Fertilizer combined application proportion method and application of fertilizer combined application proportion method in agricultural greenhouse gas emission reduction

Abstract

The invention discloses a fertilizer combined application proportion method and application of the fertilizer combined application proportion method in agricultural greenhouse gas emission reduction, and relates to the field of combined application proportion configuration.The method comprises the steps that image data of each kind of crops are collected, and then feature data of the image data are extracted; therefore, the growth quality of the crops can be analyzed and judged according to the feature data of the crops, and finally, the support vector machine classification model has good classification performance for the growth quality grades of the crops obtained according to the feature data of the crops. The growth quality grade data of the crops to be cultivated can be subsequently acquired, so that the application proportion of the fertilizer sample can be fed back and optimized; by comprehensively considering the growth quality grade of the crops and the greenhouse gas emission amount, the finally obtained optimal fertilizer application proportion can meet the requirements of the growth quality grade of the crops and the greenhouse gas emission amount at the same time, so that the optimal fertilizer application proportion has higher use value.

Description

Technical Field

[0001] The present invention belongs to the field of fertilizer application ratio configuration, and in particular, relates to a fertilizer application ratio method and its application in agricultural greenhouse gas emission reduction. Background Art

[0002] Nitrogen-potassium fertilizers are the collective name for nitrogen fertilizers and potassium fertilizers. They are two commonly used fertilizers in agricultural production. The use of nitrogen fertilizers enhances crop growth and yield, but it also intensifies the nitrogen conversion process in the soil, leading to large emissions of greenhouse gases such as N2O. In addition, improper application of nitrogen fertilizers may also cause soil degradation, water pollution and a reduction in biodiversity. At the same time, potassium fertilizer (KF), as an important nutrient element, plays an irreplaceable role in maintaining the normal growth of crops and improving their stress resistance. After potassium fertilizer is applied, K + Can be obtained by mixing with NH4 + Competition for fixed sites directly changes the nitrogen nutrient content in the soil, or indirectly affects N2O emissions by affecting plant nitrogen utilization and the activity of microorganisms involved in nitrogen conversion. Therefore, in agricultural production, the specific amount of fertilizer applied needs to be considered comprehensively based on factors such as soil, crop type and growth stage to achieve the best effect. At the same time, attention should be paid to the rational use of fertilizers to avoid environmental pollution and waste of resources caused by excessive fertilization.

[0003] Chinese patent CN107306765A discloses a water-fertilizer integrated irrigation system and irrigation method, including a variable frequency water pump, a main electromagnetic valve and a proportional integral regulating valve arranged on a water flow pipeline, the first water outlet of the proportional integral regulating valve is connected to the fertilization pipeline via a proportional fertilization pump, the second water outlet of the proportional integral regulating valve is connected to a flow sensor, the flow sensor is connected to the sub-electromagnetic valves of each area via a PH / EC sensor, the sub-electromagnetic valves are arranged on the irrigation pipelines of each area, each irrigation area is also provided with a soil moisture sensor, the soil moisture sensor, the PH / EC sensor, the flow sensor, the variable frequency water pump, the main electromagnetic valve, the proportional integral regulating valve and the proportional fertilization pump are connected to the controller respectively. The invention has a simple structure and is easy to operate. It can realize accurate field irrigation and fertilization operations without pre-configuring fertilizers, greatly improves the efficiency of irrigation and fertilization, and can effectively avoid environmental pollution and land salinization problems caused by fertilizer waste.

[0004] The current fertilizer application ratio method is often based on farmers' experience and habits and lacks scientific basis; this method may lead to inaccurate fertilizer application and affect crop growth and yield. At the same time, the existing fertilizer application ratio configuration method is often based on a single crop model, ignoring the differences in fertilizer requirements of different crops. This may lead to, in some cases, the fertilizer application ratio being unsuitable for the growth needs of specific crops, resulting in low fertilizer utilization and causing environmental pollution. Summary of the invention

[0005] In view of the problems in the related art, the present invention proposes a fertilizer application ratio method and its application in agricultural greenhouse gas emission reduction to overcome the above-mentioned technical problems existing in the existing related technology.

[0006] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0007] The present invention is a method for proportioning fertilizers, comprising the following steps:

[0008] S1. Set the types of crops to be collected, and collect image data of each crop at multiple time points according to the set types of crops; then extract multiple feature data of each image data to obtain crop feature data;

[0009] S2, constructing multiple initial support vector machine classification models; using multiple feature data of crops of the same type at the same time point in the crop feature data to train and test each initial support vector machine classification model, to obtain multiple final support vector machine classification models;

[0010] S3, setting multiple groups of crops to be cultivated and initializing multiple fertilizer samples with different application ratios; applying fertilizer samples with different application ratios to each group of crops to be cultivated, using the multiple final support vector machine classification models to obtain growth quality grade data of the crops to be cultivated and collect greenhouse gas emission data, and performing feedback optimization on the fertilizer samples according to the growth quality grade data and the greenhouse gas emission data to obtain the optimal fertilizer application ratio;

