A summer maize large trumpet mouth period precision topdressing decision system based on leaf age dynamic tracking and accumulated temperature model

Abstract

This invention discloses a precision topdressing decision-making system for summer maize at the large trumpet stage based on dynamic leaf age tracking and accumulated temperature model, involving crop growth monitoring and precision fertilization decision-making. The system consists of an observation terminal, a temperature acquisition device, and a decision processor. It uses the records of the last unrevealed and first revealed leaves on the same plant to form the accumulated temperature range for leaf expansion, constraining the accumulated temperature model parameter set for each sample plant. Combined with temperature forecasts, the latest entry and earliest exit times from the target leaf age range, and the operation duration, a common operable time window is formed. When the time window is empty, two revealing results are pre-simulated for sample plants within the constraint boundary. Only leaf positions where both branches are not empty and at least one branch changes the operation boundary in a feasible direction are selected for supplementary measurement, and the calculation is recalculated based on the measured results. When the time window is not empty, the amount of fertilizer already applied is deducted, generating a topdressing prescription that is constrained by the time window. This invention maintains a correlation between the output of additional observations, the appropriate timing for complete operation, and the remaining topdressing amount.

Description

Technical Field

[0001] This invention relates to the field of crop growth process monitoring and precision fertilization decision-making technology, and in particular to a precision topdressing decision-making system for summer maize at the large trumpet stage based on dynamic leaf age tracking and accumulated temperature model. Background Technology

[0002] Topdressing during the tasseling stage of summer maize needs to be planned in conjunction with the plant's growth progress and the timing of the application. Leaf age observations are usually spaced out, and the leaf expansion process varies among different plants in the same plot. Temperature changes can also alter the timing of subsequent leaf emergence. When using single leaf age or average prediction stages to plan applications, the timing of events within the observation interval and the differences in appropriate timing between different plants may not be fully reflected.

[0003] Chinese invention patent application CN120510524A discloses a method and system for determining crop leaf age based on image recognition, which uses leaf characteristics and accumulated temperature information to determine or correct leaf age; CN117994680A discloses a method and system for monitoring crop growth based on multi-source images, which combines meteorological information and phenological stages to form a water and fertilizer management plan. These technologies provide ways to link leaf age, growth stage, and crop management.

[0004] In scenarios requiring continuous topdressing operations to be completed on the same plot, obtaining only a phase estimate is insufficient to determine whether the entire operation falls within a common optimal period. When existing observations cannot provide an executable timeframe, deciding which plant and which leaf position to continue observing will also affect whether uncertainties affecting the operation schedule can be eliminated. Therefore, it is necessary to link the range of phase uncertainties formed by interval observations, the differences among sample plants, and the actual operation duration, and to ensure that additional observations can serve to reassess the operation conditions. Summary of the Invention

[0005] The purpose of this invention is to provide a precise topdressing decision system for summer maize at the large trumpet stage based on dynamic leaf age tracking and accumulated temperature model, in order to solve the problems of difficulty in confirming the common appropriate time for complete topdressing operations under conditions of intermittent leaf age observation and temperature changes, as well as the unclear relationship between additional observations and operation access.

[0006] To achieve the above objectives, the system of the present invention includes an observation terminal, a temperature acquisition device, and a decision processor connected to both; the observation terminal records the leaf position and leaf sheath exposure status of fixed sample plants; the decision processor forms a leaf expansion accumulated temperature range based on the last unexposed and first exposed times of the same leaf position of the same fixed sample plant and the temperature data from the temperature acquisition device, and retains a set of accumulated temperature model parameters that satisfy the constraints of the leaf expansion accumulated temperature range for each fixed sample plant; based on the accumulated temperature model parameter set and temperature forecast, the system determines the target leaf age range for each fixed sample plant entering the large trumpet stage. The maximum value of the latest time of the encirclement is the starting point of the time window, and the minimum value of the earliest departure time minus the fertilization operation time is the ending point of the time window, forming a common workable time window; when the time window is empty, the results of the exposed and unexposed leaves of the fixed sample plants at the restricted boundary are pre-rehearsed for each leaf position, forming a two-branch parameter set. Only the leaf position commanded by the instruction that both sets are not empty and at least one branch advances the starting point or delays the ending point is selected for supplementary measurement at the observation terminal. The parameter set is updated and the time window is recalculated based on the supplementary measurement results; when the time window is not empty, the fertilization prescription with the start time limited to the time window is generated by deducting the amount of fertilizer already applied according to the stage fertilization budget.

[0007] Furthermore, the observation terminal assigns a fixed sample plant identifier to each fixed sample plant, numbers the leaf positions consecutively from bottom to top, and stores the fixed sample plant identifier, the leaf position, the observation time, and the leaf pulvinus exposure state in association; the observation terminal saves the leaf unfolding record of the exposed leaf positions, and continues the leaf position numbering by using the leaf unfolding record after the lower leaves wither, so that the leaf position number of the same fixed sample plant remains unchanged at different observation times.

[0008] Furthermore, the decision processor calculates the cumulative effective accumulated temperature using a unified accumulated temperature starting point based on the temperature data, and uses the cumulative effective accumulated temperature at the last unexposed and first exposed times of the same leaf position of the same fixed sample plant as the lower boundary and upper boundary of the accumulated temperature range for leaf expansion, respectively; the lower boundary of the accumulated temperature is an open boundary, and the upper boundary of the accumulated temperature is a closed boundary; when only unexposed or exposed records are obtained, unilateral constraints are retained, and the accumulated temperature range for leaf expansion is formed after obtaining the records on the other side.

[0009] Furthermore, the decision processor uses the sum of the leaf expansion temperature offset and the leaf expansion temperature per unit leaf position multiplied by the leaf position number as the model leaf expansion temperature, and expands the model leaf expansion temperature according to the preset model error boundary to form the model leaf expansion temperature error band; within the parameter boundary where the leaf expansion temperature per unit leaf position is positive, it retains parameter pairs composed of the leaf expansion temperature offset and the leaf expansion temperature per unit leaf position that make the model leaf expansion temperature error band intersect with the leaf expansion temperature intervals of at least two different leaf positions of the same fixed sample plant, forming the temperature accumulation model parameter set.

[0010] Furthermore, the target leaf age range for the large trumpet stage is defined by the exposure of the leaf pulvinus at the entrance leaf position as the entry condition and the exposure of the leaf pulvinus at the exit leaf position as the exit condition. The decision processor, based on the accumulated temperature model parameter set for each fixed sample plant, takes the supremacy of the upper boundary of the model leaf expansion accumulated temperature error band at the entrance leaf position as the entry-stage accumulated temperature threshold, and takes the inframance of the lower boundary of the model leaf expansion accumulated temperature error band at the exit leaf position as the exit-stage accumulated temperature threshold. The entry-stage accumulated temperature threshold and the exit-stage accumulated temperature threshold are then converted into the latest time and the earliest time, respectively. The maximum value of the latest time for each fixed sample plant is used as the starting point of the time window, and the minimum value of the earliest time for each fixed sample plant minus the topdressing operation duration is used as the ending point of the time window, forming a common operable time window that includes the starting point but not the ending point.

[0011] Further, the decision processor identifies the fixed sample plants that generate the start or end point of the time window as candidate samples for retesting. For the leaf positions to be observed in the candidate samples, the lower boundary of the model leaf expansion accumulated temperature error band is not greater than the cumulative effective accumulated temperature at the retesting time as an exposure constraint, and the upper boundary of the model leaf expansion accumulated temperature error band is greater than the cumulative effective accumulated temperature at the retesting time as a non-exposure constraint. The accumulated temperature model parameter set is intersected with the exposure constraint and the non-exposure constraint to form an exposure branch set and a non-exposure branch set, respectively. Only the leaf positions to be observed that are not empty in both the exposure branch set and the non-exposure branch set, and whose recalculation causes the start point of the time window to be advanced or the end point of the time window to be delayed, are included in the candidate leaf positions for retesting, and the retesting object is selected from the candidate leaf positions for retesting.

