CN114391351A

CN114391351A - Variable fertilization decision method and device

Info

Publication number: CN114391351A
Application number: CN202111642208.XA
Authority: CN
Inventors: 宋晓宇; 杨贵军; 李伟国; 龙慧灵; 冯海宽; 孟炀
Original assignee: Research Center of Information Technology of Beijing Academy of Agriculture and Forestry Sciences
Current assignee: Research Center of Information Technology of Beijing Academy of Agriculture and Forestry Sciences
Priority date: 2021-12-29
Filing date: 2021-12-29
Publication date: 2022-04-26

Abstract

The invention provides a variable fertilization decision method and a device, wherein the method comprises the following steps: clustering the target area according to the yield data to obtain a plurality of partitions; determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil; determining the target fertilizing amount of each subarea according to the target nutrient amount and the fertilizing mode; and generating a vector diagram of the management subarea according to the target fertilizing amount of each subarea, and using the vector diagram to guide the operation of the agricultural machinery. According to the method, the target area is clustered through the yield data to obtain a plurality of subareas, and the target nutrient quantity is determined according to the target yield and the relevant parameters, so that the yield potential and the nutrient comprehensive management requirements of crops are considered, and meanwhile, variable fertilization decision is carried out on different management subareas of a farmland according to the spatial difference of farmland soil nutrients, so that the cost can be saved and the crop growth efficiency can be improved.

Description

Variable fertilization decision method and device

Technical Field

The invention relates to the field of agricultural fertilization, in particular to a variable fertilization decision method and a variable fertilization decision device.

Background

The fertilizer industry is a typical high-energy-consumption industry, and raw materials and fuels required by the production of nitrogen fertilizers accounting for more than seven percent of the total amount of the fertilizer are directly from energy sources. The fertilizer is seriously wasted, and the agricultural cost saving, quality improvement and efficiency improvement can be really realized only by comprehensively applying the intelligent agricultural machinery equipment technology according to the growth space difference of farmland soil and crops and carrying out variable fertilization decision aiming at different operation units of the farmland. However, how to couple the obtained farmland soil, crop growth and management partition information to construct an accurate fertilization decision model and actually fall the variable fertilization technology to the ground is urgent to be deeply researched.

At present, the most widely applied method is a soil testing and formulated fertilization method, which is a metering fertilization technology developed on the basis of soil fertility chemistry, and a fertilization suggestion is provided through the determination of available nutrients of soil. The method comprises the steps of dividing soil fertility into a plurality of grades according to the soil fertility, taking an area with equal fertility as a formula area, and estimating the more appropriate fertilizer type and the fertilizing amount in the whole formula area according to the soil nutrient test result and the field test result of the area.

However, this method requires a large scale of field-by-field sampling tests for soil nutrients, and requires a lot of manpower and material resources. Therefore, in practical popularization, several soil nutrient data are often tested selectively on one group of farmlands in one village to represent the fertility level of one group of farmlands in one village, and only one sample is generally taken for a single field, so that the information of soil nutrients among all the fields and inside the fields cannot be mastered. In addition, because the soil nutrient test data of one year are used for many years, the annual change of soil nutrients is ignored; errors and waste in the last kilometer in the soil testing formula fertilization can be caused, and the targeted fertilization decision can not be really realized.

Disclosure of Invention

Aiming at the problems in the prior art, the invention provides a variable fertilization decision method and a variable fertilization decision device.

The invention provides a variable fertilization decision method, which comprises the following steps: clustering the target area according to the yield data to obtain a plurality of partitions; determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil; determining the target fertilizing amount of each subarea according to the target nutrient amount and the fertilizing mode; and generating a vector diagram of the management subarea according to the target fertilizing amount of each subarea, and using the vector diagram to guide the operation of the agricultural machinery.

According to the variable fertilization decision method of one embodiment of the present invention, the clustering a target area according to yield data to obtain a plurality of partitions includes: calculating fuzzy effect indexes and normalized classification entropies of partition results with different partition quantities according to the output data, and determining the quantity of target partitions according to the fuzzy effect indexes and the normalized classification entropies; correspondingly, clustering the target area according to the yield data to obtain a plurality of partitions, specifically comprising: and clustering the target area according to the yield data and the number of the target partitions to obtain a plurality of partitions corresponding to the number of the target partitions.

According to the variable fertilization decision method, the determining of the number of target partitions according to the fuzzy effect index and the normalized classification entropy includes: and determining the number of the partitions when the sum of the fuzzy effect index and the normalized classification entropy is minimum as the number of the target partitions.

According to the variable fertilization decision method of an embodiment of the present invention, after clustering the target area according to the yield data to obtain a plurality of partitions, the method further includes: and determining an optimal filtering window and filtering times based on mode filtering according to the clustering partitioning result, and adjusting the clustering result of the partition based on the optimal filtering window and filtering times.

According to the variable fertilization decision method, the determining of the target fertilization amount of each partition according to the target nutrient amount and the fertilization mode comprises the following steps: and determining the target fertilizing amount according to the target nutrient amount and the content of the nutrients which can be provided by the soil under the condition of determining that the crop is sowed at one time by variable decision, and combining the fertilizer utilization rate.

According to the variable fertilization decision method, the determining of the target fertilization amount of each partition according to the target nutrient amount and the fertilization mode comprises the following steps: and determining that the base fertilizer is evenly applied when the crops are sown, and determining the target fertilizing amount according to the target nutrient amount, the nutrient content of the applied base fertilizer and the nutrient content available in the soil and the utilization rate of the fertilizer under the condition of variable decision-making application of the top dressing for the key growth period of the crops.

According to the variable fertilization decision method, the determining of the target fertilization amount of each partition according to the target nutrient amount and the fertilization mode comprises the following steps: determining that the variable decision-making of the time-base fertilizer is applied during crop sowing, and under the condition that the variable decision-making of the top dressing is applied during the key growth period of the crops: determining the fertilizing amount of the base fertilizer according to the target nutrient amount, the content of the nutrients which can be provided by the soil in the sowing period and the current-season utilization rate of the fertilizer; determining the crop topdressing amount according to the target nutrient amount, the nutrient content of the applied base fertilizer amount and the nutrient content which can be provided by the soil in the crop topdressing growth period, and combining the in-season utilization rate of the fertilizer; wherein the target fertilizing amount comprises a base fertilizer fertilizing amount and a crop topdressing amount.