[0011] Since different types of crops require different fertilizer ratios, this solution takes into account a variety of crop types and can obtain the corresponding optimal fertilizer ratio for each crop, making the application of the solution universal and increasing the scope of application of the solution; by collecting image data of each crop and then extracting the feature data of the image data, the growth quality of the crop can be analyzed and judged based on the feature data of the crop; that is, the support vector machine classification model is trained and tested using the feature data of the crop to obtain the final support vector machine classification model; so that the final support vector machine classification model is The method has good classification performance in obtaining the growth quality grade of crops according to the characteristic data of crops, and provides an accurate model for the subsequent acquisition of growth quality grade data of crops to be cultivated, so as to facilitate the subsequent acquisition of growth quality grade data of crops to be cultivated for feedback optimization of the application ratio of fertilizer samples; secondly, in this scheme, by comprehensively considering the growth quality grade of crops and greenhouse gas emissions, the optimal fertilizer application ratio finally obtained can meet the needs of both the growth quality grade of crops and greenhouse gas emissions, making the optimal fertilizer application ratio more valuable.

[0012] Preferably, the fertilizer comprises nitrogen and potassium fertilizer.

[0013] Preferably, the S1 comprises the following steps:

[0014] S11, set the crop type set a = {a1, a2, ..., a i ,...a a′}, a i represents the i-th crop type set, a′ represents the total number of crop types set; according to the crop type set a={a1, a2, ..., a i ,...a a′}Get the corresponding crop cultivation sample set represents the crop cultivation sample corresponding to the i-th crop type;

[0015] Re-set the static influencing factor set and the set of dynamic influencing factors c = {c1, c2}; represents the i-th dynamic influencing factor set, Indicates the total number of dynamic influencing factors set; c1 and c2 represent the crop growth quality grade and greenhouse gas emissions respectively;

[0016] S12, in the static influencing factor set Under the same conditions, the crop cultivation sample set Each crop cultivation sample is fertilized and cultivated;

[0017] In the crop cultivation sample collection In the growth cycle of each crop cultivation sample, c′ time points are uniformly selected to obtain the data collection time point matrix as follows,

[0018]

[0019] in, represents the jth data collection time point selected within the growth cycle of the i-th crop cultivation sample;

[0020] S13. Set the crop growth quality level set d i represents the set i-th crop growth quality level, Indicates the total number of set crop growth quality levels; according to the data collection time point matrix Crop growth quality grade set And the dynamic influencing factor set c = {c1, c2} on the crop cultivation sample set The data of each crop cultivation sample in the data collection time point is collected to obtain the crop growth image set matrix And the corresponding crop growth quality grade dataset matrix e; respectively as follows,

[0021]

[0022] in, represents the crop image set collected at the jth data collection time point selected within the growth cycle of the i-th crop cultivation sample, express The kth crop image in ij represents the growth quality grade dataset corresponding to the crop image set collected at the jth data collection time point selected within the growth cycle of the i-th crop cultivation sample, e ijk Indicates e ij The kth crop growth quality grade data in ; Represents the total number of crop images collected at each data collection time point;

[0023] S14. Set the extraction feature type set d′ i represents the set i-th extraction feature type, Indicates the total number of set extraction feature types;

[0024] According to the feature type set extracted The crop growth image matrix Extract feature data from each crop growth image in the image to obtain the growth image feature data set matrix as follows,

[0025]

[0026] in, It represents the characteristic data matrix of the crop image collected at the jth data collection time point selected during the growth period of the i-th crop cultivation sample, as follows:

[0027]

[0028] in, express The k1th feature data of the kth crop growth image in ;

[0029] Crop types include, for example, food crops, such as rice, wheat, corn, sorghum, etc.; vegetables and fruits, such as tomatoes, cucumbers, eggplants, peppers, apples, pears, peaches, etc.; oil crops, such as peanuts, soybeans, rapeseed, cotton, etc.; economic crops, such as sugarcane, tea, coffee, spices, medicinal plants, etc., by taking all the above-mentioned crops into consideration, this scheme has better universality; static influencing factors include, for example, light, water and soil nutrients, etc., by quantifying these influences, it is convenient to analyze the quantitative relationship between the growth quality grade of crops and greenhouse gas emissions and the fertilizer application ratio; extracted feature types include, for example, leaf color, leaf size and shape, stem diameter size and shape, etc. These feature data can better reflect the growth quality grade of crops, which is convenient for subsequent analysis of the growth quality grade of crops.

[0030] Preferably, S2 comprises the following steps:

[0031] S21, construct a′×c′ initial support vector machine classification models, each of which is provided with a calling interface, and obtain the support vector machine classification model calling interface matrix e′; as follows

[0032]

[0033] Among them, e′ ij express The calling interface of the corresponding initial support vector machine classification model; setting the kernel functions of the a′×c′ initial support vector machine classification models to be radial basis functions;

[0034] S22, using the growth image feature data set matrix and the crop growth quality grade data set matrix e to train and test a′×c′ initial support vector machine classification models, and obtain a′×c′ final support vector machine classification models;

[0035] The support vector machine classification model is a machine learning classification model with excellent performance. By using the acquired characteristic data of crops to train and test the support vector machine classification model, the support vector machine classification model can have a good classification and prediction ability for the growth quality grade of crops based on the characteristic data of crops, providing an accurate model for the subsequent acquisition of growth quality grade data of crops to be cultivated.