[0012] Furthermore, the temperature forecast includes a lower bound and an upper bound; the decision processor converts the lower bound and the upper bound of the temperature forecast into a lower bound and an upper bound of cumulative effective accumulated temperature, respectively, determines the latest time when the lower bound of the cumulative effective accumulated temperature reaches the entry accumulated temperature threshold, determines the earliest time when the upper bound of the cumulative effective accumulated temperature reaches the exit accumulated temperature threshold, and restricts the common workable time window to the work period covered by the temperature forecast.

[0013] Furthermore, the stage topdressing budget and the amount of fertilizer already applied are measured by nitrogen mass for the same plot and the same topdressing stage; the decision processor uses the non-negative difference between the stage topdressing budget and the amount of fertilizer already applied as the remaining nitrogen amount, receives the nitrogen mass fraction of the fertilizer, divides the remaining nitrogen amount by the nitrogen mass fraction of the fertilizer to obtain the remaining fertilizer amount, and writes the remaining fertilizer amount, plot identifier and the common workable time window into the topdressing prescription.

[0014] Furthermore, the decision processor obtains the time window margin by subtracting the time window start point from the time window end point, and preferentially selects the supplementary measurement candidate leaf positions that, after recalculation of at least one branch, make the common workable time window non-empty; if there are no supplementary measurement candidate leaf positions that meet the priority selection conditions, the selection is made in descending order of the smaller value of the time window margin obtained by recalculating the exposed branch set and the unexposed branch set; when the supplementary measurement candidate leaf positions are empty, the topdressing prescription is kept in an ungenerated state, and the common workable time window is recalculated after adding new observation records or updating the temperature forecast.

[0015] Furthermore, when the accumulated temperature model parameter set is empty, the decision processor retains the leaf expansion accumulated temperature interval that causes constraint incompatibility and sends a verification instruction to the observation terminal; when the leaf position record and temperature record still cannot form a non-empty accumulated temperature model parameter set after verification, a new tracking leaf position segment is established with the newly recent continuous leaf position, and the accumulated temperature model parameter set is reconstructed with the leaf expansion accumulated temperature interval in the new tracking leaf position segment; when the new tracking leaf position segment has not yet contained at least two leaf expansion accumulated temperature intervals with different leaf positions, the topdressing prescription is kept in an ungenerated state.

[0016] This invention limits the accumulated temperature model parameters to the actual exposure events of the same leaf position on the same plant. It forms a common operable time window for the entire operation by using the entry and exit boundaries of each permissible model, ensuring that the output start time is consistent with observational uncertainty, plant differences, and operation duration. Two retest results are pre-simulated for plants within the time window, and the retest targets are selected based on changes in the operation boundaries, allowing the retests to apply to the prescription admission conditions. The actual retest feedback is then used to update the original parameter set, forming a continuous process from growth observation to the time-window-constrained topdressing prescription. Attached Figure Description

[0017] The accompanying drawings illustrate the system composition and main processing relationships of this invention.

[0018] Figure 1 This is a schematic diagram illustrating the system composition and the relationship between observation, temperature, and topdressing prescription transmission in this invention.

[0019] Figure 2 This is a schematic diagram of the processing flow of the present invention, which forms a set of accumulated temperature model parameters and a common operable time window from the accumulated temperature range of the leaf expansion.

[0020] Figure 3 This is a schematic diagram illustrating the feedback processing of the present invention, which selects supplementary test objects based on time window boundaries and exposed / unexposed branches. Detailed Implementation

[0021] like Figure 1As shown, this invention provides a precision topdressing decision-making system for summer maize at the large trumpet stage based on dynamic leaf age tracking and accumulated temperature model. The system includes an observation terminal, a temperature acquisition device, and a decision processor connected to both. The observation terminal provides an interface for manual input and photo verification of leaf sheath status. The temperature acquisition device provides the plot temperature with the sampling time. The decision processor receives both types of records and performs parameter constraints, operation windows, and supplementary measurement calculations. The connection can be wired or wireless via a data gateway. The recorded plant identification and sampling time are transmitted along with the data.

[0022] The observation terminal can be a handheld terminal with a camera; the temperature acquisition device uses a temperature sensor and a time recorder placed inside a ventilated radiant hood; the decision processor uses a plot-edge computer or server with storage space. The processor saves event logs, parameter boundaries, model versions, and operation records, and sends the topdressing prescription to the fertilization operation terminal. The fertilization operation terminal displays the allowed start time and fertilizer amount, or transmits the fertilizer amount to the quantitative fertilization device as the operation setting value.

[0023] Before tracking begins, plots are registered according to variety, sowing date, and management conditions. Within each plot, fixed, continuously identifiable, and dispersed sample plants are selected, and their locations and selection ranges are recorded. Separate plot records are established for areas with significant differences in conditions. A common workable window describes the common optimal period for registered sample plants under a given model and temperature boundary; the sample plant selection range is saved with the prescription. The satisfaction of conditions by a limited number of sample plants does not directly equate to every plant in the plot satisfying the conditions.

[0024] The observation terminal records the leaf position and pulvinus exposure status of the fixed sample plants. Pulvinus exposure refers to the clear identification of the connection between the leaf sheath and the leaf blade, and this leaf position is recorded as an unfolded leaf. Observers examine the pulvinus from the side of the plant, and if necessary, take additional images from another direction; obstructed, damaged, or blurry images are recorded as unreadable and require separate acquisition, and are not considered as an unexposed state. The number of visible leaf tips and the number of existing green leaves do not replace consecutive leaf position numbers.

[0025] The observation terminal assigns a fixed sample plant identifier to each fixed sample plant, numbers the leaf positions consecutively from bottom to top, and stores the fixed sample plant identifier, leaf position, observation time, and leaf sheath exposure status in a linked manner. The fixed sample plant identifier is not reused within a single tracking task. Each observation is indexed by a plot identifier, fixed sample plant identifier, leaf position, and observation time, and stores the exposed, non-exposed, or unreadable status, as well as an image index. The observation terminal is uniformly time-synchronized, the processor rearranges late records according to the observation time, and the transmission arrival time is only used for communication status management.

[0026] The observation terminal stores records of leaf unfolding at the exposed leaf positions. After the lower leaves wither, the leaf unfolding records are used to continue leaf position numbering, and the leaf position number remains unchanged for the same fixed plant at different observation times. In early observations, easily identifiable leaf positions are marked with labels that do not affect growth, and side-view photos of the plant and the relationship between adjacent leaf positions are saved. Subsequent observations continue numbering upwards from the still identifiable marked leaf positions, with withered leaf positions retaining their original numbers. When a plant is destroyed, the sequence of that plant is closed, a new label is assigned to a replacement plant, leaf unfolding events are re-accumulated, and the new plant record is not spliced ​​to the original plant record.

[0027] The target leaf age range during the large trumpet stage is defined by the emergence of the leaf sheath at the entrance leaf position as the entry condition and the emergence of the leaf sheath at the exit leaf position as the exit condition. The entrance and exit leaf positions are configured according to the variety and local agronomic scheme of the plot, with the exit leaf position being larger than the entrance leaf position. In this implementation, leaf age is expressed as the number of consecutive leaves with exposed leaf sheaths. In the calculation example described later, the entrance leaf position is taken as 11 and the exit leaf position as 13, representing the target range when 11 or 12 leaves have unfolded; these values ​​are calculation conditions, and the actual configuration can be changed according to the variety and cultivation scheme.

[0028] The temperature acquisition device can sample and upload Celsius temperatures every hour. The processor sets a uniform accumulated temperature starting point for the same plot, retrieves the corresponding time from the sowing record of that plot, and uses the model offset of each sample plant to accommodate differences in emergence and early growth. Historical and forecast temperatures use the same base temperature, upper limit temperature, and time unit to avoid adding data from different accumulated temperature calibers.

