Microbiome data-driven methods and systems for predicting soil fertility

Abstract

This invention provides a microbiome data-driven method and system for predicting soil fertility, belonging to the field of data prediction technology. The method includes: configuring a soil sampling matrix; performing intermittent stratified sampling at H frequencies and measuring microbiome data to obtain a microbial community time-series matrix; performing nutrient conversion processing and capturing dynamic evolution patterns to predict and output F soil fertility evolution curves; connecting to and calling F growth fertility demand curves; locating nutrient gaps to obtain M fertility demand deviations at M fertility demand gap nodes; and optimizing the fertilization compensation strategy for the next fertility compensation cycle to obtain a fertilization compensation execution sequence. This invention solves the technical problem of existing technologies lacking comprehensive analysis of the microbial community, leading to an inability to accurately capture the root causes of soil fertility changes, resulting in inaccurate predictions of fertilizer demand and fertilization timing, thus affecting crop production efficiency and resource utilization.

Description

Technical Field

[0001] This invention relates to the field of data prediction technology, specifically to a method and system for predicting soil fertility driven by microbiome data. Background Technology

[0002] Soil is the foundation of agricultural production, and its fertility directly affects crop growth, yield, and quality. Therefore, accurate assessment and management of soil fertility has become a key issue in modern agricultural management, especially in greenhouse cultivation environments where soil fertility management is particularly important due to the relatively enclosed and controllable environment.

[0003] In existing technologies, the role of microorganisms is often overlooked or inferred indirectly only through limited physicochemical indicators. Microbial communities not only influence nutrient transformation and plant health, but their metabolic processes are also crucial for maintaining and restoring soil fertility. Therefore, relying solely on traditional soil chemical analysis methods cannot fully utilize the functional information of microbial communities, leading to an inability to accurately capture the root causes of soil fertility changes. This, in turn, results in inaccurate predictions of fertilizer requirements and application timing, impacting crop production efficiency and resource utilization. In greenhouse environments, the limitations of traditional methods are particularly pronounced because soil changes within greenhouses are typically influenced by multiple factors, and traditional methods relying on static data cannot adapt to the rapidly changing soil fertility state. Summary of the Invention

[0004] This application provides a microbiome data-driven method and system for predicting soil fertility, aiming to solve the technical problem that the lack of comprehensive analysis of microbial communities in existing technologies leads to the inability to accurately capture the root causes of changes in soil fertility, resulting in inaccurate predictions of fertilizer demand and fertilization timing, which in turn affects crop production efficiency and resource utilization.

[0005] The first aspect disclosed in this application provides a microbiome data-driven method for predicting soil fertility. The method includes: dynamically adjusting the sampling grid size based on crop planting density and configuring a soil sampling matrix in the planting area; using 1 / H of the fertility compensation cycle as the soil sampling interval window, performing intermittent stratified sampling at H frequencies in the soil sampling matrix, and performing real-time microbiome data measurement to obtain a microbial community time-series matrix; performing fertility nutrient conversion processing on the microbial community time-series matrix, and then performing dynamic evolution law capture to predict and output F soil fertility evolution curves; using the latitude and longitude coordinates of the planting area and crop type as dual search constraints, and connecting to the network to call F growth and fertility demand curves of the target crop for F types of fertilizers; locating nutrient gaps by overlapping the F soil fertility evolution curves and the F growth and fertility demand curves to obtain M fertility demand gap nodes and M fertility demand deviations; and optimizing the fertilization compensation strategy for the next fertility compensation cycle based on the M fertility demand deviations to obtain a fertilization compensation execution sequence.

[0006] The second aspect of this application discloses a microbiome data-driven soil fertility prediction system. This system is used in the aforementioned microbiome data-driven soil fertility prediction method. The system includes: a soil sampling matrix configuration module for dynamically adjusting the sampling grid size according to crop planting density and configuring a soil sampling matrix in the planting area; a data measurement module for using 1 / H of the fertility compensation cycle as the soil sampling interval window, performing intermittent stratified sampling at frequency H times in the soil sampling matrix, and performing real-time microbiome data measurement to obtain a microbiome community time-series matrix; and a pattern capture module for performing fertility and nutrient conversion analysis on the microbiome community time-series matrix. After processing, the system performs dynamic evolution pattern capture and predicts and outputs F soil fertility evolution curves. A demand curve retrieval module uses the latitude and longitude coordinates of the planting area and crop type as dual search constraints to connect to and retrieve F growth and fertility demand curves for the target crop for F types of fertilizers. A nutrient gap location module locates nutrient gaps by overlapping the F soil fertility evolution curves and the F growth and fertility demand curves, obtaining M fertility demand deviations for M fertility demand gap nodes. A compensation strategy optimization module optimizes the fertilization compensation strategy for the next fertility compensation cycle based on the M fertility demand deviations, obtaining a fertilization compensation execution sequence.

[0007] One or more technical solutions provided in this application have at least the following beneficial effects:

[0008] By dynamically adjusting the sampling grid size, refined soil sampling can be performed based on crop planting density and soil characteristics. This ensures the representativeness and accuracy of the sampling data, comprehensively reflecting the spatial distribution of soil fertility. During the fertility compensation cycle, combining microbiome data measurement provides high-frequency soil microbial information for fertility prediction, thereby enhancing the timeliness and accuracy of the prediction results. By processing the microbial community time-series matrix for nutrient transformation, the impact of soil microbial activity on nutrient transformation can be revealed, providing a scientific basis for subsequent soil fertility prediction and making the prediction results closer to the actual changes in soil fertility. By capturing dynamic evolution patterns, soil fertility prediction is not limited to current data but also considers the trend of soil fertility changes over time. The predicted soil fertility evolution curve reflects… It can accurately predict soil fertility changes at different time points, demonstrating a high degree of time-series forecasting capability. By combining the latitude and longitude coordinates of the planting area and crop type, and using a dual retrieval constraint method, it can accurately obtain the fertilizer demand curve of the target crop. This allows for optimization of fertilization plans based on specific crop requirements and growth cycles. By overlaying the soil fertility evolution curve with the crop fertility demand curve, it can pinpoint nutrient gaps. This process can accurately identify the points of soil fertility insufficiency and fertilizer demand gaps, thereby guiding the formulation of subsequent fertilization strategies. By analyzing the deviation in fertility demand and optimizing fertilization compensation strategies based on this data, fertilization becomes more precise and effective. The optimized fertilization compensation execution sequence ensures timely adjustments to fertilizer supply based on the differences between soil fertility demand and crop demand, avoiding resource waste and improving fertilization efficiency.

[0009] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0010] Figure 1 A schematic diagram of the microbiome data-driven soil fertility prediction method provided in the embodiments of this application.

[0011] Figure 2 A schematic diagram of the structure of a microbiome data-driven soil fertility prediction system provided in an embodiment of this application.

