A method for early identification and drenching control of rhizome rot of polygonatum

Abstract

This invention discloses a method for early identification and root drenching control of root rot in Polygonatum sibiricum, relating to the field of disease monitoring and precise control in traditional Chinese medicinal materials. The method involves setting up an electrochemical monitoring unit, a segmented valve control unit, a diversion sampling unit, a nucleic acid confirmation unit, an oxygen regulation unit, and a microcapsule control unit in the root zone of the Polygonatum sibiricum planting bed, and establishing a shared situation vector (RZS) in the data fusion center. The method incorporates missing fingerprint MDF, phenotypic event items, nucleic acid confidence items, and reachability items into a unified gating link, completing the sequential control of early identification, nucleic acid confirmation, oxygen regulation pulsed root drenching, and microcapsule injection. Through write-back updates of threshold library objects, non-disease drift library objects, and waterway maintenance objects, a closed-loop operation of identification and control is achieved.

Description

Technical Field

[0001] This invention relates to the field of disease monitoring and precise control of Chinese medicinal materials, specifically a method for early identification and root irrigation control of root rot in Polygonatum sibiricum. Background Technology

[0002] Polygonatum is a perennial medicinal plant, and its roots and rhizomes are key components for yield and quality. Root rot typically begins with an imbalance in the root zone's microenvironment. Early symptoms are mainly concentrated in the underground root system and rhizosphere, with above-ground symptoms appearing relatively late. At this stage, it is often difficult to establish stable identification conclusions through routine field inspections, frequently resulting in late detection, a shortened control window, and the expansion of localized lesions before further treatment.

[0003] Current technical approaches primarily rely on manual inspection and experience-based judgment, using observations of plant growth, leaf color changes, and root and stem abnormalities for diagnosis. While this approach is feasible in the mid-to-late stages of disease, it lacks continuous monitoring methods for changes in the disease progression in the early stages, making it difficult to establish repeatable risk assessment criteria at the bed level. Inconsistent judgment standards among different operators lead to unstable triggering conditions. Alternatively, warnings may be issued using a single signal or a few indicator thresholds, such as relying solely on ORP, conductivity, or local environmental parameters. While these solutions are structurally simple, they are prone to false or missed triggers under conditions of irrigation disturbance, temperature fluctuations, or missing sampling. Even with subsequent detection steps, identification, confirmation, and control are often disconnected, failing to form a unified state-driven closed-loop decision-making process. Another approach separates detection from root irrigation application, with the relationship between detection results and control actions relying on manual interpretation or static rule mapping. This approach has two problems in engineering deployment: the detection results are not modeled in a unified manner with the control permission conditions, resulting in the coexistence of detectable but not safe to control and controllable but lacking sufficient evidence; after control is implemented, there is a lack of a structured write-back mechanism, and historical negative samples, missing detection mechanisms and maintenance information fail to continuously feed back into threshold and strategy updates.

[0004] For the case of root rot in Polygonatum sibiricum, current technologies lack systematic governance for missing data states. The field data acquisition process is affected by sensor contamination, pipeline conditions, agricultural activities, and environmental disturbances, resulting in frequent instances of continuous and shared missing data. If missing data information is merely noted as an anomaly and not incorporated into the main decision-making chain, the system cannot distinguish between insufficient evidence and genuinely low risk, thus affecting the consistency of confirmation triggering, control authorization, and downgrade maintenance.

[0005] Furthermore, existing technologies for using phenotypic evidence are mostly limited to isolated recordings, failing to form a closed loop with electrochemical status, nucleic acid confirmation, and the object of rewriting. For underground diseases like root rot in Polygonatum sibiricum, without the linkage constraint between phenotypic events and nucleic acid confirmation results, negative sample rewriting can easily introduce contaminated samples, leading to deviations in subsequent threshold updates and insufficient long-term operational stability.

[0006] Therefore, there is an urgent need for an early identification and root irrigation control method for Polygonatum root rot that can use the shared situation vector RZS as a unified interaction object, incorporate missing fingerprint MDF and phenotypic event items into the gating link, and connect identification, verification, control, judgment, re-examination and maintenance write-back, so as to solve the shortcomings of existing technologies in terms of early identification stability, control verifiability and long-term self-calibration capability. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and propose a method for early identification and root irrigation control of root rot in Polygonatum sibiricum, so as to solve the above-mentioned problems.

[0008] The objective of this invention is achieved through the following technical solution: a method for early identification and root drenching control of root rot in Polygonatum sibiricum, comprising the following steps: S1 sets up unit modules in the root zone of the Polygonatum planting bed. The unit modules include an electrochemical monitoring unit, a segmented valve control unit, a shunt sampling unit, a nucleic acid confirmation unit, an oxygen regulation unit, and a microcapsule control unit. A shared situation vector RZS is established in the data fusion center. The unit modules only operate collaboratively by reading and writing the shared situation vector RZS and do not directly exchange raw observation data and control quantities. S2 establishes a baseline window and drift rate upper limit based on the healthy period fingerprint database of the bed segment and initializes the shared situation vector RZS. The shared situation vector RZS includes ORP deviation, impedance distance, conductivity disturbance, water content deviation, temperature stress, dissolved oxygen deficiency, acidification anomaly, enzyme proxy, nucleic acid confidence, reachability, missing fingerprint MDF and phenotypic event. The missing fingerprint MDF includes missing location, missing duration, missing phase and missing correlation and is used to limit the missing domain and control domain. The phenotypic event includes necrotic spot indication, softening indication, mold layer indication, pull-out resistance indication, growth potential indication and chlorosis indication. S3 continuously integrates and updates the shared situation vector RZS and performs early identification gating: Under the condition that the missing fingerprint MDF has not entered the missing domain and the conductivity perturbation term has not exceeded the perturbation threshold, when the cooperative drift of the ORP deviation term and the impedance distance term meets the consistency criterion and the necrotic spot indication, softening indication and pull-out resistance indication in the phenotypic event term meet the positive criterion, a bed segment risk marker is generated and the shunting sampling unit is triggered to shun the leachate according to the shunting parameters determined by RZS; Under the condition that the missing fingerprint MDF restricts the updating of the ORP deviation term or the impedance distance term, when the acidification anomaly term and the enzyme proxy term show a cooperative abnormal trend and the phenotypic event term meets the confirmation gating condition, a missing-driven confirmation trigger is generated and the trigger path is recorded. The S4 nucleic acid confirmation unit performs nucleic acid identification on the diverted leachate to obtain the nucleic acid confirmation result and writes it into the RZS with confidence. When the nucleic acid confirmation is negative, it is written into the non-pathological drift library under the condition that the phenotypic event item is negative and the threshold library and consistency criteria are updated. When the nucleic acid confirmation is invalid, the missing mechanism identifier is written and the retest strategy is updated. S5 When the RZS characterization of nucleic acid confidence terms reaches the positive threshold, reachability terms meet the control permission domain, and the missing fingerprint MDF is in the control verifiable domain, oxygen-regulating pulsed root irrigation is first performed, and the regression rates of ORP deviation terms and dissolved oxygen deficit terms are written back to the RZS as the gating conditions for entering microcapsule control; when the root region enters the controllable window and acidification anomaly terms and enzyme proxy terms are in an observable state, microcapsule injection is performed, and the injection parameters are determined by the RZS and corrected according to reachability terms and missing mechanism identifiers; After S6 control is applied, RZS is continuously updated and stopped and re-examined: the risk mark is removed when the baseline is returned and the phenotypic event item turns negative; when the abnormality and missing fingerprint MDF are allowed, re-examine and repeat control in the order of oxygen regulation-microcapsule; when the missing fingerprint MDF enters the degraded state, enter the degraded mode and prohibit oxygen regulation and microcapsule control, and write back the maintenance object.