The invention also provides a variable fertilization decision device, comprising: the target partitioning module is used for clustering the target area according to the yield data to obtain a plurality of partitions; the nutrient determining module is used for determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil; the fertilization determining module is used for determining the target fertilization amount of each subarea according to the target nutrient amount and the fertilization mode; and the data storage module is used for generating a management partition vector diagram according to the target fertilizing amount of each partition and guiding the agricultural machinery to operate.

The invention also provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the steps of any one of the variable fertilization decision methods.

The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the variable fertilization decision method as described in any one of the above.

According to the variable fertilization decision method and the device, the target area is clustered through the yield data to obtain a plurality of partitions, the yield data and the soil nutrients have an incidence relation, so that the information of the soil nutrients among all plots and inside the plots can be effectively mastered, the fertilization amount can be changed according to the yield data in different periods, the annual change of the soil nutrients is fully considered, the target nutrient amount is determined according to the target yield and relevant parameters, the crop yield potential and the comprehensive nutrient management requirement are considered, and the variable fertilization decision is performed on different management partitions of a farmland according to the spatial difference of the soil nutrients of the farmland, so that the cost can be saved and the crop growth efficiency can be improved.

Drawings

In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.

Fig. 1 is one of the flow diagrams of a variable fertilization decision method provided by the present invention;

fig. 2 is a second schematic flow chart of a variable fertilization decision method provided by the present invention;

FIG. 3 is a schematic diagram of the changes of FPI and NCE of FCM clustering result provided by the present invention;

FIG. 4 is a schematic diagram of the variation of the spatial consistency index of the mode filtering provided by the present invention;

FIG. 5 is a schematic diagram of the variation of the partition fragmentation index provided by the present invention under different smoothing windows;

FIG. 6a is a schematic diagram illustrating changes in spatial fragmentation parameters after multiple filtering of FCM cluster partition results according to the present invention;

FIG. 6b is a second schematic diagram illustrating the variation of spatial fragmentation parameter after multiple filtering of FCM clustering partition results provided by the present invention;

FIG. 7 is a schematic diagram of the optimal partitioning result provided by the present invention;

FIG. 8 is a schematic diagram of the target yield for the optimal partitioning result provided by the present invention;

fig. 9 is a schematic structural diagram of a variable fertilization decision device provided by the invention;

fig. 10 is a schematic structural diagram of an electronic device provided by the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

In recent years, research on precise variable fertilization technology and matched machinery has become the key point in the field of international agricultural production, and related machinery is developing towards large-scale, informatization, automation and intellectualization. At present, a relatively complete information network is established, the data of crop yield, soil moisture content, nitrogen, phosphorus and potassium fertilizer amount and the like over the years are input into a precise agricultural network, and a variable fertilization technology is adopted to precisely allocate the fertilizer application amount, so that the fertilizer utilization rate is effectively improved, and the environmental pollution is reduced.

Scientific and effective nutrient management can not only improve crop yield, fertilizer utilization efficiency and economic benefits, but also reduce environmental pollution. Aiming at the difficult problem of the current precision agricultural variable fertilization decision, the invention aims at researching the plot, utilizes the annual yield data of crops to perform management partition division, obtains the soil nutrient data of the crop at the key fertilization period, performs nitrogen, phosphorus and potassium topdressing variable fertilization decision in the farmland range based on a farmland nutrient balance model, and realizes the simplified landing of the precision agricultural variable decision technology so as to meet the requirement of guiding agricultural production by the precision agricultural large-area application.

The invention provides a variable fertilization decision method and a device based on combination of yield data and soil nutrient data, aiming at the problems of poor timeliness and no difference in fertilization decision existing in the current soil testing formula fertilization, particularly in the soil fertility partition (level) formula fertilization work.

The variable fertilization decision method and apparatus of the present invention are described below with reference to fig. 1-10. Fig. 1 is a schematic flow chart of a variable fertilization decision method provided by the present invention, and as shown in fig. 1, the present invention provides a variable fertilization decision method, including:

101. and clustering the target area according to the yield data to obtain a plurality of partitions.

Crop yield data (crop yield measurements) are determined prior to 101. The traditional crop yield measurement is generally based on manual acquisition, a plurality of yield measurement sample parties are set in a farmland, and the yield data of the sample parties are acquired in the crop maturity stage to represent the yield level of the plot. In recent years, more and more production-planning harvesters assist in acquiring crop yield data at a field scale. The yield measuring system is assembled on the yield measuring harvester and used for collecting crop yield data, and generally comprises a global positioning system, a grain flow sensor, a net grain elevator shaft speed sensor, an advancing speed sensor, a grain sensor, a header height potentiometer, a data card, graphic software and other modules.

The main factors influencing the accuracy of the yield data acquired by the yield measuring harvester mainly comprise the flow measured by the sensors of grains and the accuracy of the field moving speed of the harvester. The yield abnormal value is processed by adopting the following ideas: 1) the abnormal value is directly deleted, and the subsequent analysis is not participated in; 2) replacing the minimum value in the special values with the minimum value of the normal data, and replacing the maximum value with the maximum value of the normal data; 3) all outliers were replaced with the average of the normal data.

In 101, yield data is used for management zoning, which is an economical and efficient method for accurate agricultural variable input, and the zoning manages the fields with different spatial heterogeneity characteristics according to the spatial variability based on various factors limiting crop yield. The management partitioning algorithm is mainly a clustering algorithm, and common clustering algorithms comprise a K-means (K-means), a Fuzzy C-means (FCM), an iterative self-organizing data analysis technology algorithm (ISODATA) and the like.

102. And determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil.

According to different crop management needs, the following three forms of fertilization modes can be generally selected: firstly, the method comprises the following steps: when crops are sowed, fertilizers required by crop growth are used as base fertilizers and are applied at one time by adopting variable decision; secondly, the method comprises the following steps: when crops are sowed, one part of the fertilizer is uniformly applied as a base fertilizer; the other fertilizers are applied in the crop key growth period after fertilization in a variable decision manner; thirdly, the method comprises the following steps: when crops are sown, one part of the fertilizer is used as a base fertilizer and is applied by adopting a variable decision; and the other fertilizers are applied in the key growth period of the crops through topdressing variable decision.

