Fertilizer proportion adjusting method based on walnut tree growth stage

Abstract

The invention relates to a fertilizer ratio adjusting method based on the growth stage of walnut trees, and belongs to the technical field of walnut tree fertilization. The method comprises the following steps: determining fertilizer demand factors and environmental factors according to the category of target walnut trees; importance factors are screened from the fertilizer demand factors and the environmental factors according to historical planting data, and the growth weight of each importance factor is calculated; constructing a multi-growth-stage function according to each importance factor and the growth weight thereof; according to the multi-growth-stage function of the target walnut tree, calculating a growth demand coefficient of an important fertilizer demand factor in the current growth stage, and determining an original fertilizer ratio in the current growth stage; and predicting the future yield based on the original fertilizer ratio of the current growth stage, and adjusting the original fertilizer ratio of the current growth stage according to the predicted future yield to obtain the optimal fertilizer ratio of the current growth stage. According to the method, the optimal fertilizer ratio of the target walnut tree in the current growth stage and the environment can be accurately determined.

Description

Technical Field

[0001] This invention relates to the field of walnut tree fertilization technology, and in particular to a method for adjusting fertilizer ratios based on the growth stages of walnut trees. Background Technology

[0002] Walnut trees are a type of nut crop with high economic value and rich nutrition. They have a long growth cycle and usually require different fertilizer ratios at different growth stages to achieve high yields.

[0003] Traditional fertilizer application methods often rely on manual experience and fail to fully consider the characteristics of walnut tree growth stages and environmental factors. In addition, the fertilizer requirements of walnut trees vary greatly at different growth stages, and the required ratios of different fertilizers also differ at different growth stages. Failure to consider the characteristics of walnut tree growth stages can easily lead to fertilizer waste or nutrient deficiency, thereby affecting the yield of walnut trees. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method for adjusting fertilizer ratios based on the growth stages of walnut trees. The technical solution of this invention is as follows: A method for adjusting fertilizer ratios based on walnut tree growth stages includes: S1, determine multiple fertilizer requirement factors and multiple environmental factors based on the category of the target walnut tree; S2. Based on historical planting data of multiple batches of reference target walnut trees from seed planting to fruiting period, multiple important factors are screened from multiple fertilizer demand factors and multiple environmental factors, and the growth weight of each important factor is calculated. S3, construct a multi-stage function for the target walnut tree based on each important factor and its growth weight; S4, determine the current growth stage of the target walnut tree, calculate the growth requirement coefficient of the important fertilizer requirement factors in the current growth stage according to the multi-growth stage function of the target walnut tree, and determine the original fertilizer ratio for the current growth stage according to the growth requirement coefficient of the important fertilizer requirement factors. The important fertilizer requirement factors are the fertilizer requirement factors among the importance factors. S5. Based on the original fertilizer ratio at the current growth stage, predict the future yield, compare the future yield with the standard yield, and adjust the original fertilizer ratio at the current growth stage according to the comparison results to obtain the optimal fertilizer ratio at the current growth stage.

[0005] Preferably, S2 includes: S21. Based on fertilizer demand factors and environmental factors, extract multiple fertilizer demand characteristics, multiple environmental characteristics, yield data and growth status data from the historical planting data of multiple batches of reference target walnut trees from seed planting to fruiting period. S22, calculate the correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of each batch of reference target walnut trees, and select multiple important factors from multiple fertilizer requirement factors and multiple environmental factors based on the correlation calculation results; S23. Perform linear analysis on all important factors and growth status data to obtain the linear coefficient of each important factor, and use the linear coefficient of each important factor as its respective growth weight.

[0006] Preferably, S22 includes: S221, Calculate the original correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of each batch of reference target walnut trees according to the correlation algorithm; S222, the original correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of all batches of reference target walnut trees is averaged to obtain the correlation between each fertilizer requirement feature and each environmental feature and yield data. S223, sort all fertilizer demand features and all environmental features in descending order of their correlation with yield data, and select the fertilizer demand factors and environmental factors corresponding to the first preset number of fertilizer demand features and environmental features in the descending order as the preset number of importance factors.

