A method for predicting release of nutrients under water and fertilizer coupling conditions of straw returning to field

Abstract

The present application relates to the field of data analysis, and more particularly to a method for predicting nutrient release of straw returning to field under water and fertilizer coupling conditions, which comprises: acquiring and segmenting soil environment time series data to obtain environmental segment feature data set; quantitatively evaluating the temperature and humidity synergy of the environmental segment to obtain a correction factor of the proportion of the synergy window; analyzing the main limiting factor structure of the environmental segment to obtain a correction factor of the main limiting factor structure; jointly correcting the standard Euclidean distance according to the synergy and the limitation to obtain a modified distance of the decomposition mode; and performing decomposition mode clustering and calibration analysis on the environmental segment to obtain a nutrient release prediction result, thereby solving the mode confusion problem caused by the fact that the existing clustering based on statistical characteristics cannot distinguish the difference in synergy and the difference in main limiting factor.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method for predicting nutrient release from straw returned to the field under water-fertilizer coupling conditions. Background Technology

[0002] In modern intensive and precision agricultural production, straw return to the field has become an important technical approach to increase soil organic matter content, improve soil physicochemical properties, and realize the resource utilization of agricultural by-products. Especially with the increasing application of coupled irrigation models such as integrated water and fertilizer management, drip irrigation, and soil moisture control in field production, the rhizosphere environment of farmland exhibits significant temporal fluctuations and multi-factor coupling. The decomposition process of straw in the soil is no longer a slow, stable, single-factor controlled process, but is simultaneously affected by multiple factors such as temperature, humidity, tillage system, and fertilization methods. In this scenario, accurately predicting the mineralization release level of organic nutrients (such as nitrogen, phosphorus, and potassium) in straw over a certain time scale based on real-time or near-real-time soil environmental monitoring data has direct production guidance significance and application value for determining the timing and amount of topdressing, correcting crop nutrient balance models, reducing redundant fertilizer input, and mitigating the resulting non-point source pollution. Therefore, developing an engineeringable, deployable, and interoperable technical route for predicting nutrient release from straw return based on environmental monitoring data is an important direction in current precision planting management. In existing technologies, in order to extract representative environmental patterns of straw decomposition from complex, multi-period soil environmental data, a common data mining process of "sensor acquisition - time window division - feature extraction - cluster recognition - pattern calibration" is usually adopted. A typical approach involves deploying soil environmental monitoring sensors in the target farmland area to continuously collect soil temperature and moisture data at fixed time intervals (e.g., 1 hour, 30 minutes), forming a multivariate time series. This long-term series is then divided into several independent environmental segments according to fixed time window lengths (e.g., 24h, 48h, or 72h). Subsequently, statistical measures reflecting the overall state of each environmental segment, such as mean temperature, temperature variance, mean humidity, and humidity variance, are calculated to form a feature vector describing that segment. Finally, the feature vectors of all environmental segments are input into unsupervised clustering algorithms such as K-means for automatic classification. Environmental segments grouped into the same cluster are considered to be under similar decomposition conditions. Through plot experiments or historical field calibration, a correspondence is established between this type of environmental pattern and a certain range of nutrient release rates, thus completing the modeling process of inferring nutrient release levels from environmental patterns. This approach has the advantages of clear structure, low implementation cost, and ease of integration with existing agricultural IoT and database systems, and has therefore been widely adopted in agricultural big data analysis, intelligent irrigation scheduling, and variable fertilization decision-making systems. However, the above-mentioned technical approach based on statistical characteristics and standard K-means clustering has a fundamental mismatch with the mechanism of straw decomposition.The decomposition of straw in soil is a biochemical process dominated by soil microbial activity. This process's dependence on temperature and humidity is not an independent linear relationship, but rather a typical two-factor synergistic relationship: the decomposition rate can only be maintained at a high level when both soil temperature and humidity are within the optimal range for microbial metabolism and activity. If either factor fails to meet the optimal range for a given period, the effective decomposition contribution for that period will be significantly suppressed, exhibiting a typical bottleneck effect. However, the standard Euclidean distance commonly used in existing technologies is a dimensionally independent additive metric. When comparing the similarity of two environmental segments, it only calculates the difference in mean temperature, the difference in mean humidity, and the difference in their respective variances, and then sums the squares. This fails to identify whether temperature and humidity simultaneously meet the optimal range at the same time point, a crucial factor for decomposition efficiency. This leads to a recurring error in actual production: two environmental segments are statistically very similar. One is an asynchronous, limited pattern where temperature and humidity alternately reach their optimal ranges at different times; the other is a synchronous, optimal pattern where temperature and humidity are simultaneously within their optimal ranges for most of the window. However, due to their similar means and variances, they are incorrectly grouped into the same clustering class in existing algorithms. The incorrectly merged class actually contains samples with different biological mechanisms and significantly different actual decomposition capabilities, causing a systematic bias in the subsequent nutrient release rate mapping based on that class. This weakens the predictive value for guiding field water and fertilizer regulation and fertilizer formulation optimization. Furthermore, for environmental segments also identified as having low synergy, existing methods cannot further identify the true cause of the low synergy—whether it's due to persistently low temperature, insufficient water, or excessive humidity and hypoxia. This further limits the direct conversion of clustering results into specific agronomic measures (increased watering, heat preservation, drainage, etc.). Therefore, existing environmental segment clustering methods based on general distance metrics are insufficient to meet the practical needs of predicting nutrient release from straw return under water-fertilizer coupling conditions, which require perceptibility of temperature and humidity synergy, distinguishability of principal constraint factors, and interpretability of clustering results. There is an urgent need to make specific improvements to the distance metric mechanism for this scenario. Summary of the Invention

[0003] In view of this, the present invention aims to propose a method for predicting nutrient release from straw returned to the field under water-fertilizer coupling conditions, in order to solve the pattern confusion problem caused by the inability of existing clustering based on statistical features to distinguish between differences in synergy and differences in principal constraint factors.