According to the result of the partition division of the yield data management, according to the three fertilization decision modes, selecting crop fertilization key periods (crop seeding period and crop topdressing key period), carrying out soil nutrient data acquisition in different management partitions, wherein the soil sampling depth is 1-20cm, taking the sampling center of each sampling point as the center of a circle, taking 3 soil samples in total within the range of 5 m of the radius, the interval between every two samples is not less than 2m, the total soil samples of the sampling point are obtained by adding 3 soil samples together, the soil samples are used for testing other soil indexes (quick-acting potassium, effective phosphorus, alkaline-hydrolyzable nitrogen, and the like) after being air-dried and sieved in an oven, and a sampling point is accurately positioned by utilizing a hand-held Differential Global Positioning System (DGPS).

The calculation method of the target nutrient amount comprises the following steps:

wherein z is the target nutrient amount (kg) required by the crop; y is the target yield (kg) of the crop; c₁The amount of nutrients (kg) required for an economic yield of 1 kg per production; c₂The utilization rate of a certain nutrient fertilizer; and x is the nutrient content provided by the soil in the crop fertilization period and is generally determined according to the soil available nutrient index.

The target yield is the yield of crops expected to be achieved on a certain soil, and an estimation method is adopted, namely the average yield of the last years is used as the base number, and on the basis, the target yield is increased by 20% -80%.

The amount of nutrients (C) required for each 1 kg of crop seeds produced₁) And is a relative value that varies depending on the kind of crop, the type of soil, and the level of yield. On the same crop and on the same soil type, the value decreases instead as the yield level increases. Crop nutrient requirements are generally related to environmental conditions, cultivation techniques, and particularly yield levels. With the increase of the crop yield, the absorption amount of nitrogen, phosphorus and potassium is correspondingly increased. Due to the selective absorption of nutrients by crops and the relatively stable chemical structure of crop tissues, the absorption amount of nutrients per yield of crops can vary within a certain range. In general, the nutrient requirement per unit yield is often considered as a constant in the study. The general rule is that about 3 kg of pure nitrogen, 1-1.5 kg of phosphorus and 2-4 kg of potassium are needed for producing about 100 kg of wheat grains, wherein the ratio of nitrogen, phosphorus and potassium is 3:1: 3. Generally, the determination can be carried out by field tests and plant analysis.

Utilization ratio of nutrient fertilizer (C)₂) The parameter refers to the utilization of a certain fertilizer applied by the current season crop. It is mainly influenced by the variety of the fertilizer, the type of the crop, the type of the soil, the level of the yield andthe influence of factors such as annual climate change and the like. The availability of certain fertilizers on certain crops and fertilizer varieties decreases with increasing yield levels and soil fertility. For example, the utilization rates of nitrogen fertilizer, phosphate fertilizer and potassium fertilizer for wheat are 32%, 19% and 44% respectively, the utilization rates of nitrogen fertilizer, phosphate fertilizer and potassium fertilizer for rice are 35%, 25% and 41% respectively, and the utilization rates of nitrogen fertilizer, phosphate fertilizer and potassium fertilizer for corn are 32%, 25% and 43% respectively.

103. And determining the target fertilizing amount of each subarea according to the target nutrient amount and the fertilizing mode.

According to different crop management needs, the following three forms of fertilization modes can be generally selected: firstly, the method comprises the following steps: when crops are sowed, fertilizers required by crop growth are used as base fertilizers and are applied at one time by adopting variable decision; secondly, the method comprises the following steps: when crops are sowed, one part of the fertilizer is uniformly applied as a base fertilizer; the rest fertilizers are used as top dressing variable decisions to be applied in the key growth period of crops; thirdly, the method comprises the following steps: when crops are sown, one part of the fertilizer is used as a base fertilizer and is applied by adopting a variable decision; and the other fertilizers are applied in the key growth period of the crops through topdressing variable decision. And respectively calculating the target fertilizing amount by considering the three conditions.

104. And generating a vector diagram of the management subarea according to the target fertilizing amount of each subarea, and using the vector diagram to guide the operation of the agricultural machinery.

And converting the grid graph of the research plot clustering partitioning result obtained by the process into vector data to generate a management partitioning vector graph. And adding the data of different partition fertilizing amounts obtained by the calculation in the steps into corresponding fields of the yield partition vector file, and generating a variable fertilizing decision file required by agricultural machinery operation so as to guide the field fertilizing operation of the agricultural machinery.

According to the variable fertilization decision method, the target area is clustered through the yield data to obtain a plurality of partitions, the yield data and the soil nutrients have an incidence relation, so that the information of the soil nutrients among all plots and inside the plots can be effectively mastered, the fertilization amount can be changed according to the yield data in different periods, the annual season change of the soil nutrients is fully considered, the target nutrient amount is determined according to the target yield and relevant parameters, the crop yield potential and the comprehensive nutrient management requirements are considered, and the variable fertilization decision is performed on different management partitions of a farmland according to the spatial difference of the soil nutrients of the farmland, so that the cost can be saved and the crop growth efficiency can be improved.

In one embodiment, the clustering the target area according to the production data to obtain a plurality of partitions includes: calculating fuzzy effect indexes and normalized classification entropies of partition results with different partition quantities according to the output data, and determining the quantity of target partitions according to the fuzzy effect indexes and the normalized classification entropies; correspondingly, clustering the target area according to the yield data to obtain a plurality of partitions, specifically comprising: and clustering the target area according to the yield data and the number of the target partitions to obtain a plurality of partitions corresponding to the number of the target partitions.

In the embodiment of the invention, when the yield data is used for managing the partitions, a Fuzzy C-means (FCM) clustering algorithm is adopted, and the appropriate number of partitions is determined according to the spatial distribution and variation condition of the yield data participating in the partitions and the actual area size of a research area (namely a target area). FCM clustering is the most classical one in a clustering algorithm, the core principle of the FCM clustering is shown in a formula, and a cost function J represents the sum of squares of errors of an integral data set.

