Orchard fertilizer injection machine control method and system based on big data

Abstract

The invention discloses an orchard fertilizer injection machine control method and system based on big data, relates to the technical field of orchard fertilizer injection, and provides the following scheme: a fruit tree growth period is divided into a plurality of monitoring sub-time sequences according to a time sequence, each monitoring sub-time sequence corresponds to a specific phenological period, and the monitoring sub-time sequences are used for monitoring the fruit tree growth period; uniformly numbering all the monitoring sub-time sequences in the growth cycle; and synchronously collecting multi-dimensional data of each monitoring sub-time sequence, wherein the multi-dimensional data comprises soil fertility data, fruit tree physiological data and environmental data. A machine learning model is used to predict the start and end time of a phenological period, errors introduced by artificial experience are effectively avoided, key parameters are calculated through independent marking and weighting for possible overlapping time periods of adjacent phenological periods, the division precision of monitoring sub-time sequences is remarkably improved, and the canopy temperature and the soil moisture content are combined with dynamic changes of the canopy temperature and the soil moisture content. And the slice length of the sub-time sequence is adaptively adjusted, so that the data acquisition efficiency is further optimized.

Description

Technical Field

[0001] This invention relates to the field of orchard fertilization technology, specifically to a control method and system for orchard fertilization machines based on big data. Background Technology

[0002] Current orchard fertilization technology mainly adopts timed and quantitative fertilization based on preset programs, or simple threshold control by detecting basic soil parameters such as NPK content and pH value. Mainstream fertilization machine systems are usually equipped with soil sensor arrays, which can realize fixed-point and quantitative fertilizer injection. In terms of control strategy, most equipment uses PID control algorithm to adjust the amount of fertilizer injected, and realizes remote monitoring and data recording through Internet of Things technology.

[0003] The existing orchard fertilization technology has shortcomings. In terms of phenological period division, the reliance on manual experience leads to significant errors in timing. At the same time, relying solely on soil indicators or thresholds set by static models makes it difficult to accurately reflect the actual needs of fruit trees, making the preset quantitative fertilization method unable to effectively meet actual production needs. Therefore, we propose an orchard fertilization machine control method and system based on big data to solve this problem. Summary of the Invention

[0004] To address the aforementioned technical problems, this paper provides a control method and system for orchard fertilizer injection machines based on big data. This technical solution solves the defects of existing orchard fertilizer injection technologies mentioned in the background. In terms of phenological period division, the reliance on manual experience leads to significant errors in time nodes. At the same time, relying solely on soil indicators or thresholds set by static models makes it difficult to accurately reflect the actual needs of fruit trees, resulting in the inability of preset quantitative fertilization methods to effectively meet actual production needs.

[0005] To achieve the above objectives, the technical solution adopted by this invention is as follows: This invention provides a control method for orchard fertilizer injection machines based on big data, the method comprising: The fruit tree growth cycle is divided into several monitoring sub-time series according to the time series. Each monitoring sub-time series corresponds to a specific phenological period, and all monitoring sub-time series within the growth cycle are uniformly numbered. Multi-dimensional data from each monitoring sub-time series are collected synchronously, including soil fertility data, fruit tree physiological data, and environmental data; Based on the coupled analysis of soil fertility data and fruit tree physiological data, a dynamic fertilizer requirement assessment model was constructed, and fertilizer requirement early warning indices for each monitoring sub-time series were generated. A multidimensional constraint system is established based on environmental data. The weighted geometric average method is used to generate environmental constraint factors. The fertilizer demand early warning index is nonlinearly corrected to obtain the fertilizer injection control coefficient. Based on the comparison results between the fertilization control coefficient and the dynamic threshold range, a graded fertilization strategy is triggered, and the fertilization parameters are adjusted in real time based on the change rate of soil electrical conductivity.