[0004] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0005] A method for predicting nutrient release from straw return to the field under water-fertilizer coupling conditions, the method comprising:

[0006] Step S1: Obtain environmental segment feature datasets by collecting and segmenting soil environmental time-series data;

[0007] Step S2: Obtain the correction factor for the proportion of the synergistic window by quantitatively evaluating the temperature and humidity synergy of environmental segments;

[0008] Step S3: Obtain the principal constraint structure correction factor by analyzing the principal constraint structure of the environmental fragment;

[0009] Step S4: Obtain the decomposition mode correction distance by sequentially multiplying the standard Euclidean distance with the environmental fragment co-window proportion correction factor and the environmental fragment principal limit factor structure correction factor;

[0010] Step S5: Obtain nutrient release prediction results by performing decomposition pattern clustering and calibration analysis based on the distance corrected by the decomposition pattern;

[0011] The step of obtaining the coordination window ratio correction factor by quantitatively evaluating the temperature and humidity coordination of environmental segments includes: obtaining environmental segment coordination time integral data by performing optimal interval matching processing on the temperature and humidity observation data of environmental segments; and obtaining the environmental segment coordination window ratio correction factor by performing inter-segment difference quantification processing on the environmental segment coordination time integral data.

[0012] The step of obtaining principal constraint factor structure correction factors by analyzing the principal constraint factor structure of environmental segments includes: identifying and processing non-cooperative time point data of environmental segments to obtain a set of non-cooperative time point data of environmental segments; performing constraint state classification and statistical processing on the set of non-cooperative time point data of environmental segments to obtain constraint factor contribution spectrum data of environmental segments; and performing inter-segment structural difference quantification processing on the constraint factor contribution spectrum data of environmental segments to obtain principal constraint factor structure correction factors of environmental segments.

[0013] Furthermore, the step of acquiring environmental segment feature datasets by collecting and segmenting soil environmental time-series data includes:

[0014] A sampling frequency and time window length are set, and soil environmental monitoring sensors are deployed in the soil of the target farmland area. The soil environmental monitoring sensors continuously collect time-series data of the soil environment in the target farmland area at the set sampling frequency. The soil environmental time-series data includes at least soil temperature time-series data and soil moisture time-series data. The continuously collected soil temperature time-series data and soil moisture time-series data are segmented according to the time window length to obtain multiple independent environmental segments. For each environmental segment, the mean temperature observation value, the variance of the temperature observation value, the mean humidity observation value, and the variance of the humidity observation value are calculated. The mean temperature observation value, the variance of the temperature observation value, the mean humidity observation value, and the variance of the humidity observation value are combined to form the feature vector of the environmental segment, so as to form the environmental segment feature dataset.

[0015] Furthermore, the step of obtaining environmental segment coordinated time integral data by performing optimal interval matching processing on the environmental segment temperature and humidity observation data includes:

[0016] Optimal ranges for soil temperature and soil moisture are defined. These optimal ranges are then used as the optimal temperature and humidity thresholds, respectively. For any target time within any target environmental segment, the temperature and humidity observation values ​​corresponding to that time are matched against the optimal temperature and humidity thresholds to obtain the temperature and humidity suitability indicators for that time. When both the temperature and humidity indicators simultaneously meet the optimal conditions, the target time is recorded as a cooperatively suitable time. The ratio of the number of cooperatively suitable time points across all times within the environmental segment to the total number of time points within the segment is calculated, and this ratio is used as the cooperative time integral data for the environmental segment.

[0017] Furthermore, the step of obtaining the environmental segment cooperative window proportion correction factor by quantifying the inter-segment differences in the environmental segment cooperative time integral data includes:

[0018] Set sensitivity adjustment parameters; for any two target environment segments, the absolute value of the result of subtracting the integral data of the coordination time of the first target environment segment from the integral data of the coordination time of the second target environment segment is used as the assessment of the coordination time difference between the first and second target environment segments; the result of multiplying the sensitivity adjustment parameters and the assessment of the coordination time difference between the first and second target environment segments is mapped using a hyperbolic tangent function, and the result of adding the constant 1 to the corresponding mapping result is used as the correction factor for the proportion of the coordination window of the environment segments of the first and second target environment segments.

[0019] Furthermore, the step of identifying and processing non-cooperative time point data of environmental segments to obtain a set of non-cooperative time point data of environmental segments includes:

[0020] For any target environmental segment, collaborative determination is performed on all moments within the target environmental segment one by one; when at least one of the temperature suitability indicator data and humidity suitability indicator data at any moment does not meet the optimal condition, the moment is recorded as a non-cooperative time point; all non-cooperative time points determined within the target environmental segment are summarized to obtain the corresponding set of non-cooperative time points for the environmental segment.