[0007] Preferably, S23 includes: S231, construct a polynomial linear function for each batch of target walnut trees based on all importance factors and the growth status data in the historical planting data of each batch of reference target walnut trees, wherein the original linear coefficients in the polynomial linear function correspond one-to-one with the importance factors; S232, concatenate all the original linear coefficients of the polynomial linear function of all batches of reference target walnut trees to obtain the linear coefficient matrix; S233, normalize the elements in each column of the linear coefficient matrix to obtain the linear coefficient of each importance factor, and use the linear coefficient of each importance factor as its respective growth weight.

[0008] Preferably, the growth stages of the target walnut tree include the budding stage, the trunk growth stage, the fruit development stage, and the fruit ripening stage, and S3 includes: S31. Based on the growth characteristics of each growth stage of the target walnut tree, a linear function is used as the basic function for the budding stage, an exponential function is used as the basic function for the trunk growth stage, a logarithmic function is used as the basic function for the fruit development stage, and a saturation function is used as the basic function for the fruit ripening stage. The independent variables of the basic function for each growth stage are the growth time and importance factors of the target walnut tree, and the dependent variable is the standard growth amount of the target walnut tree. S32, input the growth weight of each important factor into the basic function of each growth stage to obtain the growth requirement function of each growth stage, and connect the growth requirement functions of each growth stage in the order of growth stages to obtain the multi-growth stage function of the target walnut tree.

[0009] Preferably, S4 includes: S41, Determine the current growth stage of the target walnut tree based on its current growth time and basic information; S42, input the current growth time of the target walnut tree, the environmental parameters in the basic information and the measured standard growth into the multi-growth stage function to obtain multiple growth requirement groups of important fertilizer requirement factors of the target walnut tree; S43, input the environmental parameters, each growth requirement group and the current growth time into the multi-growth stage function to obtain the predicted standard growth amount of each growth requirement group, calculate the squared error between the predicted standard growth amount and the measured standard growth amount of each growth requirement group, and select the growth requirement group with the smallest squared error as the growth requirement coefficient of multiple important fertilizer requirement factors of the target walnut tree in the current growth stage. S44, take the growth requirement coefficient of each important fertilizer requirement factor as the ratio of each important fertilizer requirement factor, and take all important fertilizer requirement factors and their ratios as the original fertilizer ratio for the current growth stage of the target walnut tree.

[0010] Preferably, step S43 involves setting environmental parameters. Any growth requirement group and current growth time Input to multi-stage growth function In this process, the predicted standard growth for each growth requirement group is obtained, and the predicted standard growth for that growth requirement group is compared with the measured standard growth. When the square of the error is calculated, it is achieved through formula (1): (1); In formula (1), Represents the regularization coefficient. This indicates the predicted standard growth rate.

[0011] Preferably, S5 includes: S51, fit the average yield of multiple important fertilizer requirements and multiple important environmental characteristics, standard growth, growth time and yield data corresponding to the historical planting data of multiple batches of reference target walnut trees, and obtain the average yield function of the target walnut trees. S52, input the growth requirement coefficients, environmental parameters, growth time and standard growth amount of several important fertilizer requirements of the target walnut tree at the current growth stage into the yield mean function to obtain the future yield of the target walnut tree; S53, calculate the deviation between the future yield and the standard yield. If the deviation is negative, adjust the original fertilizer ratio according to the deviation to obtain the optimal fertilizer ratio. If the deviation is positive or 0, use the original fertilizer ratio as the optimal fertilizer ratio.

[0012] Preferably, step S53, when adjusting the original fertilizer ratio according to the deviation value to obtain the optimal fertilizer ratio, includes: S531, construct a yield loss function based on the standard yield, and calculate the partial derivatives of each important fertilizer demand factor in the original fertilizer ratio based on the yield loss function; S532: Based on the partial derivatives of each important fertilizer demand factor, use the gradient descent method to iteratively update the ratio of each important fertilizer demand factor until the yield loss function converges and stop iterating. The last updated ratio of each important fertilizer demand factor is taken as the optimal fertilizer ratio.

[0013] All of the above-mentioned optional technical solutions can be combined arbitrarily, and the present invention will not provide a detailed description of the structure after each combination.