[0009] The healthy period fingerprint database consists of root zone ORP data, rhizosphere impedance spectrum data, conductivity data, root zone water content data, root zone temperature data, leachate acidification signal, and extracellular enzyme activity proxy signal continuously collected during the disease-free period of the same bed segment. ORP deviation terms, impedance distance terms, conductivity perturbation terms, water content deviation terms, temperature stress terms, dissolved oxygen deficit terms, acidification anomaly terms, enzyme proxy terms, nucleic acid confidence terms, and reachability terms are written into the shared situation vector RZS to form electrochemical deviation term group and environmental perturbation term group.

[0010] The fusion update includes time updates and measurement updates. The time update decays the historical situation according to the forgetting factor to maintain trend continuity. The measurement update writes the new observations into the shared situation vector RZS and adjusts the fusion weights according to the conductance perturbation term and temperature stress term. At the same time, consistency checks are performed on the ORP deviation term, impedance distance term and water content deviation term, and the inconsistency window is written into the environmental perturbation flag for the suppression determination of the consistency criterion.

[0011] Missing fingerprint MDF is written into the shared situation vector RZS and includes the missing location, missing duration, missing phase, and missing correlation, and generates a missing mechanism identifier. The missing location is set for root zone ORP data, rhizosphere impedance spectrum data, root zone water content data, root zone temperature data, conductivity data, leachate acidification signal, extracellular enzyme activity proxy signal, and water flow and pressure difference signal, and distinguishes between continuous and intermittent missing. The missing duration includes the longest continuous missing duration, cumulative missing duration, and missing duty cycle. The missing phase includes intraday phase, irrigation phase, agricultural phase, and fertility phase, and records the phase peak and repeatability. The missing correlation includes the set of co-missing pairs, the number of co-missing times, and the co-missing delay, and is used to determine the link co-missing structure. The missing domain and the control domain are jointly determined by the missing duration and missing correlation.

[0012] Phenotypic event items are written into the shared situation vector RZS and include phenotypic indicator sub-items corresponding to water-soaked brown necrotic spots on the root or rhizome surface, softening of root or rhizome tissue, mold layer on the rhizome surface, reduced uprooting resistance, reduced growth potential, and chlorosis and yellowing of lower leaves, respectively. Phenotypic event items are generated from root zone sampling records, rhizome surface image discrimination output, and uprooting resistance discrimination output and written into the shared situation vector RZS with timestamps.

[0013] The consistency criteria include combined drift gating and phenotypic gating. Combined drift gating is determined by the co-drift of the ORP deviation term and the impedance distance term satisfying the consistency check and the conductivity perturbation term not exceeding the perturbation threshold. Phenotypic gating is determined by the necrotic spot indication, softening indication and pull-out resistance indication in the phenotypic event term satisfying the positive criteria. The confirmatory gating condition is determined by the missing fingerprint MDF restricting the update of the ORP deviation term or the impedance distance term and the acidification abnormality term and the enzyme proxy term showing a co-abnormal trend.

[0014] The nucleic acid confirmation unit performs nucleic acid release processing on the shunt filtrate within the detection branch and performs nucleic acid recognition and signal amplification reactions to output nucleic acid confirmation results. When the nucleic acid confirmation result is negative, under the condition that the phenotypic event item is negative, the corresponding combination drift sample is written into the non-disease drift library and the threshold library and consistency criteria are updated. When the nucleic acid confirmation result is invalid, the timing and number of retests are determined based on the missing phase, missing duration, missing correlation and reachability, and the invalidity reason is written into the missing mechanism identifier to update the retest strategy.

[0015] The control permission domain includes nucleic acid confidence terms reaching the positive threshold, reachability terms meeting the control permission conditions, and missing fingerprint MDF being in the control verifiable domain. The target window, pulse duration, and interval of oxygen-regulating pulse root irrigation are calculated from ORP deviation terms, dissolved oxygen deficit terms, water content deviation terms, and reachability terms. The regression rates of ORP deviation terms and dissolved oxygen deficit terms are written back to the shared situation vector RZS as the gating condition for entering microcapsule control. When the missing fingerprint MDF sets the water flow and pressure difference signal or the reachability terms do not meet the conditions, oxygen-regulating pulse root irrigation is prohibited, and the reason for prohibition is written into the missing mechanism identifier to trigger maintenance writeback.

[0016] Microcapsule injection is performed after the root zone enters a controllable window, the acidification anomaly and enzyme proxy are in an observable state, and the missing fingerprint MDF has not entered a degraded state. The microcapsule root irrigation agent used by the microcapsule control unit is unlocked and released when the acidification environment and the extracellular enzyme enrichment environment are simultaneously satisfied, so as to form a local high gradient in the lesion microenvironment and inhibit non-target release in the healthy root zone. The dosage, concentration and injection rate of microcapsule injection are determined by the impedance distance term, acidification anomaly term, enzyme proxy term and reachability term, and are dynamically corrected during the injection process based on the reachability term and missing mechanism identifier.

[0017] The write-back objects include threshold library objects, diversion and confirmation triggering strategy table objects, oxygenation parameter objects, microcapsule dose mapping objects, bed segment risk map objects, phase blacklist library objects, non-disease drift library objects, and waterway maintenance objects. Among them, the threshold library objects are used to update the early screening threshold, disturbance exclusion threshold, missing domain boundary, and control domain boundary. The diversion and confirmation triggering strategy table objects are used to update diversion parameters, retest intervals, and retesting strategies. The oxygenation parameter objects are used to update the target window, pulse parameters, and phase avoidance strategies. The microcapsule dose mapping objects are used to update the situation and dose-response mapping relationship and use missing fingerprint MDF and phenotypic event items as dose correction factors. The phase blacklist library objects are used to solidify missing phase contamination windows and sampling scheduling avoidance rules. The waterway maintenance objects generate maintenance marks based on the missing mechanism identifier and, after maintenance, use missing measurement recovery boundary events to write back, remove the degradation mode, and relearn the baseline window and drift rate upper limit.

[0018] The beneficial effects of this invention are: This invention integrates the electrochemical monitoring unit, shunt sampling unit, nucleic acid confirmation unit, oxygen regulation unit, and microcapsule control unit into a shared situation vector (RZS) driving framework. Each unit operates collaboratively only by reading and writing the shared situation vector (RZS), without directly exchanging raw observation data and control variables. This structure enables identification, confirmation, and control to operate under the same semantic state, reducing triggering deviations caused by semantic inconsistencies between links and improving the consistency of bed-level process execution.

[0019] This invention simultaneously incorporates ORP deviation, impedance distance, conductivity disturbance, water content deviation, temperature stress, dissolved oxygen deficit, acidification anomaly, enzyme proxy, nucleic acid confidence, and reachability terms into the shared situation vector RZS, and organizes early identification gating using consistency criteria. Compared to single-index judgment methods, this scheme, through multiple situational coordination constraints on triggering conditions, can maintain the stability of the identification link under conditions where irrigation disturbances and environmental fluctuations coexist, reducing non-target triggering caused by occasional fluctuations.

[0020] This invention incorporates the missing fingerprint MDF as a decision variable into the shared situation vector RZS, and defines the missing domain and control domain by missing location, missing duration, missing phase, and missing correlation. When critical observation channels are restricted, the system does not simply shut down, but instead switches to missing-driven confirmation triggering and degradation strategy determination. This mechanism can transform missing information into executable gating conditions, improving interpretability and operational security in missing scenario scenarios.

