first death range and the second death range, representing the range of nutrient elements and water supply that the plants can still survive without reaching the death threshold;

[0114] It should be noted that the standard cultivation scheme of water and fertilizer amounts for plants in the prior art is collected. The standard scheme provides a suitable environment for plant growth. On this basis, the water and fertilizer amounts are adjusted so that the experiment can be carried out within a reasonable range.

[0115] The average number of leaves reflects the growth vitality and photosynthesis efficiency of the plants. The average number of leaves is compared with the standard number of leaves. The standard number of leaves is the average number of leaves per acre based on the standard cultivation plan. The average height of the plants reflects the overall growth status and nutrient absorption capacity of the plants. The average height of the plants is compared with the standard height of the plants. The standard height of the plants is the average height of the plants per acre based on the standard cultivation plan. Thus, the health status of the plants is comprehensively evaluated. Among them, the ratio of the number of leaves to the height of the plants is used as a key parameter, which is further corrected by a nonlinear function, providing a quantitative basis for plant health assessment.

[0116] The reflectivity of the near-infrared band of each acre of plants is used to evaluate the water content and structural changes of the leaves. The water content and structural changes of the leaves directly affect the synthesis and accumulation of active ingredients in the plants. The reflectivity of the near-infrared band of each acre of plants is compared with the reflectivity of the standard near-infrared band of each acre of plants. The reflectivity of the standard near-infrared band is the reflectivity of the near-infrared band of each acre of plants based on the standard cultivation plan; the reflectivity of the green band of each acre of plants is used to evaluate the chlorophyll content of the leaves. The photosynthesis efficiency of chlorophyll directly affects the synthesis of plant secondary metabolites. The reflectivity of the green band of each acre of plants is compared with the reflectivity of the standard near-infrared band of each acre of plants. The standard green band reflectance of each acre of plants is compared with the standard green band reflectance, which is the green band reflectance of each acre of plants based on the standard cultivation plan; the blue light band reflectance of each acre of plants is used to evaluate the synthesis of plant growth hormones and secondary metabolites, and the blue light band reflectance of each acre of plants is compared with the blue light band reflectance of each acre of plants, which is the standard blue light band reflectance of each acre of plants based on the standard cultivation plan; the concentration of medicinal ingredients in plants is inferred by comparing the green band reflectance, near-infrared band reflectance and blue light band reflectance with standard values ​​and adjusting them nonlinearly.

[0117] During the target time period, an initialized particle swarm with R individuals that meet the constraints is randomly generated, and each particle swarm represents a possible water-fertilizer dosing scheme. Each element of the particle swarm represents the dosing of water and various nutrient elements, which must meet certain constraints. For example, the dosing of each nutrient element must be within the first death range, and the dosing of water must be within the second death range to ensure that the generated scheme will not cause death to the plant; then, the preset prediction model is used to calculate the fitness value of each individual, and the output of the fitness function represents the concentration of the medicinal ingredient. This fitness value is used to evaluate the pros and cons of each water-fertilizer dosing scheme; and They correspond to the attack frequency and cruise weight respectively, and control the balance between exploration and development in the optimization process. Through the current number of iterations and the maximum number of iterations, the weight of exploration (extensive search for new solutions) is gradually reduced, and the weight of development (local optimization solution) is increased. As time goes by, the attack frequency (used to explore new solutions) and cruise weight (used to develop existing solutions) will be dynamically adjusted, gradually shifting from extensive exploration of the solution space to fine optimization of the optimal solution. By continuously updating the solution of each individual, the particle swarm gradually approaches the global optimal water and fertilizer dosing scheme. Each round of iteration will recalculate the fitness value of the updated particle and determine whether the maximum number of iterations has been reached. If the maximum number of iterations has been reached, the iteration will be stopped and the water and fertilizer dosing for the target time will be determined; otherwise, the iteration will continue.

[0118] In one embodiment of the present invention, at the pth time, the feature vector of each node is updated according to the collected p+1th group of data, including:

[0119] According to the newly collected data of environment, soil, water and fertilizer at the p+1th moment, the feature vectors constructed with the newly collected data are used to replace the feature vectors of the environment, soil and water and fertilizer nodes at the pth moment.