[0012] Figure labeling: Soil sampling matrix configuration module 10, data measurement module 20, pattern capture module 30, demand curve calling module 40, nutrient gap location module 50, compensation strategy optimization module 60. Detailed Implementation

[0013] This application provides a method and system for predicting soil fertility based on microbiome data. It solves the technical problem that the lack of comprehensive analysis of microbial communities in the prior art leads to the inability to accurately capture the root causes of changes in soil fertility, which in turn results in inaccurate prediction of fertilizer demand and fertilization time, affecting crop production efficiency and resource utilization.

[0014] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0015] Example 1, as Figure 1 As shown in the embodiments of this application, a soil fertility prediction method driven by microbiome data is provided, the method comprising:

[0016] The sampling grid size is dynamically adjusted according to the crop planting density, and a soil sampling matrix is ​​configured in the planting area.

[0017] Crop planting density is determined based on the spatial distribution of crop planting, such as the number of crop plants per hectare. In greenhouse cultivation environments, crop planting density has a significant impact on soil microbial communities, nutrient distribution, and overall soil health. Because the greenhouse environment is relatively closed and controllable, soil fertility changes are relatively stable. Therefore, changes in crop planting density within the greenhouse more directly affect soil nutrient demand and distribution. Specifically, higher crop planting densities lead to more intense competition among crops, resulting in greater fluctuations in soil nutrients and moisture. To obtain more accurate soil information, a dynamic method is used to adjust the soil sampling grid size according to planting density. In high-density areas, the grid size can be appropriately reduced to ensure better sampling of soil homogeneity within a smaller area; while in low-density areas, the grid size can be appropriately increased to reduce the number of sampling points while maintaining representativeness. After determining the sampling grids for different areas, a soil sampling matrix is ​​configured within the planting area. Each grid contains a certain number of sampling points used to measure relevant soil properties.

[0018] Using 1 / H of the fertility compensation cycle as the soil sampling interval window, intermittent stratified sampling with a frequency of H was performed on the soil sampling matrix, and real-time microbiome data was measured to obtain the microbial community time series matrix.

[0019] The fertility compensation cycle H refers to the adjustment and implementation of fertilization strategies based on changes in soil fertility within a certain time interval. For example, if the fertility of a certain nutrient element in the soil needs to be replenished every three months, then H is three months. The reciprocal of the fertility compensation cycle 1 / H is used as the interval window for soil sampling. For example, if H is 3 months, then the soil sampling interval is 1 / 3 months, that is, once a month, to monitor changes in soil fertility more frequently.

[0020] Stratified sampling is a sampling method used to reduce the impact of soil heterogeneity. Based on different soil depths, multiple sampling layers are set, such as dividing the soil into surface (0-20cm), middle (20-40cm), and deep (40-60cm). Within each sampling cycle, different soil layers are sampled intermittently according to a set frequency H. This approach can effectively capture the temporal and depth-related changes in soil fertility. After each sampling, a soil sample is obtained.

[0021] The soil microbial community has a significant impact on soil health and crop growth. During each soil sampling, real-time microbiome data analysis of the sampled soil can be performed. Techniques such as DNA sequencing and PCR analysis can be used to detect the microbial populations and their diversity in the soil. The results are used to analyze the interrelationships between soil health, fertilization effectiveness, and crop growth.

[0022] Each time microbiome data is measured, a new set of data is generated. Over time, all the microbiome data are organized into a matrix in chronological order to obtain the microbiome time series matrix. The microbiome time series matrix is ​​constructed based on microbiome data collected at different time points and in different sampling areas. This matrix reflects the changing trend of the microbiome over time and can provide a basis for soil management and fertilization decisions.

[0023] After performing fertility nutrient conversion processing on the microbial community time series matrix, dynamic evolution law capture is performed to predict and output F soil fertility evolution curves.

[0024] Transforming data from the microbial community time series matrix into soil fertility and nutrient information can be achieved by analyzing the relationship between the microbial community and major nutrients in the soil (such as nitrogen, phosphorus, potassium, sulfur, etc.) and establishing a fertility and nutrient conversion model. For example, based on the quantitative indicators of microbial activities (such as decomposition of organic matter, nitrogen fixation, phosphorus hydrolysis, etc.), the microbial community time series matrix can be converted into a corresponding fertility and nutrient time series matrix.

[0025] Soil fertility evolution is a dynamic process influenced by various factors, including fertilization, climate change, and crop growth stages. Based on existing fertility nutrient time series matrices, statistical methods or machine learning algorithms, such as time series analysis, regression analysis, and neural networks, are used to model soil fertility, capture its evolution patterns, and predict future trends in soil fertility based on these patterns. The result of this process is the generation of F soil fertility evolution curves, each representing the trend of soil fertility change under different fertility states.

[0026] Using the latitude and longitude coordinates of the planting area and the crop type as dual search constraints, the network retrieves the F growth fertility requirement curves of the target crop for F types of fertilizers.

[0027] The latitude and longitude coordinates of the planting area determine the soil's climate characteristics, soil type, and natural fertility. The crop type affects the demand for specific fertilizers (such as nitrogen fertilizer, phosphorus fertilizer, etc.). Using this information as search constraints, the growth fertility demand curve of the target crop can be retrieved online. For example, some agricultural service platforms or databases can provide a large amount of crop growth data, soil requirements, and fertilization guidelines. By accessing this data online, it is helpful to formulate precise fertilization strategies based on local conditions.

[0028] Specifically, using the latitude and longitude coordinates of the planting area, local climate conditions, soil types, and other relevant agricultural information are queried. For example, different latitudes represent different climate zones, which will affect the crop's growth cycle and fertilizer requirements. Different types of crops, such as rice, wheat, and corn, have different fertilizer requirements. The crop's growth cycle, root characteristics, and requirements for different nutrients determine the fertilization strategy. For example, crops with well-developed root systems require more nitrogen fertilizer, while fruit trees require more potassium fertilizer. Each crop has a growth fertility requirement curve for different growth stages (such as germination, vegetative growth, flowering and fruiting). By establishing a crop growth model and combining climate, soil conditions, and crop growth stages, the requirements of each crop for different fertilizers (such as nitrogen, phosphorus, and potassium) at different time points are predicted, thus establishing a growth fertility requirement curve.

[0029] By overlaying the F soil fertility evolution curves and the F growth fertility demand curves, nutrient gaps are located, and M fertility demand deviations are obtained for the M fertility demand gap nodes.

[0030] The fertilizer requirements of each crop vary at different growth stages, and changes in soil fertility also differ. If soil fertility is insufficient to meet the crop's needs, a nutrient deficit occurs. By comparing F soil fertility evolution curves with F growth fertility requirement curves, and comparing the two curves point by point, we can identify areas where soil fertility is lower than the crop's requirements—the so-called nutrient deficit. Specifically, if soil fertility is greater than or equal to the crop's requirements, there is no deficit; if soil fertility is less than the crop's requirements, a fertilizer deficit exists. By analyzing these overlapping curves, we can determine which time points exhibit significant nutrient deficits.