[0006] Furthermore, a big data-based orchard fertilizer injection machine control system is proposed to implement the big data-based orchard fertilizer injection machine control method described above, comprising: The time series division module is used to divide the fruit tree growth cycle into several monitoring sub-time series according to the time series, and each monitoring sub-time series corresponds to a specific phenological period; The multi-source data acquisition module is used to synchronously acquire multi-dimensional data from each monitoring sub-time series, including soil fertility data, fruit tree physiological data, and environmental data. The dynamic assessment module is used to construct a dynamic fertilizer requirement assessment model based on the coupled analysis of soil fertility data and fruit tree physiological data, and to generate fertilizer requirement early warning indices for each monitoring sub-time series. The environmental constraint module is used to establish a multi-dimensional constraint system based on environmental data, generate environmental constraint factors using the weighted geometric average method, perform nonlinear correction on the fertilizer demand early warning index, and obtain the fertilizer injection control coefficient. The fertilization execution module is used to trigger a graded fertilization strategy based on the comparison results between the fertilization control coefficient and the dynamic threshold range, and to adjust the fertilization parameters in real time based on the change rate of soil electrical conductivity.

[0007] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By using machine learning models to predict the start and end times of phenological periods, errors introduced by human experience can be effectively avoided. For the possible overlapping periods of adjacent phenological periods, key parameters are independently labeled and weighted to significantly improve the segmentation accuracy of monitoring sub-time series. Combined with the dynamic changes of canopy temperature and soil moisture, the slice length of sub-time series is adaptively adjusted to further optimize data collection efficiency.

[0008] 2. By accurately quantifying the synergistic and antagonistic effects of nutrient elements through the nutrient balance term, simulating the nonlinear inhibition of microbial activity by acid-base stress using an exponential decay model for the pH buffer term, and characterizing the marginal diminishing effect of organic matter on fertility through a logarithmic function for the organic matter gain term, a multi-factor synergistic soil fertility index calculation model was constructed. This effectively reduced the misjudgment rate of nutrient requirements. Furthermore, by introducing the photosynthetic sensitivity coefficient and the time decay factor, the weight allocation of the fruit tree physiological index was dynamically adjusted to adapt to the physiological characteristics changes at different phenological stages, thereby more accurately reflecting the true nutritional needs of fruit trees.

[0009] 3. By triggering graded fertilization through dynamic threshold intervals and adjusting fertilization parameters based on real-time monitoring of soil conductivity changes, a closed-loop control system for prediction, execution, verification, and optimization is constructed. By dynamically adjusting fertilization strategies, fertilizer utilization is significantly improved, while leaching losses caused by excessive fertilization are effectively reduced, thus lowering the risk of agricultural non-point source pollution. Attached Figure Description

[0010] Figure 1 This is a flowchart of a big data-based orchard fertilizer injection machine control method proposed in this invention; Figure 2 This is a flowchart of the monitoring sub-time series partitioning method in this invention; Figure 3 This is a flowchart of the graded fertilization method in this invention; Figure 4 This is a structural block diagram of a big data-based orchard fertilizer injection machine control system proposed in this invention. Detailed Implementation

[0011] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0012] Reference Figure 1-3 As shown, a method for controlling an orchard fertilizer injection machine based on big data is described, the method comprising: The fruit tree growth cycle is divided into several monitoring sub-time series according to the time series. Each monitoring sub-time series corresponds to a specific phenological period, and all monitoring sub-time series within the growth cycle are uniformly numbered. Multi-dimensional data from each monitoring sub-time series are collected synchronously, including soil fertility data, fruit tree physiological data, and environmental data; Based on the coupled analysis of soil fertility data and fruit tree physiological data, a dynamic fertilizer requirement assessment model was constructed, and fertilizer requirement early warning indices for each monitoring sub-time series were generated. A multidimensional constraint system is established based on environmental data. The weighted geometric average method is used to generate environmental constraint factors. The fertilizer demand early warning index is nonlinearly corrected to obtain the fertilizer injection control coefficient. Based on the comparison results between the fertilization control coefficient and the dynamic threshold range, a graded fertilization strategy is triggered, and the fertilization parameters are adjusted in real time based on the change rate of soil electrical conductivity.

[0013] Reference Figure 2 As shown, the division of the growth cycle into monitoring sub-time series according to the phenological stages of fruit trees specifically includes: Acquire genomic data of fruit tree varieties, local historical climate data, and historical phenological observation data, input them into a machine learning model, and obtain the start and end times of each phenological period; Based on the start and end times of each phenological period, determine whether there is an overlapping interval between adjacent phenological periods. If so, change the end time of the previous phenological period to the start time of the overlapping interval, change the start time of the next phenological period to the end time of the overlapping interval, and divide the overlapping interval into a new phenological period. It should be noted that, in this embodiment, when the values ​​are matched with phenological periods, the values ​​of the overlapping area are the weighted average of the values ​​of adjacent phenological periods, and the weight ratio is the proportion of the overlapping area to the number of days in the two phenological periods.