[0120] The feature vector of each node is adjusted by gradual updating; specifically, when the system collects new environmental, soil, and water and fertilizer data at the k+1th time point, these new data are used to generate new feature vectors. These new feature vectors are then used to replace the feature vectors at the previous kth time point. The state of the node is constantly refreshed and updated with the latest data to more accurately reflect the current situation.

[0121] By calculating the characteristic relationships between water, fertilizer, soil and environmental nodes and performing clustering based on the characteristic relationships, the interaction between different elements in the plant growth environment can be captured and reflected. At different times, the overall health of the plant and the accumulation of medicinal ingredients are jointly determined by the interaction between water, fertilizer, soil and the environment. In the clustering process, different weights are first assigned to each characteristic relationship by calculating the ratio of each characteristic relationship to the overall comprehensive characteristic relationship, so that the clustering vector can more accurately reflect the actual situation of the current environment.

[0122] In one embodiment of the present invention, generating a training sample set of a preset prediction model and generating a training sample set of a preset verification model include:

[0123] The training of the preset prediction model includes: pairing the clustering vector of the a-th plant node at the i-th moment with the feature vector of the a-th plant node at the i+u-th moment as a set of training samples, repeating the process to obtain a prediction model training sample set; training the prediction model using the prediction model training sample set;

[0124] The training of the preset verification model includes: pairing the similarity of the clustering vectors of the a-th plant node and the b-th plant node at the i-th moment with the similarity of the feature vectors of the i-th plant node and the j-th plant node at the i+u-th moment, and if the similarity of the clustering vectors of the a-th plant node and the b-th plant node at the i-th moment is greater than a first similarity threshold, taking them as a group of training samples, repeating the process to obtain a verification model training sample set; using the verification model training sample set to train the verification model;

[0125] During the training process, the mean square error is used as the loss function for the preset prediction model and the preset verification model to optimize the prediction model;

[0126] The preset prediction model and the preset verification model are both neural network models; wherein the clustering vector of the a-th plant node at the i-th moment is used as the input of the prediction model, the feature vector of the a-th plant node at the i+u-th moment is used as the sample label of the prediction model, and the mean square error is used as the loss function to reversely optimize the weights and bias parameters of the hidden layer of the prediction model; wherein the ratio of the clustering vectors of the a-th plant node and the b-th plant node at the i-th moment is used as the input of the verification model (the feature similarity of the clustering vectors of the a-th plant node and the b-th plant node is greater than the first similarity threshold), the ratio of the feature vectors of the a-th plant node and the b-th plant node at the i+u-th moment is used as the sample label of the verification model, and the mean square error is used as the loss function to reversely optimize the weights and bias parameters of the hidden layer of the verification model;

[0127] A second preset time period G is set; the environment, soil, water and fertilizer data and plant images of the P mu experimental field are collected at Q equally spaced moments; based on the environment, soil and water and fertilizer data collected at the i-th moment in the second preset time period G, the health score and medicinal ingredient concentration of the plant at the i+u-th moment are predicted using the prediction model; the verification model and the prediction model are used to verify each other.

[0128] It should be noted that there is a first nonlinear mapping relationship between the clustering vectors of the environment, soil and water and fertilizer nodes at the i-th moment and the feature vectors of the corresponding plant nodes at the i+u-th moment. The first nonlinear mapping relationship can be captured by collecting the clustering features at each moment and the feature vectors of the corresponding plants to train the prediction model; wherein, the feature similarity of the feature vectors of plant node a and plant node b at the i-th moment is greater than the first similarity threshold, indicating that plant node a and plant node b are similar; similarly, the feature vector of plant node a at the i+u-th moment and the feature vector of plant node b at the i+u-th moment are also similar, and have a second nonlinear mapping relationship, which is captured by the verification model; the first nonlinear mapping relationship and the second nonlinear mapping relationship can be mutually verified; two models are used to capture the two nonlinear mapping relationships, which reduces the demand for data and accelerates the convergence speed of the model; in addition, diverse data are obtained for the plant cultivation strategy of the M-acre experimental field, and diverse data are conducive to the model learning nonlinear mapping relationships.