[0036] Preferably, the S22 comprises the following steps:

[0037] S221. Setting the first training error threshold set And the training data ratio set According to the training data ratio set For the growth image feature dataset matrix and divide each data set of the crop growth quality grade data set matrix e to obtain an image feature training data set matrix, an image feature test data set matrix, a growth quality grade training data set matrix, and a growth quality grade test data set matrix;

[0038] S222, inputting each row of the image feature training data set matrix and the growth quality grade test data matrix into a′×c′ initial support vector machine classification model for training operation;

[0039] In the process of training the i|i=1,2,...,a′×c′ initial support vector machine classification model, when the training error is less than When , the training operation is stopped, and the i|i=1,2,...,a′×c′ trained support vector machine classification model is obtained;

[0040] S223, set the accuracy threshold set f1 = {f 11 ,f 12 ,...,f 1i ,...,f 1(a′×c′)}; Input each row of the image feature test data set matrix and the growth quality grade test data matrix into a′×c′ trained support vector machine classification model for testing, and obtain the test accuracy data set f2={f 21 ,f 22 ,...,f 2i ,...,f 2(a′×c′)};

[0041] When f 2i ≥f 1i When |i=1,2,...,a′×c′, the i|i=1,2,...,a′×c′ trained support vector machine classification model is used as the i|i=1,2,...,a′×c′ final support vector machine classification model; otherwise, repeat S23 until f 2i ≥f 1i When |i=1,2,...,a′×c′, the i-th|i=1,2,...,a′×c′ final support vector machine classification model is obtained;

[0042] S224, replacing the a′×c′ initial support vector machine classification models with a′×c′ final support vector machine classification models, that is, each calling interface in the support vector machine classification model calling interface matrix e′ corresponds to a final support vector machine classification model;

[0043] Different support vector machine classification models are used to classify characteristic data of crops at different time points or of different types, so that each support vector machine classification model has a good classification effect for crops of a specific type at a specific time point, thereby increasing the accuracy of subsequent classification of growth quality grade data of specific types of crops to be cultivated, thereby making the optimal fertilizer application ratio more accurate.

[0044] Preferably, S3 comprises the following steps:

[0045] S31, construct a sparrow population; set the size of the sparrow population to be f′, and the sparrow population is expressed as represents the i-th sparrow in the sparrow population; setting the search space dimension of the sparrow population to two dimensions, the maximum number of iterations to g′1 and the current number of iterations to g′2; and then setting the initial position set of the sparrow population;

[0046] S32, setting the matrix of crops to be cultivated according to the maximum number of iterations as g′1 as follows

[0047]

[0048] Among them, a ij represents the crops to be cultivated using the mass of nitrogen compounds and the mass of potassium compounds represented by the position of the jth sparrow in the i-th iteration of the sparrow population; b represents the total number of groups of crops to be cultivated, b>g′1; set the matrix of crops to be cultivated Each type of crop to be cultivated is a i ;

[0049] S33, setting the fitness function of the sparrow population

[0050] S34, start iteration, and use the fitness function of the sparrow population in each iteration process The fitness of each sparrow is calculated, and the position of the individual sparrow corresponding to the best fitness is selected; and the position of each sparrow in the sparrow population is updated according to the position of the individual sparrow corresponding to the best fitness;

[0051] When g′2≥g′1, stop the iteration and get the final position set of the sparrow population and Respectively represent the one-dimensional component and the two-dimensional component of the final position of the i-th sparrow in the sparrow population;

[0052] S35, and The optimum mass values ​​of nitrogen compounds and potassium compounds in fertilizers are used as the optimum mass values, respectively; and the optimum fertilizer application ratio is obtained according to the optimum mass values ​​of nitrogen compounds and potassium compounds;

[0053] By using the sparrow algorithm to simultaneously optimize the application ratios of multiple fertilizers, the optimal fertilizer application ratio can be obtained more efficiently, and at the same time, it can avoid falling into the local optimal situation during the optimization process, thereby improving the robustness of the optimal fertilizer application ratio finally obtained.

[0054] Preferably, setting the initial position set of the sparrow population in S31 comprises the following steps:

[0055] S311. Set the mass value range of nitrogen compounds in fertilizer to Recorded as the first value interval; the mass value interval of potassium compounds in fertilizer is Recorded as the second value interval; and They represent the lower and upper limits of the mass of nitrogen compounds, respectively. and They represent the lower and upper limits of the mass of potassium compounds respectively;

[0056] S312: According to the first value interval And the second value interval Initialize the position of each sparrow in the sparrow population and obtain the initial position set g = {[g 11 ,g 12 ],[g 21 ,g 22 ],...,[g i1 ,g i2],...,[g f′1 ,g f′2 ]}; g i1 and g i2 represent the one-dimensional component and the two-dimensional component of the initial position of the i-th sparrow in the sparrow population respectively; g i1 and g i2 The calculation formulas are as follows:

[0057]

[0058] The fertilizer application ratio is initialized by even distribution, and the initialized application ratio data is used as the initial position of each sparrow in the sparrow population, so that the solution space of the sparrow population is richer, thereby making it easier to find the optimal fertilizer application ratio.