Wherein m is a flexible parameter of a control algorithm, C is a cluster number, and j is a class label; u. of_ijRepresents a sample x_iMembership belonging to class j. xi denotes the ith sample, and x is a sample with d-dimensional features. c. C_jIs the center of class j, also having d dimension. I | L | can be any metric that represents data similarity (distance), most commonly the euclidean norm (also known as the euclidean norm, L2 norm, euclidean distance). The FCM algorithm is a method for continuously and iteratively calculating membership degree u_ijAnd a cluster center c_jUntil they reach the optimum.

Where for a single sample xi, its sum of membership degrees for each cluster is 1. The termination condition of the iteration is:

where k is the number of iteration steps and μ is the error threshold. The above formula means that, after iteration is continued, the membership degree does not change greatly, that is, the membership degree is considered to be unchanged, and a better (local optimal or global optimal) state is achieved.

In the process of cluster analysis, the most suitable number of partitions in the clustering result can be judged by calculating two evaluation indexes, namely Fuzzy Performance Index (FPI) and Normalized Classification Entropy (NCE).

The FPI represents how many different classes a data point can be shared in a cluster, and can be used for judging the separation degree of each class from the other classes in a clustering result, the value range of the FPI calculation result is between [0-1], and the lower the value is, the data point can be shared by fewer classes, namely, the higher the separation degree between different classes is, the larger the difference is; conversely, a higher FPI value means that a data point is shared by more classes, and the separation degree between different classes is low and the difference is smaller.

Wherein n is the number of data, m is the number of categories, mu_ij(1. ltoreq. i.ltoreq.n, 1. ltoreq. j.ltoreq.k) represents the ith sample X in the data matrix X_iTo the jth cluster center c_jK is polyThe number of classes.

In the embodiment of the present invention, the data matrix X is the yield data of the region under study, n is the data number of the yield data set, k is the category number of the yield data (at least 2, at most n is not more than the data number of the data set), μ_ijIs to represent a sample x in a fuzzy C-means clustering process_iTo the clustering center c_jDegree of membership (i.e. x)_iBelong to c_jProbability of (d).

NCE represents the degree of corruption of data organization (or similarity) due to a data set being classified into different categories. The NCE value is between [0-1], and the lower value indicates that the data set in each class has high homogeneity after clustering, and the variance in the class is smaller than that of the whole plot; conversely, higher values represent greater variance and lower homogeneity.

Wherein n is the number of data, k is the number of categories, mu_ij(1. ltoreq. i.ltoreq.n, 1. ltoreq. j.ltoreq.k) represents the ith sample X in the data matrix X_iTo the jth cluster center c_jA membership value of.

In one embodiment, the determining the number of target partitions according to the fuzzy effect index and the normalized classification entropy includes: and determining the number of the partitions when the sum of the fuzzy effect index and the normalized classification entropy is minimum as the number of the target partitions.

Specifically, referring to the following example of the present invention, the number of partitions when the sum of the two parameters is minimum is determined as the reference number of clusters by considering both of the two parameters.

In an embodiment, after clustering the target area according to the yield data to obtain a plurality of partitions, the method further includes: and determining an optimal filtering window and filtering times based on mode filtering according to the clustering partitioning result, and adjusting the clustering result of the partition based on the optimal filtering window and filtering times.

Considering that the clustering and partitioning results have certain spatial fragmentation and islanding phenomena, it is necessary to perform partitioning post-processing on the clustering results. According to the method, a mode filtering (Majority Analysis) method is adopted to smooth classified results, the continuity of each management partition after classification is increased, fragmentation is reduced, the partition results are evaluated through space consistency indexes (average value change rate, standard deviation change rate, variation coefficient change rate) and space fragmentation indexes (patch density PD, core area TCA, average core area MCA and aggregation index AI), an optimal filtering window and filtering times are determined, division and extraction of accurate agricultural management partitions are finally completed, and final clustering partition results of research plots are obtained.

In one embodiment, the determining the target fertilizing amount of each partition according to the target nutrient amount and the fertilizing mode comprises: and determining the target fertilizing amount according to the target nutrient amount and the content of the nutrients which can be provided by the soil under the condition of determining that the crop is sowed at one time by variable decision, and combining the fertilizer utilization rate.

Under the condition that all fertilizers required by crops are applied once during crop seeding, the dosage of nitrogen fertilizers, phosphorus fertilizers and potassium fertilizers required by crop topdressing or base fertilizers can be respectively calculated according to the following formula aiming at different target yields:

the amount of fertilizer applied to the crops is (the amount of nutrients required by the target yield of the crops-the content of nutrients which can be provided by the soil in the sowing period of the crops)/the utilization rate of the fertilizer in season.

In one embodiment, the determining the target fertilizing amount of each partition according to the target nutrient amount and the fertilizing mode comprises: and determining that the base fertilizer is evenly applied when the crops are sown, and determining the target fertilizing amount according to the target nutrient amount, the nutrient content of the applied base fertilizer and the nutrient content available in the soil and the utilization rate of the fertilizer under the condition of variable decision-making application of the top dressing for the key growth period of the crops. Specifically, the fertilization mode is that the time base fertilizer is uniformly applied during crop seeding, the top dressing variable decision of the key growth period of the crops is applied, and the target fertilization amount is calculated according to the following formula.

The target fertilizing amount (here, the crop topdressing amount) is (the amount of nutrients required for the target yield of crops-the nutrient content of the applied base fertilizer-the nutrient content of the soil in the key growth period of the crops)/the current-season utilization rate of the fertilizer.

According to the formula, the dosage of nitrogen fertilizer, phosphorus fertilizer and potassium fertilizer required by the crop topdressing or the base fertilizer is respectively calculated according to different target yields.

In one embodiment, the determining the target fertilizing amount of each partition according to the target nutrient amount and the fertilizing mode comprises: determining that the variable decision-making of the time-base fertilizer is applied during crop sowing, and under the condition that the variable decision-making of the top dressing is applied during the key growth period of the crops: determining the fertilizing amount of the base fertilizer according to the target nutrient amount, the content of the nutrients which can be provided by the soil in the sowing period and the current-season utilization rate of the fertilizer; determining the crop topdressing amount according to the target nutrient amount, the nutrient content of the applied base fertilizer amount and the nutrient content which can be provided by the soil in the crop topdressing growth period, and combining the in-season utilization rate of the fertilizer; wherein the target fertilizing amount comprises a base fertilizer fertilizing amount and a crop topdressing amount. The method comprises the following specific steps:

the amount of the base fertilizer applied to the crops is (the amount of nutrients required by the target yield of the crops-the content of the nutrients which can be provided by the soil in the sowing period of the crops)/the utilization rate of the fertilizer in the season.