[0029] To convert each temperature sample value into effective growth heat, the effective temperature difference is calculated based on the configured base temperature and upper limit temperature: (1) In the formula, the effective temperature difference function The input is Celsius temperature. The output is a non-negative temperature difference in °C; base temperature Starting temperature for effective heat calculation, upper limit temperature All values ​​are upper temperature cutoff values, expressed in °C. ; and These represent the smaller and larger values, respectively. A value of 0 is used when the temperature difference is below 0, and the upper limit is used when it exceeds the upper limit. A base temperature of 10℃ and an upper limit temperature of 30℃ can be used for calculations described later; the model parameters are determined using the same caliber.

[0030] To ensure that leaf expansion records and temperature records have a common scale, the effective temperature difference from the unified accumulated temperature starting point to the observation time is accumulated according to the actual sampling duration: (2) In the formula, the cumulative effective accumulated temperature Indicates the time from the unified accumulated temperature start point The amount of heat, expressed in °C·day; This represents the accumulated number of time periods. For the first The end time of the segment, which is the cumulative target observation time for this period; From 1 to Time period index; temperature For the first The segment represents the sampling temperature and duration. The number of days elapsed in this segment is greater than 0; This represents the summation of segments, the effective temperature difference function. The previous definition is retained. One hour is converted to... If the sampling interval is not completed at the end of the day, the actual elapsed time will be counted.

[0031] Historical temperature records are verified for continuity using timestamps; duplicate data with duplicate timestamps are not counted repeatedly. In the event of a gap, backup sensor records from the same site that have already undergone deviation calibration are prioritized, and alternative sources are preserved. If the gap still cannot be filled, the affected event records are temporarily excluded from prescription approval. After recovery, the data is recalculated from the last valid cumulative value before the gap. Records with temperatures exceeding sensor range or clock reversal are reviewed first and not filled at 0℃.

[0032] The decision processor forms a leaf-expansion accumulated temperature interval based on the last and first emergence times of the same leaf position in the same fixed sample plant, and the temperature data from the temperature acquisition device. It then reserves a set of accumulated temperature model parameters that satisfy the constraints of this interval for each fixed sample plant. The processor converts each readable record to a unified accumulated temperature axis, searching for paired records only within the same leaf position of the same sample plant. The leaf-expansion accumulated temperature interval represents the range of accumulated temperatures that may occur during the actual emergence event at that leaf position, and is subsequently used to exclude model parameters incompatible with the records. Different sample plants have their own accumulated temperature model parameter sets, which are then combined during plot-level calculations to determine the prediction boundaries for each sample plant.

[0033] The decision processor calculates the cumulative effective accumulated temperature using a unified starting point based on the temperature data. The cumulative effective accumulated temperature at the last unrevealed and first revealed times for the same leaf position on the same fixed plant is used as the lower and upper boundaries of the accumulated temperature interval for leaf expansion, respectively. When multiple unrevealed records exist for the same leaf position, the latest effective record is used; when multiple revealed records exist, the earliest effective record is used. Newly arrived effective records can only raise the lower boundary or lower the upper boundary of the accumulated temperature; the original record and the reason for the change are retained with the boundary version.

[0034] The lower boundary of the accumulated temperature is an open boundary, and the upper boundary of the accumulated temperature is a closed boundary. A record of no exposure indicates that leaf expansion has not yet occurred, therefore the exposure event is later than that record; a record of exposure indicates that exposure has already occurred, therefore the exposure event is no later than that record. This distinction is also used for parameter set feasibility, supplementary measurement branches, and departure deadlines to avoid misjudging interval endpoints as workable states.

[0035] To preserve the extent of leaf emergence events on the same plant and leaf, a leaf expansion accumulated temperature range is constructed based on the last time before emergence and the first time of emergence: (3) In the formula, For fixed sample plant indexing, Leaf position numbering from bottom to top; accumulated temperature range for leaf expansion. With accumulated temperature lower boundary and the upper boundary of accumulated temperature All boundary units are ℃·day; time and These represent the effective observation times for the last time a leaf at that position on the sample plant did not appear and the first time it appeared. The superscripts (negative and positive) only distinguish between the two types of observations. (Cumulative function) The value at the last unrevealed moment corresponds to the lower boundary of the accumulated temperature, and the value at the first revealed moment corresponds to the upper boundary of the accumulated temperature. The calculation method remains the same as before; parentheses indicate that the lower endpoint is not included, and square brackets indicate that the upper endpoint is included. Complete interval requirement. .

[0036] When only unexposed records or exposed records are obtained, a single-sided constraint is retained, and the leaf expansion temperature interval is formed after obtaining the other side of the record. When only unexposed records are obtained, the processor retains the condition that the exposed temperature is greater than the recorded temperature; when only exposed records are obtained, the processor retains the condition that the exposed temperature is not greater than the recorded temperature, and does not artificially fill in the other end. If the observation time of an unexposed record at the same leaf position is not earlier than the observation time of a confirmed exposed record, or if the lower boundary of the complete interval is not less than the upper boundary, the record is marked as conflicting and the original evidence is retained. The leaf position, time, and image are checked before updating. Late unexposed records whose observation time is earlier than the exposed record are still paired according to their observation time and are not judged as state reversal due to the order of reception. If the cumulative effective temperature does not increase between two valid observations but the bolster state has changed, the temperature record and the applicability of this segment of the model are checked, and the temperature is not artificially increased to form an open interval.

[0037] like Figure 2As shown, the processor converts leaf-by-leaf event records into a constraint set for each plant, and then calculates the target stage boundary. Parameter relationships are used within a limited tracking leaf position segment; this segment includes continuous leaf positions used for constraints, as well as the entry and exit leaf positions to be predicted, and does not extrapolate the entire growing season based on a single leaf expansion rate. Parameter boundary records include variety, tracking leaf position segment, temperature range, and applicable conditions; if the applicable conditions are exceeded, the process proceeds to verification and reconstruction.

[0038] The decision processor uses the sum of the leaf-spread accumulated temperature offset and the leaf-spread accumulated temperature per unit leaf position multiplied by the leaf position number as the model's leaf-spread accumulated temperature. This model's leaf-spread accumulated temperature is then extended according to a preset model error boundary to form a model leaf-spread accumulated temperature error band. The leaf-spread accumulated temperature offset represents the positional difference on a uniform accumulated temperature axis, and the leaf-spread accumulated temperature per unit leaf position represents the heat increment corresponding to adding one more leaf position. The model leaf-spread accumulated temperature error band allows for a bounded deviation between the model center and the actual exposure event of the leaf sheath, preventing discrete observation constraints from degenerating into fitting a single estimate.

[0039] To map leaf position to the accumulated temperature range required for exposure, a model error bound is applied to the affine model center of the sample plant: (4) In the formula, the accumulated temperature of the model leaf unfolding is... and its error band All units are °C·day; leaf expansion accumulated temperature offset Accumulated temperature per unit leaf position, expressed in °C·day. Leaf position number (in °C·day / leaf) The model error bound is calculated based on a 1-unit difference between adjacent leaf positions. Sample plant index (calculated in °C·day and not less than 0) Continuing from the previous text. Parameters If positive, the model center increases with the leaf position. The error band is used for event intersection judgment and calculation of upper and lower thresholds for the target stage.

[0040] The parameter boundaries are determined by registered leaf expansion records of the same variety, temperature caliber, and similar tracked leaf positions. The accumulated temperature difference between different leaf position event intervals can be normalized by leaf position difference, establishing a finite closed interval for the accumulated temperature per unit leaf position, with the lower bound strictly greater than 0. The finite range of the offset is then deduced from this interval and the accumulated temperature of each event. When using multiple calibration plants, the envelope of each plant's range is retained, along with the margin corresponding to the temperature measurement and time resolution. The parameter boundaries are re-registered when the recorded temperature caliber or variety changes.