[0059] Preferably, the S33 comprises the following steps:

[0060] S331, in the current round of iteration, according to the data collection time point matrix Collect the image set and greenhouse gas emission data of the crops to be cultivated at each data collection time point to obtain the image matrix h of the crops to be cultivated and the greenhouse gas emission data set represents the greenhouse gas emission data of the crops to be cultivated collected at the i-th data collection time point during the current round of iteration; the image matrix h of the crops to be cultivated is as follows,

[0061]

[0062] Among them, h ij represents the jth image collected from the crop to be cultivated at the i-th data collection time point during the current round of iteration;

[0063] S332: extracting feature type set according to the Feature data is extracted from each image in the crop image matrix h to obtain the crop feature data set matrix h′; as follows:

[0064]

[0065] Among them, h ij represents the feature data set of the jth image collected from the crop to be cultivated at the i-th data collection time point in the current round of iteration; h ijk Indicates h ij The kth feature data in ;

[0066] S333, calling the support vector machine classification model to call the final support vector machine classification model corresponding to each interface in the ith row of the interface matrix e′ to perform classification operations on each feature data set in the feature data set matrix h′ of the crop to be cultivated, and obtaining the growth quality grade data set of the crop to be cultivated Indicates the growth quality grade data of cultivated crops at the i-th data collection time point in the current round of iteration;

[0067] S334, according to the cultivated crop growth quality grade data set and greenhouse gas emissions datasets Constructing a fitness function of the sparrow population;

[0068] By comprehensively constructing the fitness function of the sparrow population based on the crop growth quality grade dataset and the greenhouse gas emission dataset, the sparrow population optimizes the application ratio of fertilizer samples based on the simultaneous optimization of the crop growth quality grade and the greenhouse gas emissions. The final optimal fertilizer application ratio can simultaneously meet the requirements of both crop growth quality grade and greenhouse gas emissions, making the optimal fertilizer application ratio more valuable.

[0069] Preferably, the S334 includes the following steps:

[0070] S3341, according to the cultivation of crop growth quality grade data set Calculate average growth quality grade data According to the greenhouse gas emissions data set Calculating average greenhouse gas emissions data

[0071] S3342, setting a growth quality level threshold l'1 and a greenhouse gas emission threshold l'2; constructing a fitness function of the sparrow population according to the growth quality level threshold l'1, the greenhouse gas emission threshold l'2, the average growth quality level data and the average greenhouse gas emission data as follows,

[0072]

[0073] Among them, α is a positive number, indicating the correction amount;

[0074] By setting growth quality level thresholds and greenhouse gas emission thresholds, the growth quality level and greenhouse gas emission are limited and the optimization range is determined.

[0075] A fertilizer application ratio system includes a crop image acquisition module, a crop image feature extraction module, a final support vector machine classification model construction module and a fertilizer application ratio adjustment optimization module;

[0076] The crop image acquisition module is used to collect image data of each crop at multiple time points according to the set crop types;

[0077] The crop image feature extraction module is used to extract multiple feature data from each image data to obtain crop feature data;

[0078] The final support vector machine classification model construction module is used to train and test each initial support vector machine classification model using multiple feature data of crops of the same type at the same time point in the crop feature data to obtain multiple final support vector machine classification models;

[0079] The fertilizer application ratio adjustment optimization module is used to perform feedback optimization on fertilizer samples according to the growth quality grade data and greenhouse gas emission data to obtain the optimal fertilizer application ratio.

[0080] The present invention has the following beneficial effects:

[0081] 1. The present invention sets a crop image acquisition module, a crop image feature extraction module, a final support vector machine classification model construction module and a fertilizer application ratio adjustment optimization module; by considering the types of various crops, the corresponding optimal fertilizer application ratio can be obtained for each crop, so that the application of the scheme has universality and increases the scope of application of the scheme; by collecting image data of each crop and then extracting feature data of the image data, the growth quality of the crop can be analyzed and determined according to the feature data of the crop; that is, the support vector machine classification model is trained and tested using the feature data of the crop to obtain the final support vector machine classification model ; The final support vector machine classification model has good classification performance for the growth quality grade of crops obtained according to the characteristic data of crops, and provides an accurate model for the subsequent acquisition of growth quality grade data of crops to be cultivated, so as to facilitate the subsequent acquisition of growth quality grade data of crops to be cultivated for feedback optimization of the application ratio of fertilizer samples; secondly, by accurately adjusting the application ratio of nitrogen fertilizer and potassium fertilizer in this scheme, greenhouse gas emissions can be effectively reduced, especially reducing N2O and CH4 gas emissions that have a significant impact on global warming; this will not only help alleviate the problem of global climate change, but also improve soil health and water quality, thereby promoting the protection of biodiversity. At the same time, this method emphasizes environmental protection and agricultural sustainability, and meets the needs of modern agricultural production to transform to green and environmental protection.