[0031] Based on the nutrient gap location, the size of the fertility demand gap is quantified, which is the fertility demand deviation. The fertility demand deviation represents the difference between the fertilizer required by the crop and the actual soil fertility at the gap node. The specific calculation method is to subtract the actual soil fertility from the crop fertility demand. By analyzing multiple time points, M fertility demand gap nodes are obtained, that is, M time points where there is a nutrient gap, and M fertility demand deviations. The deviation of each node represents the gap between soil fertility and crop demand at that time point.

[0032] Based on the M fertility demand deviations, the fertilization compensation strategy for the next fertility compensation cycle is optimized to obtain the fertilization compensation execution sequence.

[0033] Based on the deviation of each fertility demand gap node, a reasonable fertilization compensation strategy is set. The goal is to formulate a fertilization compensation plan based on the deviation of each gap. For example, for nodes with positive deviations, a fertilization increase plan is formulated; while for nodes with negative deviations, fertilization can be reduced. The key to optimizing the fertilization strategy is to determine the fertilization amount at each node based on the magnitude of fertilizer demand. For example, if the soil is deficient in nitrogen fertilizer at a certain stage, but the crop demand is high, the amount of nitrogen fertilizer applied can be increased.

[0034] The optimization process uses mathematical optimization methods, such as linear programming, dynamic programming, and genetic algorithms, to find the optimal fertilizer allocation scheme. These algorithms comprehensively consider the economics, environmental impact, and crop requirements of fertilization based on the deviation of each fertility demand gap, thereby determining the optimal fertilization plan. Through optimization, a fertilization compensation execution sequence is obtained. The fertilization compensation execution sequence refers to the specific plan for fertilization arranged in chronological order within the next fertility compensation cycle. This sequence includes fertilization operations at each fertility demand gap node, including the amount of fertilizer, type of fertilizer, timing, and frequency of fertilization at each time point.

[0035] Furthermore, after performing fertility nutrient conversion processing on the microbial community time series matrix, dynamic evolution law capture is performed to predict and output F soil fertility evolution curves. The method includes:

[0036] The microbial community time series matrix is ​​loaded into a microbial interaction network for fertility and nutrient correlation transformation to obtain a fertility and nutrient time series matrix; the fertility and nutrient time series matrix is ​​normalized based on linear interpolation to obtain F soil fertility time series data; the dynamic evolution law of the F soil fertility time series data is captured, and the F soil fertility evolution curves are predicted and output.

[0037] The microbial community time series matrix contains information on microbial composition and abundance at different time points. Using graph neural networks (Graph Neural Networks), this data can be transformed into a network of interactions between microorganisms. Graph Neural Networks are used to model these interactions; for example, some microorganisms can collaboratively transform specific nutrients (such as nitrogen, phosphorus, and potassium) through metabolic processes. In this way, the complex relationships within the microbial community can be captured. The microbial interaction network is specifically represented by a functional gene-nutrient transformation mapping table, which can link the functional genes of different microorganisms with their nutrient transformation capabilities. For example, some microorganisms can convert atmospheric nitrogen into usable nitrogen sources in the soil through nitrogen-fixing genes, or some microorganisms can release phosphorus from the soil by decomposing organic matter. Through this functional gene-nutrient transformation mapping table, the contribution and transformation process of the microbial community to nutrients at each time point can be inferred, generating a fertility nutrient time series matrix. This matrix reflects the changes in various nutrients (such as nitrogen, phosphorus, potassium, and trace elements) in the soil at different time points. Using this data, the temporal variation patterns of nutrients in the soil can be captured.

[0038] Because data may be missing at certain time points or the data intervals may be uneven in actual measurements, linear interpolation is used to fill in the missing values ​​in the fertility nutrient time series matrix. Linear interpolation calculates the values ​​of missing data points based on the linear relationship between known data points. After linear interpolation, to eliminate the influence of dimensions and unify the data scale, the fertility nutrient time series matrix is ​​normalized. Normalization scales the data proportionally to a specific range, such as between 0 and 1, allowing comparisons of data from different dimensions. After interpolation and normalization, the fertility nutrient time series matrix is ​​transformed into F soil fertility time series data, which are used to describe the changing trends of soil fertility at different time points.

[0039] Soil fertility change is a dynamic process influenced by various factors, including fertilization, climate change, and crop growth. To predict future soil fertility change trends, F time-series data on soil fertility are modeled to capture their dynamic evolution patterns. For example, time series analysis is used to capture the changing patterns of soil fertility over time, identifying characteristics such as trends, periodic fluctuations, and abrupt changes. This allows for the identification of long-term trends and short-term fluctuations in soil fertility. Based on the captured dynamic evolution patterns, regression analysis is used to predict future soil fertility evolution, ultimately outputting F soil fertility evolution curves. These curves demonstrate the expected change paths of soil fertility under different conditions.

[0040] Furthermore, based on the M fertility demand deviations, the fertilization compensation strategy for the next fertility compensation cycle is optimized to obtain a fertilization compensation execution sequence. The method includes:

[0041] Using the end time of the F soil fertility evolution curves as the compensation starting point, calculate the M fertility compensation time limits for the M fertility demand gap nodes; optimize the fertilization compensation strategy based on the M fertility compensation time limits and the M fertility demand deviations to obtain M benchmark compensation strategies; perform application conflict correction on the M benchmark compensation strategies and output the corrected application sequence as the fertilization compensation execution sequence.

[0042] Each soil fertility evolution curve represents the trend of soil fertility change within a certain time range. The last time point predicted is extracted from each soil fertility evolution curve and used as the starting time for compensation. This time point corresponds to the critical growth stage of the crop or the timing of the next fertilization cycle. Based on the time difference between the last time point of the soil fertility evolution curve and the fertility demand gap node, the fertility compensation time limit for each fertility demand gap node is calculated. The compensation time limit refers to how long the gap needs to be filled to ensure that the crop's fertilizer needs are met. For example, a gap needs to be filled within two weeks, while another gap may take a month.

[0043] The optimization goal of fertilization compensation strategies is to enable soil fertility to compensate for crop fertilizer needs within the specified time frame and by applying appropriate amounts and timing of fertilizer application. Based on the compensation time frame and deviation of each fertility demand gap node, optimization algorithms, such as linear programming, genetic algorithms, and particle swarm optimization, are used to find the optimal fertilization compensation strategy. The goal of these strategies is to determine the optimal amount, timing, and frequency of fertilizer application so as to meet crop fertilizer needs as efficiently as possible within the specified compensation time frame. After optimization, M baseline compensation strategies are obtained, each corresponding to a fertility demand gap node. Each baseline compensation strategy defines the amount of fertilizer to be applied and the timing of fertilization within the compensation time frame.