[0129] In one embodiment of the present invention, the verification model and the prediction model are used to verify each other, including:

[0130] Based on the clustering vector of plant node a and the clustering vector of plant node b at the i-th moment in the second preset time period G, the prediction feature vectors of node a and node b at the i+u-th moment are obtained respectively through the prediction model;

[0131] Based on the clustering vector of plant node a and the clustering vector of plant node b at the i-th moment in the second preset time period G, it is determined that the similarity between the clustering vector of plant node a and the clustering vector of plant node b is greater than a first similarity threshold;

[0132] If the similarity between the clustering vector of plant node a and the clustering vector of plant node b is greater than the first similarity threshold, the verification values ​​of node a and node b at the i+uth time are obtained respectively through the verification model;

[0133] Determine whether the difference between the similarity of the predicted feature vectors of node a and node b at the i+uth moment and the verification value is greater than a second similarity threshold;

[0134] If the difference between the similarity of the predicted feature vectors of node i and node j at the i+uth moment and the verification value is greater than the second similarity threshold, the prediction model training is completed;

[0135] If the difference between the similarity of the predicted feature vectors of node i and node j at the i+uth moment and the verification value is less than the second similarity threshold, the prediction model and the verification model continue to be trained.

[0136] For example, at the i-th moment, the clustering features of plant a and plant b are and , and The feature similarity between them is 0.95>the first similarity threshold 0.9. At the i+uth moment, the prediction feature vectors of plant a and plant b obtained by the prediction model are and , and The feature similarity is 0.84, and the verification value of plant a and plant b obtained by the verification model is 0.7; it is judged that the difference between the similarity of the predicted feature vector of node a and node b at the i+uth moment and the verification value is 0.14>the second similarity threshold 0.1, thereby judging that both the verification model and the prediction model have converged.

[0137] In another embodiment of the present invention, a time series model can also be deployed to evaluate the accuracy of the prediction model; for example, the feature vector of the ath plant node at n moments is collected, and the feature vector of the ath plant node at n+1 moments is predicted based on the feature vector of the plant node at n moments; when n+1=i+u, the outputs of the prediction model and the time series model are basically equal; if the difference between the outputs of the prediction model and the time series model is too large, it means that the prediction model has not converged or the time series model has not converged.

[0138] In one embodiment of the present invention, updating the individual based on the update factor includes:

[0139] Select the individuals whose health scores are lower than the health threshold among the R individuals as the parent generation update;

[0140] Calculate the attack vector of the i-th individual , the formula is: ;in, represents the current global optimal individual, represents the i-th individual;

[0141] Set the hyperplane of the i-th individual, which is a vector perpendicular to the attack vector of the i-th individual ;

[0142] Calculate the cruise vector of the i-th individual , the formula is: ;in, represents the projection of the attack vector on the hyperplane;

[0143] The update formula of the individual is as follows:

[0144] ;

[0145] ;

[0146] in, represents the update vector of the ith individual, represents the updated value of the i-th individual, and Represents a random number, represents the modulus length of the attack vector, Indicates the modulus of the cruise vector.

[0147] It should be noted that this update method introduces a combination of attack vectors and cruise vectors in the process of particle swarm optimization to ensure that the search of individuals in the solution space is not limited to local optimization, but can also fully explore the global solution; the attack vector guides the individual to move in the direction of the current global optimal solution to accelerate the convergence speed; the cruise vector maintains the diversity of individuals through projection on the hyperplane, avoids individuals from falling into the local optimum, and enhances the global search ability of the algorithm; and It is adjusted dynamically. As the number of iterations increases, it will gradually tend to more refined searches rather than broad searches. This allows the algorithm to conduct extensive exploration in the early stages and focus on development in the later stages, thereby accelerating convergence. The adjustment of dynamic weights ensures that the algorithm can quickly and effectively find the optimal solution in the search space, avoiding invalid search steps and improving the efficiency of the overall optimization.

[0148] In one embodiment of the present invention, based on the environment, soil data and plant images of the target area, combined with a preset model, the water and fertilizer dosage at the target time is determined, and a drip irrigation scheme is planned, including:

[0149] Collect the number of droppers and nozzles and the maximum delivery volume per unit time of the droppers and nozzles;

[0150] According to the duration of the target time and the water and fertilizer allocation at the target time, the actual delivery volume per unit time is determined;

[0151] According to the actual delivery volume per unit time, the drip irrigation scheme is obtained based on the drip irrigation priority strategy;

[0152] Among them, the drip irrigation priority strategy means that drip irrigation is used first to meet the water and fertilizer allocation requirements at the target time; when the drip irrigation delivery per unit time cannot fully meet the water and fertilizer allocation requirements at the target time, the sprinkler irrigation system is used to supplement it.