The amount of the top dressing of the crops is (the amount of nutrients required by the target yield of the crops-the nutrient content of the applied base fertilizer amount-the nutrient content of the soil in the growth period of the top dressing of the crops)/the utilization rate of the fertilizer in the season.

With reference to the foregoing embodiments, fig. 2 is a second schematic flow chart of the variable fertilization decision method provided by the present invention, and a specific example is now described. In this example, because the distribution of the yield data is relatively dense, a direct culling method was chosen to handle outliers, and a specific culling method is as follows.

1) Based on years of wheat planting experience and yield harvest, wheat yields of 750 kg/ha and 7500 kg/ha were considered as reasonable yield data, while yield data points outside this range were considered as outlier data to be filtered.

2) The forward speed of the combine harvester is taken as a target, the speed corresponding to each yield point is obtained by dividing the distance by a fixed sampling interval, the speed within the range of +/-3 times of standard deviation of the speed average value of all yield points is determined as a reasonable forward speed according to a 3 delta method, and the yield points higher than or lower than the speed range are considered as abnormal data to be filtered; in addition, in combination with the actual working condition of the harvester, if the speed change between two adjacent yield points exceeds 20%, the sudden stop or rush operation of the harvester is judged, and the yield points are also considered as abnormal data and are removed.

3) Considering the construction of the harvester, there can be filling time errors as well as emptying time errors at the time of harvesting. During the harvesting process, the flow rate of the grains is gradually increased from the starting of the harvester (namely, the harvesting is started) until the grains are kept in a stable state after the filling is finished, and the period of time is called the filling time; at the end of harvesting, the grain flow rate will gradually decrease as the harvester stops harvesting until all the grain is emptied, a period called the emptying time, during which the flow rate of grain detected by the sensor is not reasonable, since the harvester has stopped working at this time, this part is invalid data and needs to be rejected.

The three outlier elimination methods described above can be implemented by ArcGIS and SPSS/Excel, and the present invention chooses to process all yield data by ArcGIS10.1 and SPSS 22. According to the error analysis result, abnormal value processing is respectively carried out on the yield data of the four years of 2013, 2014, 2016 and 2017, and the result shows that 7% -11% of data points are removed. Considering that the wheat varieties planted in the plot are different every year and some years may be affected by non-human factors such as natural disasters, the corresponding annual overall yield fluctuation range may have difference, and the annual overall yield difference is not beneficial for managing the subareas, the annual yield data needs to be standardized to eliminate the difference, namely the yield data of each sampling point is divided by the average value of all the yield data of the year, and the standardized result is shown in table 1.

TABLE 1

According to the results shown in table 1, the four-year yield mean and median values were not substantially changed before and after the treatment, while the maximum and minimum values were largely changed because a large number of abnormal values (including a large number of maximum and minimum values) were removed, so that the maximum value was entirely small and the minimum value was entirely large, the distribution interval of the yield values was narrow, and the skewness coefficients of the data after the treatment were reduced to different degrees every year, meaning that the distribution of the data set was more close to the normal distribution. The reduction of the standard deviation and the coefficient of variation proves that after the abnormal value is processed, the yield variation generated by the abnormal value is successfully eliminated, and the real variation information of the processed yield data is well reflected.

The discrete yield data points on the continuous grid surface are interpolated using ordinary kriging interpolation. Since the swath of the harvester is 6m and the distance by which the travel speed of the harvester advances within a time interval is generally 6m, the pixel of the interpolated output result is determined to be 6m x 6 m. And according to the best fitting model of the four-year yield data obtained in the previous section, completing corresponding common kriging interpolation, and averaging the interpolation results of the four-year yield data to obtain a four-year average yield data interpolation graph.

FCM cluster analysis is carried out on the average yield data of four years to obtain the FPI and the NCE corresponding to the partition number, as shown in figure 3, it can be seen from the figure that as the partition number increases, the FPI shows a change trend which is reduced firstly and then becomes stable, while the NCE increases firstly and then becomes stable, according to the definitions of the FPI and the NCE, the partition result is better when the values of the FPI and the NCE are smaller, and under the condition that the change trends of the FPI and the NCE are different, the partition number with the minimum sum of the FPI and the NCE is regarded as the optimal partition number, namely the partition number is 5.

In order to remove the small patches and the isolated pixels and make the partitioning result smoother, the invention filters the partitioning result by respectively using rectangular windows with the dimensions of 18, 30, 42, 54, 66, 78 and 90m (equivalent to the dimensions of 3 × 3, 5 × 5, 7 × 7, 9 × 9, 11 × 11, 13 × 13 and 15 × 15 pixels respectively).

For the filtering processing result, statistical analysis is performed by using spatial consistency indicators (including a Mean change rate, a Standard Deviation (STD) change rate, a Coefficient of Variation (CV) change rate) (see fig. 4), and a block total spatial fragmentation indicator (plaque density PD, core area TCA, Mean core area MCA, and aggregation index AI) (see fig. 5).

It can be seen from the trend of the parameters in fig. 4 that the parameters change with the increase of the size of the smoothing window, when the size of the smoothing window changes from the original size to 54m, the influence of the filtering scales in different partitions on the yield Mean value Mean is not large, but the yield standard deviation change rate STD and the coefficient of variation change rate CV increase with the change of the filtering scales first, when the filtering scale is 54m (9 × 9 pixels), the changes of the two parameters tend to be stable, which indicates that the changes of the STD and CV in different partitions converge when the filtering scales increase again.

As can be seen from the trend of the parameters in fig. 5 with the increase of the size of the smoothing window, the magnitude of the change of the three parameters is very large when the smoothing window is changed from the original size to 54m size, where PD is reduced from 18.36 to 1.77, TCA is increased from 2.47 to 18.60, and AI is increased from 55.28 to 90.05; while the size of the smoothing window varies from 9 × 9 to 15 × 15, there are only slight variations in all three parameters, with PD decreasing from 1.77 to 0.83, TCA increasing from 18.60 to 20.74, and AI increasing from 90.05 to 93.29. The optimal smoothing window is selected to have a size of 9 × 9 by integrating the plurality of evaluation indexes.