[0082] 2. In the present invention, characteristic data of crops at different time points or of different types are classified using different support vector machine classification models, so that each support vector machine classification model has a good classification effect for crops of a specific type at a specific time point, thereby increasing the accuracy of subsequent classification of growth quality grade data of specific types of crops to be cultivated, thereby making the obtained optimal fertilizer application ratio more accurate, which not only reduces greenhouse gas emissions in agricultural production, but also improves crop yield and quality, thereby increasing farmers' economic benefits.

[0083] 3. In the present invention, the sparrow algorithm is used to simultaneously optimize the application ratios of multiple fertilizers, so that the optimal fertilizer application ratio can be obtained more efficiently, and at the same time, it can avoid falling into the local optimum during the optimization process, thereby improving the robustness of the optimal fertilizer application ratio finally obtained.

[0084] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0086] Figure 1 The present invention is a schematic diagram of a process for configuring a fertilizer application ratio by a fertilizer application ratio system. DETAILED DESCRIPTION

[0087] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0088] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0089] This embodiment discloses a method for proportioning fertilizers, wherein nitrogen and potassium fertilizers are used as examples for the fertilizers.

[0090] The following steps are involved:

[0091] S1. Set the types of crops to be collected, and collect image data of each crop at multiple time points according to the set types of crops; then extract multiple feature data of each image data to obtain crop feature data;

[0092] The S1 comprises the following steps:

[0093] S11, set the crop type set a = {a1, a2, ..., a i ,...a a′}, a i represents the i-th crop type set, a′ represents the total number of crop types set; according to the crop type set a={a1, a2, ..., a i ,...a a′}Get the corresponding crop cultivation sample set represents the crop cultivation sample corresponding to the i-th crop type;

[0094] Re-set the static influencing factor set and the set of dynamic influencing factors c = {c1, c2}; represents the i-th dynamic influencing factor set, Indicates the total number of dynamic influencing factors set; c1 and c2 represent the crop growth quality grade and greenhouse gas emissions respectively;

[0095] S12, in the static influencing factor set Under the same conditions, the crop cultivation sample set Each crop cultivation sample is fertilized and cultivated;

[0096] In the crop cultivation sample collection In the growth cycle of each crop cultivation sample, c′ time points are uniformly selected to obtain the data collection time point matrix as follows,

[0097]

[0098] in, represents the jth data collection time point selected within the growth cycle of the i-th crop cultivation sample;

[0099] S13. Set the crop growth quality level set d i represents the set i-th crop growth quality level, Indicates the total number of set crop growth quality levels; according to the data collection time point matrix Crop growth quality grade set And the dynamic influencing factor set c = {c1, c2} on the crop cultivation sample set The data of each crop cultivation sample in the data collection time point is collected to obtain the crop growth image set matrix And the corresponding crop growth quality grade dataset matrix e; respectively as follows,

[0100]

[0101] in, represents the crop image set collected at the jth data collection time point selected within the growth cycle of the i-th crop cultivation sample, express The kth crop image in ij represents the growth quality grade dataset corresponding to the crop image set collected at the jth data collection time point selected within the growth cycle of the i-th crop cultivation sample, e ijk Indicates e ij The kth crop growth quality grade data in ; Represents the total number of crop images collected at each data collection time point;

[0102] S14. Set the extraction feature type set d′ i represents the set i-th extraction feature type, Indicates the total number of set extraction feature types;

[0103] According to the feature type set extracted The crop growth image matrix Extract feature data from each crop growth image in the image to obtain the growth image feature data set matrix as follows,

[0104]

[0105] in, It represents the characteristic data matrix of the crop image collected at the jth data collection time point selected during the growth period of the i-th crop cultivation sample, as follows:

[0106]

[0107] in, express The k1th feature data of the kth crop growth image in ;

[0108] S2, constructing multiple initial support vector machine classification models; using multiple feature data of crops of the same type at the same time point in the crop feature data to train and test each initial support vector machine classification model, to obtain multiple final support vector machine classification models;

[0109] The S2 comprises the following steps:

[0110] S21, construct a′×c′ initial support vector machine classification models, each of which is provided with a calling interface, and obtain the support vector machine classification model calling interface matrix e′; as follows

[0111]

[0112] Among them, e′ ij express The calling interface of the corresponding initial support vector machine classification model; setting the kernel functions of the a′×c′ initial support vector machine classification models to be radial basis functions;

[0113] S22, using the growth image feature data set matrix and the crop growth quality grade data set matrix e to train and test a′×c′ initial support vector machine classification models, and obtain a′×c′ final support vector machine classification models;

[0114] The S22 comprises the following steps:

[0115] S221. Setting the first training error threshold set And the training data ratio set According to the training data ratio set For the growth image feature dataset matrix and divide each data set of the crop growth quality grade data set matrix e to obtain an image feature training data set matrix, an image feature test data set matrix, a growth quality grade training data set matrix, and a growth quality grade test data set matrix;

[0116] S222, inputting each row of the image feature training data set matrix and the growth quality grade test data matrix into a′×c′ initial support vector machine classification model for training operation;

[0117] In the process of training the i|i=1,2,...,a′×c′ initial support vector machine classification model, when the training error is less than When , the training operation is stopped, and the i|i=1,2,...,a′×c′ trained support vector machine classification model is obtained;