[0014] Within a single phenological period, based on orchard microenvironment sensor data, including canopy temperature and soil moisture, the phenological period is divided into a high-demand segment, a stable segment, and a low-demand segment, and the time slice length of the monitoring subsequence is dynamically adjusted. For example, when the canopy temperature is >28℃ and the soil moisture is <60%, it is recorded as a high-demand segment, and the slice length of the monitoring subsequence is set to 8 hours. When the canopy temperature is <22℃ and the soil moisture is >80%, the slice length is set to 48 hours. The remaining cases are recorded as stable segments, and the slice length is set to 24 hours. All monitoring subsequences are numbered sequentially and marked as high-demand, stable, and low-demand segments.

[0015] Specifically, the soil fertility data are as follows: Soil fertility data are collected through a multi-depth sensor network based on the preset acquisition nodes of each monitoring sub-time series; for example, the preset acquisition nodes are set as the four equal division points of the slice length of the monitoring sub-time series. The soil fertility data includes NPK content, pH value and organic matter content, and the soil fertility data is normalized. Soil fertility index was calculated based on a composite ecological response model; The composite ecological response model is as follows: ; In the formula, Soil fertility index, These are the normalized values ​​for each component of the NPK content. and These are the normalized values ​​for pH and organic matter content, respectively. , , These are the correction coefficient for fruit tree type, pH sensitivity parameter, and organic matter saturation threshold, respectively, where i is the monitoring sub-time series number; for example, they can be 0.015, 0.5, and 20, respectively. It should be noted that, For the nutrient balance term, the geometric mean is used to emphasize the synergistic effect of each component of NPK. It reaches its maximum value when N:P:K approaches 1:1:1. Furthermore, a non-linear penalty is applied to nutrient imbalance through the exponential part. That is, when one component of NPK is much larger than the sum of the other two components, the value of the nutrient balance term decays exponentially. This is used as a parameter to quantify the adjustment of differences in NPK utilization rates among different fruit tree varieties. As a pH buffer term, most fruit tree root zone soils in the pH range of 6.0-6.5 can maintain the high activity of key soil enzymes, such as acid phosphatase and catalase, and maximize the availability of trace elements such as iron and zinc. Therefore, 6.3 was selected as the optimal pH reference value. e is used to provide the mathematical basis for exponential decay to simulate the rapid deterioration of acid-base barriers. When the value is 0.5, it can best fit the change of soil phosphatase activity with pH. For the organic matter gain term, a logarithmic model is used to simulate the decreasing growth rate of key indicators such as soil water holding capacity and microbial activity when organic matter increases. The organic matter saturation threshold is set to 20. When the organic matter content reaches 20%, the organic matter gain term considers that the fertility contribution of organic matter has approached the physiological upper limit and can avoid the impact of excessive organic matter on the soil fertility index in special soils.

[0016] Specifically, the physiological data of the fruit trees are as follows: Based on the preset acquisition nodes of each monitoring sub-time series, leaf reflectance spectra and infrared gas analysis data are acquired synchronously. Chlorophyll content is obtained using time-series inversion and direct measurement methods. Leaf moisture content and photosynthetic rate ; Based on the genomic type and phenological stage of fruit tree varieties, the chlorophyll content saturation threshold was obtained. and physiological critical water content ; Obtain historical photosynthetic rate data for this fruit tree variety during the same phenological period, and obtain the maximum photosynthetic rate. and minimum value ; The physiological index of fruit trees was calculated based on the dynamic coupling model, and the threshold values ​​of the physiological index were set for each phenological stage. For example, the physiological index threshold can be set to 0.68, 0.82, 0.90 and 0.78 during the budding, flowering, fruit enlargement and ripening stages, respectively, and scaled according to the markings of high demand segment, stable segment and low demand segment: high demand segment +0.05, stable segment +0, low demand segment -0.03. The dynamic coupling model is as follows: ; In the formula, For fruit tree physiological index, Photosynthetic sensitivity coefficient The date of the current data collection node. The central date of the phenological period. , These are the weighting coefficients. Here, i represents the decay rate factor, and i is the monitoring sub-time series number. For example, a preset data collection node; , 0.6 and 0.4 are acceptable values; For example, the photosynthetic sensitivity coefficient is used to quantify the contribution of photosynthetic efficiency to the physiological index of fruit trees, and can be taken as 0.18, 0.25, 0.35 and 0.28 at the budding, flowering, fruit enlargement and ripening stages, respectively; the decay rate factor is used to quantify the influence of phenological timeliness on the physiological index, and can be taken as 0.04, 0.03, 0.07 and 0.11 at the budding, flowering, fruit enlargement and ripening stages, respectively.