[0041] The model error bound can be determined by the reserved leaf position verification: When there are at least 4 previous leaf expansion intervals, one leaf position is reserved each time. Using the midpoint of the interval for the remaining leaf positions as the observation, a least-squares straight line is fitted between the leaf position number and the midpoint of the interval. If the slope is positive, the fitted line is taken as the model center. If the slope is not positive, the verification is recorded and the current calibration is paused. The absolute value of the difference between the predicted accumulated temperature of the center at the reserved leaf position and the two endpoints of the interval for that leaf position is calculated. The larger value is taken, and then the maximum value of the reserved results is taken, and a temperature measurement margin is added. This process saves usage records, boundaries, and versions. When there are only 2 leaf expansion intervals and no registered error budget, the system remains in parameter preparation state and continues to acquire records.

[0042] The error bound describes the applicable scope of the current model configuration. The processor continues to check compatibility with new leaf position events, and recalibrates when the accumulated temperature model parameter set is empty, under obvious stress conditions, or when the tracking leaf position segment switches; the historical calibration range is not automatically extended to the error guarantee for all future growth conditions. The calculations described below use a model error bound of 1℃·day.

[0043] Within the parameter boundary where the accumulated temperature per unit leaf position is positive, parameter pairs consisting of the accumulated temperature offset and the accumulated temperature per unit leaf position are retained, ensuring that the model's accumulated temperature error band intersects with the accumulated temperature intervals at least two different leaf positions of the same fixed sample plant. These pairs form the accumulated temperature model parameter set. The determination is based on the intersection of the error band with each event interval, rather than requiring the entire error band to fall within the observation interval. The processor converts each complete interval into a strict lower constraint and a non-strict upper constraint; single-sided observations only add the existing one-sided constraint. The complete intervals at least two different leaf positions and the finite parameter boundary together form a computable set.

[0044] To preserve parameter pairs that can explain all registered leaf unfolding events, the parameter boundaries are intersected with the error compatibility conditions for each leaf position: (5) In the formula, the accumulated temperature model parameter set By parameter pair composition; Sample plants The finite parameter boundaries limit the offset and the positive unit leaf position expansion temperature; This is the set of leaf position numbers for this sample plant that have a complete event interval. Model center. Error bound and interval boundaries , As previously stated, all accumulated temperature measurements are expressed in °C·day. Symbols The symbol indicates that a leaf position belongs to a set of leaf positions. The symbol ∀ indicates that each leaf position in the set of leaf position numbers must satisfy the constraints. The curly braces indicate a set of parameter pairs that simultaneously satisfy the listed conditions. Existing constraints are added to single-sided records in the same direction.

[0045] Parametric intersection can be performed by a two-dimensional linear constraint solver. Due to the existence of strict inequalities, the processor first checks for strict feasibility: an auxiliary positive margin in °C·day is introduced. For each strict constraint, the difference between the upper boundary of the model and the lower boundary of the event must be at least this margin. This margin is maximized within the original non-strict constraints and parameter boundaries. The upper limit of the auxiliary margin is set to 1℃·day, used only to form bounded auxiliary problems. With strict constraints, the optimal margin must be greater than the solution tolerance to be considered non-empty; for example, the solution tolerance can be taken as... ℃·day. This tolerance only controls the numerical determination and does not characterize the accuracy of field measurements.

[0046] Once strict feasibility is established, the strict inequality is replaced with a closure constraint that includes the boundary. The upper or lower bound of the desired linear objective is then calculated to obtain the supremum or infimum of the original non-empty set. If only the closure intersection point exists without strict feasibility parameters, the set is still considered empty and not used as a candidate model. If the solver cannot confirm the boundary, the prescription remains ungenerated; the numerical margin uses conservative rounding in the time output to avoid numerical errors expanding the acceptance range.

[0047] When the accumulated temperature model parameter set is empty, the decision processor retains the leaf accumulated temperature intervals causing constraint incompatibility and sends a verification command to the observation terminal. The verification command carries the conflicting leaf position, observation time, image index, and temperature recording range. Constraints can be removed one by one for feasibility checks to locate the record combinations causing incompatibility; this trial calculation is only used to locate the cause, and a prescription is not issued after the conflicting records are deleted before verification.

[0048] If, after verification, the leaf position and temperature records still cannot form a non-empty set of accumulated temperature model parameters, a new tracking leaf position segment is established using the most recent continuous leaf positions, and the accumulated temperature model parameter set is reconstructed using the leaf expansion accumulated temperature interval within the new tracking leaf position segment. If verification confirms an error, the record is corrected and the previous version is retained; if the record is accurate but the original short segment model is no longer compatible, a new applicable interval is established using the most recent continuous leaf positions, and the parameter boundaries and error limits of this interval are reused. Old records are retained in the old tracking leaf position segment, not mixed with the new segment, and the error limits are not repeatedly expanded to forcibly form a time window.

[0049] If the new tracking leaf position segment does not yet contain accumulated temperature intervals for at least two different leaf positions, the topdressing prescription remains ungenerated. The processor continues to receive observations and save one-sided conditions until the new segment has complete event intervals, effective parameter boundaries, and error budgets for at least two different leaf positions, at which point a rigorous feasibility check is performed. The new segment also needs to cover the target inlet and outlet leaf positions, and time window calculations resume after the subsequent calculation conditions are met.

[0050] When the accumulated temperature model parameter set for all in-use sample plants is not empty, the decision processor receives the temperature forecast covering the target plot and records the release time, effective time range, lower limit of temperature, and upper limit of temperature. The forecast source can be a meteorological service interface or a plot forecast file. The input is initially standardized to Celsius temperature and continuous time period, and is linked with the current measured cumulative accumulated temperature. When the lower limit is higher than the upper limit, there is a time gap, or the forecast has expired, the system enters a forecast waiting for update state.

[0051] Based on the accumulated temperature model parameter set and temperature forecast, a common workable time window is formed by taking the maximum value of the latest time when each fixed sample plant enters the target leaf age range of the large trumpet stage as the starting point and the minimum value of the earliest departure time minus the topdressing operation time as the ending point. The latest entry and earliest departure times are considered separately for each plant, taking into account all allowable parameters and the forecast temperature range, without averaging different sample plants or different models. The topdressing operation time is the time required from the start to the completion of the current plot prescription, including fertilization travel, field turning, and necessary resupply; it is estimated and recorded based on the plot area, effective operational capacity, and operational plan. The common workable time window is recalculated when the time increases.

[0052] The decision processor, based on the accumulated temperature model parameter set for each fixed sample plant, takes the supremum of the upper boundary of the accumulated temperature error band at the inlet leaf position as the inlet accumulated temperature threshold, and the inframum of the lower boundary of the accumulated temperature error band at the outlet leaf position as the outlet accumulated temperature threshold. The inlet and outlet accumulated temperature thresholds are then converted into the latest and earliest possible times, respectively. The inlet threshold uses the accumulated temperature boundary that allows the latest possible arrival at the inlet in the model, and the outlet threshold uses the accumulated temperature boundary that allows the earliest possible arrival at the outlet. The two thresholds can be reversed, indicating that a common optimal time cannot be confirmed under the existing model uncertainty and sample plant differences; therefore, the thresholds cannot be interchanged or the mean value cannot be used to eliminate the inversion.

[0053] To transform the permissible model for each plant into conservative entry and exit conditions, the supremacy of the upper boundary of the entry error band and the infimy of the lower boundary of the exit error band are calculated separately: (6) In the formula, the accumulated temperature threshold for the initial period is... With the threshold of accumulated temperature during the departure period All measurements are in °C·day; inlet blade position Exit blade position It is the registered target leaf age boundary, superscript or subscript. , For marking entrances and exits. , Let the supremum and infimum be the upper bound and the lower bound, respectively, and the range of values ​​be calculated for the sample plants. Non-empty parameter set Model Center ,parameter , and error limits Following the previous text, the threshold is used for time-based inversion of the predicted accumulated temperature. Among them, The accumulated temperature at the center of the model plant at the inlet leaf position is the sample plant. The accumulated temperature at the model center of the sample plant at the exit leaf position is ℃·day; the former is used to calculate the accumulated temperature threshold at the beginning of the period, and the latter is used to calculate the accumulated temperature threshold at the end of the period.