[0014] By means of the above solution, the beneficial effects of the present invention are as follows: By determining fertilizer requirements and environmental factors based on the type of target walnut trees, and screening multiple important factors based on historical planting data of several batches of reference target walnut trees, and calculating the growth weight of each important factor, a multi-growth stage function for the target walnut trees is constructed based on the important factors and their growth weights. By comprehensively considering the fertilizer requirements and environmental factors of the target walnut trees, and selecting the most important factors among these two types of factors, the multi-growth stage function for the target walnut trees can be constructed based on the important factors. This allows for accurate prediction of the fertilizer requirements of the target walnut trees at each growth stage under the influence of environmental factors, thereby optimizing fertilizer ratios and increasing yield.

[0015] By calculating the growth requirement coefficients of key fertilizer demand factors in the current growth stage based on the multi-stage function of the target walnut tree, and determining the original fertilizer ratio for the current growth stage based on the growth requirement coefficients of key fertilizer demand factors, the future yield is predicted based on the original fertilizer ratio. After comparing the future yield with the standard yield, the original fertilizer ratio is adjusted to obtain the optimal fertilizer ratio. This allows for timely evaluation of the effectiveness of the original fertilizer ratio, enabling the optimization of the original fertilizer ratio based on the predicted future yield. It can accurately determine the optimal fertilizer ratio for the current growth stage and environment of the target walnut tree, avoiding fertilizer waste or nutrient deficiency, and thus increasing the yield of the target walnut tree.

[0016] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for adjusting fertilizer ratios based on the growth stages of walnut trees, as provided in an embodiment of the present invention. Detailed Implementation

[0018] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0019] like Figure 1 As shown, this embodiment of the invention provides a method for adjusting fertilizer ratios based on the growth stages of walnut trees, including S1-S5: S1, determine multiple fertilizer requirement factors and multiple environmental factors based on the category of the target walnut tree.

[0020] Specifically, the target walnut tree refers to a certain type of walnut tree. Fertilizer requirements refer to the fertilizer elements needed for the growth of the target walnut tree, including nitrogen, phosphorus, potassium, and calcium. Environmental factors refer to environmental parameters related to the growth of the target walnut tree, including soil pH, rainfall, temperature, and light.

[0021] S2. Based on historical planting data of multiple batches of reference target walnut trees from seed planting to fruiting period, multiple important factors are selected from multiple fertilizer demand factors and multiple environmental factors, and the growth weight of each important factor is calculated.

[0022] Specifically, reference target walnut trees refer to walnut trees of the same category as the target walnut trees. Multiple batches of reference target walnut trees refer to different batches of reference target walnut trees planted at different times (e.g., different years). Each batch of reference target walnut trees includes multiple reference target walnut trees. Historical planting data includes fertilizer requirements, environmental data, yield data, and growth status data of the reference target walnut trees from seed planting to fruiting.

[0023] In one specific embodiment, S2 includes: S21. Based on fertilizer demand factors and environmental factors, extract multiple fertilizer demand characteristics, multiple environmental characteristics, yield data and growth status data from the historical planting data of multiple batches of reference target walnut trees from seed planting to fruiting period.

[0024] Specifically, fertilizer demand characteristics are standardized values ​​obtained by standardizing fertilizer data collected from the growth process of target reference walnut trees based on fertilizer demand factors. For example, if the data corresponding to potassium, a fertilizer demand factor, is denoted as 'k' during the growth of a batch of target reference walnut trees, then the standardized value of data 'k' corresponding to potassium, calculated using extreme value standardization, is the fertilizer demand characteristic of potassium. Environmental characteristics can be obtained similarly. Growth status data includes disease data, growth defect data, and disaster data. Among them, disease data includes disease type, disease severity, and disease occurrence time; growth defect data includes the degree of stunting or poor leaf development; disaster data includes disaster type, disaster severity, and disaster time. The yield data of a batch of target reference walnut trees is the sum of the yields of all target reference walnut trees included in that batch. It should be noted that, for the convenience of subsequent analysis and calculation, this embodiment of the invention sets each batch of target reference walnut trees to include the same number of target reference walnut trees.

[0025] S22, calculate the correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of each batch of reference target walnut trees, and select multiple important factors from multiple fertilizer requirement factors and multiple environmental factors based on the correlation calculation results.

[0026] Specifically, correlation represents the degree of linear association between fertilizer demand characteristics or environmental characteristics and yield data, ranging from [-1, 1].

[0027] In one specific embodiment, S22 includes: S221, calculate the original correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of each batch of reference target walnut trees according to the correlation algorithm.