[0017] Specifically, the coupled analysis based on soil fertility data and fruit tree physiological data constructs a dynamic fertilizer requirement assessment model and generates fertilizer requirement early warning indices for each monitoring sub-time series, specifically including: Historical soil fertility data of fruit tree varieties at specific phenological stages were obtained, a soil fertility index was generated, and then... The principle is to eliminate unreasonable values; Calculate the maximum value of all soil fertility indices during this phenological period. and minimum value If there is no historical soil fertility data, then soil fertility data of the same soil type in the area where the orchard is located will be called, and data from different sources or units will be normalized. For each monitoring sub-time series, a fertilizer demand early warning index is calculated by combining the dynamic fertilizer demand assessment model; The dynamic fertilizer demand assessment model is as follows: ; In the formula, This is a fertilizer demand early warning index. denoted as the physiological index threshold for each phenological period, and i as the monitoring sub-time series number.

[0018] Specifically, the process of introducing environmental constraint factors generated based on environmental data to adaptively correct the fertilizer demand early warning index, obtaining a fertilizer injection control coefficient, and adjusting the fertilizer application rate and timing of the fertilizer injection machine according to the fertilizer injection control coefficient includes: The environmental data include photosynthetically active radiation, soil moisture content, and soil temperature. The degree of water stress is quantified by the logistic function relationship between soil moisture content and physiological critical moisture content to obtain the water factor. The light factor is generated by the Michaelis saturation curve of photosynthetically active radiation. The temperature factor is obtained by averaging the values ​​assigned to the soil layer according to the temperature. For example, a soil temperature <10℃ is assigned a value of 0.2, a soil temperature 10℃ ≤ soil temperature <15℃ is assigned a value of 0.5, a soil temperature 15℃ ≤ soil temperature <25℃ is assigned a value of 0.5, and a soil temperature ≥25℃ is assigned a value of 0.7. Based on the normalized values ​​of water factor, light factor and temperature factor, the environmental constraint factor is obtained by weighted geometric mean calculation method; for example, the weight ratio of weighted geometric mean calculation method is different in each phenological stage, 5:3:2 for budding stage and swelling stage, 4:4:2 for flowering stage, and balanced for maturity stage. The product of the fertilizer demand early warning index and the environmental constraint factor is used as the fertilizer injection control coefficient.

[0019] See Figure 3 As shown, the step of triggering a tiered fertilization strategy based on the comparison results between the fertilization control coefficient and the dynamic threshold range, and adjusting fertilization parameters in real time based on the change rate of soil electrical conductivity, specifically includes: Based on the genomic type and phenological stage of fruit tree varieties, and combined with markers for high-demand, stable, and low-demand periods, the fertilization trigger threshold range and unit fertilization amount are dynamically set. The unit fertilization amount is the basic fertilization requirement determined by fruit tree genomic data. For example, by formula Analyze the upper and lower limits of the fertilizer injection trigger threshold range, when Pick t is the upper limit, t is the lower limit, where t is the lower limit, The coefficients for high, stable, and low demand segments can be set to 1.25, 1.125, and 1, respectively. This is the decay rate factor; When the fertilizer injection control coefficient is greater than the upper limit of the fertilizer injection trigger threshold range, it is directly executed within the current monitoring sub-time series node according to the unit fertilizer application amount. When the fertilizer injection control coefficient is within the fertilizer injection trigger threshold range, it is executed according to the high fixed ratio of the preset unit fertilizer application amount. When the fertilizer injection control coefficient is lower than the lower limit of the fertilizer injection trigger threshold range, it will not be executed in the current monitoring sub-time series node. If the fertilizer injection control coefficient rises in the next monitoring sub-time series but does not enter the fertilizer injection trigger threshold range, it will be executed at a low fixed ratio of the unit fertilizer application amount. If it falls, fertilizer injection will be suspended. For example, the high fixed ratio is 100-120%, and the low fixed ratio is 60-80%; After fertilization, soil electrical conductivity data is collected. If the daily change rate of soil electrical conductivity is greater than 5%, the high or low fixed ratio of the next fertilization operation will be reduced by 10%. If the change rate of soil electrical conductivity is less than -5%, supplementary fertilization will be triggered.