[0021] Furthermore, the step of obtaining environmental segment constraint factor contribution spectrum data by performing constraint state classification statistical processing on the set of non-cooperative time points of environmental segments includes:

[0022] A set of constraint state categories for non-coordinated time points is defined, including at least low-temperature constraint state, high-temperature constraint state, drought constraint state, and excessively wet constraint state. For any non-coordinated time point in the set of non-coordinated time points of any target environmental segment, the soil temperature and soil moisture observation values ​​corresponding to the non-coordinated time point are obtained. The constraint state of the non-coordinated time point is determined according to the optimal range of soil temperature and optimal range of soil moisture. Non-coordinated time points that meet the low-temperature constraint condition are recorded as low-temperature constraint state, non-coordinated time points that meet the high-temperature constraint condition are recorded as high-temperature constraint state, non-coordinated time points that meet the drought constraint condition are recorded as drought constraint state, and non-coordinated time points that meet the excessively wet constraint condition are recorded as excessively wet constraint state. The number or duration of non-coordinated time points under each constraint state is statistically analyzed to obtain the statistical data of non-coordinated time points corresponding to each constraint state. The statistical data of non-coordinated time points corresponding to each constraint state and the sum of the statistical data of non-coordinated time points corresponding to all constraint states are normalized, and the proportion data corresponding to each constraint state after normalization is used as the constraint factor contribution spectrum data of the target environmental segment.

[0023] Furthermore, the step of quantifying the structural differences between environmental fragments by analyzing the contribution spectrum data of environmental fragment constraint factors to obtain the structural correction factor of the environmental fragment principal constraint factor includes:

[0024] For any two target environment segments, the first target environment segment and the second target environment segment, the square root operation is performed on the constraint factor contribution spectrum data of the first target environment segment and the constraint factor contribution spectrum data of the second target environment segment, and the sum of the squares of the differences is calculated. The square root of the sum is used as the evaluation of the structural difference of the constraint factor contribution spectrum between the first target environment segment and the second target environment segment.

[0025] Subtracting the environmental segment co-time integral data of the first target environmental segment from the constant 1 yields the first co-dependency weight component; subtracting the environmental segment co-time integral data of the second target environmental segment from the constant 1 yields the second co-dependency weight component; multiplying the first co-dependency weight component and the second co-dependency weight component results in the co-dependency weight for assessing the difference in the structure of the constraint factor contribution spectrum; adding the result of multiplying the co-dependency weight and the assessment of the difference in the structure of the constraint factor contribution spectrum to the constant 1 results in the correction factor for the principal constraint factor structure of the environmental segments of the first and second target environmental segments.

[0026] Furthermore, the step of obtaining the decomposition mode correction distance by sequentially multiplying the standard Euclidean distance by the environmental fragment co-window proportion correction factor and the environmental fragment principal limit factor structure correction factor includes:

[0027] For any two target environment segments, the first target environment segment and the second target environment segment are obtained respectively. The standard Euclidean distance between the first target environment segment and the second target environment segment is calculated based on the environment segment feature vector. The standard Euclidean distance is then multiplied by the environment segment collaborative window proportion correction factor and the environment segment principal limit factor structure correction factor in sequence. The result of the multiplication is used as the decomposition mode correction distance between the first target environment segment and the second target environment segment.

[0028] Compared with the prior art, the present invention has the following advantages:

[0029] This invention presents a method for predicting nutrient release from straw return under water-fertilizer coupling conditions. By explicitly introducing a quantitative index of the proportion of synergistic windows where temperature and humidity simultaneously meet the criteria into the clustering distance, the similarity between environmental segments is no longer determined solely by time-series-neutral statistics such as mean and variance. Instead, it accurately reflects the biologically crucial quantity—the proportion of observation periods during which the optimal temperature and humidity conditions required for straw decomposition can be simultaneously met. Consequently, asynchronously constrained segments and synchronously optimal segments, which would otherwise be misclassified using traditional Euclidean distance, can be effectively separated under the distance metric of this invention. The clustering results show higher consistency with the actual decomposition capacity in the field, thus providing a cleaner, single-mechanism sample set for subsequent nutrient release rate calibration. Furthermore, after identifying synergistic differences, this invention further constructs a principal constraint factor structure correction mechanism based on the constraint factor contribution spectrum, enabling environmental segments exhibiting low synergistic characteristics to be finely distinguished according to their constraint sources (low temperature, high temperature, drought, excessive humidity, etc.). This distinction not only ensures the clarity of the mechanisms among the various models in the decomposition model library, but also provides direct guidance for differentiated regulation in the field: when the clustering results show that the field is mainly limited by drought, the irrigation strategy can be adjusted first; when it shows that it is mainly limited by low temperature, there is no need to blindly increase irrigation or fertilizer application, thereby avoiding excessive fertilization or water regulation mismatch caused by model confusion, and improving the resource utilization efficiency and decision-making accuracy of straw return to the field under the conditions of integrated water and fertilizer management. Attached Figure Description

[0030] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0031] Figure 1 This is a flowchart illustrating a method for predicting nutrient release from straw returned to the field under water-fertilizer coupling conditions, as described in an embodiment of the present invention. Detailed Implementation

[0032] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0033] See Figure 1 This is a flowchart of a method for predicting nutrient release from straw returned to the field under water-fertilizer coupling conditions, as provided in Embodiment 1 of the present invention. Figure 1 As shown, a method for predicting nutrient release from straw returned to the field under water-fertilizer coupling conditions may include:

[0034] Step S1: Obtain environmental segment feature dataset by collecting and segmenting soil environmental time series data.