[0021] In this invention, upon risk triggering, the triage sampling unit performs triage according to the triage parameters determined by the shared situation vector RZS, and the nucleic acid confirmation unit outputs the nucleic acid confirmation result and writes it into the nucleic acid confidence term. When the nucleic acid confirmation is negative, the sample is only written into the non-pathological drift database and written back to update the threshold database and consistency criterion, provided that the phenotypic event term is negative. This linkage constraint makes the negative write-back more reliable and reduces the risk of threshold updates being disturbed by atypical samples.

[0022] This invention sets the control permission domain as a joint determination of nucleic acid confidence terms, reachability terms, and missing fingerprint MDF, and adopts a sequential control path of first oxygen-regulating pulse root perfusion and then microcapsule injection. After oxygen regulation, the regression rates of ORP deviation terms and dissolved oxygen deficit terms are written back to the shared situation vector RZS as the gating condition for entering microcapsule control, so that the control action is driven by real-time situation feedback rather than fixed time, which helps to improve the verifiability and controllability of the control process.

[0023] The microcapsule injection unit of this invention performs microcapsule injection when the root zone enters a controllable window and acidification anomalies and enzyme proxy terms are observable, and dynamically corrects injection parameters based on reachable terms and missing mechanism identifiers. This design can couple and match the microenvironment state with the control intensity, avoiding high-risk injection actions in unverifiable states and enhancing the engineering adaptability of root dredging control strategies in complex field conditions.

[0024] This invention constructs a write-back object system comprising threshold library objects, diversion and confirmation triggering strategy table objects, oxygenation parameter objects, microcapsule dosage mapping objects, bed section risk map objects, phase blacklist library objects, non-disease drift library objects, and waterway maintenance objects. This system continuously writes back and updates itself after shutdown, re-inspection, downgrade, and maintenance are completed. This closed-loop mechanism allows operational experience to be distilled into reusable strategies, ensuring that the system maintains iterable strategies, traceable maintenance, and recoverable state during long-term deployment. Attached Figure Description

[0025] Figure 1 The process of this invention Figure 1 ; Figure 2 The process of this invention Figure 2 ; Figure 3 The process of this invention Figure 3 . Detailed Implementation

[0026] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1 like Figure 1 As shown, in this embodiment 1, an electrochemical monitoring unit, a segmented valve control unit, a shunt sampling unit, a nucleic acid confirmation unit, an oxygen regulation unit, and a microcapsule control unit are deployed in the root zone of the Polygonatum planting bed. A shared situation vector RZS is maintained in the data fusion center. The bed length is set to 8 m, and each bed is divided into 4 root zones. Each root zone maintains a set of RZS and independently performs identification, confirmation, and control. The units do not directly exchange raw observation data and control quantities, but only cooperate by reading and writing RZS. The RZS is written with 12 state quantities in a fixed order, namely, ORP deviation, impedance distance, conductivity disturbance, water content deviation, temperature stress, dissolved oxygen deficit, acidification anomaly, enzyme proxy, nucleic acid confidence, reachability, missing fingerprint MDF, and phenotypic event.

[0028] After root zone ORP, rhizosphere impedance spectrum, conductivity, water content, temperature, leachate acidification signal, extracellular enzyme activity proxy signal, and water flow and pressure difference signal enter the fusion center, they are first aligned in time and unified in dimensions. The unified sampling period is 5 min, the impedance spectrum sweep frequency period is 30 min, and the fusion update period is 10 min. The sampling period and fusion update period are derived from the continuous observability constraint of the healthy period fingerprint database. After statistical analysis based on a 15-day continuous disease-free window, 5 min can cover more than 95% of the effective dynamic changes, 10 min can meet the valve control execution delay constraint, and 15 days is derived from the shortest available database construction length during the stable period of the same plot and season. To avoid ambiguity in the time reference, all subsequent data with subscripts are... , The window and gating formulas are based on a 10-minute fusion cycle; the 5-minute original channels are first aggregated into aligned observations within each fusion cycle, and the 30-minute impedance spectrum is maintained using the most recent effective frequency sweep result within adjacent fusion cycles. Normalization is based on... ; Execution, among which, For the first Signal type at time The normalized value, For the first Signal type at time The original observations, The first in the healthy period fingerprint database Mean of the signal The first in the healthy period fingerprint database Standard deviation of signal type.

[0029] The impedance distance term uses Mahalanobis distance and five fixed frequency points: 100 Hz, 1 kHz, 10 kHz, 50 kHz, and 100 kHz. These frequency points are derived from the first five frequency bands used in the pre-test sensitivity scan to distinguish between disease disturbances and irrigation disturbances. The initial value of the five-dimensional impedance eigenvector is taken from the pre-health period. The mean of the sampling points, i.e. , This is derived from the fact that a complete irrigation phase can be covered within a 3-hour window. The impedance characteristic moving average vector is calculated according to... Update, in which, For a moment Impedance characteristic moving mean vector The moving average vector of impedance characteristics at the previous time step. For a moment Five-dimensional impedance eigenvectors The mean forgetting factor is set to 0.90, based on the calibration results of minimizing the mean drift of impedance during the healthy period. The initial value of the covariance matrix is ​​set according to... set up, For covariance operators, Used for numerical regularization and derived from condition number stability constraints. Impedance distance according to ; Calculate, where, For a moment Impedance distance term, For the transpose operator, This is the inverse of the covariance matrix. The covariance matrix is... Update, in which, Let be the covariance forgetting factor and set it to 0.95. Let be the covariance matrix of the previous time step. It is an identity matrix. and All values ​​were derived from the numerical stability calibration during the healthy library construction phase, with the goal of maintaining the condition number less than 1. .

[0030] RZS time updates and measurement updates are performed in combination. Initial values ​​for the state vector are set according to... Settings, in which The healthy reference state vector is composed of the average values ​​of all state variables in the healthy period fingerprint database. The state transition matrix is ​​taken as... ,in It is a 12-dimensional identity matrix. This data is derived from the point where the mean absolute error of the one-step prediction was minimized during the health period playback validation. Time updates are based on... Execution, among which, For a moment Prior state vector, For a moment Posterior state vector. Prior covariance matrix according to dissemination, among which, For a moment Prior covariance matrix, For a moment Posterior covariance matrix, Let be the process noise covariance matrix. Initial value according to The setting is derived from the covariance of the healthy state samples; Take diagonal form ,in ; For the scalar intensity of process noise, For the first Each fusion cycle updates the residual vector of the state. This is a L2 norm operator. This update method is used to follow the intensity of operating condition disturbances and avoid underestimation of covariance.

[0031] Measurement update by Execution, among which, For a moment Posterior state vector For a moment Gain matrix, An eight-dimensional observation vector. This is the observation mapping matrix. A fixed sparse mapping is used, initialized by a one-to-one mapping of directly observable states and a rule that sets unobservable states to zero. Values ​​are determined based on a correspondence table between sensors and defined states. The observation residual is defined as... The residual samples during the healthy period are denoted as , The initial values ​​of the observation noise covariance are 8-dimensional identity matrices, and are calculated as follows: Set, and press Update, in which, For a moment Observation noise covariance matrix, The observation noise covariance matrix at the previous time step. The noise forgetting factor is set to 0.90, based on the observed noise stationarity fitting results. The base gain value is set according to... ; Calculate, and according to After completing the covariance posterior update, among which, For a moment Posterior covariance matrix. Gain is set to the default value. Conductivity disturbance term season Temperature stress items On the basis of the above, further orders The thresholds of 0.45 and 0.40 are derived from the 75th percentile of the corresponding indicator in the health database, and the scaling factors of 0.6 and 0.7 are derived from the erroneous trigger minimization calibration of the fallback of extreme irrigation disturbances.