[0020] See Figure 4 As shown, this solution proposes a big data-based orchard fertilizer injection machine control system to implement the aforementioned big data-based orchard fertilizer injection machine control method, including: The time series division module is used to divide the fruit tree growth cycle into several monitoring sub-time series according to the time series, and each monitoring sub-time series corresponds to a specific phenological period; The multi-source data acquisition module is used to synchronously acquire multi-dimensional data from each monitoring sub-time series, including soil fertility data, fruit tree physiological data, and environmental data. The dynamic assessment module is used to construct a dynamic fertilizer requirement assessment model based on the coupled analysis of soil fertility data and fruit tree physiological data, and to generate fertilizer requirement early warning indices for each monitoring sub-time series. The environmental constraint module is used to establish a multi-dimensional constraint system based on environmental data, generate environmental constraint factors using the weighted geometric average method, perform nonlinear correction on the fertilizer demand early warning index, and obtain the fertilizer injection control coefficient. The fertilization execution module is used to trigger a graded fertilization strategy based on the comparison results between the fertilization control coefficient and the dynamic threshold range, and to adjust the fertilization parameters in real time based on the change rate of soil electrical conductivity.

[0021] The dynamic evaluation module specifically includes: The data preprocessing subunit is used to obtain historical soil fertility data of fruit tree varieties during specific phenological periods, generate soil fertility indices, and remove unreasonable values ​​according to principles. The coupling analysis subunit is used to calculate the maximum and minimum values ​​of all soil fertility indices during the phenological period. If there is no historical soil fertility data, it will call the soil fertility data of the same soil type in the area where the orchard is located, and normalize the data from different sources or in different dimensions. The early warning generation subunit is used to calculate the fertilizer demand early warning index for each monitoring sub-time series in conjunction with the dynamic fertilizer demand assessment model.

[0022] The environmental constraint module specifically includes: The factor calculation subunit is used to quantify the degree of water stress based on the logistic function relationship between soil moisture content and physiological critical moisture content, and obtain the water factor. The light factor is generated by using the Michaelis saturation curve of photosynthetically active radiation, and the temperature factor is obtained by averaging the values ​​assigned to the soil layer according to the temperature. The constrained synthesis subunit is used to calculate the environmental constraint factor based on the normalized values ​​of the moisture factor, light factor, and temperature factor using a weighted geometric mean method.

[0023] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A big data-based orchard fertilizer injection machine control method, characterized in that, The method comprises: dividing the fruit tree growth cycle into a plurality of monitoring time sequences in time sequence, each monitoring time sequence corresponding to a specific phenophase, and uniformly numbering all monitoring time sequences in the growth cycle; synchronously collecting multi-dimensional data of each monitoring time sequence, the multi-dimensional data including soil fertility data, fruit tree physiological data and environmental data; based on coupling analysis of the soil fertility data and the fruit tree physiological data, constructing a dynamic fertilizer demand evaluation model, and generating a fertilizer demand early warning index for each monitoring time sequence; based on the environmental data, establishing a multi-dimensional constraint system, generating an environmental constraint factor by using a weighted geometric mean method, and nonlinearly correcting the fertilizer demand early warning index to obtain a fertilization control coefficient; based on a comparison result of the fertilization control coefficient and a dynamic threshold interval, triggering a grading fertilization strategy, and based on a soil conductivity rate, real-time feedback adjusting fertilization parameters.