[0044] In practice, multiple baseline compensation strategies may conflict. For example, at the same time point, multiple compensation strategies may require fertilizer application, or the types and quantities of fertilizer applied at the same time may conflict. Therefore, conflict correction is needed to ensure that the execution order and amount of fertilizer application do not conflict. Specifically, conflicts between baseline compensation strategies are identified and corrected. For example, if multiple fertilizer demand gap nodes require fertilization at the same time, the conflict can be resolved by adjusting the fertilization time or the amount of fertilizer applied. The goal of conflict correction is to ensure that all fertilization needs can be executed smoothly without unnecessary waste or inefficiency due to excessive fertilizer application or overlapping times. After conflict correction, the final fertilization compensation execution sequence is obtained. This sequence ensures that at each time point, fertilization needs are reasonably arranged, all fertilizer needs can be met, and the fertilization process is efficient and conflict-free.

[0045] Furthermore, the microbial community time series matrix is ​​loaded into a microbial interaction network for fertility and nutrient association transformation to obtain a fertility and nutrient time series matrix. Prior to this, the method includes:

[0046] Based on fertility contribution evaluation, a first group of associated microorganisms for the first fertilizer requirement is selected; the first fertilizer requirement and the first group of associated microorganisms are used as dual search constraints, and multiple sample first fertility values ​​and multiple associated microbiome data are locally retrieved; multiple regression analysis is performed on the multiple sample first fertility values ​​and multiple associated microbiome data to output a first single fertilizer function; similarly, F single fertilizer functions for the F types of fertilizer requirements are constructed, where F≥M and F is a positive integer; multi-function collaborative verification is performed on the F single fertilizer functions to output F single fertilizer correction functions; F parallel microbial interaction channels are constructed, and after using the F types of fertilizer requirements to identify the F microbial interaction channels, the F single fertilizer correction functions are mapped and migrated to the F microbial interaction channels to complete the construction of the microbial interaction network.

[0047] Fertility contribution assessment evaluates the role of microorganisms in fertility formation and maintenance by analyzing the impact of microbial communities on nutrient transformation under different soil conditions. Different microorganisms promote soil fertility through different metabolic pathways (such as nitrogen fixation, phosphorus hydrolysis, and decomposition of organic matter). Fertility contribution assessment can be conducted in various ways, including experimental data analysis (such as the determination of microbial activity in soil samples), functional gene expression analysis (such as nitrogen fixation genes, phosphorus solubilization genes, etc.), and microbial community structure analysis. Based on this, by calculating the degree of contribution of microbial communities to soil fertility, microbial communities that make a significant contribution to the primary fertilizer requirement are selected to obtain the first group of associated microorganisms. For example, if the primary fertilizer requirement is nitrogen, then those microorganisms that can participate in nitrogen fixation, nitrogen transformation, or nitrogen release are screened. These microorganisms directly or indirectly improve the effectiveness of the fertilizer in the soil through their metabolic activities, thereby affecting crop growth.

[0048] Using the primary fertilizer requirement and the first group of associated microorganisms as dual search constraints, local data is retrieved. This involves combining fertilizer type and microbiome data to find relevant sample data, ensuring that the obtained fertilizer and microbiome data are specific to particular fertilizers and microbiome groups. By retrieving and calling multiple sample data from the database, relevant fertility values ​​(such as concentration data of elements like nitrogen, phosphorus, and potassium) and microbiome data (i.e., soil microbial community data related to the first group of microorganisms) are obtained. This data can come from different geographical locations, different time points, different soil types, and different crop species. By calling this data, a more refined analysis of the relationship between fertilizers and microorganisms can be performed.

[0049] Multiple regression analysis is a commonly used statistical method to reveal the relationship between multiple independent variables (in this case, multiple microbial community data) and dependent variables (such as soil fertility values ​​or fertilizer requirements). Through regression analysis, the contribution of each variable in the microbial community to fertilizer requirements can be found. Through multiple regression analysis, a mathematical model of the relationship between fertilizer requirements and the microbial community can be established, namely the first single fertilizer function. This function can describe the changes in fertilizer requirements in the soil under different microbial community configurations. For example, some microorganisms can improve nitrogen availability and reduce fertilizer requirements, while other microorganisms are positively correlated with the requirements of a certain specific fertilizer.

[0050] Different types of fertilizers (such as nitrogen, phosphorus, and potassium) have different effects on soil fertility and microbial communities. Therefore, in addition to the first type of fertilizer required, for F types of fertilizers required, F single-fertilizer functions are constructed in the same way. Each single-fertilizer function is generated through multiple regression analysis, specifically depending on the interaction between various fertilizers and the microbial community. Each single-fertilizer function describes the relationship between the corresponding fertilizer and the microbial community and can quantify the impact of the microbial community on the demand for a specific fertilizer. In this step, F represents the number of fertilizer types required, and M represents the number of fertility demand gap nodes. Typically, F is greater than or equal to M, and F is a positive integer, which means that more than M types of fertilizers need to be modeled.

[0051] Multi-function co-validation of F single-fertilizer functions aims to verify whether these functions can effectively cooperate within the same system. The application of different fertilizers is not only affected by their respective microbial communities but may also generate feedback effects through interactions with other fertilizers. During the co-validation process, the functions are adjusted based on multiple factors, such as the influence of microbial communities, fertilizer type interactions, and fertilization time. Through co-validation, each single-fertilizer function can be optimized to complement each other and ensure that there are no conflicts in the same fertilization strategy. Finally, F single-fertilizer correction functions are output. After optimization, the single-fertilizer correction functions can more accurately reflect the synergistic relationship between microbial communities and fertilizer requirements, thereby improving the effectiveness and sustainability of the fertilization strategy.

[0052] For each type of fertilizer requirement, a corresponding independent microbial interaction channel is constructed. Each channel expresses the relationship between a specific fertilizer and the microbial community, reflecting how microorganisms participate in fertilizer transformation and release. Through these channels, the transformation process of different fertilizers in the soil can be accurately simulated, and the microbial community's response to fertilizers can be optimized. In each microbial interaction channel, the F types of fertilizers required are used to identify the corresponding channel. This step associates each fertilizer requirement with its corresponding microbial channel and migrates the previously co-validated F single-fertilizer correction functions to the corresponding channels. This process ensures that the correction function for each fertilizer matches its corresponding microbial community interaction. After the above steps are completed, the microbial interaction network is constructed. This network describes the interactions between different fertilizer types and the microbial community, enabling dynamic adjustment of fertilizer use based on soil conditions, crop needs, and fertilizer application plans to achieve precision fertilization and optimize the crop growth environment.

[0053] Furthermore, the microbial community time series matrix is ​​loaded into a microbial interaction network for fertility and nutrient association transformation to obtain a fertility and nutrient time series matrix. The method includes:

[0054] After initial intermittent stratified sampling is performed on the soil sampling matrix, real-time microbiome data is measured to obtain a first microbial community matrix, wherein each matrix node records node microbiome data. The node microbiome data of each matrix node in the first microbial community matrix is ​​synchronized to the microbial interaction network for fertility and nutrient association transformation, and the node fertility value array of each matrix node is calculated and output to form a first fertility and nutrient matrix. In this way, after obtaining H fertility and nutrient matrices, the H fertility and nutrient matrices are spliced ​​in time to obtain the fertility and nutrient time-series matrix.