[0028] Specifically, in this embodiment of the invention, the correlation algorithm is implemented using the Pearson correlation coefficient algorithm.

[0029] S222, the original correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of all batches of reference target walnut trees is averaged to obtain the correlation between each fertilizer requirement feature and each environmental feature and yield data.

[0030] For example, assuming the original correlations between potassium fertilizer requirement characteristics and yield data for all batches (taking four batches as an example) of the target walnut trees are 0.5, 0.8, 1, and -0.1 respectively, then the correlation between potassium fertilizer requirement characteristics and yield data is 0.55.

[0031] S223, sort all fertilizer demand features and all environmental features in descending order of their correlation with yield data, and select the fertilizer demand factors and environmental factors corresponding to the first preset number of fertilizer demand features and environmental features in the descending order as the preset number of importance factors.

[0032] Specifically, the preset quantity is 10 in this embodiment of the invention. For example, if the first preset quantity of fertilizer demand features and environmental features in the reverse sorting includes 7 fertilizer demand features and 3 environmental features, then the 7 fertilizer demand factors and 3 environmental factors corresponding to these 7 fertilizer demand features and 3 environmental features are respectively regarded as 10 importance factors.

[0033] S23. Perform linear analysis on all important factors and growth status data to obtain the linear coefficient of each important factor, and use the linear coefficient of each important factor as its respective growth weight.

[0034] Specifically, the linear analysis in this embodiment of the invention employs a polynomial linear function.

[0035] In one specific embodiment, S23 includes: S231, construct a polynomial linear function for each batch of target walnut trees based on all importance factors and the growth status data in the historical planting data of each batch of reference target walnut trees, wherein the original linear coefficients in the polynomial linear function correspond one-to-one with the importance factors.

[0036] Specifically, based on the disease data in the growth status data, the ratio of disease occurrence time to the growth time of the reference target walnut tree is calculated as the disease ratio. This disease ratio is then multiplied by the disease severity to obtain the disease value of the reference target walnut tree. Expert review is conducted based on the degree of stunting or poor leaf development in the growth defect data to obtain the growth defect value of the reference target walnut tree. The ratio of disaster time to the growth time of the reference target walnut tree is calculated based on the disaster data in the growth status data. This disaster ratio is then multiplied by the disaster severity to obtain the disaster value of the reference target walnut tree. All types of growth status data are weighted according to preset weights (different preset weights for different disease types and disaster types; the preset weight for growth defect data is 0.3; if disease data, growth status data, or growth defect data are not available, the corresponding preset weight is modified to 0). This yields the growth status value for each batch of reference target walnut trees. The growth status values ​​are then... As the dependent variable, all important factors As independent variables, the fertilizer requirement characteristics and environmental characteristics corresponding to all important factors of each batch of reference target walnut trees are fitted with growth status values ​​to obtain a polynomial linear function for each batch of reference target walnut trees. For example, the polynomial linear function of a certain batch of target walnut trees can be expressed as: ;in, Represents the original linear coefficients.

[0037] S232, concatenate all the original linear coefficients of the polynomial linear function of all batches of reference target walnut trees to obtain the linear coefficient matrix.

[0038] Specifically, in the linear coefficient matrix, each row represents the original linear coefficients of different types of importance factors in the same batch of reference target walnut trees, and each column represents the original linear coefficients of the same type of importance factors in different batches of reference target walnut trees.

[0039] S233, normalize the elements in each column of the linear coefficient matrix to obtain the linear coefficient of each importance factor, and use the linear coefficient of each importance factor as its respective growth weight.

[0040] Specifically, after normalization, the linear coefficients of each importance factor range from [0,1].

[0041] S3, construct a multi-stage growth function for the target walnut tree based on each important factor and its growth weight.

[0042] Specifically, the multi-stage growth function is a function that is constructed based on the four stages of the target walnut tree: budding stage, trunk growth stage, fruit development stage, and fruit ripening stage, and contains four piecewise functions.

[0043] In one specific embodiment, the growth stages of the target walnut tree include the budding stage, the trunk growth stage, the fruit development stage, and the fruit ripening stage, and S3 includes: S31. Based on the growth characteristics of each growth stage of the target walnut tree, a linear function is used as the basic function for the budding stage, an exponential function is used as the basic function for the trunk growth stage, a logarithmic function is used as the basic function for the fruit development stage, and a saturation function is used as the basic function for the fruit ripening stage. The independent variables of the basic function for each growth stage are the growth time and importance factors of the target walnut tree, and the dependent variable is the standard growth amount of the target walnut tree.