[0032] The missing fingerprint MDF write rules are described using computable logic rather than enumeration. Let... This represents the longest consecutive period of missing data for a critical channel. The number of test pairs missing for key channels. Mark missing fields. To implement verifiable domain labeling, then in, For indicator functions, For logical AND operators. Let Indicates the first Class-critical channels at time Are there any missing tests? This indicates a missing test result. Indicates availability; the key channel set includes eight categories: ORP, impedance spectrum, water content, temperature, conductivity, acidification, enzyme proxy, and water pressure differential. The longest consecutive missing measurement duration is indicated by pressing [the command / method]. ; Calculate the total number of missing test pairs according to ; Calculate, where, min represents the fusion period. This represents a pairwise combination of critical pathways. The 40 min, 20 min, 3, and 1 values ​​are derived from the boundary statistics of confirmatory feasibility and control safety in the missing data playback experiment. Phenotypic events are indicated by necrotic plaques. Softening Instructions Mold Indicator Pull-up resistance indicator Growth potential indicator Greening indicator Write, all Furthermore, the data originates from on-site sampling records, image discrimination output, and pull-out resistance discrimination output.

[0033] The ORP deviation term, conductivity disturbance term, water content deviation term, temperature stress term, acidification anomaly term, and enzyme proxy term are all derived from the original observations according to a fixed window. Let... For a moment Measured ORP values ​​(unit: mV, sampling frequency: 5 min). The values ​​are measured conductivity values ​​(unit: mS / cm, sampling frequency: 5 min). The measured value of volumetric water content in the root zone (unit: %), with a sampling frequency of 5 min. Measured root zone temperature (unit: ...) C, sampling frequency 5 min), The leachate acidification signal (unit pH equivalent deviation, sampling frequency 10 min) is used. If the signal is an extracellular enzyme surrogate signal (equivalent value in U / L, sampling frequency 10 min), then ； ； ； ; ； ; in, and These are the mean and standard deviation of ORP during the healthy period, respectively. and These are the mean and standard deviation of root zone temperature during the healthy period. For the median operator, For the absolute median difference operator, , and These are the 0.9 quantile, 0.5 quantile, and 0.1 quantile, respectively. To prevent division by a constant and take , This is the interval truncation operator. That is, the water content deviation term. This refers to the temperature stress term. Under the aforementioned 10-minute fusion index, the 12-point window corresponds to 2 hours, and the 18-point window corresponds to 3 hours, derived from the shortest coverage duration of a round of irrigation and root zone chemical response.

[0034] Early identification gating employs a quantization consistency criterion. The difference between the ORP deviation term and the impedance distance term is defined as follows: and ,in This is a fusion period index. Calculated over the most recent 12 periods. ; in, The difference correlation coefficient, For the same sign ratio, For the correlation coefficient operator, For the summation operator, For sign functions. The overall consistency score is defined as follows: in, For consistency scoring. Phenotypic positive score is defined as follows: in, Phenotypic positive score. Risk trigger marker. according to Calculate, where, This represents the conductivity perturbation term. The thresholds of 0.50, 0.75, and 0.70 are derived from the optimal point of the Youden index obtained through combined ROC analysis of the database samples and field replays.

[0035] when At that time, the split sampling unit performs splitting and proceeds to nucleic acid confirmation. The split volume is set to 50 mL, the split time to 35 s, and the delivery delay to the test is 20 s, derived from the required test reagent volume and valve-controlled stabilization time calibration. Nucleic acid confidence items are set according to... ; Calculate, where, For nucleic acid confidence terms, It is an exponential function. Let be the slope parameter and set to 0.85. The threshold number of cycles for the nucleic acid target channel. The reference threshold number of cycles is set to 32. and The classification boundary is derived from the fitted positive / negative control samples. When both nucleic acid and phenotype are negative, a non-pathological drift library write-back is performed. When the nucleic acid is invalid, a missing mechanism marker is written and a retest is triggered. The retest interval is 30 minutes, and the maximum number of retests is 3. The data is derived from the nucleic acid degradation time statistics of samples from the same batch.

[0036] The control permission domain is jointly quantified by nucleic acid confidence terms, reachability terms, and MDF. Reachability terms are quantified by... Calculate, where, For reachable items, For stress consistency scoring, Scoring for traffic consistency The success rate of valve control execution is scored. The pressure was directly measured by a root zone pressure sensor, with a sampling frequency of 1 Hz and the median value taken within a 10-minute window. ; Calculate, where, The median pressure value over a 10-minute window. In order to manage target pressure, The upper limit of the allowable deviation is set at 0.20 bar, and the value is based on the steady-state pressure control accuracy of the root irrigation pump. Measured directly by the flow meter, with a sampling frequency of 1 Hz and the average value taken within a 10-minute window, according to... ; Calculate, where, This is the average measured flow rate. For target traffic, The upper limit of the allowable deviation is set at 0.15 L / min, derived from the pipeline calibration report. Statistics are directly compiled from the valve execution log, based on the most recent data. Calculation of control command success rate: ; in, This represents the number of times feedback was received out of the most recent 20 valve control commands. The weights of 0.45, 0.35, and 0.20 are derived from regression analysis of the contribution of historical control failure cases. (Permitted trigger variable) according to Calculate, where, The control permit thresholds of 0.70 and 0.65 are derived from the combined minimum safety boundary of nucleic acid confirmation reliability and pipeline accessibility.

[0037] Oxygen-controlled pulsed root irrigation under controlled permit markings Even if the conditions are met, a waterway executability check is still required. Let... For the original flow rate observations, These are the original observed values ​​of the pressure difference. and These are the flow rate and differential pressure quality indicator bits, respectively. The water circuit missing measurement marker is then set according to... Calculation. Oxygen regulation prohibition marker pressed. Calculate, where, This indicates that the oxygen adjustment pulse is prohibited and the reason for the prohibition will be written to the missing mechanism flag to trigger a maintenance writeback. Only when and Oxygen-regulating pulses are executed at specific times. The target window, pulse duration, and pulse interval for oxygen-regulating pulse root irrigation are all determined by the ORP deviation term. Dissolved oxygen deficit Water content deviation term With reachable items Joint calculation. The median oxygenation target is calculated according to... Calculate, where, For a moment Oxygen target median (unit: mg / L). Target window upper and lower bounds are set as follows: Calculate, where, and These represent the lower and upper bounds of the oxygenation target window, respectively. The single pulse duration and pulse interval are set as follows: Calculate, where, The duration of a single pulse (in minutes). The pulse interval (in minutes) represents the number of pulses in this embodiment. The value is fixed at 3, derived from the fitting of the dissolved oxygen recovery time constant in the root zone and the constraints of stable pump operation. The regression gating score is calculated after oxygen adjustment. in, The regression gating score is calculated after oxygen adjustment. The regression rate of the ORP deviation term. The regression rate is for the dissolved oxygen deficit term. The regression rate is calculated over a 6-fusion period window: in, min represents the fusion period. The dissolved oxygen deficit term is defined as follows: , The values ​​are actual dissolved oxygen probe readings (unit: mg / L, sampling frequency: 5 min). This represents the median oxygenation target for the current fusion cycle. The denominators 0.010 and 0.008 are the regression reference rates for ORP and dissolved oxygen targets, respectively, derived from the median regression rate during the healthy recovery phase. Microcapsule-controlled trigger variables. according to Calculate, where, This is an acidification anomaly. For enzyme proxy terms, the thresholds of 1.0, 0.60, and 0.55 are derived from the release conditions defined in the lesion microenvironment. To achieve dynamic correction based on reachability terms and missing mechanism identifiers, missing mechanism identifiers are defined. ,in Indicates no risk of missing items. Indicates the risk of partial missing measurements. This indicates the risk of shared link failure. The classification rules are as follows: ; Confirmed. Correction factors are as follows: Calculate, where, Correction coefficients for missing mechanisms. Injection concentration. Total Injection Dose With injection rate Determined jointly by impedance distance term, acidification anomaly term, enzyme proxy term, and reachability term, and by Dynamic correction, using ; in, The unit is g / L. The unit is L / min. The unit is g. The duration of a single injection is set to 8 minutes. Indicates will Cut off to interval The values ​​of 0.8 to 1.6 and 0.6 to 1.2 are derived from the reagent safety window and the pump's rated operating condition calibration. The results are derived from the calibration of the uniformity of bed irrigation.