[0055] Soil fertility is influenced by the microbial community, and the composition and function of the microbial community may vary across different soil layers and sampling points. Initial intermittent stratified sampling is performed on a soil sampling matrix. This method ensures reasonable data collection from different soil layers, thereby capturing soil diversity and heterogeneity. The microbial community data recorded at each matrix node reflects the microbial composition of that soil layer. After sampling, real-time microbiome data analysis is performed, for example using high-throughput sequencing technologies such as 16S rRNA sequencing or metagenomic sequencing, to analyze the microbiome at each matrix node. This data includes the diversity, abundance, and functional characteristics of the microbial population. The data from each matrix node constitutes a node in the first microbial community matrix.

[0056] The matrix microbiome data of each matrix node in the first microbial community matrix is ​​synchronized to the microbial interaction network. The microbial interaction network evaluates the contribution of the microbial community to fertilizer transformation and soil fertility based on the interactions between different microorganisms, such as competition, cooperation, and inhibition, as well as the functional characteristics of the microorganisms. It calculates and outputs the node fertility value array of each matrix node. This node fertility value array reflects the soil fertility status of each node, taking into account the nutrient content in the soil and the role of the microbial community. Finally, the node fertility value arrays of multiple nodes constitute the first fertility nutrient matrix.

[0057] Following the steps described above, a corresponding fertility and nutrient matrix is ​​generated for the microbial community data at each time point. This matrix describes the fertility status of each sampling point and incorporates the influence of the microbial community. This process is repeated H times, with a new fertility and nutrient matrix generated each time based on the sampling data and microbial assay results from different time points. The resulting H fertility and nutrient matrices are then concatenated in chronological order to form a fertility and nutrient time-series matrix. This time-series matrix demonstrates the dynamic changes in soil fertility at multiple time points, providing data support for subsequent fertilization strategies and soil management.

[0058] Furthermore, based on linear interpolation normalization of the fertility nutrient time series matrix, F soil fertility time series data are obtained, the method comprising:

[0059] Based on the F types of fertilizer requirements, the fertility nutrient time series matrix is ​​decomposed to obtain F single fertilizer nutrient time series matrices; a sampling confidence weight matrix is ​​assigned to the soil sampling matrix; after performing single nutrient single node interpolation on the F single fertilizer nutrient time series matrices, the interpolation results are normalized based on the sampling confidence weight matrix, and the F soil fertility time series data are output.

[0060] The fertility nutrient time series matrix contains the nutrient changes of various fertilizers. Based on F types of fertilizer requirements, the fertility nutrient time series matrix is ​​decomposed into F single fertilizer nutrient time series matrices according to different fertilizer requirements. For example, if the fertilizer requirements include nitrogen, phosphorus and potassium fertilizers, each single fertilizer nutrient time series matrix only records the nutrient changes related to nitrogen fertilizer, phosphorus fertilizer or potassium fertilizer.

[0061] The sampling confidence weight matrix is ​​used to represent the data reliability of each sampling point. The confidence level of a sampling point is affected by various factors, such as soil heterogeneity, sampling depth, and climatic conditions. Confidence weights are obtained through soil type, historical data, or model predictions. Specifically, each sampling point is assigned a confidence value to represent the reliability of the data. For example, sampling points from stable areas or with historical monitoring data have higher confidence weights, while areas with scarce data or poor conditions may have lower confidence levels. Based on the reliability of the sampling points, a sampling confidence weight matrix is ​​constructed. Each element in the matrix represents the confidence weight of the corresponding sampling point, which can be a value between 0 and 1, with values ​​closer to 1 indicating higher data reliability.

[0062] To ensure the continuity and accuracy of soil fertility data, interpolation is performed on the data in each single fertilizer nutrient time-series matrix. Interpolation methods such as linear interpolation and spline interpolation can be used to fill in blank data caused by missing data or uneven sampling points. After interpolation, the interpolation results are normalized according to the sampling confidence weight matrix corresponding to the soil sampling matrix. The normalization process adjusts the weights based on the sampling confidence weight matrix; that is, sampling points with higher weights (points with higher confidence) have a larger proportion in the interpolation results, while points with lower confidence have a smaller impact on the results. After interpolation and normalization, F soil fertility time-series data are output. These data reflect the impact of different fertilizers on soil fertility and are continuous in time, providing a basis for subsequent fertilization strategies and soil management.

[0063] Furthermore, the method involves capturing the dynamic evolution patterns of the F soil fertility time-series data and predicting and outputting the F soil fertility evolution curves.

[0064] The F soil fertility time-series data are processed graphically to output F crop fertility time-series curves; fertility state transition analysis is performed on the F crop fertility time-series curves to output F fertility time-series dynamic features; the evolution and update of the F crop fertility time-series curves are performed based on the F fertility time-series dynamic features to output the F soil fertility evolution curves.

[0065] Each soil fertility time series data point represents the change in soil fertility at different points in time. To facilitate analysis and presentation, these time series data points are visualized using line graphs, curve graphs, and other methods to show the change of each fertility data point over time. For example, with time as the x-axis and soil fertility value as the y-axis, adjacent data points are connected by lines in the coordinate system to form a crop fertility time series curve. A separate crop fertility time series curve can be generated for each type of fertilizer. These curves show the trend of soil fertility change over time.

[0066] During the crop growth cycle, soil fertility changes in different states, such as increasing, stabilizing, or decreasing. Fertility state transition analysis reveals the patterns of soil fertility change by identifying transformation patterns in fertility data. For example, time series models are used to analyze the trends and periodic changes in fertility time-series curves, and abrupt changes or points of change in the soil fertility curves are detected to identify fertility state transitions. After completing the fertility state transition analysis, F dynamic features of fertility time series are extracted, including information on soil fertility change patterns, inflection points, and periodic fluctuations. These features can help determine the trend of soil fertility changes over different time periods, identify key fertilization opportunities, and provide support for fertilization decisions.

[0067] Based on the extracted fertility time-series dynamic characteristics (such as fertility change patterns, inflection points, periodic fluctuations, etc.), the fertility time-series curve for each crop is updated. This means that the shape of each time-series curve is adjusted in combination with dynamic characteristics to more accurately reflect the actual evolution of soil fertility. The update can be achieved by correcting the slope, inflection points, or period of the curve to ensure that the curve better matches the actual situation of soil fertility changes. After the update is completed, F soil fertility evolution curves are obtained. These curves show the evolution trend of soil fertility throughout the entire crop growth cycle, including the influence of factors such as fertilization, soil restoration, and microbial activity.