[0044] Specifically, during the budding stage, the standard growth of the target walnut tree changes linearly over time, so a linear function is chosen as the basic function for this stage. During the trunk growth stage, the target walnut tree enters a rapid growth period, with rapid growth in the root system and above-ground parts, resulting in a sharp increase in the standard growth; therefore, an exponential function is chosen as the basic function for this stage. During the fruit development stage, the growth rate of the target walnut tree begins to slow down, and the increase in the standard growth becomes stable; therefore, a logarithmic function is chosen as the basic function for this stage. During the fruit ripening stage, the growth of the target walnut tree is close to saturation, and the standard growth will maintain a relatively stable constant value; therefore, a saturation function is chosen as the basic function for this stage.

[0045] The standard growth amount refers to the standardized value obtained by standardizing the growth amount of the target walnut tree at different growth stages. For example, during the budding stage, the growth amount of the target walnut tree at the current growth time (i.e., the measurement time) is calculated by referencing the leaves and plant height of the target walnut tree through experiments or expert measurements. Similarly, the growth amounts at different growth times during the trunk growth stage, fruit development stage, and fruit ripening stage can be obtained. By standardizing the growth amounts at different growth times for all growth stages, the standard growth amount of the target walnut tree at different growth times is obtained. Here, growth time refers to the time difference between the target walnut tree's budding and the current time (i.e., the time for measuring growth).

[0046] Assuming the standard growth of the target walnut tree is The target walnut tree's growth period is The important factors are Then the basic function of the budding stage can be expressed as: ; in, These represent the initial growth weights of each important factor. The growth coefficient for growth time is set to 0.8 by default in this embodiment of the invention (because growth time is the most important factor in the variation of standard growth amount). The basic function of the tree trunk growth stage can be expressed as: ; in, This represents the initial standard growth during the trunk growth stage, where e represents the exponent base. The basic functions of fruit development stages can be expressed as: ;in, This indicates the initial standard growth rate at each stage of fruit development. The basic function of fruit ripening stages can be expressed as: ;in, This represents the initial standard growth rate at the fruit ripening stage. This represents the theoretical upper limit of the standard growth rate. This represents the preset growth scale coefficient. The initial standard growth is obtained based on historical experience values, and represents the preset standard growth that should be achieved at each growth stage.

[0047] S32, input the growth weight of each important factor into the basic function of each growth stage to obtain the growth requirement function of each growth stage, and connect the growth requirement functions of each growth stage in the order of growth stages to obtain the multi-growth stage function of the target walnut tree.

[0048] Specifically, after inputting the growth weight of each importance factor into the basic function of each growth stage, the growth weight will replace the initial growth weight of each importance factor. For example, if the growth weight of potassium among the importance factors is 0.1, then the basic function of each growth stage will replace the initial growth weight of potassium. Replace with 0.1.

[0049] The multi-stage growth function, based on the fundamental function of the above embodiments, can be expressed as: ; in, These represent the growth weights of each important factor. This indicates a conditional statement.

[0050] S4. Determine the current growth stage of the target walnut tree, calculate the growth requirement coefficient of the important fertilizer requirement factors in the current growth stage according to the multi-growth stage function of the target walnut tree, and determine the original fertilizer ratio for the current growth stage according to the growth requirement coefficient of the important fertilizer requirement factors. The important fertilizer requirement factors are the fertilizer requirement factors among the importance factors.

[0051] Specifically, the growth requirement coefficient refers to the value corresponding to the important fertilizer requirement factors in the multi-stage growth function. Taking the multi-stage growth function above as an example, potassium, among the important factors, corresponds to the value in the multi-stage growth function. Then the calculated current growth stage The value is the growth requirement coefficient corresponding to the importance factor potassium.

[0052] In one specific embodiment, S4 includes: S41, determine the current growth stage of the target walnut tree based on its current growth time and basic information.

[0053] Specifically, the basic information for the target walnut tree includes environmental parameters, measured standard growth, and plant condition. Based on the plant condition (leaves, fruit, or branches) and the current growth period, experts determine the current growth stage of the target walnut tree.