[0035] First, the sampling frequency and time window length are set. In this embodiment of the invention, the sampling frequency is set to once per hour, and the time window length is set to 72 hours, meaning that data from every three days constitutes an independent environmental segment. Soil environmental monitoring sensors are deployed in the soil of the target farmland area. The soil environmental monitoring sensors continuously collect time-series data of the soil environment of the target farmland area at the set sampling frequency. The soil environmental time-series data includes at least soil temperature time-series data and soil moisture time-series data. The continuously collected soil temperature time-series data and soil moisture time-series data are segmented according to the time window length to obtain multiple independent environmental segments. For each environmental segment, the mean temperature observation value, the variance of the temperature observation value, the mean humidity observation value, and the variance of the humidity observation value are calculated. The mean temperature observation value, the variance of the temperature observation value, the mean humidity observation value, and the variance of the humidity observation value are combined to form the feature vector of the environmental segment, thereby forming the environmental segment feature dataset.

[0036] This completes the acquisition of environmental segment feature datasets through the collection and segmentation of soil environmental time-series data.

[0037] Step S2: Obtain the correction factor for the proportion of the synergistic window by quantitatively evaluating the temperature and humidity synergy of environmental segments.

[0038] Standard Euclidean distance, when calculating the similarity between two environmental segments, treats the differences in temperature and humidity as independent quantities and sums them up. This calculation method completely fails to perceive the synergistic differences in temperature and humidity changes within the two segments. Therefore, when faced with an asynchronous temperature and humidity segment where temperature and humidity alternately meet optimal conditions over time, and a synchronous temperature and humidity segment where temperature and humidity are always synchronously met optimal conditions over time, even if their statistical characteristics such as means are highly similar, the former will result in severely inhibited microbial decomposition activity due to the persistent bottleneck effect, while the latter will promote efficient and continuous nutrient release. The blind spot of the synergistic effect in existing algorithms will incorrectly classify these two patterns with vastly different decomposition efficiencies as similar, leading to serious distortion in the clustering results. The core logic of this step is: first, to construct a quantitative index for each environmental segment that can objectively reflect the synergistic relationship between its internal temperature and humidity. Then, when measuring the similarity between any two segments, the core is to compare their differences in this synergistic index. When the synergy index values ​​of two segments are similar, regardless of whether their synchronicity is both high or low, they can be preliminarily considered to have some similarity in the biological effects they induce. Conversely, if the calculation results show that the synergy indexes of the two segments are significantly different, it indicates that they represent two fundamentally different decomposition modes. In this case, even if their original feature vectors are very close in numerical Euclidean distance, a sufficiently large penalty gain must be applied to their calculated distance through a correction factor to ensure that they are effectively separated in the subsequent clustering process.

[0039] In summary, this invention first obtains the coordinated time integral data of environmental segments by performing optimal interval matching processing on the temperature and humidity observation data of environmental segments. Specifically, optimal intervals for soil temperature and soil humidity are set. In this embodiment, the optimal interval for soil temperature is set to 25 degrees Celsius to 35 degrees Celsius, and the optimal interval for soil humidity is set to 60% to 80%. The optimal intervals for soil temperature and soil humidity are used as the optimal temperature and optimal humidity judgment thresholds, respectively. For any target time within any target environmental segment, the temperature and humidity observation values ​​corresponding to the target time are matched according to the optimal temperature and optimal humidity judgment thresholds to obtain the temperature suitability indication data and humidity suitability indication data for the target time. When the temperature and humidity suitability indication data for the target time simultaneously meet the optimal conditions, the target time is recorded as the coordinated suitable time. The ratio of the number of coordinated suitable time points to the total number of time points within the environmental segment is calculated, and the obtained ratio is used as the coordinated time integral data of the environmental segment.

[0040] After obtaining the collaborative time integral data of the environmental segments, the inter-segment difference quantification processing is performed on the collaborative time integral data of the environmental segments to obtain the environmental segment collaborative window ratio correction factor. Specifically, a sensitivity adjustment parameter is set. In this embodiment of the invention, the sensitivity adjustment parameter is set to 5. This parameter can be adjusted according to the actual scenario and is not required. For any two target environmental segments, the first target environmental segment and the second target environmental segment, the absolute value of the result of subtracting the collaborative time integral data of the first target environmental segment from the collaborative time integral data of the second target environmental segment is used as the collaborative time difference assessment between the first target environmental segment and the second target environmental segment. The result of multiplying the sensitivity adjustment parameter and the collaborative time difference assessment between the first target environmental segment and the second target environmental segment is mapped using a hyperbolic tangent function. The result of adding the constant 1 to the corresponding mapping result is used as the environmental segment collaborative window ratio correction factor between the first target environmental segment and the second target environmental segment.