[0118] S223, set the accuracy threshold set f1 = {f 11 ,f 12 ,...,f 1i ,...,f 1(a′×c′)}; Input each row of the image feature test data set matrix and the growth quality grade test data matrix into a′×c′ trained support vector machine classification model for testing, and obtain the test accuracy data set f2={f 21 ,f 22 ,...,f 2i ,...,f 2(a′×c′)};

[0119] When f 2i ≥f 1i When |i=1,2,...,a′×c′, the i|i=1,2,...,a′×c′ trained support vector machine classification model is used as the i|i=1,2,...,a′×c′ final support vector machine classification model; otherwise, repeat S23 until f 2i ≥f 1i When |i=1,2,...,a′×c′, the i-th|i=1,2,...,a′×c′ final support vector machine classification model is obtained;

[0120] S224, replacing the a′×c′ initial support vector machine classification models with a′×c′ final support vector machine classification models, that is, each calling interface in the support vector machine classification model calling interface matrix e′ corresponds to a final support vector machine classification model;

[0121] S3, setting multiple groups of crops to be cultivated and initializing multiple nitrogen-potassium fertilizer samples with different application ratios; applying nitrogen-potassium fertilizer samples with different application ratios to each group of crops to be cultivated, using the multiple final support vector machine classification models to obtain growth quality grade data of the crops to be cultivated and collect greenhouse gas emission data, and performing feedback optimization on the nitrogen-potassium fertilizer samples according to the growth quality grade data and the greenhouse gas emission data to obtain the optimal nitrogen-potassium fertilizer application ratio;

[0122] The S3 comprises the following steps:

[0123] S31, construct a sparrow population; set the size of the sparrow population to be f′, and the sparrow population is expressed as represents the i-th sparrow in the sparrow population; setting the search space dimension of the sparrow population to two dimensions, the maximum number of iterations to g′1 and the current number of iterations to g′2; and then setting the initial position set of the sparrow population;

[0124] The step of setting the initial position set of the sparrow population in S31 comprises the following steps:

[0125] S311, set the mass value range of nitrogen compounds in nitrogen-potassium fertilizers to Recorded as the first value interval; the mass value interval of potassium compounds in nitrogen-potassium fertilizers is Recorded as the second value interval; and They represent the lower and upper limits of the mass of nitrogen compounds, respectively. and They represent the lower and upper limits of the mass of potassium compounds respectively;

[0126] S312: According to the first value interval And the second value interval Initialize the position of each sparrow in the sparrow population and obtain the initial position set g = {[g 11 ,g 12 ],[g 21 ,g 22 ],...,[g i1 ,g i2 ],...,[g f′1 ,g f′2 ]}; g i1 and g i2 represent the one-dimensional component and the two-dimensional component of the initial position of the i-th sparrow in the sparrow population respectively; g i1 and g i2 The calculation formulas are as follows:

[0127]

[0128] S32, setting the matrix of crops to be cultivated according to the maximum number of iterations as g′1 as follows

[0129]

[0130] Among them, a ij represents the crops to be cultivated using the mass of nitrogen compounds and the mass of potassium compounds represented by the position of the jth sparrow in the i-th iteration of the sparrow population; b represents the total number of groups of crops to be cultivated, b>g′1; set the matrix of crops to be cultivated Each type of crop to be cultivated is a i ;

[0131] S33, setting the fitness function of the sparrow population

[0132] The S33 comprises the following steps:

[0133] S331, in the current round of iteration, according to the data collection time point matrix Collect the image set and greenhouse gas emission data of the crops to be cultivated at each data collection time point to obtain the image matrix h of the crops to be cultivated and the greenhouse gas emission data set represents the greenhouse gas emission data of the crops to be cultivated collected at the i-th data collection time point during the current round of iteration; the image matrix h of the crops to be cultivated is as follows,

[0134]

[0135] Among them, h ij represents the jth image collected from the crop to be cultivated at the i-th data collection time point during the current round of iteration;

[0136] S332: extracting feature type set according to the Feature data is extracted from each image in the crop image matrix h to obtain the crop feature data set matrix h′; as follows:

[0137]

[0138] Among them, h ij represents the feature data set of the jth image collected from the crop to be cultivated at the i-th data collection time point in the current round of iteration; h ijk Indicates h ij The kth feature data in ;

[0139] S333, calling the support vector machine classification model to call the final support vector machine classification model corresponding to each interface in the ith row of the interface matrix e′ to perform classification operations on each feature data set in the feature data set matrix h′ of the crop to be cultivated, and obtaining the growth quality grade data set of the crop to be cultivated Indicates the growth quality grade data of cultivated crops at the i-th data collection time point in the current round of iteration;

[0140] S334, according to the cultivated crop growth quality grade data set and greenhouse gas emissions datasets Constructing a fitness function of the sparrow population;

[0141] The S334 comprises the following steps:

[0142] S3341, according to the cultivation of crop growth quality grade data set Calculate average growth quality grade data According to the greenhouse gas emissions data set Calculating average greenhouse gas emissions data