[0054] S42 inputs the current growth time of the target walnut tree, the environmental parameters in the basic information, and the measured standard growth into the multi-growth stage function to obtain multiple growth requirement groups of the important fertilizer requirement factors of the target walnut tree.

[0055] Specifically, the current growth time is input into the multi-stage growth function. In the process, the measured standard growth is input into the multi-stage growth function. In the process, environmental parameters such as temperature parameters from the basic information are input into the multi-growth stage function. In the process, the precipitation parameter is input into the multi-stage growth function. In the process, until the last environmental parameter is input; after inputting the current growth time of the target walnut tree, the environmental parameters in the basic information and the measured standard growth amount into the multi-growth stage function, since the multi-growth stage function is a multi-variable function, it can obtain multiple sets of solutions for important fertilizer demand factors, and each set of solutions is a growth demand set.

[0056] S43. Input the environmental parameters, each growth requirement group and the current growth time into the multi-growth stage function to obtain the predicted standard growth amount of each growth requirement group. Calculate the squared error between the predicted standard growth amount and the measured standard growth amount of each growth requirement group. Select the growth requirement group with the smallest squared error as the growth requirement coefficient of multiple important fertilizer requirement factors in the current growth stage of the target walnut tree.

[0057] Specifically, the squared error is obtained by calculating the square of the difference between the predicted and measured standard growth for each growth requirement group. If the difference is used directly, negative and positive differences may cancel each other out, making it impossible to accurately assess the magnitude of the difference for each growth requirement group. Therefore, the square of the difference is chosen to improve its significance. By squaring the difference, all differences are ensured to become positive, thus eliminating the influence of the sign.

[0058] In one specific embodiment, S43 involves setting environmental parameters. Any growth requirement group and current growth time Input to multi-stage growth function In this process, the predicted standard growth for each growth requirement group is obtained, and the predicted standard growth for that growth requirement group is compared with the measured standard growth. When the square of the error is calculated, it is achieved through formula (1): (1); In formula (1), Represents the regularization coefficient. This indicates the predicted standard growth rate.

[0059] Specifically, formula (1) adds a penalty term through regularization coefficient to rationally calculate the squared error between the predicted standard growth and the measured standard growth for each growth demand group.

[0060] S44, take the growth requirement coefficient of each important fertilizer requirement factor as the ratio of each important fertilizer requirement factor, and take all important fertilizer requirement factors and their ratios as the original fertilizer ratio for the current growth stage of the target walnut tree.

[0061] Specifically, after standardizing all growth requirement coefficients, standard coefficients are obtained. All standard coefficients are used to form the ratio of all important fertilizer requirements. For example, if the standard coefficients of the standardized growth requirement coefficients are: potassium 0.2, phosphorus 0.5, and nitrogen 0.3, then the original fertilizer ratio is 20% potassium, 50% phosphorus, and 30% nitrogen.

[0062] S5. Based on the original fertilizer ratio at the current growth stage, predict the future yield, compare the future yield with the standard yield, and adjust the original fertilizer ratio at the current growth stage according to the comparison results to obtain the optimal fertilizer ratio at the current growth stage.

[0063] Specifically, the comparison result refers to the deviation between future output and standard output.

[0064] In one specific embodiment, S5 includes: S51, fit the average yield of multiple important fertilizer requirements and multiple important environmental characteristics, standard growth, growth time and yield data corresponding to the historical planting data of multiple batches of reference target walnut trees, and obtain the average yield function of the target walnut trees.

[0065] Specifically, in this embodiment of the invention, the fitting method selected is the support vector machine regression algorithm. First, the average yield of multiple important fertilizer requirements, multiple important environmental features, standard growth amount, growth time and yield data corresponding to important factors in all batches of reference target walnut trees is mapped to a high-dimensional space to obtain high-dimensional data. Then, an initial mean function is trained based on the high-dimensional data (in this embodiment of the invention, the radial basis function commonly used in the vector machine regression algorithm is selected) to obtain the average yield function.

[0066] S52 inputs the growth requirement coefficients, environmental parameters, growth time and standard growth amount of several important fertilizer requirements of the target walnut tree at the current growth stage into the yield mean function to obtain the future yield of the target walnut tree.