2. The method of claim 1, wherein, The fruit tree growth cycle is divided into a plurality of monitoring time sequences in time sequence, each monitoring time sequence corresponding to a specific phenophase, and all monitoring time sequences in the growth cycle are uniformly numbered, specifically comprising: obtaining genomic data of fruit tree varieties, local historical climate data and historical phenological observation data, inputting a machine learning model, and obtaining the start and end time of each phenophase; based on the start and end time of each phenophase, determining whether there is an overlapping interval between adjacent phenophases, if there is, changing the end time of the previous phenophase to the start time of the overlapping interval, changing the start time of the next phenophase to the end time of the overlapping interval, and separately dividing the overlapping interval into a new phenophase; within a single phenophase, based on the real-time collected canopy temperature and soil moisture data, dynamically dividing the phenophase into a high demand segment, a stable segment and a low demand segment, and adaptively adjusting the slice length of the monitoring time sequence; numbering all monitoring sub-sequences in time sequence, and marking the high demand segment, the stable segment and the low demand segment.

3. The method of claim 1, wherein, The soil fertility data specifically comprises: based on the preset collection nodes of each monitoring time sequence, collecting soil fertility data through a multi-depth sensor network; the soil fertility data includes NPK content, pH value and organic matter content, and the soil fertility data is normalized; calculating a soil fertility index according to a composite ecological response model; the composite ecological response model is: ； wherein, is the soil fertility index, is the normalized value of each sub-item of NPK content, and are the normalized values of pH and organic matter content, respectively, , , are the fruit tree type correction coefficient, the pH sensitivity parameter and the organic matter saturation threshold, respectively, and i is the monitoring sub-time series number.

4. The method of claim 1, wherein, The fruit tree physiological data specifically comprises: Based on the preset acquisition nodes of each monitoring sub-time sequence, the leaf reflectance spectrum and infrared gas analysis data are synchronously acquired, and the chlorophyll content , leaf water content and photosynthetic rate are obtained by using time series inversion method and direct determination method. Based on the genotypes of fruit tree varieties and phenological stages, the saturation threshold of chlorophyll content is obtained and the physiological critical moisture content ; Obtain the historical photosynthetic rate data of the same phenophase of the fruit tree variety, and obtain the maximum and minimum values of the photosynthetic rate and minimum values ; calculating a fruit tree physiological index according to a dynamic coupling model, and setting a physiological index threshold in each phenophase; the dynamic coupling model is: ； In the formula, is a fruit tree physiological index, is a photosynthetic sensitivity coefficient, is a current collection node date, is a phenophase center date, , is a weight coefficient, is a decay rate factor, is a monitoring sub-time sequence number, is a preset collection node.

5. The method of claim 4, wherein, Based on coupling analysis of the soil fertility data and the fruit tree physiological data, a dynamic fertilizer demand evaluation model is constructed, and a fertilizer demand early warning index for each monitoring time sequence is generated, specifically comprising: Obtaining historical soil fertility data of fruit tree varieties at a specific phenological period, generating a soil fertility index, and adopting Principle of eliminating unreasonable values; The maximum value of all soil fertility indexes in the phenophase is counted and the minimum value If there is no historical soil fertility data, the soil fertility data of the same soil type in the region where the orchard is located is called, and data of different sources or dimensions is normalized. for each monitoring time sequence, a fertilizer demand early warning index is calculated in combination with the dynamic fertilizer demand evaluation model; the dynamic fertilizer demand evaluation model is: ； In the formula, is a pre-fertilizer warning index, is a physiological index threshold value in each phenological period, is a monitoring sub-time series number.

6. The method of claim 1, wherein, based on the environmental data, a multi-dimensional constraint system is established, an environmental constraint factor is generated by using a weighted geometric mean method, and the fertilizer demand early warning index is nonlinearly corrected to obtain a fertilization control coefficient, specifically comprising: the environmental data includes photosynthetically active radiation, soil moisture content and soil temperature; The degree of water stress is quantified according to a logistic function relationship between soil water content and physiological critical water content, and a water factor is obtained; a light factor is generated by using a Michaelis-Menten saturation curve of photosynthetically active radiation; and a temperature factor is obtained by segmenting and averaging the temperature of the soil layer; Based on the normalized values of the water factor, the light factor and the temperature factor, an environmental constraint factor is obtained by using a weighted geometric mean calculation method; The product of the fertilizer demand early warning index and the environmental constraint factor is taken as the fertilizer injection control coefficient.