Observing the results after one-time filtering, the filtering results of each scale, although largely eliminating a large amount of broken small patches and isolated image elements existing in the original partitioning results, still have some unsmooth burrs and isolated image elements at the edges of each partition (especially in the partitioning results with small smooth scales), and influence the continuity of the partitions. It is considered that the partitioning results of FCM clustering are filtered multiple times (from one to five times) over smooth windows of all sizes in an attempt to reduce this occurrence.

And analyzing the spatial crushing degree indexes under different filtering times and the change of parameters in the interval, wherein specific filtering results are shown in fig. 6a and 6 b. The variation trends of the spatial fragmentation parameter and the parameters in the intervals under different filtering times are comprehensively considered, and the fact that when the filtering times are three times, the variation of the spatial fragmentation parameter in a larger range and the variation of the standard deviation, the variation coefficient and the area ratio in each interval can be better balanced is considered, namely, the filtering result is considered to be the best when the filtering times are three times.

In summary, the optimal partition result is determined, and as shown in fig. 7, the processing result is converted from the raster file into a vector file (. shp file) by the ENVI software, and the vector file can be opened and edited by the ArcView, ArcGIS and other software.

According to the method, for the optimal yield partition result, the four-year yield average conditions of the research areas 2013, 2014, 2016 and 2017 are counted for different partitions, and the result is shown in the following table 2.

TABLE 2

According to the data provided by the soil and fertilizer workstation, the nutrient demand of wheat per kilogram at different yield levels in the target area is sorted out, and the nutrient demand is shown in the following table 3.

TABLE 3

Yield horizon (kilogram/mu)	Pure N (kilogram/mu)	Need pure P (kilogram/mu)	Need pure K (kilogram/mu)
				500	0.029	0.015	0.026
450	0.029	0.015	0.025
				400	0.029	0.015	0.024
350	0.029	0.016	0.022
				300	0.029	0.016	0.021

In the 4-month wheat jointing period in 2019, a soil sampling experiment is carried out on the test area, 48 sampling points are in total, and a hand-held Differential Global Positioning System (DGPS) is used for accurately positioning the sampling points so as to ensure that the sampling points are uniformly distributed in the whole test field.

The research area is divided into 5 yield subareas based on the yield data of many years, the target yields of different yield subareas are consistent, but the spatial variation of soil nutrients exists in the same subarea due to different spatial distribution positions of different subareas in the research area. Table 4 shows the results of analysis of soil nutrient data for a total of 15 plots for 5 different plots.

TABLE 4

In the research area of the invention, 120kg/ha of base fertilizer urea (with the N content of 46%) and 30kg/ha of calcium superphosphate (with the P2O5 content of 20%) are applied during wheat sowing. And (3) performing variable top dressing application according to target yield of different subareas and nutrient content conditions of farmland soil in the wheat jointing stage, and calculating the top dressing amount by referring to a second mode (uniform top dressing application during crop seeding and variable top dressing decision application during key crop growth) provided by the invention:

the amount of top dressing of the crops is (the amount of nutrients required by the target yield of the crops-the amount of nutrients contained in the applied base fertilizer-the amount of nutrients available to the soil in the key growth period of the crops)/the utilization rate of the fertilizers in season.

Determining the amount of nutrients required by the target yield of crops:

the nutrient pure element amount required per mu of farmland under different target yields converted from the nutrient demand per kg of wheat of different yield levels in the target area (table 3) is shown in table 5.

TABLE 5

Yield horizon (kilogram/mu)	Pure N (kilogram/mu)	Need P₂O₅(kilogram/mu)	Need K₂O (kilogram/mu)
				333	9.73	5.25	7.31
345	10.07	5.39	7.68
				362	10.58	5.60	8.25
379	11.07	5.81	8.79
				390	11.41	5.94	9.16

In the research area of the invention, 45kg/ha of base fertilizer urea (with the N content of 46 percent), 75kg/ha of calcium superphosphate (with the P2O5 content of 20 percent) and 30kg/ha of potassium sulfate (with the K2O content of 50 percent) are applied during wheat seeding. The nutrient content of the applied base fertilizer is calculated according to the base fertilizer application amount during wheat seeding in the research area:

applied base fertilizer has N content (kg/mu) 45/15X 46% ═ 1.38 (kg/mu)

The content of the applied base fertilizer P (kilogram/mu) is 75/15 multiplied by 20% ═ 1 (kilogram/mu)

The content of K in the applied base fertilizer (kg/mu) is 30/15X 50% ═ 1 (kg/mu)

Calculating the content of nutrients provided by the soil in the key growth period of the crops:

the supply of certain nutrient in the soil is obtained by converting a soil test value, and the equation is as follows:

soil nutrient supply (kg/mu) is equal to the measured value of soil nutrient (mg/kg). times.0.15 times soil available nutrient correction coefficient

Wherein the soil available nutrient correction coefficient is obtained by field tests. Experiments show that the correction coefficient is not a fixed value and has an obvious negative correlation with a soil test value. The invention utilizes a relational expression of a correction coefficient of available nutrients of brown soil and the content of the available nutrients of soil to determine the content of nutrients which can be provided by the soil in a key growth period of crops, wherein:

a. calculating the nitrogen supply amount of the soil:

the soil nitrogen supply (kg/mu) is equal to the soil alkaline hydrolysis nitrogen content (mg/kg) multiplied by 0.15 multiplied by the soil available nitrogen nutrient correction coefficient (%).

In the formula, 0.15 is a conversion coefficient, that is, a multiplier obtained by converting a measured value (mg/kg) of soil nutrients of 0 to 20cm of surface soil (a volume weight of 1.12g/cm3) into kg/mu (unit of nutrient supply amount and fertilizing amount in soil).