[0041] In one implementation, assume the first The environmental segment in the first The temperature observation value at each time point is ;No. The environmental segment in the first The humidity observation value at each moment is ;No. The environmental segment in the first The temperature observation value at each time point is ;No. The environmental segment in the first The humidity observation value at each moment is The length of the environmental fragment is The sensitivity adjustment parameter is: The temperature suitability step function is: The humidity suitability step function is: Then the first The environmental fragment and the first The formula for calculating the environmental fragment collaborative window proportion correction factor for each environmental fragment is:

[0042]

[0043] in, Indicates the first The environmental fragment and the first Correction factor for the proportion of environmental fragment collaborative windows in each environmental fragment; Represents the hyperbolic tangent function; Indicates the first The environmental segment in the first Temperature observation values ​​at each time point; Indicates the first The environmental segment in the first Humidity observation values ​​at each moment; Indicates the first The environmental segment in the first Temperature observation values ​​at each time point; Indicates the first The environmental segment in the first Humidity observation values ​​at each moment; Indicates the length of the environmental fragment; This indicates the sensitivity adjustment parameter; This represents the temperature suitability step function. The function value is 1 when the input temperature value is within the optimal range of soil temperature, and 0 otherwise. This represents the humidity suitability step function. The function value is 1 when the input humidity value is within the optimal range of soil humidity, and 0 otherwise.

[0044] It should be noted that the correction factor \alpha_{ij} constructed in this invention can directly quantify and compare the potential decomposition efficiency differences between different environmental segments determined by the temperature and humidity coupling synergy effect, thereby making up for the blind spot of the synergy effect of the standard Euclidean distance.

[0045] The core part of the formula is the cooperative time integral. ,Right now In this design, the product term It is a cooperative state function that utilizes the properties of a step function; it only applies when... The result of this term is 1 only when both the temperature and humidity at a given time point fall within their respective optimal ranges; otherwise, the result will be 0 if any single factor fails to meet the condition. This is obtained by summing the results over all time points in the entire segment and dividing by the total duration. The value quantitatively reveals the proportion of time within the optimal decomposition window of that segment, namely, the synergistic suitability of temperature and humidity. A segment with optimal temperature and humidity synchronization... The value will approach 1; while a temperature and humidity asynchronously constrained segment, because at any given time at least one factor is not in an optimal state, its The value will approach 0. This is after obtaining a value capable of characterizing the macroscopic synergy of each segment. and Then, by calculating their absolute differences This allows for the quantification of the degree of difference between the two fragments in their core decomposition patterns. This represents a fundamental advancement over existing techniques that only compare disordered statistics such as the mean. Finally, to ensure that the correction factor functions effectively while maintaining the numerical stability of the algorithm, this invention employs... The structure is subjected to a bounded nonlinear mapping. When the cooperability of two segments is similar, Approaching 0, The distance will approach 1, at which point the factor is in a silent state and has no impact on the original distance, correctly identifying segments with similar cooperability into the same class. Conversely, when comparing a high-cooperability segment with a low-cooperability segment, It will be a large value. It will smoothly increase and approach its upper limit of 2. This significantly increased penalty gain corrects for the excessively small Euclidean distance caused by accidental similarity in statistical features, thus effectively separating these two distinctly different biological effects in the cluster space. In this way, This enables distance metrics to respond to temperature and humidity coupling synergies, solving the blind spot problem of synergy effects in existing technologies.

[0046] Thus, the correction factor for the proportion of the synergistic window was obtained by quantitatively evaluating the synergistic effect of temperature and humidity on environmental segments.

[0047] Step S3: Obtain the principal constraint structure correction factor by analyzing the principal constraint structure of the environmental fragment.

[0048] Through the correction factor in step S2 Distance measurement mechanisms can macroscopically identify and distinguish environmental segments with different temperature and humidity synergies, initially solving the blind spot problem of synergy effects in existing technologies. However, this only addresses the first level of the problem. In the actual process of straw decomposition, for environmental segments also identified as having low synergy, the root causes leading to low synergy and thus inhibiting the decomposition rate may differ. Specifically, one low-synergy segment's decomposition process is mainly limited by continuous low temperatures, while humidity is suitable most of the time; another low-synergy segment is mainly limited by continuous drought stress, while temperature is mostly within the optimal range. Although both situations result in low final decomposition efficiency, they reveal that the main limiting factors faced by the field environment are different, corresponding to low-temperature limiting and drought limiting modes, respectively. In precision agriculture management, accurately distinguishing these two modes is instructive for diagnosing the root cause of the problem and formulating targeted field control measures. After solving the macroscopic synergy problem, existing technologies still lack mechanisms to further distinguish the internal structure of these limiting modes. Therefore, in order to achieve more refined identification of decomposition modes... We need to focus on two environmental segments that both exhibit low synergy and further quantify the differences in their respective primary limiting factors. If both segments have low synergy, but one is dominated by temperature and the other by humidity, a correction factor should be applied to penalize their calculated distance accordingly.

[0049] In summary, this invention first identifies and processes non-cooperative time point data of environmental segments to obtain a set of non-cooperative time point data of environmental segments. Specifically, for any target environmental segment, cooperative determination is performed on all moments within the target environmental segment one by one; when at least one of the temperature suitability indicator data and humidity suitability indicator data at any moment does not meet the optimal condition, that moment is recorded as a non-cooperative time point; all non-cooperative time points determined within the target environmental segment are summarized to obtain the corresponding set of non-cooperative time point data of environmental segments.