[0068] Furthermore, by overlaying the F soil fertility evolution curves and the F growth fertility demand curves, nutrient gaps are located to obtain M fertility demand deviations at M fertility demand gap nodes. The method includes:

[0069] The system interactively obtains F demand deviation thresholds for the F types of fertilizers; based on the fertility compensation cycle, it overlays the F soil fertility evolution curves and F growth fertility demand curves, locates nutrient gaps within the fertility compensation cycle according to the F demand deviation thresholds, and outputs the M fertility demand gap nodes; it performs demand deviation extension analysis on the M fertility demand gap nodes and outputs the M fertility demand deviation amounts, wherein the M fertility demand deviation amounts are mapped to values ​​greater than the M demand deviation thresholds.

[0070] The demand deviation threshold refers to the maximum acceptable deviation between soil fertility and crop fertilizer requirements during crop growth. When the deviation exceeds this threshold, it indicates that the soil fertility can no longer meet the crop's needs and compensation is required. The demand deviation threshold is determined based on historical data, expert experience, or simulation results. For example, some crops are more sensitive to nitrogen and phosphorus requirements at certain growth stages, so setting a smaller deviation threshold is more appropriate. Specifically, the demand deviation threshold for each fertilizer can be flexibly adjusted according to variables such as soil type, crop type, and climate.

[0071] By overlaying F soil fertility evolution curves with F growth fertility demand curves, a one-to-one correspondence is established between the soil fertility and crop fertility demand curves within the fertility compensation cycle. The resulting curves are then compared, and fertility demand gaps are identified based on the demand deviation threshold for each fertilizer. In other words, if soil fertility deficiency exceeds the set demand deviation threshold at a given time point, that time point is considered to have a fertility demand gap. This method identifies which periods throughout the crop's growth cycle require additional fertilizer. Based on the location of these fertility demand gaps, M fertility demand gap nodes are output. Each node represents a time point where soil fertility cannot meet the crop's needs, and the gap exceeds the set deviation threshold.

[0072] At each fertility demand gap node, an extended demand deviation analysis is performed. The goal is to assess the persistence of fertility demand and the trend of gap changes after the gap occurs. This extended demand deviation analysis can be conducted using dynamic analysis methods, such as time series analysis, regression models, or other predictive methods, to estimate the duration and magnitude of the fertility gap. Based on the extended analysis results, M fertility demand deviation quantities are output. These deviation quantities represent the specific differences in each fertility demand gap, indicating the severity of the gap. Each deviation quantity is mapped to a corresponding demand deviation threshold, indicating which gaps have deviations exceeding the set threshold. This helps identify which regions or stages require larger-scale fertilizer supplementation.

[0073] Furthermore, the method includes performing application conflict correction on the M baseline compensation strategies and outputting a corrected application sequence as the fertilizer compensation execution sequence.

[0074] The system interactively obtains F fertilization suppression conflict restrictions for the F types of fertilizers required; it calls M fertilization suppression conflict restrictions from the F fertilization suppression conflict restrictions according to the M benchmark compensation strategies; it performs application conflict correction on the M benchmark compensation strategies according to the M fertilization suppression conflict restrictions, and outputs the fertilization compensation execution sequence.

[0075] Fertilizer antagonism limits refer to the potential mutual antagonistic effects that may occur when different fertilizers are applied within the same time period or spatial range. The composition or application method of different fertilizers may cause mutual interference or inhibition, affecting fertilizer conversion or absorption. For example, the application of nitrogen and potassium fertilizers may antagonize each other under certain soil conditions, or the application of phosphorus fertilizer may conflict with the application of certain micronutrients. Fertilizer antagonism limits for each fertilizer (such as nitrogen, phosphorus, and potassium) are obtained by combining inputs from agricultural expert systems, soil management databases, climate models, or farm managers. These limits are generated based on scientific literature, field experiment data, or simulation results from soil and crop models.

[0076] For each baseline compensation strategy (i.e., the previously determined fertilization plan), the corresponding fertilization inhibition conflict constraint is invoked from F fertilization inhibition conflict constraints based on its fertilization content (i.e., the type and timing of the applied fertilizer). For example, if a baseline compensation strategy includes the application of nitrogen and potassium fertilizers, and the application of these two fertilizers at the same time would cause a conflict, then this conflict should be identified and adjusted according to the fertilization inhibition conflict constraint. After analyzing each baseline compensation strategy and invoking the corresponding fertilization inhibition conflict constraint, M fertilization inhibition conflict constraints are output. Each constraint contains specific constraints on fertilization, such as fertilization time, fertilizer amount, and fertilizer type.

[0077] Based on known fertilization inhibition conflict constraints, application conflict correction is performed on M baseline compensation strategies. The goal is to eliminate or mitigate the mutual inhibition effects of different fertilizers, thereby improving fertilization efficiency and ensuring that crop nutrient requirements are met. For example, if two fertilizers applied at the same time would inhibit each other, the application time interval can be adjusted; if the application of certain fertilizers is excessive, potentially causing conflicts, the application amount of the corresponding fertilizer needs to be reduced. Based on the conflict constraints, the optimal fertilizer combination is selected to avoid the co-application of incompatible fertilizers. After conflict correction, a final fertilization compensation execution sequence is generated. This sequence includes information such as the time, amount, and type of fertilization operation, ensuring that the fertilization process meets the crop's nutrient requirements while avoiding conflicts and inhibition effects between fertilizers.

[0078] In summary, the microbiome data-driven soil fertility prediction method provided in this application has the following technical effects:

[0079] By dynamically adjusting the sampling grid size, refined soil sampling can be performed based on crop planting density and soil characteristics. This ensures the representativeness and accuracy of the sampling data, comprehensively reflecting the spatial distribution of soil fertility. During the fertility compensation cycle, combining microbiome data measurement provides high-frequency soil microbial information for fertility prediction, thereby enhancing the timeliness and accuracy of the prediction results. By processing the microbial community time-series matrix for nutrient transformation, the impact of soil microbial activity on nutrient transformation can be revealed, providing a scientific basis for subsequent soil fertility prediction and making the prediction results closer to the actual changes in soil fertility. By capturing dynamic evolution patterns, soil fertility prediction is not limited to current data but also considers the trend of soil fertility changes over time. The predicted soil fertility evolution curve reflects… It can accurately predict soil fertility changes at different time points, demonstrating a high degree of time-series forecasting capability. By combining the latitude and longitude coordinates of the planting area and crop type, and using a dual retrieval constraint method, it can accurately obtain the fertilizer demand curve of the target crop. This allows for optimization of fertilization plans based on specific crop requirements and growth cycles. By overlaying the soil fertility evolution curve with the crop fertility demand curve, it can pinpoint nutrient gaps. This process can accurately identify the points of soil fertility insufficiency and fertilizer demand gaps, thereby guiding the formulation of subsequent fertilization strategies. By analyzing the deviation in fertility demand and optimizing fertilization compensation strategies based on this data, fertilization becomes more precise and effective. The optimized fertilization compensation execution sequence ensures timely adjustments to fertilizer supply based on the differences between soil fertility demand and crop demand, avoiding resource waste and improving fertilization efficiency.