The soil available nitrogen nutrient correction coefficient is determined by the relation between the brown soil available nutrient correction coefficient and the soil available nutrient content:

y＝0.8×73.736x^-1.0241

(r＝－0.7346^**，n＝128)

wherein y is the soil available nitrogen nutrient correction coefficient (%), x is the soil alkaline hydrolysis nitrogen content (mg/kg), and 0.8 is the correction coefficient set according to the test area condition.

b. Calculating the phosphorus supply amount of the soil:

the phosphorus supply amount (kg/mu) of the soil is equal to the available phosphorus content (mg/kg) of the soil multiplied by 0.15 multiplied by the available phosphorus nutrient correction coefficient (%) of the soil.

The soil available phosphorus nutrient correction coefficient is determined by the relation between the brown soil available nutrient correction coefficient and the soil available nutrient content:

y＝0.8×100.43x^-1.0962

(r＝－0.8959^**，n＝119)

wherein y is the soil available phosphorus nutrient correction coefficient (%), x is the soil available phosphorus content (mg/kg), and 0.8 is the correction coefficient set according to the test area condition.

c. Calculating the potassium supply amount of the soil:

the soil potassium supply amount is equal to the soil quick-acting potassium content multiplied by 0.15 multiplied by the soil quick-acting potassium nutrient correction coefficient (%).

The soil available potassium nutrient correction coefficient is determined by the relation between the brown soil available nutrient correction coefficient and the soil available nutrient content:

y＝0.6×7.9924x^-0.8318

(r＝－0.9026＊＊，n＝113)

wherein y is the soil available potassium nutrient correction coefficient (%), x is the soil available potassium content (mg/kg), and 0.6 is the correction coefficient set according to the test area condition.

The final soil nutrient correction factors and soil nutrient supply for different management subareas in the research area are shown in table 6.

TABLE 6

Calculating the season utilization rate of the fertilizer:

analysis of a large number of test results shows that the season utilization rate of the fertilizer is obviously related to the wheat yield level and the soil nutrient content. Under different yield levels, the current utilization rate of nitrogen, phosphorus and potassium and soil nitrogen, phosphorus and potassium nutrients have extremely obvious logarithmic negative correlation, the invention utilizes the current utilization rate of the brown soil nitrogen, phosphorus and potassium fertilizers and the content relation of the available nutrients in the soil to calculate the current utilization rate of the fertilizers in different management subareas, and because the target yield of wheat in different management subareas is less than 400 kg, the nitrogen, phosphorus and potassium fertilizers in a research area are calculated as follows:

the utilization rate of the nitrogen fertilizer in season:

the season utilization rate of the nitrogen fertilizer is determined by the relation between the season utilization rate of the brown soil nitrogen fertilizer and the content of available nutrients in soil, and when the wheat yield Y is less than 6000 kilograms per hectare, the season utilization rate of the nitrogen fertilizer is calculated by the following formula:

y＝4.117-0.8467×lnx

(r＝－0.624^**，n＝63)

wherein y is the utilization rate of the nitrogen fertilizer in the season, and x is the content (mg/kg) of alkaline hydrolysis nitrogen in the soil.

b. The utilization rate of the phosphate fertilizer in season:

the season availability of the phosphate fertilizer is determined by the relation between the season availability of the brown soil P2O5 and the content of available nutrients in the soil, when the yield Y of the wheat is less than 6000 kg/hectare:

y＝0.583-0.1144×lnx

(r＝－0.811^**，n＝63)

wherein y is the utilization rate of P2O5 in season, and x is the content (mg/kg) of available phosphorus in soil.

c. Utilization rate of potash fertilizer in season

The season availability of the potash fertilizer is determined by the relation between the season availability of the brown soil K2O and the content of available nutrients in the soil, and when the wheat yield Y is less than 6000 kg/hectare:

y＝2.289-0.3889×lnx

(r＝－0.791＊＊，n＝63)

wherein y is the season utilization rate of K2O, and x is the soil available potassium content (mg/kg)

The season utilization of nitrogen, phosphorus and potassium fertilizers in different management subareas of the research area is shown in table 7.

TABLE 7 season utilization rate of N, P and K fertilizers in different management subareas of research district

Calculating the final fertilizing amount of three elementary fertilizers of nitrogen, phosphorus and potassium:

the final fertilizing amount calculation formulas of the three elementary fertilizers of nitrogen, phosphorus and potassium are as follows:

the top dressing amount of the nitrogen fertilizer (the nitrogen absorption amount of the wheat with the target yield, the nitrogen content of the base fertilizer and the nitrogen supply amount of the soil)/the current season utilization rate of the nitrogen fertilizer.

The topdressing amount of the phosphate fertilizer is (the phosphorus absorption amount of the wheat with the target yield-the phosphorus content of the base fertilizer-the phosphorus supply amount of the soil)/the current utilization rate of the phosphate fertilizer.

The top dressing amount of the potash fertilizer (the potassium absorption amount of the wheat with the target yield-the potassium content of the base fertilizer-the potassium supply amount of the soil)/the current-season utilization rate of the potash fertilizer.

Based on the above related parameter calculation results, the final fertilization decision results of different management partitions in the research area of the present invention are shown in table 8:

TABLE 8

And (3) introducing the fertilizing amount data of different calculated subareas into a management subarea division vector file by utilizing an attribute adding function in ARCVIew software or ArcGIS software, and generating a nitrogen, phosphorus and potassium fertilizer fertilizing prescription map of a research area to provide guidance for agricultural machinery operation.

In the following, the variable fertilization decision device provided by the present invention is described, and the variable fertilization decision device described below and the variable fertilization decision method described above may be referred to correspondingly.

Fig. 9 is a schematic structural diagram of a variable fertilization decision device provided by the present invention, and as shown in fig. 9, the variable fertilization decision device includes: a target partition module 901, a nutrient determination module 902, a fertilization determination module 903, and a data storage module 904. The target partitioning module 901 is configured to cluster the target area according to the yield data to obtain a plurality of partitions; the nutrient determination module 902 is used for determining the target nutrient amount required by the partitioned crops according to the target yield of each partitioned crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil; the fertilization determining module 903 is used for determining the target fertilization amount of each partition according to the target nutrient amount and the fertilization mode; the data storage module 904 is used for generating a vector diagram of the management subarea according to the target fertilizing amount of each subarea, and the vector diagram is used for guiding the operation of the agricultural machinery.