[0038] Stop variable according to Calculate, where, To indicate a stop, This is an ORP deviation term. For impedance distance, the symbol is... This indicates a decrease over two consecutive update cycles. The thresholds of 0.20, 1.20, and 0.30 are derived from the 90% coverage boundary of the post-illness recovery sample distribution. Remove the risk marker at the appropriate time. and Repeat the oxygenation-microcapsule treatment sequence and apply control measures repeatedly. It will then enter a downgrade mode and prohibit oxygen regulation and microcapsule control.

[0039] In this embodiment, the synergistic drift of the ORP deviation term and the impedance distance term simultaneously constrains electrochemical disturbances and changes in tissue conductivity paths, reducing false alarms caused by irrigation disturbances to single indicators. Therefore, the false alarm rate decreases under the same detection rate conditions. The sequence of adjusting oxygen first and then microcapsules improves the effective concentration gradient of the agent reaching the lesion microenvironment by enhancing root zone oxygen supply and mass transfer efficiency, thus significantly increasing the drug concentration ratio between the lesion area and the healthy area. The joint gating of MDF and reachability terms simultaneously writes detectability and executableness into the control trigger condition, blocking high-risk control actions in scenarios of missing detection and pipeline anomalies, thus reducing the false control rate. The quantitative indicators corresponding to the above causal links are the false alarm rate, the lesion area / healthy area concentration ratio, and the false control rate, respectively.

[0040] In this embodiment, field measurements were conducted on 160 bed segments in the same plot of land. These included 40 segments in a closed-loop control group, 40 segments in a single ORP threshold group, 40 segments in a fixed-dose root irrigation group after nucleic acid confirmation, and 40 segments in a manual inspection group, with a monitoring period of 21 days. Percentage indicators were statistically analyzed based on bed segment-fusion cycle event-level samples (10-minute cycle, with approximately [number missing] effective event samples per group). Regarding early identification effectiveness, the closed-loop protocol group achieved an average early identification time of 2.6 days compared to manual inspection, corresponding to a detection rate of 92.5% and a false alarm rate of 7.8%; the single ORP threshold group achieved a detection rate of 76.3% and a false alarm rate of 18.6%; and the fixed-dose root drenching group achieved a detection rate of 80.1% and a false alarm rate of 14.9%. Regarding confirmatory reliability, the nucleic acid confirmatory sensitivity was 94.1%, the specificity was 96.2%, and the invalidity rate was 3.5%. Regarding control results, the root rot incidence rate at the end of 21 days was 11.4% in the closed-loop protocol group, 25.7% in the single ORP threshold group, 19.8% in the fixed-dose root drenching group, and 31.6% in the manual inspection group. Regarding targeted release effectiveness, the drug concentration ratio between the lesion area and the healthy area was 3.4 in the closed-loop protocol group and 1.7 in the fixed-dose root drenching group. Regarding the safety performance under the simulated missing test conditions, the false control rate of the closed-loop scheme group was 4.2%, and the false detection rate was 8.1%, while the false control rate of the single ORP threshold group was 12.9%, and the false detection rate was 15.4%. All of the above indicators were statistically analyzed according to the same sampling rule and written back to the threshold library objects, diversion and confirmation trigger strategy table objects, oxygen adjustment parameter objects, and microcapsule dose mapping objects for the next round of parameter updates.

[0041] Example 2 This embodiment, while maintaining the same system architecture and shared situation vector RZS writing order as in Embodiment 1, enhances the robust gating, missing driver verification, control prohibition, and degradation recovery mechanisms driven by the missing fingerprint MDF. This ensures that the identification and control link remains verifiable even when critical observation channels experience continuous or shared missing measurements. The electrochemical monitoring unit, segmented valve control unit, shunt sampling unit, nucleic acid verification unit, oxygen regulation unit, and microcapsule control unit still coordinate solely through reading and writing RZS, without adding direct data connection paths between modules. To ensure reproducibility, this embodiment quantifies and writes all four types of quantities—missing location, missing duration, missing phase, and missing correlation—into the MDF, which then directly determines the missing domain, verifiable control domain, and degradation mode triggering.

[0042] Let the first Key observation channels at time The original observations are Key channel set The data includes eight categories: root zone ORP, rhizosphere impedance spectrum, root zone water content, root zone temperature, conductivity, leachate acidification signal, extracellular enzyme activity proxy signal, and water flow rate and pressure difference signal. The ORP channel is measured in mV with a sampling frequency of 5 min; the impedance spectrum channel has a frequency range of 100 Hz to 100 kHz with a sweep period of 30 min; the water content channel is measured in % with a sampling frequency of 5 min; and the temperature channel is measured in mV. The sampling frequency for each channel is 5 min, with the conductivity channel measured in mS / cm and sampled at 5 min. The acidification signal channel is measured in pH equivalent deviation and sampled at 10 min. The enzyme surrogate signal channel is measured in U / L equivalent value and sampled at 10 min. The water pressure differential channel is measured in kPa and sampled at 1 Hz, and aggregation occurs in each fusion cycle. Missing measurement locations for each channel are marked according to... Calculate, where, Indicates the first Channel-like time Are there any missing tests? This indicates that there are no valid observations. This indicates the quality flag for that channel. This indicates a quality control failure. Real-time observations from the corresponding sensor or detection link; The results are derived from the data acquisition driver return code and range verification results, without relying on manual annotation.

[0043] The duration of missing tests is determined by the longest consecutive duration of missing tests. Cumulative untested time Duty cycle with missing measurement Common characteristics. Based on the fusion cycle. For time steps, the longest consecutive missing test duration is set as follows: ; Calculate the cumulative missed test time according to... ; Calculate the missing duty cycle according to ; Calculate, where, This indicates a 4h window length. Indicates the number of critical channels. Based on statistics of the shortest closed-loop coverage time of irrigation-infiltration-root zone response, a complete operation phase of the same bed section can be covered within 4 hours.

[0044] The missing phases consist of intraday phases, irrigation phases, agricultural phases, and fertility phase codes. Let... For intraday phase encoding, Encoding the irrigation phase, Encoding agricultural phases, For fertility phase encoding, the phase vector is defined as follows: The four phase types are directly derived from clock data, irrigation control logs, agricultural task logs, and fertility period records. Irrigation phases are automatically generated based on valve opening / closing and flow step events; agricultural phases are automatically mapped based on timestamps from manual operation records; and fertility phases are automatically mapped based on planting dates and agronomic staging tables. To meet the requirements for recording the main peak and repeatability of missing phases, a third phase is defined. The discrete state set of phase-like structures is The statistics window is For each fusion cycle, the count of missing events for this type of phase is calculated as follows: ; Calculation, phase main peak according to ; Confirm, repeatability ; Calculate, where, For the first The main peak state of phase-like phase. For the first Phase repeatability index, To prevent division by a constant. The corresponding 12-hour window originates from the need to identify clusters of missing data across day and night. and All were written to the missing fingerprint MDF.

[0045] Missing test correlation is based on the number of co-missing test pairs With common missing measurement delay intensity Calculation. Let the set of channel pairs be... Delayed set Corresponding to 0, 10, and 20 min, then ; ; in, This indicates a 2-hour correlation window. This is the correlation coefficient operator. This data is derived from field statistics showing that the most common fault structure is most likely to form in the untested propagation link within two hours.