7. The method of claim 6, wherein, According to the comparison result of the fertilizer injection control coefficient and the dynamic threshold interval, a hierarchical fertilization strategy is triggered, and the fertilization parameters are adjusted in real time based on the soil conductivity change rate, and specifically comprising: Based on the genotypes and phenological stages of fruit tree varieties, combined with the markers of the high demand segment, the stable segment and the low demand segment, the fertilizer injection trigger threshold interval and the unit fertilization amount are dynamically set; When the fertilizer injection control coefficient is greater than the upper limit of the fertilizer injection trigger threshold interval, the unit fertilization amount is directly executed in the current monitoring sub-time sequence node; when the fertilizer injection control coefficient is within the fertilizer injection trigger threshold interval, the high fixed proportion of the preset unit fertilization amount is executed; When the fertilizer injection control coefficient is lower than the lower limit of the fertilizer injection trigger threshold interval, no execution is performed in the current monitoring sub-time sequence node; if the fertilizer injection control coefficient rises in the next monitoring sub-time sequence but does not enter the fertilizer injection trigger threshold interval, the low fixed proportion of the unit fertilization amount is executed; if it decreases, the fertilizer injection is suspended; After the fertilizer injection, the soil conductivity data is collected, if the soil conductivity change rate is greater than 5% within a day, the high fixed proportion or the low fixed proportion of the next fertilizer injection operation is reduced by 10%, if the soil conductivity change rate is less than-5%, the supplementary fertilizer injection is triggered.

8. A big data-based orchard fertilizer injection machine control system, characterized in that, A big data-based orchard fertilizer injection machine control method according to any one of claims 1-7, comprising: a time sequence division module for dividing the fruit tree growth cycle into a plurality of monitoring sub-time sequences according to time sequence, each monitoring sub-time sequence corresponding to a specific phenological stage; a multi-source data acquisition module for synchronously acquiring multi-dimensional data of each monitoring sub-time sequence, the multi-dimensional data including soil fertility data, fruit tree physiological data and environmental data; a dynamic evaluation module for coupling analysis based on soil fertility data and fruit tree physiological data, constructing a dynamic fertilizer demand evaluation model, and generating a fertilizer demand early warning index for each monitoring sub-time sequence; an environmental constraint module for establishing a multi-dimensional constraint system based on environmental data, generating an environmental constraint factor by using a weighted geometric mean method, and nonlinearly correcting the fertilizer demand early warning index to obtain a fertilizer injection control coefficient; a fertilizer injection execution module for triggering a hierarchical fertilization strategy according to the comparison result of the fertilizer injection control coefficient and the dynamic threshold interval, and adjusting the fertilization parameters in real time based on the soil conductivity change rate.

9. The big data based orchard fertilizer injection machine control system according to claim 8, wherein, The dynamic evaluation module specifically comprises: The data preprocessing subunit is configured to acquire historical soil fertility data of the fruit tree variety at a specific phenological stage, generate a soil fertility index, and adopt Unreasonable values are removed in principle. a coupling analysis subunit for counting the maximum and minimum values of all soil fertility indexes in the phenological period If there is no historical soil fertility data, soil fertility data of the same soil type in the region where the orchard is located is called, and data of different sources or dimensions are normalized.​ an early warning generation subunit for calculating the fertilizer demand early warning index for each monitoring sub-time sequence based on the dynamic fertilizer demand evaluation model.

10. The big data based orchard fertilizer injection machine control system according to claim 8, wherein, The environmental constraint module specifically comprises: The factor calculation subunit is configured to quantify the degree of water stress according to a logistic function relationship between the soil water content and the physiological critical water content, to obtain a water factor, to generate a light factor by using a Michaelis-Menten saturation curve of photosynthetically active radiation, and to obtain a temperature factor by segmentally assigning and averaging the soil layer temperature. The constraint synthesis subunit is configured to obtain an environmental constraint factor by using a weighted geometric mean calculation method based on the normalized values of the water factor, the light factor and the temperature factor.

Images