[0153] For example, there is a target area of ​​10 mu, in which the medicinal plant Astragalus is planted; based on the environment, soil data and plant images of the target area, combined with the preset growth model, the system calculates that the water and fertilizer demand of Astragalus in the next day is 5,000 liters; through the system's sensor network, the parameters of the drip irrigation and sprinkler irrigation systems deployed in the target area are collected; the drip irrigation system is equipped with 200 drip tubes, and the maximum delivery volume per unit time of each drip tube is 10 liters; the sprinkler irrigation system is equipped with 5 nozzles, and the maximum delivery volume per unit time of each nozzle is 100 liters; according to the target time The duration (10 hours) and the calculated water and fertilizer demand of 5000 liters are taken into account, and it is calculated that the amount of water and fertilizer that needs to be transported per unit time is 500 liters; based on the drip irrigation priority strategy, the maximum delivery volume per unit time of the drip irrigation system is 2000 liters, which is much higher than the actual delivery volume per unit time of 500 liters. Therefore, the drip irrigation system is used first to fully meet the water and fertilizer demand within the target time; since the drip irrigation system can fully meet the demand, there is no need to activate the sprinkler irrigation system in this plan, and each drip irrigation transports 2.5 liters of water and fertilizer per unit time.

[0154] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. A water-fertilizer integrated drip irrigation control system, characterized in that: include: The data collection module (101) is used to collect environment, soil, water and fertilizer data and plant images of an M mu experimental field at t equal intervals within a first preset time period T; wherein the cultivation strategy for the M mu experimental field includes a first strategy and a second strategy; The first strategy is to determine the dosage of each nutrient element by the first random parameter based on the standard cultivation scheme during the growth and maturity period of M acres of plants; The second strategy is to determine the water dosage by the second random parameter based on the standard cultivation scheme during the growth and maturity period of M acres of plants; The first random parameter and the second random parameter linearly expand their value ranges as the planting time of the plants increases; until the number of dead plants in the M mu of plants reaches the death threshold, the value ranges of the first random parameter and the second random parameter are restored to the initial values ​​and recorded as the fth moment; the value ranges of the first random parameter and the second random parameter at the futh moment are obtained and set as the first death range and the second death range; The graph construction module (102) is used to map the environment, soil, water, fertilizer and plants of each acre of experimental land into nodes, and construct the initial feature vectors of the environment, soil and water and fertilizer nodes based on the data collected at the first moment, specifically including: The average height and average number of leaves per acre of plants are obtained by identifying plant images; The health score of plants per acre was obtained based on the average height and average number of leaves per acre. The calculation formula is as follows: ;in, represents the average number of leaves of the x-th acre of plants at the u-th moment, represents the average height of the x-th acre of plants at the u-th moment, represents the standard number of leaves of the plant at the uth moment, represents the standard height of the plant at the uth moment, and are all exponential parameters, is the proportionality coefficient; By analyzing the green band, near infrared band and blue light band of the plant image, the green band reflectance, near infrared band reflectance and blue light band reflectance of each acre of plant were obtained respectively; The concentration of medicinal ingredients per acre of plants is obtained based on the green band reflectance, near infrared band reflectance and blue light band reflectance of plants per acre. The concentration of medicinal ingredients indicates the concentration of medicinal ingredients of medicinal plants. The calculation formula is as follows: ;in, It represents the green band reflectance of the leaves of the x-th acre plant at the u-th moment, represents the standard value of the green band reflectance of the leaves at the uth moment, represents the near-infrared reflectivity of the leaves of the x-mu plant at the u-th moment, represents the standard value of the near-infrared reflectivity of the leaf at the uth moment, represents the reflectance of the blue light band of the leaves of the x-th acre plant at the u-th moment, represents the standard value of the blue light band reflectivity of the blade at the u-th moment, represents the sine function, and All represent proportionality coefficients; The average time for plant nodes to absorb water and fertilizer is determined by fluorescent dye tracing method as u moments, and the initial feature vector of the plant node is constructed based on the health score and medicinal ingredient