[0046] The MDF domain partitioning is directly determined by the quantities mentioned above. Let the missing domain boundary parameters be denoted as... and The boundary parameters of the verifiable domain under control are denoted as and Its initialization is , , and Missing field marker Verifiable domain markers for control With the downgrade trigger flag According to The calculations, where the initial threshold values ​​of 40 min, 20 min, 80 min, 3, and 5 are derived from the risk boundary statistics in the missed detection replay experiment, aiming to simultaneously constrain missed detections and erroneous control measures. This embodiment... Statistical rules including delayed co-missed tests are adopted. With the downgrade threshold All were recalibrated according to this rule, and the no-delay definition was not directly adopted. , , and As a threshold library object, boundary parameters can be written back for updating. In this embodiment, if there is no write-back during runtime, the initial values ​​are retained. The MDF state vector is defined as follows: And write the missing fingerprint MDF item to RZS, any gating decision is based on This is a prerequisite input. Nucleic acid confidence terms. Acidification Abnormalities Enzyme proxy items Phenotypic positive score and reachable items In this embodiment, it is explicitly redefined as a computable quantity. Nucleic acid confidence terms are... ; Calculate, where, The threshold cycle number for the nucleic acid target channel is derived from the nucleic acid confirmatory unit test results. and Derived from the fitting boundary of positive / negative control samples, the nucleic acid positive threshold was taken Acidification anomalies and enzyme proxy items are categorized as follows: ； ; Calculate, where, and The original observations of acidification and enzyme surrogate signals are shown, with 18 points corresponding to a 3-hour time span. To prevent division by a constant, the phenotypic positive score is calculated as follows: Calculate, where, , , These are necrotic spot indicators, softening indicators, and pull-out resistance indicators, derived from root zone sampling records, image discrimination outputs, and mechanical force measurement records. Reachable items are categorized by... Calculate, where, Pressure consistency score, derived from the median value of a 1 Hz pressure sensor within a 10-minute window; The flow consistency score is derived from the average of the 1 Hz flow meter over a 10-minute window. The valve control execution success rate score is derived from the most recent 20 valve control command logs. The weights of 0.45, 0.35, and 0.20 are derived from the regression analysis of the contribution of control failure.

[0047] In this embodiment, the consistency criterion maintains the combined drift gating and phenotypic gating structure of Embodiment 1, but adds an observation availability mask to handle cases where ORP or impedance updates are limited by MDF. Let... This is a marker for missing ORP channel data. For impedance path missing measurement marking, further define , .like Then execute the standard combined drift gating, if Then the missing driver confirmation trigger is executed, triggering the variable. according to Calculate, where, and These are acidification anomalies and enzyme proxy terms, respectively. This represents the phenotypic positivity score. Thresholds of 0.58, 0.52, and 0.60 are derived from the F1 optimal value of the replay set of nucleic acid-positive samples under test-deficient conditions. The trigger path vector will be activated upon triggering. Write the traffic splitting and confirmation triggering strategy table object as the basis for subsequent retesting strategy updates.

[0048] When nucleic acid testing is invalid, the timing and number of retests are no longer fixed, but are determined jointly by the missing phase, the duration of the missing test, the correlation of the missing test, and reachability factors. Let the retest interval be... and retest limit According to Calculate, where, The set of highly disturbed phases for irrigation is generated according to the rule that the proportion of invalid confirmations for that phase in the logs of the most recent 30 days is not less than 0.25. For reachable terms, the constants 20 and 10, and the threshold 0.70, are derived from the recovery delay statistics and sampling reachability statistics of nucleic acid link failure samples, while 0.25 is derived from the retest resource occupancy inflection point analysis. The retest strategy is written into the missing mechanism identifier and synchronously written back to the phase blacklist database object to avoid repeatedly triggering invalid submissions in high-interference phases.

[0049] Security is jointly gated by the MDF and reachability terms. Let... Mark the flow channel as missing, according to Calculate; set For missing differential pressure signal markers, follow these steps: Calculation, where and These are the original observation values ​​for flow rate and differential pressure, respectively. and These are the corresponding quality identification positions. Facility control prohibition markings. according to Calculate. If Then, oxygen regulation and microcapsule control should be prohibited and written into the water system maintenance object; if and Then, the oxygen-adjusted microcapsule sequential control is maintained as in Example 1. The threshold of 0.65 is derived from the statistics of the inflection point of control failure probability in the low-value segment of the reachable terms.

[0050] Both entering and exiting the downgrade mode are based on calculable conditions, not manual judgment. The entry condition is... Once logged in, the system will only retain monitoring, diversion, confirmation, and maintenance write-back functions; oxygen adjustment and microcapsule control will be disabled. The deactivation conditions are as follows: ; Calculate, where, The downgrade recovery markers are derived from the statistics of the shortest stable time when the false trigger rate falls back to near the baseline after recovery, representing six consecutive fusion cycles. The degradation is lifted and baseline relearning is triggered. The relearning window takes the most recent 72 fusion cycles, which are derived from the parameter restabilization time statistics after missing test recovery.

[0051] Missing tests are not merely used as data cleaning markers, but rather enter the RZS through the MDF and are pre-gated to the entire identification-confirmation-implementation chain gating, mechanistically binding observability and execution capability to the same state space. When ORP or impedance is unavailable, the system uses the synergistic anomaly of acidification anomalies and enzyme proxy terms as alternative evidence, along with phenotypic gating, thereby maintaining the continuity of the confirmation chain. Under conditions of high co-missing tests or long-term missing tests, the system switches to a degradation mode and prohibits high-risk implementation actions, directly reducing the risk of misimplementation. The corresponding quantifiable performance indicators are confirmation-triggered recall rate, invalid retest closure completion rate, and misimplementation rate.

[0052] This embodiment uses data from 160 bed sections in the same plot and period as in Embodiment 1, and overlays missing test playback scenarios for field verification. All four sets of settings are kept consistent. Within each set, three levels of random missing test data (10%, 20%, and 30%) and two levels of shared missing test scenarios are added. Percentage indicators are statistically analyzed based on event-level samples of bed section-operating condition playback segments (each set of valid event samples is no less than...). Using the ability to maintain confirmation under missing test conditions as an indicator, the recall rates for missing-driven confirmation triggers were 91.3%, 88.7%, and 84.5%, respectively, with corresponding false trigger rates of 7.1%, 8.6%, and 10.9%. Using the ability to retest and close the loop after invalidation as an indicator, the proportion of valid confirmation within three retests was 93.4%, with an average retest completion time of 47 minutes. Using control safety as an indicator, the false control rate under missing test conditions was 4.6%, a decrease of 8.2 percentage points compared to the control process without MDF gating (12.8%). Using degradation recovery stability as an indicator, the number of unauthorized control triggers after degradation was 0, and the threshold write-back drift amplitude was controlled within 0.62 times the standard deviation of the healthy baseline within 72 cycles after recovery. The above results were recorded and written back to the threshold library objects, diversion and confirmation trigger strategy table objects, phase blacklist library objects, and waterway maintenance objects according to unified statistical rules.

[0053] Example 3 This embodiment, building upon the closed-loop structures of Embodiments 1 and 2, enhances the contribution of phenotypic event items to the specificity of early identification. It integrates threshold library objects, diversion and confirmation triggering strategy table objects, oxygenation parameter objects, microcapsule dose mapping objects, bed section risk map objects, phase blacklist library objects, non-disease drift library objects, and waterway maintenance objects into a unified write-back chain, enabling the system to maintain interpretable self-calibration capabilities under complex operating conditions. The 12-item structure of the shared situation vector RZS remains unchanged, and the phenotypic event items still consist of necrotic spot indicators, softening indicators, mold layer indicators, pull-out resistance indicators, growth potential indicators, and chlorosis indicators. However, this embodiment makes the generation algorithms, time alignment, and write-back constraints of all six indicators explicit to ensure that phenotypic evidence is not replaced by electrochemical drift alone.