[0067] S53, calculate the deviation between the future yield and the standard yield. If the deviation is negative, adjust the original fertilizer ratio according to the deviation to obtain the optimal fertilizer ratio. If the deviation is positive or 0, use the original fertilizer ratio as the optimal fertilizer ratio.

[0068] Specifically, the standard yield is the pre-harvest yield of fruit determined in advance based on the variety of the target walnut tree.

[0069] In a specific embodiment, step S53, when adjusting the original fertilizer ratio according to the deviation value to obtain the optimal fertilizer ratio, includes: S531, construct a yield loss function based on the standard yield, and calculate the partial derivatives of each important fertilizer demand factor in the original fertilizer ratio based on the yield loss function.

[0070] Specifically, in this embodiment of the invention, the production loss function can be implemented using the mean squared error loss function, based on the production loss function. Calculate a key fertilizer requirement factor in the original fertilizer formulation. When the partial derivative is taken, the formula is: .

[0071] S532: Based on the partial derivatives of each important fertilizer demand factor, use the gradient descent method to iteratively update the ratio of each important fertilizer demand factor until the yield loss function converges and stop iterating. The last updated ratio of each important fertilizer demand factor is taken as the optimal fertilizer ratio.

[0072] Specifically, the ratio of each important fertilizer requirement factor in the original fertilizer formula is subtracted from its partial derivative to obtain the latest ratio. Then, the convergence of the yield loss function is evaluated. If the yield loss function does not converge, the latest ratio is subtracted from its latest partial derivative until the yield loss function converges, at which point the iteration stops. The last updated ratio of each important fertilizer requirement factor is taken as the optimal fertilizer formula.

[0073] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for adjusting fertilizer ratios based on the growth stages of walnut trees, characterized in that, include: S1, determine multiple fertilizer requirement factors and multiple environmental factors based on the category of the target walnut tree; S2. Based on historical planting data of multiple batches of reference target walnut trees from seed planting to fruiting period, multiple important factors are screened from multiple fertilizer demand factors and multiple environmental factors, and the growth weight of each important factor is calculated. S3, construct a multi-stage function for the target walnut tree based on each important factor and its growth weight; S4, determine the current growth stage of the target walnut tree, calculate the growth requirement coefficient of the important fertilizer requirement factors in the current growth stage according to the multi-growth stage function of the target walnut tree, and determine the original fertilizer ratio for the current growth stage according to the growth requirement coefficient of the important fertilizer requirement factors. The important fertilizer requirement factors are the fertilizer requirement factors among the importance factors. S5. Based on the original fertilizer ratio at the current growth stage, predict the future yield, compare the future yield with the standard yield, and adjust the original fertilizer ratio at the current growth stage according to the comparison results to obtain the optimal fertilizer ratio at the current growth stage.

2. The method for adjusting fertilizer ratio based on the growth stage of walnut trees according to claim 1, characterized in that, S2 includes: S21. Based on fertilizer demand factors and environmental factors, extract multiple fertilizer demand characteristics, multiple environmental characteristics, yield data and growth status data from the historical planting data of multiple batches of reference target walnut trees from seed planting to fruiting period. S22, calculate the correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of each batch of reference target walnut trees, and select multiple important factors from multiple fertilizer requirement factors and multiple environmental factors based on the correlation calculation results; S23. Perform linear analysis on all important factors and growth status data to obtain the linear coefficient of each important factor, and use the linear coefficient of each important factor as its respective growth weight.

3. The method for adjusting fertilizer ratio based on the growth stage of walnut trees according to claim 2, characterized in that, S22 includes: S221, Calculate the original correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of each batch of reference target walnut trees according to the correlation algorithm; S222, the original correlation between each fertilizer requirement feature and each environmental feature and yield data in the historical planting data of all batches of reference target walnut trees is averaged to obtain the correlation between each fertilizer requirement feature and each environmental feature and yield data. S223, sort all fertilizer demand features and all environmental features in descending order of their correlation with yield data, and select the fertilizer demand factors and environmental factors corresponding to the first preset number of fertilizer demand features and environmental features in the descending order as the preset number of importance factors.