[0054] The temperature forecast includes a lower bound and an upper bound. The upper and lower bounds are stored in pairs for the same forecast time period. Point forecasts can be used as a special case where the upper and lower bounds are equal; when unequal upper and lower bounds are used, the source and applicable conditions of the meteorological service given interval or the interval formed by forecast error verification should be preserved. The interval only represents the configured forecast range; the prescription time window is recalculated as the forecast is updated.

[0055] The decision processor converts the lower and upper bounds of the temperature forecast into a lower and upper bound of cumulative effective accumulated temperature, respectively. The latest time is determined when the lower bound of the cumulative effective accumulated temperature reaches the entry-period accumulated temperature threshold, and the earliest time is determined when the upper bound of the cumulative effective accumulated temperature reaches the exit-period accumulated temperature threshold. The common workable window is then limited to the workable period covered by the temperature forecast. The effective temperature difference cutoff function is monotonic; therefore, the lower bound temperature accumulates into a slower heat process, while the upper bound temperature accumulates into a faster heat process. These two processes are retained separately, neither mixing the upper and lower bounds into a single mean nor using the faster process as a basis for guaranteeing entry.

[0056] To extend the accumulated temperature range from the current measured conditions to the forecast period, the effective temperature difference corresponding to the upper and lower bounds of the future temperature is accumulated separately: (7) In the formula, For the current moment The measured cumulative effective accumulated temperature, in °C·day; and They are time points The lower and upper limits of the cumulative effective accumulated temperature are both determined by This is obtained by adding the predicted accumulated temperature increment after the current time, with units of °C·day. In the forecast time division of this formula, This represents the cumulative number of forecast periods. For the first The end time of each forecast period; This is the forecast time period index, starting from the first segment after the current time up to the [number missing]th segment. part; For the first The duration of each forecast period is in days and is greater than 0. and The first The lower and upper limits of the temperature forecast for each forecast period are in °C. and The effective temperature difference function is obtained by substituting the lower and upper bounds of the temperature above into equation (1). The obtained non-negative effective temperature difference is in °C. In this formula, the superscripts “−” and “+” for temperature and cumulative effective accumulated temperature only indicate the lower and upper bounds, respectively, and do not represent positive or negative numbers or addition and subtraction operations; the lower bound of cumulative effective accumulated temperature is accumulated only according to the lower bound of temperature, and the upper bound is accumulated only according to the upper bound of temperature.

[0057] The processor locates the time period when the first entry threshold is crossed in the lower bound accumulated temperature sequence and the time period when the first exit threshold is crossed in the upper bound accumulated temperature sequence, and then interpolates within the non-zero increment segment. The time when the exit threshold is reached is used as the cutoff boundary; if the predicted valid endpoint is earlier than the calculable exit time, the predicted valid endpoint is used as the confirmed completion cutoff.

[0058] To convert the accumulated temperature thresholds for the arrival and departure periods into calendar times for scheduled operations, the first arrival times are calculated for both cumulative forecast sequences: (8) In the formula, Sample plants The latest entry time is determined by the moment when the lower limit of the cumulative effective accumulated temperature first reaches the entry accumulated temperature threshold. The earliest time of departure for this sample plant is determined by the moment when the upper limit of the cumulative effective accumulated temperature first reaches the threshold of the accumulated temperature for departure. The time variable used in the initial arrival determination has a range of values. ; This is the valid end time for this round of temperature forecasts. , The current time of equation (7) Use the same time zone and time scale. and Representing time respectively The lower and upper limits of the cumulative effective accumulated temperature are taken at the end of each forecast period, and the corresponding calculated values ​​of equation (7) are obtained. The values ​​are calculated according to the aforementioned interpolation rules within the time period. and The definition in equation (6) is used. The set of times within the curly braces represents the time range within which the corresponding accumulated temperature arrival condition is met, and the vertical line is used to separate the range of values ​​of the time variable from the accumulated temperature condition to be met; The infimum of the set of times is taken to determine the time boundary of the first arrival. Time inversion is only performed within the forecast coverage area; if the threshold is not reached, it is handled according to the following rules.

[0059] When the entry threshold is not higher than the current measured accumulated temperature, the latest time is taken as the current time; when the exit threshold is not higher than the current measured accumulated temperature, it is marked as not being able to confirm that the plant is still in the target stage, and no joint operation prescription will be issued in this round. Recalculation is allowed after updating the parameter set based on newly added valid observations; the plant is only marked as having passed the target stage when valid observations confirm that the leaf sheath at the exit leaf position has appeared. Future temperature rates are not used to extrapolate unknown historical dates. If the threshold is not reached within the forecast period, the first arrival result is recorded as... This is used as a marker for the date not yet reached, not as the actual predicted date. If the lower bound accumulated temperature has not reached the entry threshold within the forecast period, the prescription remains ungenerated; if the upper bound accumulated temperature has not reached the exit threshold at the forecast end, only the on-time condition up to the forecast end is confirmed. If the accumulated temperature is a plateau and the threshold has not been reached, subsequent time periods will continue to be scanned without division by zero interpolation.

[0060] The maximum value of the latest time for each fixed sample plant is used as the starting point of the time window, and the minimum value of the earliest time for each fixed sample plant minus the fertilization operation duration is used as the ending point of the time window, forming a common operable time window that includes the starting point but not the ending point. The common operable time window represents the set of allowed start times, not just the leaf age state at a single instant. The start time plus the fertilization operation duration must be strictly earlier than the earliest departure boundary of all sample plants and must also be within the effective forecast range; equal starting and ending points are treated as empty time windows.

[0061] To ensure that the entire topdressing operation falls within the common suitable period and forecast validity period of the registered sample plants, the intersection of the time points for each sample plant is taken and the operation time is deducted: (9) In the formula, The set of indices of the fixed sample plants in use is non-empty; the starting point of the time window. For each latest time The maximum value, the end of the time window For the earliest time and forecast valid endpoint The earliest one to deduct the time for topdressing operations The moment after; It also uses the same time unit as the time difference. Commonly operable time window. Includes the starting point, but excludes the ending point; , To obtain the maximum and minimum values, only... Continue to formulate physical fertilizer prescriptions.

[0062] Before outputting to the work terminal, the time window start point is shifted upwards to the terminal's executable time grid, and the last grid moment, strictly less than the theoretical time window end point, is taken as the last allowed start time. The terminal uses a discrete set of allowed start times, including the first and last grid moments mentioned above; the theoretically common workable time window remains left-closed and right-open. The time margin required for equipment response is included in the topdressing operation duration before recalculating the time window, without repeatedly deducting the same margin. The time grid can be 1 second, and the actual resolution is determined by the terminal. If the first allowed grid moment is later than the last allowed grid moment, the prescription is not generated; if they are equal, work is only allowed to start at that grid moment. Time rounding does not expand the theoretically common workable time window.

[0063] When the time window is empty, the results of revealing and not revealing leaves are pre-simulated for each leaf position of the fixed sample plant with the restricted boundary, forming a two-branch parameter set. Only the leaf position commanded by the instruction, where both sets are not empty and at least one branch advances the start point or delays the end point, is selected for supplementary measurement at the observation terminal. The parameter set is updated and the time window is recalculated based on the supplementary measurement results. Figure 3 As shown, supplementary measurements expand from the sample plants within the current time window to specific leaf positions. The rehearsal only calculates the impact of hypothetical observations on decision-making and does not record actual leaf expansion; only after the terminal returns a valid supplementary measurement record are the parameter constraints for that sample plant updated and the common workable time window recalculated. Forecasts, operation durations, and other sample plant sets remain unchanged throughout the same round of branch comparisons.