[0054] Root and stem surface images were acquired from a fixed position at a sampling frequency of 30 minutes, with the probability of necrotic spots output for each sampling. Softening probability Mold layer probability and the probability of greening Its value range is The pulling resistance was collected by a mechanical force gauge and recorded as... The unit is N, and the sampling frequency is 4 h; the growth potential is denoted as The values ​​were obtained by fusing daily plant height increments and leaf area increments, with a sampling frequency of 24 h. The six phenotypic sub-items were... Calculate, where, For the first Each phenotypic sub-item at time The indicated value, Image probability threshold, To raise the resistance threshold, This represents the growth potential threshold. Growth potential is determined by... Calculate, where, For a moment Measured plant height (unit: cm, sampling frequency: 24 h). For a moment Measured leaf area (unit: cm²) (Sampling frequency 24 h) To prevent division by zero, the six initial threshold values ​​are respectively set to... The optimal Youden index was derived from a healthy-disease control sample.

[0055] The writing of phenotypic event items in RZS does not use single-point decision, but rather employs time-window robust fusion. Let... Indicating the most recent 24 fusion cycle windows (4 hours), the phenotypic positivity score is calculated as follows: Calculate, where, For the first The window mean of each sub-item. Weight vector. The initial values ​​are derived from the normalized logistic regression coefficients of historical confirmed samples and re-estimated once per monthly playback data. To avoid duplicate counting of the 4-hour and 24-hour slow channels within the short window, Only when new or new The sample fusion cycle is refreshed; the remaining cycles reuse the previous cycle. And it does not trigger a new type write. The write condition for the type event item is: This threshold is derived from the inflection point under the constraints of recall priority and controlled false alarm rate. To ensure variable closure within this embodiment, the nucleic acid confidence term, impedance distance term, missing domain marker, acidification anomaly term, enzyme proxy term, reachability term, consistency score, and flow / pressure difference missing marker are redefined in this section as follows. The nucleic acid confidence term is... Calculate, where, This represents the number of cycles required to reach the nucleic acid testing threshold. , Derived from the positive / negative control fitting boundary. Impedance distance term. Using the Mahalanobis distance definition and five-frequency point from Example 1 The covariance is updated recursively and is no longer rewritten in truncated normalized form. To further describe impedance amplitude variations, an auxiliary feature is defined. in, For a moment At frequency The impedance amplitude, For policy write-back diagnostics only, not a substitute. Gating determination. Acidification anomalies and enzyme proxy items are determined according to... ； ; Calculate, where, and Data are derived from raw observations of acidification and enzyme proxy channels, with 18 points corresponding to 3 hours. Reachable items are listed below. Calculate, where, Scoring from the 10-minute window median value of a 1 Hz pressure sensor. Scoring based on the 10-minute window mean from a 1 Hz flow meter. Success rate from the most recent 20 valve control execution logs. Missing test flags used for missing domain determination. Calculate, where, For key channel indexes, , This is the original observation of the channel. This is a quality control identifier. The longest consecutive missing test duration and the total number of missing test pairs are listed in order. ; Calculate, where, The unit is min. For the set of critical channel pairs, the 12-point window corresponds to 2 hours. Considering that this embodiment will... The statistical rules are limited to 6 types of key channel pairs, and missing field marking is based on... The thresholds of 40 min and 2 were calculated based on risk boundary statistics from replay experiments with missing tests under the same rules. The consistency score was calculated according to... ; Calculate, where, For the first difference of ORP, The ORP observations are aligned according to a 10-minute fusion period (original sampling frequency 5 min). For the first-order difference of the impedance distance term, and These are the 75th percentile thresholds of the healthy fingerprint database. The weights of 0.6 and 0.4 in the consistency score are derived from regression analysis of the contribution of historical confirmed samples to correlation and common threshold exceedance. Missing flow and differential pressure measurements are marked according to... Calculate, where, and These are the raw observations of flow rate and differential pressure, respectively. and These are the corresponding quality control identifiers.

[0056] This embodiment sets strong constraints on writing nucleic acid-negative samples into the non-disease drift library to avoid mistakenly learning disease precursor drift as healthy drift. Let the nucleic acid confidence term be... The negative threshold is Negative write library gate variables according to Calculate, where, For impedance distance, Mark missing fields. Only when Only when this is the case is it allowed to send sample vectors Write to non-disease-related drift library objects. Threshold The impedance distance from the 95th percentile boundary was derived from non-disease samples. It originates from the upper limit of the negative control distribution.

[0057] In this embodiment, microcapsule injection, in addition to using the control permission conditions of Example 1, adds a phenotypic-chemical dual consistency constraint. Let the acidification anomaly be... The enzyme proxy item is Phenotypic positive score is The microcapsule unlock mark according to Calculation, only Microcapsule injection is only permitted under certain conditions. Threshold The minimum feasible boundary for the concentration ratio of lesion area / healthy area to be greater than 2.5 in the lesion microenvironment release test.

[0058] The microcapsule dose mapping object is updated online using a parameter vector. Let the dose prediction parameter vector be... The input feature vector is ,in The phenotypic positivity score was used as a correction factor for the phenotypic event item; the missing mechanism was identified as... ,in Indicates no risk of missing items. Indicates the risk of partial missing measurements. This indicates the risk of shared link failure. The classification rules are as follows: ; Determined. The missing mechanism correction coefficient is calculated according to... Calculate, where, Derived from the missing fingerprint MDF, the missing test level and the missing waterway test marker, For missing mechanism correction coefficients. Phenotypic correction coefficients are calculated as follows: Calculate, where, This is a phenotypic correction factor. The predicted injection concentration is calculated according to... Calculate, where, The concentration is the injection concentration, expressed in g / L. The injection rate and duration are respectively calculated as follows: Calculate, where, The unit is L / min. The unit is per minute. The total injection dose is calculated as follows: calculate, The unit is g. The upper and lower limits of rate, 0.6 and 1.2, are derived from the stable operating range of the pump body, and the upper and lower limits of duration, 6 and 12, are derived from the calibration of bed irrigation uniformity and permeation safety window. The initial values ​​were obtained by linear regression fitting between healthy controls and lesion samples, denoted as . Parameter updates are pressed. Execution, among which, This represents the effective injection concentration measured for the current batch. The learning rate is set to 0.015, derived from the convergence and stability interval during offline playback.

[0059] Threshold library objects are written back exponentially. Let the early screening threshold, disturbance exclusion threshold, and control domain boundary be respectively... And let the boundary parameter of the missing domain be . The boundary parameters of the control domain are Their initial values ​​are respectively set as , , and The first three types of threshold updates are based on... Execution, among which, The candidate thresholds are given for the current validation data. For write-back gain, take respectively This stems from sensitivity analysis under stability-priority constraints. The boundaries of the missing domain and the governing domain are determined according to... Update, in which, To verify the candidate boundaries obtained by reverse inference from the set, , The runtime judgment is based on... and The triage and confirmation triggering strategy table object is updated according to the three-dimensional key-value pair of trigger type-phase-missing test level, and the key vector is denoted as... ,in For trigger type encoding, it is defined as follows: regular combination drift trigger takes 1, missing driver confirmation trigger takes 2, and re-examination trigger takes 3. The irrigation phase encoding is derived from the mapping between valve control opening / closing logs and flow step events; The missing test level code is defined as follows: and Take 0, and Take 1, Take 2. The updated value includes the diversion volume, delivery delay, retest interval, and retest limit, all of which are derived from the current success rate to determine the minimum risk combination.