4. A method for adjusting fertilizer ratios based on the growth stages of walnut trees according to claim 2 or 3, characterized in that, S23 includes: S231, construct a polynomial linear function for each batch of target walnut trees based on all importance factors and the growth status data in the historical planting data of each batch of reference target walnut trees, wherein the original linear coefficients in the polynomial linear function correspond one-to-one with the importance factors; S232, concatenate all the original linear coefficients of the polynomial linear function of all batches of reference target walnut trees to obtain the linear coefficient matrix; S233, normalize the elements in each column of the linear coefficient matrix to obtain the linear coefficient of each importance factor, and use the linear coefficient of each importance factor as its respective growth weight.

5. The method for adjusting fertilizer ratio based on the growth stage of walnut trees according to claim 1, characterized in that, The growth stages of the target walnut tree include the budding stage, trunk growth stage, fruit development stage, and fruit ripening stage, and S3 includes: S31. Based on the growth characteristics of each growth stage of the target walnut tree, a linear function is used as the basic function for the budding stage, an exponential function is used as the basic function for the trunk growth stage, a logarithmic function is used as the basic function for the fruit development stage, and a saturation function is used as the basic function for the fruit ripening stage. The independent variables of the basic function for each growth stage are the growth time and importance factors of the target walnut tree, and the dependent variable is the standard growth amount of the target walnut tree. S32, input the growth weight of each important factor into the basic function of each growth stage to obtain the growth requirement function of each growth stage, and connect the growth requirement functions of each growth stage in the order of growth stages to obtain the multi-growth stage function of the target walnut tree.

6. The method for adjusting fertilizer ratio based on the growth stage of walnut trees according to claim 5, characterized in that, S4 includes: S41, Determine the current growth stage of the target walnut tree based on its current growth time and basic information; S42, input the current growth time of the target walnut tree, the environmental parameters in the basic information and the measured standard growth into the multi-growth stage function to obtain multiple growth requirement groups of important fertilizer requirement factors of the target walnut tree; S43, input the environmental parameters, each growth requirement group and the current growth time into the multi-growth stage function to obtain the predicted standard growth amount of each growth requirement group, calculate the squared error between the predicted standard growth amount and the measured standard growth amount of each growth requirement group, and select the growth requirement group with the smallest squared error as the growth requirement coefficient of multiple important fertilizer requirement factors of the target walnut tree in the current growth stage. S44, take the growth requirement coefficient of each important fertilizer requirement factor as the ratio of each important fertilizer requirement factor, and take all important fertilizer requirement factors and their ratios as the original fertilizer ratio for the current growth stage of the target walnut tree.

7. The method for adjusting fertilizer ratio based on the growth stage of walnut trees according to claim 6, characterized in that, S43 involves environmental parameters Any growth requirement group and current growth time Input to multi-stage growth function In this process, the predicted standard growth for each growth requirement group is obtained, and the predicted standard growth for that growth requirement group is compared with the measured standard growth. When the square of the error is calculated, it is achieved through formula (1): （1）； In formula (1), Represents the regularization coefficient. This indicates the predicted standard growth rate.

8. The method for adjusting fertilizer ratio based on the growth stage of walnut trees according to claim 6, characterized in that, S5 includes: S51, fit the average yield of multiple important fertilizer requirements and multiple important environmental characteristics, standard growth, growth time and yield data corresponding to the historical planting data of multiple batches of reference target walnut trees, and obtain the average yield function of the target walnut trees. S52, input the growth requirement coefficients, environmental parameters, growth time and standard growth amount of several important fertilizer requirements of the target walnut tree at the current growth stage into the yield mean function to obtain the future yield of the target walnut tree; S53, calculate the deviation between the future yield and the standard yield. If the deviation is negative, adjust the original fertilizer ratio according to the deviation to obtain the optimal fertilizer ratio. If the deviation is positive or 0, use the original fertilizer ratio as the optimal fertilizer ratio.

9. The method for adjusting fertilizer ratio based on the growth stage of walnut trees according to claim 8, characterized in that, When S53 adjusts the original fertilizer ratio according to the deviation value to obtain the optimal fertilizer ratio, it includes: S531, construct a yield loss function based on the standard yield, and calculate the partial derivatives of each important fertilizer demand factor in the original fertilizer ratio based on the yield loss function; S532: Based on the partial derivatives of each important fertilizer demand factor, use the gradient descent method to iteratively update the ratio of each important fertilizer demand factor until the yield loss function converges and stop iterating. The last updated ratio of each important fertilizer demand factor is taken as the optimal fertilizer ratio.

Images