[0143] S3342, setting a growth quality level threshold l'1 and a greenhouse gas emission threshold l'2; constructing a fitness function of the sparrow population according to the growth quality level threshold l'1, the greenhouse gas emission threshold l'2, the average growth quality level data and the average greenhouse gas emission data as follows,

[0144]

[0145] Among them, α is a positive number, indicating the correction amount;

[0146] S34, start iteration, and use the fitness function of the sparrow population in each iteration process The fitness of each sparrow is calculated, and the position of the individual sparrow corresponding to the best fitness is selected; and the position of each sparrow in the sparrow population is updated according to the position of the individual sparrow corresponding to the best fitness;

[0147] When g′2≥g′1, stop the iteration and get the final position set of the sparrow population and Respectively represent the one-dimensional component and the two-dimensional component of the final position of the i-th sparrow in the sparrow population;

[0148] S35, and The optimal mass values ​​of nitrogen compounds and potassium compounds in nitrogen-potassium fertilizers are respectively used; and the optimal nitrogen-potassium fertilizer application ratio is obtained according to the optimal mass values ​​of nitrogen compounds and potassium compounds.

[0149] A nitrogen-potassium fertilizer application ratio system, comprising a crop image acquisition module, a crop image feature extraction module, a final support vector machine classification model construction module and a nitrogen-potassium fertilizer application ratio adjustment optimization module;

[0150] The crop image acquisition module is used to collect image data of each crop at multiple time points according to the set crop types;

[0151] The crop image feature extraction module is used to extract multiple feature data from each image data to obtain crop feature data;

[0152] The final support vector machine classification model construction module is used to train and test each initial support vector machine classification model using multiple feature data of crops of the same type at the same time point in the crop feature data to obtain multiple final support vector machine classification models;

[0153] The nitrogen-potassium fertilizer application ratio adjustment optimization module is used to perform feedback optimization on nitrogen-potassium fertilizer samples according to the growth quality grade data and greenhouse gas emission data to obtain the optimal nitrogen-potassium fertilizer application ratio.

[0154] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0155] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A method for proportioning fertilizers, characterized in that: The following steps are involved: S1. Set the types of crops to be collected, and collect image data of each crop at multiple time points according to the set types of crops; then extract multiple feature data of each image data to obtain crop feature data; S2, constructing multiple initial support vector machine classification models; using multiple feature data of crops of the same type at the same time point in the crop feature data to train and test each initial support vector machine classification model, to obtain multiple final support vector machine classification models; S3, setting multiple groups of crops to be cultivated and initializing multiple fertilizer samples with different application ratios; Fertilizer samples with different application ratios are applied to each group of crops to be cultivated, and the multiple final support vector machine classification models are used to obtain growth quality grade data of the crops to be cultivated and collect greenhouse gas emission data. The fertilizer samples are feedback optimized according to the growth quality grade data and the greenhouse gas emission data to obtain the optimal fertilizer application ratio.

2. A fertilizer application ratio method according to claim 1, characterized in that: The S1 comprises the following steps: S11, set the crop type set a = {a1, a2, ..., a i ,...a a′ }, a i represents the i-th crop type set, a′ represents the total number of crop types set; according to the crop type set a={a1, a2, ..., a i ,...a a′ }Get the corresponding crop cultivation sample set represents the crop cultivation sample corresponding to the i-th crop type; Re-set the static influencing factor set and the set of dynamic influencing factors c = {c1, c2}; represents the i-th dynamic influencing factor set, Indicates the total number of dynamic influencing factors set; c1 and c2 represent the crop growth quality grade and greenhouse gas emissions respectively; S12, in the static influencing factor set Under the same conditions, the crop cultivation sample set Each crop cultivation sample in the crop cultivation sample set is subjected to fertilization cultivation; In the growth cycle of each crop cultivation sample, c′ time points are uniformly selected to obtain the data collection time point matrix S13. Set the crop growth quality level set d i represents the set i-th crop growth quality level, Indicates the total number of set crop growth quality levels; according to the data collection time point matrix Crop growth quality grade set And the dynamic influencing factor set c = {c1, c2} on the crop cultivation sample set The data of each crop cultivation sample in the data collection time point is collected to obtain the crop growth image set matrix And the corresponding crop growth quality grade dataset matrix e; S14. Set the extraction feature type set d i ′ represents the set i-th extraction feature type, Indicates the total number of set extraction feature types; According to the feature type set extracted The crop growth image matrix Extract feature data from each crop growth image in the image to obtain the growth image feature data set matrix 3. A fertilizer application ratio method according to claim 1, characterized in that: The S2 comprises the following steps: S21, constructing a′×c′ initial support vector machine classification models, each of which is provided with a calling interface, and obtaining a support vector machine classification model calling interface matrix e′; setting the kernel functions of the a′×c′ initial support vector machine classification models to be radial basis functions; S22, using the growth image feature data set matrix d and the crop growth quality grade data set matrix e to train and test a′×c′ initial support vector machine classification models respectively, to obtain a′×c′ final support vector machine classification models.