[0080] Example 2, based on the same inventive concept as the microbiome data-driven soil fertility prediction method in the foregoing examples, such as... Figure 2 As shown in the embodiments of this application, a soil fertility prediction system driven by microbiome data is provided, the system comprising:

[0081] The soil sampling matrix configuration module 10 is used to dynamically adjust the sampling grid size according to the crop planting density and configure the soil sampling matrix in the planting area.

[0082] The data measurement module 20 is used to take 1 / H of the fertility compensation cycle as the soil sampling interval window, perform intermittent stratified sampling at frequency H in the soil sampling matrix, and perform real-time microbiome data measurement to obtain a microbial community time series matrix.

[0083] The pattern capture module 30 is used to capture the dynamic evolution pattern after processing the microbial community time series matrix for fertility nutrient conversion, and predict and output F soil fertility evolution curves.

[0084] The demand curve calling module 40 is used to use the latitude and longitude coordinates of the planting area and the crop type as dual search constraints to call the F growth fertility demand curves of the target crop for F types of fertilizers.

[0085] The nutrient gap location module 50 is used to locate the nutrient gap by overlapping the F soil fertility evolution curves and the F growth fertility demand curves, and to obtain the M fertility demand deviations of the M fertility demand gap nodes.

[0086] The compensation strategy optimization module 60 is used to optimize the fertilization compensation strategy for the next fertility compensation cycle based on the M fertility demand deviations, and obtain the fertilization compensation execution sequence.

[0087] Furthermore, the pattern capture module 30 is used to perform the following operation steps:

[0088] The microbial community time series matrix is ​​loaded into a microbial interaction network for fertility and nutrient correlation transformation to obtain a fertility and nutrient time series matrix; the fertility and nutrient time series matrix is ​​normalized based on linear interpolation to obtain F soil fertility time series data; the dynamic evolution law of the F soil fertility time series data is captured, and the F soil fertility evolution curves are predicted and output.

[0089] Furthermore, the compensation strategy optimization module 60 is used to perform the following operation steps:

[0090] Using the end time of the F soil fertility evolution curves as the compensation starting point, calculate the M fertility compensation time limits for the M fertility demand gap nodes; optimize the fertilization compensation strategy based on the M fertility compensation time limits and the M fertility demand deviations to obtain M benchmark compensation strategies; perform application conflict correction on the M benchmark compensation strategies and output the corrected application sequence as the fertilization compensation execution sequence.

[0091] Furthermore, the pattern capture module 30 is used to perform the following operation steps:

[0092] Based on fertility contribution evaluation, a first group of associated microorganisms for the first fertilizer requirement is selected; the first fertilizer requirement and the first group of associated microorganisms are used as dual search constraints, and multiple sample first fertility values ​​and multiple associated microbiome data are locally retrieved; multiple regression analysis is performed on the multiple sample first fertility values ​​and multiple associated microbiome data to output a first single fertilizer function; similarly, F single fertilizer functions for the F types of fertilizer requirements are constructed, where F≥M and F is a positive integer; multi-function collaborative verification is performed on the F single fertilizer functions to output F single fertilizer correction functions; F parallel microbial interaction channels are constructed, and after using the F types of fertilizer requirements to identify the F microbial interaction channels, the F single fertilizer correction functions are mapped and migrated to the F microbial interaction channels to complete the construction of the microbial interaction network.

[0093] Furthermore, the pattern capture module 30 is used to perform the following operation steps:

[0094] After initial intermittent stratified sampling is performed on the soil sampling matrix, real-time microbiome data is measured to obtain a first microbial community matrix, wherein each matrix node records node microbiome data. The node microbiome data of each matrix node in the first microbial community matrix is ​​synchronized to the microbial interaction network for fertility and nutrient association transformation, and the node fertility value array of each matrix node is calculated and output to form a first fertility and nutrient matrix. In this way, after obtaining H fertility and nutrient matrices, the H fertility and nutrient matrices are spliced ​​in time to obtain the fertility and nutrient time-series matrix.

[0095] Furthermore, the pattern capture module 30 is used to perform the following operation steps:

[0096] Based on the F types of fertilizer requirements, the fertility nutrient time series matrix is ​​decomposed to obtain F single fertilizer nutrient time series matrices; a sampling confidence weight matrix is ​​assigned to the soil sampling matrix; after performing single nutrient single node interpolation on the F single fertilizer nutrient time series matrices, the interpolation results are normalized based on the sampling confidence weight matrix, and the F soil fertility time series data are output.

[0097] Furthermore, the pattern capture module 30 is used to perform the following operation steps:

[0098] The F soil fertility time-series data are processed graphically to output F crop fertility time-series curves; fertility state transition analysis is performed on the F crop fertility time-series curves to output F fertility time-series dynamic features; the evolution and update of the F crop fertility time-series curves are performed based on the F fertility time-series dynamic features to output the F soil fertility evolution curves.

[0099] Furthermore, the nutrient gap location module 50 is used to perform the following operation steps:

[0100] The system interactively obtains F demand deviation thresholds for the F types of fertilizers; based on the fertility compensation cycle, it overlays the F soil fertility evolution curves and F growth fertility demand curves, locates nutrient gaps within the fertility compensation cycle according to the F demand deviation thresholds, and outputs the M fertility demand gap nodes; it performs demand deviation extension analysis on the M fertility demand gap nodes and outputs the M fertility demand deviation amounts, wherein the M fertility demand deviation amounts are mapped to values ​​greater than the M demand deviation thresholds.

[0101] Furthermore, the compensation strategy optimization module 60 is used to perform the following operation steps:

[0102] The system interactively obtains F fertilization suppression conflict restrictions for the F types of fertilizers required; it calls M fertilization suppression conflict restrictions from the F fertilization suppression conflict restrictions according to the M benchmark compensation strategies; it performs application conflict correction on the M benchmark compensation strategies according to the M fertilization suppression conflict restrictions, and outputs the fertilization compensation execution sequence.