The device embodiment provided in the embodiments of the present invention is for implementing the above method embodiments, and for details of the process and the details, reference is made to the above method embodiments, which are not described herein again.

According to the variable fertilization decision device provided by the embodiment of the invention, the yield data and the soil nutrients have an incidence relation, so that the information of the soil nutrients among all plots and inside the plots can be effectively mastered, the fertilizing amount can be changed according to the yield data at different periods, the annual change of the soil nutrients is fully considered, the crop yield potential and the comprehensive nutrient management requirements are considered, and meanwhile, the variable fertilization decision is carried out on different management partitions of a farmland according to the spatial difference of the soil nutrients of the farmland, so that the cost can be saved and the crop growth efficiency can be improved.

Fig. 10 is a schematic structural diagram of an electronic device provided in the present invention, and as shown in fig. 10, the electronic device may include: a processor (processor)1001, a communication Interface (communication Interface)1002, a memory (memory)1003 and a communication bus 1004, wherein the processor 1001, the communication Interface 1002 and the memory 1003 complete communication with each other through the communication bus 1004. Processor 1001 may call logic instructions in memory 1003 to perform a variable fertilization decision method comprising: clustering the target area according to the yield data to obtain a plurality of partitions; determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil; determining the target fertilizing amount of each subarea according to the target nutrient amount and the fertilizing mode; and generating a vector diagram of the management subarea according to the target fertilizing amount of each subarea, and using the vector diagram to guide the operation of the agricultural machinery.

In addition, the logic instructions in the memory 1003 may be implemented in the form of software functional units and may be stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

In another aspect, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform a method for variable fertilization decision provided by the above methods, the method comprising: clustering the target area according to the yield data to obtain a plurality of partitions; determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil; determining the target fertilizing amount of each subarea according to the target nutrient amount and the fertilizing mode; and generating a vector diagram of the management subarea according to the target fertilizing amount of each subarea, and using the vector diagram to guide the operation of the agricultural machinery.

In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, the computer program being implemented by a processor to perform the variable fertilization decision method provided in the above embodiments, the method including: clustering the target area according to the yield data to obtain a plurality of partitions; determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil; determining the target fertilizing amount of each subarea according to the target nutrient amount and the fertilizing mode; and generating a vector diagram of the management subarea according to the target fertilizing amount of each subarea, and using the vector diagram to guide the operation of the agricultural machinery.

The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.

Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.

Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. A variable fertilization decision method is characterized by comprising the following steps:

clustering the target area according to the yield data to obtain a plurality of partitions;

determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil;

determining the target fertilizing amount of each subarea according to the target nutrient amount and the fertilizing mode;

and generating a vector diagram of the management subarea according to the target fertilizing amount of each subarea, and using the vector diagram to guide the operation of the agricultural machinery.

2. The variable fertilization decision method of claim 1, wherein the clustering target regions according to yield data to obtain a plurality of partitions comprises:

calculating fuzzy effect indexes and normalized classification entropies of partition results with different partition quantities according to the output data, and determining the quantity of target partitions according to the fuzzy effect indexes and the normalized classification entropies;

correspondingly, clustering the target area according to the yield data to obtain a plurality of partitions, specifically comprising:

and clustering the target area according to the yield data and the number of the target partitions to obtain a plurality of partitions corresponding to the number of the target partitions.

3. The variable fertilization decision method of claim 2, wherein the determining a number of target partitions based on the fuzzy effect index and the normalized classification entropy comprises:

and determining the number of the partitions when the sum of the fuzzy effect index and the normalized classification entropy is minimum as the number of the target partitions.

4. The variable fertilization decision method of claim 1, wherein clustering target regions according to yield data to obtain a plurality of partitions further comprises:

and determining an optimal filtering window and filtering times based on mode filtering according to the clustering partitioning result, and adjusting the clustering result of the partition based on the optimal filtering window and filtering times.

5. The variable fertilization decision method of claim 1, wherein determining a target amount of fertilizer to be applied for each zone based on the target nutrient amount and fertilization pattern comprises:

and determining the target fertilizing amount according to the target nutrient amount and the content of the nutrients which can be provided by the soil under the condition of determining that the crop is sowed at one time by variable decision, and combining the fertilizer utilization rate.

6. The variable fertilization decision method of claim 1, wherein determining a target amount of fertilizer to be applied for each zone based on the target nutrient amount and fertilization pattern comprises:

and determining that the base fertilizer is evenly applied when the crops are sown, and determining the target fertilizing amount according to the target nutrient amount, the nutrient content of the applied base fertilizer and the nutrient content available in the soil and the utilization rate of the fertilizer under the condition of variable decision-making application of the top dressing for the key growth period of the crops.

7. The variable fertilization decision method of claim 1, wherein determining a target amount of fertilizer to be applied for each zone based on the target nutrient amount and fertilization pattern comprises:

determining that the variable decision-making of the time-base fertilizer is applied during crop sowing, and under the condition that the variable decision-making of the top dressing is applied during the key growth period of the crops:

determining the fertilizing amount of the base fertilizer according to the target nutrient amount, the content of the nutrients which can be provided by the soil in the sowing period and the current-season utilization rate of the fertilizer;

determining the crop topdressing amount according to the target nutrient amount, the nutrient content of the applied base fertilizer amount and the nutrient content which can be provided by the soil in the crop topdressing growth period, and combining the in-season utilization rate of the fertilizer;

wherein the target fertilizing amount comprises a base fertilizer fertilizing amount and a crop topdressing amount.

8. A variable rate fertilization decision device, comprising:

the target partitioning module is used for clustering the target area according to the yield data to obtain a plurality of partitions;

the nutrient determining module is used for determining the target nutrient amount required by the subarea crops according to the target yield of each subarea crop, the nutrient amount required by unit yield, the utilization rate of nutrient fertilizers and the nutrient content capable of being provided by soil;

the fertilization determining module is used for determining the target fertilization amount of each subarea according to the target nutrient amount and the fertilization mode;

and the data storage module is used for generating a management partition vector diagram according to the target fertilizing amount of each partition and guiding the agricultural machinery to operate.

9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program implements the steps of the variable fertilization decision method of any one of claims 1 through 7.

10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the variable fertilization decision method of any one of claims 1 to 7.