4. A fertilizer application ratio method according to claim 3, characterized in that: The S22 comprises the following steps: S221. Setting the first training error threshold set And the training data ratio set According to the training data ratio set For the growth image feature dataset matrix and divide each data set of the crop growth quality grade data set matrix e to obtain an image feature training data set matrix, an image feature test data set matrix, a growth quality grade training data set matrix, and a growth quality grade test data set matrix; S222, inputting each row of the image feature training data set matrix and the growth quality grade test data matrix into a′×c′ initial support vector machine classification model for training operation; In the process of training the i|i=1,2,...,a′×c′ initial support vector machine classification model, when the training error is less than When , the training operation is stopped, and the i|i=1,2,...,a′×c′ trained support vector machine classification model is obtained; S223, set the accuracy threshold set f1 = {f 11 ,f 12 ,...,f 1i ,...,f 1(a′×c′) }; Input each row of the image feature test data set matrix and the growth quality grade test data matrix into a′×c′ trained support vector machine classification model for testing, and obtain the test accuracy data set f2={f 21 ,f 22 ,...,f 2i ,...,f 2(a′×c′) }; When f 2i ≥f 1i When |i=1,2,...,a′×c′, the i|i=1,2,...,a′×c′ trained support vector machine classification model is used as the i|i=1,2,...,a′×c′ final support vector machine classification model; otherwise, repeat S23 until f 2i ≥f 1i When |i=1,2,...,a′×c′, the i-th|i=1,2,...,a′×c′ final support vector machine classification model is obtained; S224. Replace the a′×c′ initial support vector machine classification models with a′×c′ final support vector machine classification models, that is, each calling interface in the support vector machine classification model calling interface matrix e′ corresponds to a final support vector machine classification model.

5. A fertilizer application ratio method according to claim 1, characterized in that: The S3 comprises the following steps: S31, construct a sparrow population; set the size of the sparrow population to be f′, and the sparrow population is expressed as represents the i-th sparrow in the sparrow population; setting the search space dimension of the sparrow population to two dimensions, the maximum number of iterations to g1′ and the current number of iterations to g′2; and then setting the initial position set of the sparrow population; S32, setting the matrix of crops to be cultivated according to the maximum number of iterations g1′ Setting the crop matrix to be cultivated Each type of crop to be cultivated is a i ; S33, setting the fitness function of the sparrow population S34, start iteration, and use the fitness function of the sparrow population in each iteration process The fitness of each sparrow is calculated, and the position of the individual sparrow corresponding to the best fitness is selected; then the position of each sparrow in the sparrow population is updated according to the position of the individual sparrow corresponding to the best fitness; when g′2≥g1′, the iteration is stopped, and the final position set of the sparrow population is obtained. and Respectively represent the one-dimensional component and the two-dimensional component of the final position of the i-th sparrow in the sparrow population; S35, and The optimum mass values ​​of nitrogen compounds and potassium compounds in fertilizers are used respectively; and the optimum fertilizer application ratio is obtained according to the optimum mass values ​​of nitrogen compounds and potassium compounds.

6. A fertilizer application ratio method according to claim 5, characterized in that: The step S3 includes the following steps: Setting the initial position set of the sparrow population in S31 includes the following steps: S311. Set the mass value range of nitrogen compounds in fertilizer to Recorded as the first value interval; the mass value interval of potassium compounds in fertilizer is Recorded as the second value interval; and They represent the lower and upper limits of the mass of nitrogen compounds, respectively. and They represent the lower and upper limits of the mass of potassium compounds respectively; S312: According to the first value interval And the second value interval Initialize the position of each sparrow in the sparrow population to obtain an initial position set.

7. A fertilizer application ratio method according to claim 5, characterized in that: The S33 comprises the following steps: S331, in the current round of iteration, according to the data collection time point matrix Collect image sets and greenhouse gas emission data of the crops to be cultivated at various data collection time points to obtain an image matrix h of the crops to be cultivated and a greenhouse gas emission data set; S332, extracting feature data from each image in the image matrix of the crops to be cultivated according to the extracted feature type set, to obtain a feature data set matrix of the crops to be cultivated; S333, calling the support vector machine classification model to call the final support vector machine classification model corresponding to each interface in the ith row of the interface matrix to perform classification operations on each feature data set in the feature data set matrix of the crop to be cultivated, and obtaining a growth quality grade data set of the crop to be cultivated; S334. Constructing a fitness function of the sparrow population according to the cultivated crop growth quality grade dataset and the greenhouse gas emission dataset.

8. A fertilizer application ratio method according to claim 7, characterized in that: The S334 comprises the following steps: S3341, calculating average growth quality grade data according to the cultivated crop growth quality grade data set; and then calculating average greenhouse gas emission data according to the greenhouse gas emission data set; S3342, setting a growth quality level threshold l1' and a greenhouse gas emission threshold l2'; constructing a fitness function of the sparrow population according to the growth quality level threshold l1', the greenhouse gas emission threshold l2', the average growth quality level data and the average greenhouse gas emission data 9. A system for implementing the fertilizer application ratio method according to any one of claims 1 to 8, characterized in that: It includes a crop image acquisition module, a crop image feature extraction module, a final support vector machine classification model construction module and a fertilizer application ratio adjustment optimization module.

10. Application of the fertilizer application ratio method according to any one of claims 1 to 8 in reducing agricultural greenhouse gas emissions.

Images