concentration obtained by judging the plant image at the uth moment; and the edge is constructed according to the entity relationship, including: Build edges based on entity relationships, extract entity relationships among environment, soil, water and fertilizer, and plants, and build edges between environment, soil, and water and fertilizer nodes and their corresponding plant nodes based on entity relationships; build edges between plant nodes based on the similarity of their feature vectors; A graph updating module (103) is used to update the feature vector of each node according to the collected p+1th group of data at the pth time; The clustering module (104) is used to cluster the environment, soil and water and fertilizer nodes corresponding to each plant node at time t to obtain a clustering vector, which specifically includes: Calculate the comprehensive characteristic relationship between the environment, soil and water and fertilizer nodes at the i-th moment; Comprehensive characteristic relationship The calculation formula is as follows: ;in, , and Represent the characteristic vectors of water, fertilizer, soil and environment respectively, , and Represent the weight vectors of the characteristic vectors of water, fertilizer, soil and environment, respectively, represents the bias parameter of the comprehensive feature relationship, represents the comprehensive feature relationship vector, express Activation function, Represents element-wise multiplication; Calculate the characteristic relationship between water and fertilizer and soil, soil and environment, and environment and water and fertilizer at the i-th moment; ;in, represents the characteristic relationship between water, fertilizer and soil at the i-th moment, The bias parameter representing the characteristic relationship between water, fertilizer and soil, Represents the characteristic relationship vector between water, fertilizer and soil, Represents a splicing operation; ;in, represents the characteristic relationship between soil and environment at the i-th moment, Bias parameters representing the characteristic relationship between soil and environment, Represents the characteristic relationship vector between soil and environment; ;in, represents the characteristic relationship between the environment and water and fertilizer at the i-th moment, The bias parameter representing the characteristic relationship between the environment and water and fertilizer, Represents the characteristic relationship vector between environment and water and fertilizer; According to the characteristic relationship between water fertilizer and soil, soil and environment, environment and water fertilizer and the comprehensive characteristic relationship between environment, soil and water fertilizer nodes at the i-th moment, the clustering vector is determined. The calculation formula is as follows: ;in, Represents the clustering vector at the i-th moment; Collect the clustering vector of the ath plant node and the feature vector of the ath plant node to generate a training sample set of a preset prediction model; collect the similarity of the clustering vectors between the ath and bth plant nodes and the similarity of the feature vectors between the ath and bth nodes to generate a training sample set of a preset verification model; The planning module (105) is used to determine the water and fertilizer allocation at the target time according to the environment, soil data and plant images of the target area in combination with a preset prediction model, and plan a drip irrigation plan, specifically including: Randomly generate an initial particle swarm with R individuals that meet the constraints, and initialize the current iteration number Time to 1; The particle group is encoded, and each element of the code is expressed as the amount of water and various nutrient elements; The constraint condition is that the allocation of each nutrient element in the individual solution is within the first death range, and the allocation of water is within the second death range; The prediction model is set as a fitness function, the fitness function value represents the concentration of the medicinal ingredient, and the fitness values ​​of X individuals are obtained; Setting the update factor ; ;in, represents the initial attack frequency, used for exploration; Indicates the final attack frequency, used for convergence; represents the attack weight; initial cruising breadth, for exploration; Indicates the final cruising width, used for convergence; represents the cruise weight; Time represents the current number of iterations; Tmax represents the maximum number of iterations; Update individuals based on update factors; Determine Time≥Tmax; if Time≥Tmax, determine the water and fertilizer dosage for the target time, otherwise the cycle updates the individuals based on the update factor.

2. The water-fertilizer integrated drip irrigation control system according to claim 1 is characterized in that: At the pth time, based on the collected p+1th group of data, the feature vector of each node is updated, including: According to the newly collected data of environment, soil, water and fertilizer at the p+1th moment, the feature vectors constructed with the newly collected data are used to replace the feature vectors of the environment, soil and water and fertilizer nodes at the pth moment.