[0064] The decision processor identifies fixed sample plants at the start or end of the time window as candidate samples for retesting. For the leaf positions to be observed on these candidate samples, the lower boundary of the model's leaf expansion accumulated temperature error band is no greater than the cumulative effective accumulated temperature at the retesting time, serving as an exposure constraint. The upper boundary of the model's leaf expansion accumulated temperature error band is greater than the cumulative effective accumulated temperature at the retesting time, serving as a non-exposure constraint. The accumulated temperature model parameter set is intersected with both the exposure and non-exposure constraints to form an exposure branch set and a non-exposure branch set. The processor retains all sample plants sharing a common boundary and enumerates the leaf positions that are still identifiable and whose current state is not yet determined. For immediate retesting, the current on-site accumulated temperature is used; if there is a delay in retesting, the terminal first transmits the arrival time and temperature record from the site, and the processor re-performs the simulation based on the actual accumulated temperature before determining the leaf positions to be tested in this round. Each branch only replaces the parameter set of the selected sample plant; other sample plants retain their original sets.

[0065] To determine whether the two possible results of a supplementary measurement at a certain leaf position affect the operation boundary, the current parameter set is intersected with both revealed and unrevealed conditions: (10) In the formula, the set of exposed branches is revealed. and the set of unrevealed branches Corresponding sample plants leaf position ; , These are status markers indicating whether something is visible or not. This represents the cumulative effective accumulated temperature at the time of this round of supplementary measurements, expressed in °C·day, with subscripts indicating the time elapsed. Used to mark the current round of supplementary measurements. This represents finding the intersection of sets; This is the set of allowable parameter pairs for the fixed sample plant before the supplementary testing; This is the model offset, expressed in °C·day; The accumulated temperature required for each additional leaf position is expressed in °C·day / leaf. The accumulated temperature at the center of the model at the fixed leaf position of the sample plant is expressed in °C·day; This serves as the model error bound, expressed in °C·day. The two branches may overlap, as the same model error band may simultaneously allow for two states; non-empty branches are still judged according to strict feasibility rules.

[0066] Only leaf positions to be observed that are not empty in both the exposed branch set and the unexposed branch set, and where at least one branch, after recalculation, advances the start point of the time window or delays the end point of the time window, are included in the candidate leaf positions for supplementary testing. Supplementary testing targets are then selected from these candidate leaf positions. If a branch is infeasible, and there are no indistinguishable results for that leaf position under the current constraints, it is not included in this round of supplementary testing. If two branches reduce the parameter range but do not change the common operational boundary, they are also not included. Parallel boundary sample plants are retained in the trial calculation to prevent improving only one plant while ignoring the other, which still restricts the same boundary. Boundary changes must also reach a level that is identifiable by a unified time grid.

[0067] The decision processor obtains the time window margin by subtracting the time window start point from the time window end point, and preferentially selects the candidate leaf positions for supplementary measurements that, after recalculation of at least one branch, ensure that the common workable time window is not empty. The processor preferentially selects candidate leaf positions for which at least one valid observation result can change the rejection state to an executable state; this does not mean that the favorable result will definitely occur. When multiple candidates of the same level exist, the smaller branch margin of each candidate can be compared first, and then a stable order can be determined according to the fixed sample plant identifier and leaf position number.

[0068] If no candidate leaf position meets the priority selection criteria, the selection is made in descending order of the smaller value of the time window margin recalculated from the exposed branch set and the unexposed branch set. This selection maximizes the time window margin corresponding to the more unfavorable outcome of the two observations. A negative margin indicates that the starting point is still later than the ending point and can be used for comparison between candidates that still have no window, but it is not a job permission. The candidate results and the versions of observations, forecasts, and models used for comparison are cached together.

[0069] To compare the less favorable outcomes of the two observations when there are no candidates for direct job release, the branch margin for each candidate leaf position is calculated: (11) In the formula, the candidate comparison quantity Sample plants Leaf position The smaller of the two branch window margins, with the unit consistent with the operation duration; , To reveal the start and end points of the time window for branch recalculation, , This represents the corresponding boundary for recalculation of branches that are not explicitly shown. , Continue using branch status markers, This indicates that the smaller value is taken. Only candidate leaf positions for supplementary measurements that are feasible in both branches are calculated, and selected in descending order of this comparison value.

[0070] When the candidate leaf position for supplementary testing is empty, the topdressing prescription remains ungenerated, and the common operable time window is recalculated after new observation records are added or temperature forecasts are updated. The system does not change rejection to acceptance by deleting restricted sample plants, shortening the actual operation time, or arbitrarily widening the target leaf age range. Asynchronous actual growth may result in no common optimal period under all allowable models; in this case, continued supplementary testing cannot eliminate differences between plants. When zoning operations are required, the zoning criteria and sample plant range are re-registered, and the calculation is performed independently for each new plot, without using the common prescription from the original plot.

[0071] The supplementary measurement instruction includes the plot identifier, fixed sample plant identifier, leaf position, request number, observation time range, and record version. The terminal returns the same request number, sample plant, leaf position, actual observation time, and leaf sheath status; the processor verifies the identity and time before submitting the corresponding branch constraints. If the result is unreadable, times out, identity mismatch, or the returned time differs from the pre-recorded time, the valid temperature and time are updated first, and then recalculated. Results that fail the verification are not included in the leaf unfolding event.

[0072] Leave positions that have already been calculated under the same data version and have no executable boundary changes will not be resent with supplementary testing instructions. If the actual supplementary testing still does not form a non-empty common workable window, the restricted samples and candidates will be re-identified in the updated version; if the candidates are empty, the effective observation has confirmed the emergence of the exit leaf position, or the forecast is insufficient, the current round will end and wait for effective input to update. In this way, the results of one supplementary test are linked to the next defined state, avoiding cyclical requests under unchanged conditions.

[0073] When the time window is not empty, the topdressing budget for the current stage is used to deduct the amount of fertilizer already applied to generate a topdressing prescription with an operating time limited to the time window. The prescription is only generated when observations, models, forecasts, and budgets are all valid. The stage topdressing budget is used to determine the amount of topdressing still needed for this stage, while leaf age and accumulated temperature treatments are used to determine the timing of operations; the processor does not infer the degree of nitrogen deficiency solely based on leaf age lag. The operating conditions in the prescription are for verification by the operational terminal, and must be recalculated with the new operating duration if the execution capacity changes.

[0074] The stage-based topdressing budget and the amount of fertilizer already applied are measured by nitrogen mass for the same plot and the same topdressing stage. The stage-based topdressing budget comes from the stage-based nitrogen application plan registered for the plot, which can be formed based on soil nitrogen supply, previous fertilization, and cultivation schemes, and records the applicable plot, stage, and unit of measurement. If the input is the amount of nitrogen applied per unit area, it is first multiplied by the registered area to convert it into the total nitrogen mass of the plot; the amount of fertilizer already applied is calculated and accumulated based on the actual amount of fertilizer applied and the corresponding nitrogen mass fraction, and is not deducted from prescriptions that have been issued but not yet completed.

[0075] The decision processor uses the non-negative difference between the stage topdressing budget and the amount of fertilizer already applied as the remaining nitrogen application amount. It receives the nitrogen mass fraction of the fertilizer, divides the remaining nitrogen application amount by the nitrogen mass fraction to obtain the remaining fertilizer amount, and writes the remaining fertilizer amount, plot identifier, and common workable time window into the topdressing prescription. The nitrogen mass fraction of the fertilizer is read from the effective ingredient label or mass record of the fertilizer used; it must be greater than 0 and not greater than 1, and consistent with the fertilizer to be applied. If no effective budget is registered, the stage topdressing budget is negative, the amount of fertilizer already applied is negative, or the fertilizer composition is illegal, the physical prescription remains ungenerated. Completed operations with different fertilizers are first converted to nitrogen mass separately, and then accumulated within the same stage.