[0050] After obtaining the set of non-coordinated time points for environmental segments, the data is further processed by classifying and statistically analyzing the constraint states to obtain the contribution spectrum data of constraint factors for the environmental segments. Specifically, a set of constraint state categories for non-coordinated time points is defined, which includes at least low-temperature constraint, high-temperature constraint, drought constraint, and excessively wet constraint. For any non-coordinated time point in the set of non-coordinated time points for any target environmental segment, the corresponding soil temperature and soil moisture observation values ​​are obtained. Based on the optimal range of soil temperature and optimal range of soil moisture, the constraint state of the non-coordinated time point is determined, and the constraint state is then determined. Non-cooperative time points meeting the low-temperature constraint condition are recorded as low-temperature constraint states; non-cooperative time points meeting the high-temperature constraint condition are recorded as high-temperature constraint states; non-cooperative time points meeting the drought constraint condition are recorded as drought constraint states; and non-cooperative time points meeting the excessively wet constraint condition are recorded as excessively wet constraint states. The number or duration of non-cooperative time points under each constraint state is statistically analyzed to obtain statistical data on non-cooperative time points corresponding to each constraint state. The sum of the statistical data on non-cooperative time points corresponding to each constraint state and the statistical data on non-cooperative time points corresponding to all constraint states is normalized. The normalized percentage data corresponding to each constraint state is used as the constraint factor contribution spectrum data of the target environmental segment.

[0051] After obtaining the constraint factor contribution spectrum data of the environmental segments, the final step is to quantify the structural differences between segments by performing inter-segment structural difference quantification on the constraint factor contribution spectrum data of the environmental segments to obtain the environmental segment principal constraint factor structure correction factor. Specifically, for any two target environmental segments, the first target environmental segment and the second target environmental segment, the square root of the corresponding constraint state proportion data in the constraint factor contribution spectrum data of the first target environmental segment and the constraint factor contribution spectrum data of the second target environmental segment are calculated, and the square root of the sum is used as the constraint factor contribution spectrum structural difference assessment between the first target environmental segment and the second target environmental segment. The first collaborative dependency weight component is obtained by subtracting the environmental segment collaborative time integral data of the first target environmental segment from the constant 1. The second collaborative dependency weight component is obtained by subtracting the environmental segment collaborative time integral data of the second target environmental segment from the constant 1. The result of multiplying the first collaborative dependency weight component and the second collaborative dependency weight component is used as the collaborative dependency weight for the constraint factor contribution spectrum structural difference assessment. The result of multiplying the collaborative dependency weight and the constraint factor contribution spectrum structural difference assessment and adding the result of adding the constant 1 is used as the environmental segment principal constraint factor structure correction factor for the first target environmental segment and the second target environmental segment.

[0052] In one implementation, assume the first The first environmental fragment's limiting factor contribution spectrum Each component is ;No. The first environmental fragment's limiting factor contribution spectrum Each component is The total number of categories in the constraint factor contribution spectrum is ;No. The environmental segment coordination time integral data of each environmental segment are: ;No. The environmental segment coordination time integral data of each environmental segment are: Then the first The environmental fragment and the first The formula for calculating the principal constraint factor structure correction factor of an environmental segment is as follows:

[0053]

[0054] in, Indicates the first The environmental fragment and the first Environmental fragment principal constraint factor structure correction factor; Indicates the first Environmental segment collaborative time integral data for each environmental segment, i.e. ; Indicates the first Environmental segment co-location time integral data for each environmental segment; This represents the total number of categories in the contribution spectrum of the limiting factor; Indicates the first The first environmental fragment's limiting factor contribution spectrum One component; Indicates the first The first environmental fragment's limiting factor contribution spectrum Each component.

[0055] It should be noted that the core part of the formula is the contribution spectrum of the restriction factor for the two segments. and The Hellinger distance between them, i.e. The Hellinger distance, as a robust measure of the difference in probability distributions, can capture the difference between two contribution spectra. and The differences lie in their structures. For example, a segment dominated by drought will have a peak in its drought component in its contribution spectrum, while a segment dominated by low temperature will have a peak in its low temperature component. The Hellinger distance between these two spectral vectors will be a large value, accurately reflecting the difference in their principal constraint factors. However, comparing principal constraint factors is only meaningful when the coherence of both segments is low. Therefore, this invention designs a coherence-dependent weighting term. This is the key to achieving progressive correction. Represents the first The degree of non-cooperation of each environmental segment. Only when the first... The environmental fragment and the first When multiple environmental fragments are simultaneously determined to have low synergy by step S2 (i.e.) and (All approach 0) and The weights approach 1, causing the weight term to also approach 1, thus fully activating the penalty effect of Hellinger distance. Conversely, as long as at least one of the two segments is highly cooperative (e.g., ... As the weights approach 1, the weights will approach 0, making... It approaches 1 precisely without any correction. This design constructs a logic switch that ensures... Factors are activated only in scenarios where low-cooperation fragments undergo secondary subdivision, thus avoiding erroneous interference with other pattern pairs.

[0056] Thus, the process of obtaining principal constraint structure correction factors by analyzing the principal constraint structure of environmental fragments has been completed.

[0057] Step S4: Obtain the decomposition mode correction distance by sequentially multiplying the standard Euclidean distance with the environmental fragment co-window proportion correction factor and the environmental fragment principal limit factor structure correction factor.