3. The water-fertilizer integrated drip irrigation control system according to claim 1 is characterized in that: Generate a training sample set of a preset prediction model and generate a training sample set of a preset verification model, including: The training of the preset prediction model includes: pairing the clustering vector of the a-th plant node at the i-th moment with the feature vector of the a-th plant node at the i+u-th moment as a set of training samples, repeating the process to obtain a prediction model training sample set; training the prediction model using the prediction model training sample set; The training of the preset verification model includes: pairing the similarity of the clustering vectors of the a-th plant node and the b-th plant node at the i-th moment with the similarity of the feature vectors of the i-th plant node and the j-th plant node at the i+u-th moment, and if the similarity of the clustering vectors of the a-th plant node and the b-th plant node at the i-th moment is greater than a first similarity threshold, taking them as a group of training samples, repeating the process to obtain a verification model training sample set; using the verification model training sample set to train the verification model; During the training process, the mean square error is used as the loss function for the preset prediction model and the preset verification model to optimize the prediction model; A second preset time period G is set; the environment, soil, water and fertilizer data and plant images of the P mu experimental field are collected at Q equally spaced moments; based on the environment, soil and water and fertilizer data collected at the i-th moment in the second preset time period G, the health score and medicinal ingredient concentration of the plant at the i+u-th moment are predicted using the prediction model; the verification model and the prediction model are used to verify each other.

4. The water-fertilizer integrated drip irrigation control system according to claim 3 is characterized in that: Use validation models and prediction models to validate each other, including: Based on the clustering vector of plant node a and the clustering vector of plant node b at the i-th moment in the second preset time period G, the prediction feature vectors of node a and node b at the i+u-th moment are obtained respectively through the prediction model; Based on the clustering vector of plant node a and the clustering vector of plant node b at the i-th moment in the second preset time period G, it is determined that the similarity between the clustering vector of plant node a and the clustering vector of plant node b is greater than a first similarity threshold; If the similarity between the clustering vector of plant node a and the clustering vector of plant node b is greater than the first similarity threshold, the verification values ​​of node a and node b at the i+uth time are obtained respectively through the verification model; Determine whether the difference between the similarity of the predicted feature vectors of node a and node b at the i+uth moment and the verification value is greater than a second similarity threshold; If the difference between the similarity of the predicted feature vectors of node i and node j at the i+uth moment and the verification value is greater than the second similarity threshold, the prediction model training is completed; If the difference between the similarity of the predicted feature vectors of node i and node j at the i+uth moment and the verification value is less than the second similarity threshold, the prediction model and the verification model continue to be trained.

5. The water-fertilizer integrated drip irrigation control system according to claim 1 is characterized in that: Individual updates based on update factors include: Select the individuals whose health scores are lower than the health threshold among the R individuals as the parent generation update; Calculate the attack vector of the i-th individual , the formula is: ;in, represents the current global optimal individual, represents the i-th individual; Set the hyperplane of the i-th individual, which is a vector perpendicular to the attack vector of the i-th individual ; Calculate the cruise vector of the i-th individual , the formula is: ;in, represents the projection of the attack vector on the hyperplane; The update formula of the individual is as follows: ; ; in, represents the update vector of the ith individual, represents the updated value of the i-th individual, and Represents a random number, represents the modulus length of the attack vector, Indicates the modulus of the cruise vector.

6. The water-fertilizer integrated drip irrigation control system according to claim 1, characterized in that: According to the environment, soil data and plant images of the target area, combined with the preset model, the water and fertilizer dosage at the target time is determined, and the drip irrigation plan is planned, including: Collect the number of droppers and nozzles and the maximum delivery volume per unit time of the droppers and nozzles; According to the duration of the target time and the water and fertilizer allocation at the target time, the actual delivery volume per unit time is determined; According to the actual delivery volume per unit time, based on the drip irrigation priority strategy, the drip irrigation scheme is obtained; Among them, the drip irrigation priority strategy means that drip irrigation is used first to meet the water and fertilizer allocation requirements at the target time; when the drip irrigation delivery per unit time cannot fully meet the water and fertilizer allocation requirements at the target time, the sprinkler irrigation system is used to supplement it.

Citation Information

Patent Citations

  • An irrigation area water-saving potential measurement method and device based on big data

    CN109816207A

  • Water and fertilizer irrigation system

    CN110178518A

  • Rice field irrigation water and fertilizer variable control system and device

    CN111670672A

  • Water-fertilizer-pesticide integrated intelligent irrigation system based on machine learning algorithm

    CN112450056A

  • Water and fertilizer monitoring and analyzing system and method for greenhouse grapes

    CN115184353A