[0076] To convert the remaining nitrogen application plan for the same plot at the same stage into the current physical amount of fertilizer, first deduct the amount of nitrogen already applied, and then convert it according to the nitrogen content by mass fraction: (12) In the formula, the budget for staged topdressing is... Amount of fertilizer already applied and remaining nitrogen application All figures represent the total nitrogen mass of the plot at this stage, in kilograms; remaining fertilizer amount. The total mass of fertilizer is expressed in kilograms; the nitrogen content of the fertilizer is expressed as a mass fraction. Dimensionless and satisfying . This indicates that negative differences are truncated to 0. When the remaining nitrogen application is 0, no additional fertilization is required for this stage; when it is positive, the remaining fertilizer amount and the common operable time window are used to form the topdressing prescription.

[0077] Each topdressing prescription is associated with the observation version, model parameter version, temperature forecast version, budget version, and operation duration. Before starting the operation, the terminal re-verifies the prescription version, permitted time, remaining fertilizer amount, and fertilizer composition. When it receives updates to the observations, models, forecasts, budgets, or operation duration that affect the above, it marks the old prescription as invalid and regenerates the prescription according to the updated time window and remaining amount, even if the original start time is still within the new time window, it does not use the old fertilizer amount that has not been re-verified. After the operation is completed, the actual fertilizer amount and fertilizer composition are returned according to the unique operation number. The processor removes duplicates and updates the fertilizer amount applied. Partially completed tasks are recorded according to the actual amount completed, and the remaining tasks are recalculated with the new duration and time window.

[0078] The following example, using a single plot of land, illustrates the complete processing procedure. The plot area is 1 hectare. Fixed sample plants A and B are registered, with target inlet leaf position 11 and outlet leaf position 13. The current measured cumulative temperature is 500℃·day, the forecast validity period is 10 days from the current date, the upper and lower bounds of the forecast temperature are both 20℃, the base temperature is 10℃, the upper limit temperature is 30℃, the topdressing operation duration is 0.5 days, and the model error bounds are all 1℃·day. These values ​​are used to calculate various constraints and state changes.

[0079] Leaf positions 8 and 9 of the fixed sample plant A each have accumulated temperature ranges for leaf expansion. ℃·Hiwa ℃·day, the range of accumulated temperature per unit leaf position is ℃·day / leaf; the same leaf position of fixed sample plant B has a range ℃·Hiwa ℃·day, the range of accumulated temperature per unit leaf position is ℃·day / leaf. The offset range for both plants is taken as... ℃·day. The complete intervals are all formed from the current and previous observations. The parameter set is obtained by intersecting these intervals with the boundary. No models are arbitrarily selected in advance to replace all allowed parameters.

[0080] Based on the above constraints, the accumulated temperature thresholds for entry and exit of fixed sample plant A are 572℃·day and 547℃·day, respectively, while those for fixed sample plant B are 506℃·day and 554℃·day, respectively. The effective accumulated temperature increases by 10℃·day per day, therefore the latest possible entry date is 7.2 days from now, and the earliest possible exit date is 4.7 days from now. After deducting 0.5 days of working time, the allowable working interval is... On the day the starting point is later than the ending point, the processor refuses to issue the physical prescription and initiates a supplementary test option.

[0081] The original time window boundary was restricted by fixing the sample plant A. Using the current accumulated temperature of 500℃·day, a leaf position of 10 was pre-calculated, revealing branches and reducing the entry threshold of plant A to 552℃·day, thus changing the allowable start-up interval. The day remains empty; the absence of branching causes the isolation threshold for plant A to become 607.5℃·day, but the 554℃·day of the fixed sample plant B becomes the common isolation limit, with the interval being... The day is still empty. This leaf position can change the boundary, but neither result can directly release the job.

[0082] When the leaf position of fixed sample plant A was 11, the exposure of branches reduced the entry threshold of A to 502℃·day, while the exit threshold of A remained at 547℃·day. The 506℃·day of fixed sample plant B became the common entry limit, resulting in a non-empty interval. Day. Without revealing any branches, the threshold for the departure period of A is 558℃·day, and the common time window remains [missing information]. Leaf position 11 satisfies the priority condition that at least one branch makes the operation feasible, and therefore takes priority over leaf position 10 for supplementary testing.

[0083] If the terminal returns that leaf position 11 of the fixed sample plant has been exposed, the processor submits the exposure constraint, recalculates the allowable start date range from 0.6 days to 4.2 days after the current date, and can schedule the start date to be 0.6 days after the current date and the completion date to be 1.1 days after the current date, with the earliest departure date being earlier than 4.7 days. If the terminal returns that the leaf position has not been exposed, the application will continue to be rejected. Unexposed branches at leaf positions 8 and 9 are not feasible, and exposed branches at leaf positions 12 and higher are not feasible; neither is included in the candidate for this round of double-branch supplementary testing.

[0084] If the lower and upper bounds of the future effective accumulated temperature for the same calculation are changed to 9°C·day and 11°C·day respectively, after the leaf position 11 explicit constraint is submitted, the latest entry point will be the current one. The earliest departure time is approximately 0.6667 days. The estimated timeframe is approximately 4.2727 days. Subtracting 0.5 days of operating time, the allowable start and end dates are approximately 3.7727 days, leaving a non-empty time window. Using slower temperature accumulation at the inlet and faster temperature accumulation at the outlet will correspondingly tighten the confirmed operating range as the forecast area expands.

[0085] Assuming a valid work window is available, the budgeted nitrogen application for this 1-hectare plot is 90 kg. 30 kg of nitrogen has already been applied, and the current nitrogen content of the fertilizer is 0.46%. The calculated remaining nitrogen application is 60 kg, and the remaining fertilizer amount is approximately 130.4347826 kg. The terminal can display 130.43 kg. The calculation and work settings retain the required precision to avoid rounding errors that could negatively impact the budget. The actual amount of fertilizer applied is updated only after the work completion receipt is received.

[0086] Using a current accumulated temperature of 400℃·day and a forecast validity period of 20 days, with other temperature and operation duration conditions remaining unchanged, the fixed parameters for slow-growing plants are an offset of 0℃·day and an accumulated temperature per leaf position of 50℃·day / leaf; for fast-growing plants, these are 0℃·day and 35℃·day / leaf, with an error margin of 1℃·day for both. The previous leaf position 6 and 7 event intervals for slow-growing plants are... and ℃·day; leaf positions of fast-growing plants in the 9th and 10th intervals are and °C·day. The permissible parameters for both plants are the fixed parameters mentioned above.

[0087] The thresholds for the entry and exit periods for slow-growing plants are 551℃·day and 649℃·day, respectively, while those for fast-growing plants are 386℃·day and 454℃·day, respectively. The allowable operating range is... The current day is still empty. Even if the error band of a certain leaf position of the slow plant allows for two states at the current moment, the two branches do not change the fixed parameters and operation boundaries, so there are no effective candidates for supplementary testing. When the accumulated temperature reaches 500℃·day, the exit threshold of the fast plant has been crossed, and the processor marks it as unable to confirm that the sample plants in this plot are still in the common target stage, and stops generating the common prescription for the current target stage of this plot; if effective observation confirms that the exit leaf position of the fast plant has been revealed, then the fast plant is marked as having passed the target stage.

[0088] In another implementation, the decision processor is deployed locally at the observation terminal, and the temperature acquisition device transmits records via a short-range wireless connection. The forecast file is updated when the network is connected. The parameter set can be stored using a half-plane constraint list, or the upper and lower bounds of the linear target can be calculated using the vertices of the closure polygon after a strict feasibility determination. Both data structures retain the same open boundary information and sample plant identity, and supplementary measurements are still performed according to the same branch conditions and common operation boundaries.