[0103] Through the foregoing detailed description of the microbiome data-driven soil fertility prediction method, those skilled in the art can clearly understand the microbiome data-driven soil fertility prediction system in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0104] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for soil fertility prediction driven by microbiome data, characterized in that, The method comprises: According to the crop planting density, the size of the sampling grid is dynamically adjusted, and a soil sampling matrix is configured in the planting area; 1 / H of the fertilizer compensation period is taken as a soil sampling interval window, intermittent stratified sampling of H frequencies is performed on the soil sampling matrix, instant microbial community data measurement is performed, and a microbial community time sequence matrix is obtained; After the microbial community time sequence matrix is subjected to fertilizer nutrient conversion processing, dynamic evolution law capturing is performed, and F soil fertility evolution curves are predicted and output, specifically including: based on fertilizer contribution evaluation, a first group of associated microorganisms of a first demand fertilizer is screened; the first demand fertilizer and the first group of associated microorganisms are taken as double retrieval constraints, a plurality of sample corresponding first fertility values and a plurality of associated microbiome data are locally called; the plurality of sample corresponding first fertility values and the plurality of associated microbiome data are subjected to multivariate regression analysis, and a first single fertilizer function is output; in the same way, F single fertilizer functions of F demand fertilizers are constructed, wherein F is a positive integer; the F single fertilizer functions are subjected to multi-function collaborative verification, and F single fertilizer correction functions are output; for the F demand fertilizers, F independent microbial interaction channels corresponding to the F demand fertilizers are constructed, the microbial interaction channels are used to express the relationship between the fertilizers and the microbial community, and reflect the process of microbial participation in fertilizer conversion and release; after the F demand fertilizers are used to identify the corresponding microbial interaction channels, the F single fertilizer correction functions are mapped and migrated to the F microbial interaction channels, and the construction of the microbial interaction network is completed; then, the microbial community time sequence matrix is loaded to the microbial interaction network for fertilizer nutrient association conversion, and a fertilizer nutrient time sequence matrix is obtained, specifically including: the microbial community time sequence matrix includes a first microbial community matrix obtained by performing instant microbiome data measurement, wherein each matrix node records node microbiome data; the node microbiome data of each matrix node in the first microbial community matrix is synchronized to the microbial interaction network for fertilizer nutrient association conversion, and a node fertility value array of each matrix node is calculated and output to constitute a first fertilizer nutrient matrix; in the same way, after H fertilizer nutrient matrices are obtained, the H fertilizer nutrient matrices are time sequenced to obtain the fertilizer nutrient time sequence matrix; the fertilizer nutrient time sequence matrix is normalized based on linear interpolation to obtain F soil fertility time sequence data; the dynamic evolution law of the F soil fertility time sequence data is captured, and the F soil fertility evolution curves are predicted and output; The longitude and latitude coordinates and the crop type of the planting area are taken as double retrieval constraints, and F growth fertility demand curves of the target crop for F demand fertilizers are called on a network; By overlapping the F soil fertility evolution curves and the F growth fertility demand curves, nutrient gap positioning is performed, and M fertilizer demand gap nodes and M fertilizer demand deviation amounts are obtained, wherein F≥M; According to the M fertilizer demand deviation amounts, a fertilization compensation strategy optimization of a next fertilizer compensation period is performed, and a fertilization compensation execution sequence is obtained.

2. The microbiome data-driven soil fertility prediction method of claim 1, wherein, The fertilization compensation strategy optimization of the next fertilization compensation period is performed according to the M fertilization demand deviation amounts, and a fertilization compensation execution sequence is obtained, specifically including: The curve end time of the F soil fertility evolution curves is taken as a compensation starting point, and M fertilization compensation time limits of the M fertilization demand gap nodes are calculated; M reference compensation strategies are obtained by performing fertilization compensation strategy optimization according to the M fertilization compensation time limits and the M fertilization demand deviation amounts; The M reference compensation strategies are subjected to application conflict correction, and a corrected application sequence is output as the fertilization compensation execution sequence.

3. The microbiome data-driven soil fertility prediction method of claim 1, wherein, The F soil fertility time sequence data is obtained by normalizing the fertilization nutrient time sequence matrix based on linear interpolation, specifically including: The F soil fertility time sequence data is obtained by normalizing the fertilization nutrient time sequence matrix based on linear interpolation, specifically including: The sampling confidence weight matrix is assigned to the soil sampling matrix; After the single-nutrient single-node interpolation processing of the F single-fertilizer nutrient time sequence matrices, the interpolation result normalization processing is performed based on the sampling confidence weight matrix, and the F soil fertility time sequence data is output.

4. The microbiome data-driven soil fertility prediction method of claim 1, wherein, The F soil fertility evolution curves are predicted and output by capturing the dynamic evolution law of the F soil fertility time sequence data, specifically including: The F soil fertility time sequence data is image-processed, and F crop fertility time sequence curves are output; The F crop fertility time sequence curves are subjected to fertility state transfer analysis, and F fertility time sequence dynamic characteristics are output; The F soil fertility evolution curves are obtained by updating the F crop fertility time sequence curves according to the F fertility time sequence dynamic characteristics.

5. The microbiome data-driven soil fertility prediction method of claim 1, wherein, M fertilization demand deviation amounts of M fertilization demand gap nodes are obtained by positioning the nutrient gap by overlapping the F soil fertility evolution curves and F growth fertility demand curves, specifically including: The F demand deviation thresholds of the F demand fertilizers are obtained interactively; when the deviation exceeds the demand deviation threshold, it indicates that the soil fertility cannot meet the demand of the crops; After the F soil fertility evolution curves and F growth fertility demand curves are overlapped according to the time sequence of the fertilization compensation period, the nutrient gap is positioned according to the F demand deviation thresholds within the fertilization compensation period, and the M fertilization demand gap nodes are output; The M fertilization demand deviation amounts are output by performing demand deviation extension analysis at the M fertilization demand gap nodes.

6. The microbiome data-driven soil fertility prediction method of claim 2, wherein, The M reference compensation strategies are subjected to application conflict correction, and a corrected application sequence is output as the fertilization compensation execution sequence, specifically including: The F fertilization inhibition conflict limits of the F demand fertilizers are obtained interactively; M fertilization inhibition conflict limits are called from the F fertilization inhibition conflict limits according to the M reference compensation strategies; The M reference compensation strategies are subjected to application conflict correction according to the M fertilization inhibition conflict limits, and the fertilization compensation execution sequence is output.

7. A microbiome data-driven soil fertility prediction system characterized in that, The system for implementing the microbial community data-driven soil fertility prediction method of any one of claims 1-6, the system comprising: A soil sampling matrix configuration module is configured to dynamically adjust the sampling grid size according to the crop planting density, and configure a soil sampling matrix in the planting area; a data measurement module configured to take 1 / H of the fertility compensation period as a soil sampling interval window, perform intermittent stratified sampling of H frequency on the soil sampling matrix, and perform real-time microbiome data measurement to obtain a microbial community time series matrix; a rule capturing module configured to perform dynamic evolution rule capturing after performing fertility nutrient conversion processing on the microbial community time series matrix, and output F soil fertility evolution curves; a demand curve calling module configured to take the latitude and longitude coordinates and crop type of the planting area as dual retrieval constraints, and call F growth fertility demand curves of the target crop for F required fertilizers in a networked manner; a nutrient gap positioning module configured to perform nutrient gap positioning by overlapping the F soil fertility evolution curves and the F growth fertility demand curves, and obtain M fertility demand gap nodes and M fertility demand deviation amounts; a compensation strategy optimization module configured to perform fertilization compensation strategy optimization of the next fertility compensation period according to the M fertility demand deviation amounts, and obtain a fertilization compensation execution sequence.

Images