[0058] After completing steps S2 and S3, the collaborative window proportion correction factor and the principal limit factor structure correction factor are obtained respectively. For any two target environment segments, the first target environment segment and the second target environment segment are obtained respectively, and the environmental segment feature vectors corresponding to the first target environment segment and the second target environment segment are calculated based on the environmental segment feature vectors. The standard Euclidean distance between the first target environment segment and the second target environment segment is then multiplied by the environmental segment collaborative window proportion correction factor and the environmental segment principal limit factor structure correction factor in sequence, and the result of the multiplication is used as the decomposition mode correction distance between the first target environment segment and the second target environment segment.

[0059] Thus, the process of obtaining the decomposition mode corrected distance by performing joint cooperative and restrictive correction on the standard Euclidean distance is complete.

[0060] Step S5: Obtain nutrient release prediction results by performing decomposition pattern clustering and calibration analysis based on the distance corrected by the decomposition pattern.

[0061] This step utilizes the decomposition pattern correction distance constructed in step S4 to perform K-means clustering analysis on all environmental fragment samples to achieve intelligent identification of straw decomposition patterns and ultimately apply it to nutrient release prediction. The clustering process first requires determining the optimal number of clusters, K. In this embodiment, the elbow method is used to determine the K value. This method calculates the total sum of squares within clusters under different K values ​​and plots a curve showing its variation with the K value. The K value corresponding to the elbow point where the curve slope changes most significantly is selected as the optimal number of clusters. After determining the optimal K value, the standard K-means algorithm iterative process is executed. During this process, the output value of the metric function used by the algorithm to calculate the distance between data points and cluster centers is set to the decomposition pattern correction distance constructed in step S4. Based on this correction distance, the algorithm iteratively assigns each environmental fragment to the cluster center closest to its correction distance and updates the position of the cluster center until the algorithm meets the convergence condition, ultimately dividing the entire set of environmental fragments into K categories. After clustering, each cluster represents a straw decomposition pattern with unique temperature-humidity synergy and principal constraint factor structure. Statistical analysis of environmental segment characteristics and raw time-series data within each cluster allows for the assignment of specific meanings to each pattern. Finally, these identified decomposition patterns are correlated with nutrient release rates. The daily average nutrient release rate under each pattern is calibrated through small-scale in-situ field incubation experiments. In forecasting, after obtaining environmental forecast data for a future period, the system first determines which identified decomposition pattern cluster the environmental segment belongs to, and then uses the corresponding nutrient release rate for that cluster to accurately predict nutrient release from straw return under future water-fertilizer coupling conditions.

[0062] Thus, the prediction results of nutrient release were obtained by decomposing and clustering environmental fragments and performing calibration analysis.