[0089] The observation terminal can also use a fixed position to assist in acquiring side images of the sample plants, which are then manually verified to determine the leaf sheath exposure status. Automatic image processing, if involved in leaf position localization, is only used for auxiliary localization and prompts; images that are difficult to interpret will still be returned as uninterpretable. Changing the deployment location of the acquisition equipment or processor does not alter the relationships between event pairings, parameter set attribution, and submission constraints after actual feedback for the same plant and leaf.

Claims

1. A precision topdressing decision-making system for summer maize at the tasseling stage based on dynamic leaf age tracking and accumulated temperature model, characterized in that, This includes the observation terminal, the temperature acquisition device, and the decision processor connected to both. The observation terminal records the leaf position and leaf sheath exposure status of the fixed sample plant; The decision processor forms a leaf expansion accumulated temperature interval based on the last unrevealed and first revealed times of the same leaf position of the same fixed sample plant and the temperature data from the temperature acquisition device. It retains a set of accumulated temperature model parameters that satisfy the constraints of the leaf expansion accumulated temperature interval for each fixed sample plant. According to the accumulated temperature model parameter set and temperature forecast, the maximum value of the latest time when each fixed sample plant enters the target leaf age range of the large trumpet stage is taken as the starting point of the time window, and the minimum value of the earliest time of departure minus the topdressing operation time is taken as the ending point of the time window, forming a common workable time window. When the time window is empty, the results of revealing and not revealing are pre-rehearsed for each leaf position of the fixed sample plant with the limiting boundary, forming a two-branch parameter set. Only leaf positions where both sets are not empty and at least one branch advances the starting point or delays the ending point are selected to instruct the observation terminal to conduct supplementary measurements. The parameter set is updated and the time window is recalculated based on the supplementary measurement results. When the time window is not empty, the topdressing prescription with the start time limited to the time window is generated by deducting the amount of fertilizer already applied according to the stage topdressing budget.

2. The system according to claim 1, characterized in that, The observation terminal assigns a fixed sample plant identifier to each fixed sample plant, numbers the leaf positions consecutively from bottom to top, and stores the fixed sample plant identifier, the leaf position, the observation time, and the leaf pulvinus exposure state in association; the observation terminal saves the leaf unfolding record of the exposed leaf position, and continues the leaf position numbering by using the leaf unfolding record after the lower leaves wither, and the leaf position number of the same fixed sample plant remains unchanged at different observation times.

3. The system according to claim 2, characterized in that, The decision processor calculates the cumulative effective accumulated temperature using a unified accumulated temperature starting point based on the temperature data. The cumulative effective accumulated temperature at the last unexposed and first exposed times of the same leaf position of the same fixed sample plant is used as the lower boundary and upper boundary of the accumulated temperature range for leaf expansion, respectively. The lower boundary of the accumulated temperature is an open boundary, and the upper boundary of the accumulated temperature is a closed boundary. When only unexposed or exposed records are obtained, a one-sided constraint is retained, and the accumulated temperature range for leaf expansion is formed after obtaining the other side record.

4. The system according to claim 3, characterized in that, The decision processor uses the sum of the leaf expansion temperature offset and the leaf expansion temperature per unit leaf position multiplied by the leaf position number as the model leaf expansion temperature. It expands the model leaf expansion temperature according to the preset model error boundary to form the model leaf expansion temperature error band. Within the parameter boundary where the leaf expansion temperature per unit leaf position is positive, it retains parameter pairs consisting of the leaf expansion temperature offset and the leaf expansion temperature per unit leaf position that make the model leaf expansion temperature error band intersect with the leaf expansion temperature intervals of at least two different leaf positions of the same fixed sample plant, forming the temperature accumulation model parameter set.

5. The system according to claim 4, characterized in that, The target leaf age range for the large trumpet stage is defined by the exposure of the leaf pulvinus at the entrance leaf position as the entry condition and the exposure of the leaf pulvinus at the exit leaf position as the exit condition. The decision processor takes the supremacy of the upper boundary of the model leaf expansion accumulated temperature error band at the entrance leaf position as the entry accumulated temperature threshold and the inframance of the lower boundary of the model leaf expansion accumulated temperature error band at the exit leaf position as the exit accumulated temperature threshold, based on the accumulated temperature model parameter set for each fixed sample plant. The entry accumulated temperature threshold and the exit accumulated temperature threshold are then converted into the latest time and the earliest time, respectively. The maximum value of the latest time for each fixed sample plant is taken as the starting point of the time window, and the minimum value of the earliest time for each fixed sample plant minus the topdressing operation time is taken as the ending point of the time window, forming a common workable time window that includes the starting point of the time window but does not include the ending point of the time window.

6. The system according to claim 5, characterized in that, The decision processor identifies the fixed sample plants that generate the start or end point of the time window as candidate samples for retesting. For the leaf positions to be observed in the candidate samples, the lower boundary of the model leaf expansion accumulated temperature error band is not greater than the cumulative effective accumulated temperature at the retesting time as an exposure constraint, and the upper boundary of the model leaf expansion accumulated temperature error band is greater than the cumulative effective accumulated temperature at the retesting time as a non-exposure constraint. The accumulated temperature model parameter set is intersected with the exposure constraint and the non-exposure constraint to form an exposure branch set and a non-exposure branch set, respectively. Only the leaf positions to be observed that are not empty in both the exposure branch set and the non-exposure branch set, and whose recalculation results in the start of the time window being advanced or the end point of the time window being delayed, are included in the candidate leaf positions for retesting, and the retesting object is selected from the candidate leaf positions for retesting.

7. The system according to claim 5, characterized in that, The temperature forecast includes a lower bound and an upper bound. The decision processor converts the lower bound and the upper bound of the temperature forecast into a lower bound and an upper bound of cumulative effective accumulated temperature, respectively. The latest time is determined when the lower bound of the cumulative effective accumulated temperature reaches the entry accumulated temperature threshold, and the earliest time is determined when the upper bound of the cumulative effective accumulated temperature reaches the exit accumulated temperature threshold. The common workable time window is then restricted to the work period covered by the temperature forecast.

8. The system according to claim 1, characterized in that, The stage topdressing budget and the amount of fertilizer already applied are measured by nitrogen mass in the same plot and the same topdressing stage; the decision processor uses the non-negative difference between the stage topdressing budget and the amount of fertilizer already applied as the remaining nitrogen amount, receives the nitrogen mass fraction of fertilizer, divides the remaining nitrogen amount by the nitrogen mass fraction of fertilizer to obtain the remaining fertilizer amount, and writes the remaining fertilizer amount, plot identifier and the common workable time window into the topdressing prescription.

9. The system according to claim 6, characterized in that, The decision processor obtains the time window margin by subtracting the time window start point from the time window end point, and preferentially selects the supplementary measurement candidate leaf positions that, after recalculation of at least one branch, make the common workable time window non-empty; if there are no supplementary measurement candidate leaf positions that meet the priority selection conditions, the selection is made in descending order of the smaller value of the time window margin obtained by recalculating the exposed branch set and the unexposed branch set; when the supplementary measurement candidate leaf positions are empty, the topdressing prescription is kept in an ungenerated state, and the common workable time window is recalculated after new observation records or temperature forecast updates.

10. The system according to claim 4, characterized in that, When the accumulated temperature model parameter set is empty, the decision processor retains the leaf expansion accumulated temperature interval that causes constraint incompatibility and sends a verification instruction to the observation terminal; when the leaf position record and temperature record still cannot form a non-empty accumulated temperature model parameter set after verification, a new tracking leaf position segment is established with the newly recent continuous leaf position, and the accumulated temperature model parameter set is reconstructed with the leaf expansion accumulated temperature interval in the new tracking leaf position segment; when the new tracking leaf position segment has not yet contained at least two leaf expansion accumulated temperature intervals with different leaf positions, the topdressing prescription is kept in an ungenerated state.

Images