[0063] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting nutrient release from straw returned to the field under water-fertilizer coupling conditions, characterized in that, The method includes: Step S1: Obtain environmental segment feature datasets by collecting and segmenting soil environmental time-series data; Step S2: Obtain the correction factor for the proportion of the synergistic window by quantitatively evaluating the temperature and humidity synergy of environmental segments; Step S3: Obtain the principal constraint structure correction factor by analyzing the principal constraint structure of the environmental fragment; Step S4: Obtain the decomposition mode correction distance by sequentially multiplying the standard Euclidean distance with the environmental fragment co-window proportion correction factor and the environmental fragment principal limit factor structure correction factor; Step S5: Obtain nutrient release prediction results by performing decomposition pattern clustering and calibration analysis based on the distance corrected by the decomposition pattern; The method of acquiring environmental segment feature datasets by collecting and segmenting soil environmental time-series data includes: setting a sampling frequency and time window length; deploying soil environmental monitoring sensors in the soil of the target farmland area; continuously collecting soil environmental time-series data of the target farmland area through the soil environmental monitoring sensors at the set sampling frequency; the soil environmental time-series data including at least soil temperature time-series data and soil moisture time-series data; segmenting the continuously collected soil temperature time-series data and soil moisture time-series data according to the time window length to obtain multiple independent environmental segments; and calculating the mean temperature observation value, variance temperature observation value, mean humidity observation value, and variance humidity observation value for each environmental segment; combining the mean temperature observation value, variance temperature observation value, mean humidity observation value, and variance humidity observation value to form the feature vector of the environmental segment, thereby forming the environmental segment feature dataset. The step of obtaining the coordination window ratio correction factor by quantitatively evaluating the temperature and humidity coordination of environmental segments includes: obtaining environmental segment coordination time integral data by performing optimal interval matching processing on the temperature and humidity observation data of environmental segments; and obtaining the environmental segment coordination window ratio correction factor by performing inter-segment difference quantification processing on the environmental segment coordination time integral data. The step of obtaining principal constraint factor structure correction factors by analyzing the principal constraint factor structure of environmental segments includes: identifying and processing non-cooperative time point data of environmental segments to obtain a set of non-cooperative time point data of environmental segments; performing constraint state classification and statistical processing on the set of non-cooperative time point data of environmental segments to obtain constraint factor contribution spectrum data of environmental segments; and performing inter-segment structural difference quantification processing on the constraint factor contribution spectrum data of environmental segments to obtain principal constraint factor structure correction factors of environmental segments. The process of obtaining coordinated time integral data of environmental segments by performing optimal interval matching processing on environmental segment temperature and humidity observation data includes: setting optimal intervals for soil temperature and soil humidity; using the optimal intervals for soil temperature and soil humidity as optimal temperature and optimal humidity judgment thresholds, respectively; for any target time within any target environmental segment, using the soil temperature and soil humidity observation values ​​corresponding to the target time, performing interval matching based on the optimal temperature and optimal humidity judgment thresholds to obtain temperature suitability indication data and humidity suitability indication data for the target time; when the temperature suitability indication data and humidity suitability indication data for the target time simultaneously meet the optimal conditions, the target time is recorded as a coordinated suitable time; calculating the ratio of the number of coordinated suitable time points for all times within the environmental segment to the total number of time points within the environmental segment, and using the obtained ratio as the coordinated time integral data of the environmental segment. The step of obtaining the environmental segment collaborative window ratio correction factor by quantifying the inter-segment differences in the environmental segment collaborative time integral data includes: setting a sensitivity adjustment parameter; for any two target environmental segments, the absolute value of the result of subtracting the collaborative time integral data of the first target environmental segment from the collaborative time integral data of the second target environmental segment is used as the collaborative time difference assessment between the first and second target environmental segments; the result of multiplying the sensitivity adjustment parameter and the collaborative time difference assessment between the first and second target environmental segments is mapped using a hyperbolic tangent function, and the result of adding the constant 1 to the corresponding mapping result is used as the environmental segment collaborative window ratio correction factor between the first and second target environmental segments. The step of identifying and processing non-cooperative time point data of environmental segments to obtain a set of non-cooperative time point data of environmental segments includes: for any target environmental segment, performing cooperative determination on all moments within the target environmental segment one by one; when at least one of the temperature suitability indicator data and humidity suitability indicator data at any moment does not meet the optimal condition, the moment is recorded as a non-cooperative time point; summarizing all non-cooperative time points determined within the target environmental segment to obtain the corresponding set of non-cooperative time point data of environmental segments. The process involves classifying and statistically processing the set of non-coordinated time points of environmental segments to obtain the contribution spectrum data of environmental segment constraint factors. This includes: setting a set of constraint state categories for non-coordinated time points, which at least includes low-temperature constraint, high-temperature constraint, drought constraint, and excessively wet constraint; for any non-coordinated time point in the set of non-coordinated time points of any target environmental segment, obtaining the corresponding soil temperature and soil moisture observation values, and determining the constraint state of the non-coordinated time point based on the optimal soil temperature and optimal soil moisture ranges, classifying non-coordinated time points that meet the low-temperature constraint condition as non-coordinated... Simultaneously, time points are recorded under low-temperature constraint conditions, non-cooperative time points meeting high-temperature constraint conditions are recorded under high-temperature constraint conditions, non-cooperative time points meeting drought constraint conditions are recorded under drought constraint conditions, and non-cooperative time points meeting excessively wet constraint conditions are recorded under excessively wet constraint conditions. The number or duration of non-cooperative time points under each constraint condition is statistically analyzed to obtain statistical data on non-cooperative time points corresponding to each constraint condition. The sum of the statistical data on non-cooperative time points corresponding to each constraint condition and the statistical data on non-cooperative time points corresponding to all constraint conditions is normalized, and the normalized percentage data corresponding to each constraint condition is used as the constraint factor contribution spectrum data of the target environmental segment. The step of quantifying the structural differences between environmental fragment constraint factor contribution spectrum data to obtain environmental fragment principal constraint factor structure correction factors includes: for any two target environmental fragments, a first target environmental fragment and a second target environmental fragment, taking the square root of the proportion of each constraint state in the constraint factor contribution spectrum data of the first target environmental fragment and the constraint factor contribution spectrum data of the second target environmental fragment, and summing the squares of the differences, and taking the square root of the sum as the constraint factor contribution spectrum structural difference assessment between the first target environmental fragment and the second target environmental fragment; subtracting the environmental fragment coordination time integral data of the first target environmental fragment from a constant 1 to obtain a first coordination dependency weight component; subtracting the environmental fragment coordination time integral data of the second target environmental fragment from a constant 1 to obtain a second coordination dependency weight component; multiplying the first coordination dependency weight component and the second coordination dependency weight component as the coordination dependency weight for the constraint factor contribution spectrum structural difference assessment; and adding the result of multiplying the coordination dependency weight and the constraint factor contribution spectrum structural difference assessment to a constant 1 as the environmental fragment principal constraint factor structure correction factor for the first target environmental fragment and the second target environmental fragment.

2. The method for predicting nutrient release from straw return to the field under water-fertilizer coupling conditions according to claim 1, characterized in that, The method of obtaining the decomposition mode correction distance by sequentially multiplying the standard Euclidean distance by the environmental fragment co-window proportion correction factor and the environmental fragment principal limit factor structure correction factor includes: For any two target environment segments, the first target environment segment and the second target environment segment are obtained respectively. The standard Euclidean distance between the first target environment segment and the second target environment segment is calculated based on the environment segment feature vector. The standard Euclidean distance is then multiplied by the environment segment collaborative window proportion correction factor and the environment segment principal limit factor structure correction factor in sequence. The result of the multiplication is used as the decomposition mode correction distance between the first target environment segment and the second target environment segment.

Images