[0060] The phase blacklist database uses event frequency accumulation for objects. Let... For phase At any moment The marker for the occurrence of an invalid confirmation risk event is as follows: Calculate the blacklist count. according to Update, in which, Let be the forgetting factor and set it to 0.92. If Then the phase The phase blacklist object is written and avoidance is performed in subsequent sampling scheduling. Threshold 3 is derived from the alarm boundary of the clustering test of consecutive failure events in the same phase. The waterway maintenance object is triggered by the missing mechanism identifier, and the maintenance priority is scored. according to The coefficients 0.5, 0.3, and 0.2 are derived from the statistics of maintenance failure attribution ratios. A maintenance work order is generated and the control permission flag is frozen when the score is not lower than 0.6. The threshold of 0.6 is derived from the trade-off between maintenance work order hit rate and false alarm rate; maintenance is completed and meets the requirements for 6 consecutive fusion cycles. After thawing, the results for six consecutive cycles were derived from statistics on recovery stability.

[0061] Risk intensity is entered into the bed segment risk map objects according to bed segment number and time index. Its definition is ,in For consistency scoring, weights of 0.45, 0.35, and 0.20 were derived from the multi-bed segment confirmed label regression fitting. This object is used for multi-bed segment scheduling and ranking and serves as input for updating the oxygen regulation parameter object. The target dissolved oxygen median is used when updating the oxygen regulation parameter object. according to The coefficient 0.15 is derived from the overshoot suppression constraint, 0.5 is the risk neutral point and is derived from the hierarchical scheduling ROC balance threshold, and 6.5 and 8.0 are derived from the root zone tolerable dissolved oxygen window determined by the disease course suppression test.

[0062] Phenotypic events are directly encoded into the RZS (Regression-Zeroing System), coupling the histological and electrochemical drift characteristics of disease occurrence within the same decision space. This suppresses false positives caused solely by environmental disturbances. Only nucleic acid-negative and phenotypic-negative samples are allowed to be written into the non-disease drift library, blocking the path of mislabeled samples contaminating the threshold library and ensuring stable threshold update direction. The entire write-back process is linked, unifying identification, confirmation, control, maintenance, and scheduling within a traceable parameter update framework, ensuring the system maintains low false alarm rates and recoverability even under complex phase interference. The corresponding quantitative indicators are false alarm rate, threshold drift amplitude, and degradation recovery success rate.

[0063] This embodiment completed a 24-day field test verification on 240 bed sections in the same plot. It included 60 sections in a closed-loop full-object write-back group, 60 sections in an electrochemical-gated group, 60 sections in a group without negative library write constraints, and 60 sections in a manual inspection group. Three complex scenarios were overlaid: nighttime low temperatures, sudden large-volume irrigation, and continuous rain. Percentage indicators were statistically analyzed based on bed section-fusion cycle event-level samples (10-minute cycle, approximately [number missing] effective event samples per group). The early detection rate of the closed-loop full-object write-back group was 93.8%, with a false positive rate of 6.9%; the detection rate of the electrochemical-gated group alone was 88.1%, with a false positive rate of 14.7%; and the detection rate of the group without negative write library constraints was 90.2%, with a false positive rate of 12.9%. The drug concentration ratio between lesion areas and healthy areas was 3.7 in the closed-loop full-object write-back group and 2.2 in the group without negative write library constraints. The normalized drift amplitude of threshold library objects within 24 days was 0.18, while it was 0.41 in the group without negative write library constraints. The false control rate in complex scenarios was 3.9% in the closed-loop full-object write-back group and 11.6% in the electrochemical-gated group alone. The recovery success rate within 48 hours after downgrade was 95.4%, and the probability of re-downgrade within 72 cycles after recovery was 4.8%. The above results were written back to eight types of write-back objects according to unified statistical rules and used for the next round of parameter initialization.

[0064] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be modified within the scope of the concept described herein by means of the above teachings or the technology or knowledge in related fields.

Claims

1. A method for early identification and root drenching control of root rot in Polygonatum sibiricum, characterized in that, Includes the following steps: S1 sets up a unit module in the root zone of the Polygonatum planting bed. The unit module includes an electrochemical monitoring unit, a segmented valve control unit, a shunt sampling unit, a nucleic acid confirmation unit, an oxygen regulation unit, and a microcapsule control unit. A shared situation vector RZS is established in the data fusion center. The unit module only operates collaboratively by reading and writing the shared situation vector RZS and does not directly exchange the original observation data and control quantities. S2 establishes a baseline window and drift rate upper limit based on the healthy period fingerprint database of the bed segment and initializes the shared situation vector RZS, so that the shared situation vector RZS includes ORP deviation, impedance distance, conductivity disturbance, water content deviation, temperature stress, dissolved oxygen deficiency, acidification anomaly, enzyme proxy, nucleic acid confidence, reachability, missing fingerprint MDF and phenotypic event items. The missing fingerprint MDF includes missing location, missing duration, missing phase and missing correlation and is used to limit the missing domain and control domain. The phenotypic event items include necrotic spot indication, softening indication, mold layer indication, pull-out resistance indication, growth potential indication and chlorosis indication. S3 continuously integrates and updates the shared situation vector RZS and performs early identification gating: Under the condition that the missing fingerprint MDF has not entered the missing domain and the conductivity perturbation term has not exceeded the perturbation threshold, when the cooperative drift of the ORP deviation term and the impedance distance term satisfies the consistency criterion and the necrotic spot indication, softening indication and pull-out resistance indication in the phenotypic event term satisfy the positive criterion, a bed segment risk marker is generated and the shunting sampling unit is triggered to shun the leachate according to the shunting parameters determined by RZS; Under the condition that the missing fingerprint MDF restricts the updating of the ORP deviation term or the impedance distance term, when the acidification anomaly term and the enzyme proxy term show a cooperative abnormal trend and the phenotypic event term satisfies the confirmation gating condition, a missing-driven confirmation trigger is generated and the trigger path is recorded; The S4 nucleic acid confirmation unit performs nucleic acid identification on the diverted leachate to obtain the nucleic acid confirmation result and writes it into the RZS with confidence. When the nucleic acid confirmation is negative, it is written into the non-pathological drift library under the condition that the phenotypic event item is negative and the threshold library and consistency criteria are updated. When the nucleic acid confirmation is invalid, the missing mechanism identifier is written and the retest strategy is updated. S5 When the RZS characterization of nucleic acid confidence terms reaches the positive threshold, reachability terms meet the control permission domain, and the missing fingerprint MDF is in the control verifiable domain, oxygen-regulating pulsed root irrigation is first performed, and the regression rates of ORP deviation terms and dissolved oxygen deficit terms are written back to the RZS as the gating conditions for entering microcapsule control; when the root region enters the controllable window and acidification anomaly terms and enzyme proxy terms are in an observable state, microcapsule injection is performed, and the injection parameters are determined by the RZS and corrected according to reachability terms and missing mechanism identifiers; After S6 control is applied, RZS is continuously updated and stopped and re-examined: the risk mark is removed when the baseline is returned and the phenotypic event item turns negative; when the abnormality and missing fingerprint MDF are allowed, re-examine and repeat control in the order of oxygen regulation-microcapsule; when the missing fingerprint MDF enters the degraded state, enter the degraded mode and prohibit oxygen regulation and microcapsule control, and write back the maintenance object.

2. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The healthy period fingerprint database consists of root zone ORP data, rhizosphere impedance spectrum data, conductivity data, root zone water content data, root zone temperature data, leachate acidification signal, and extracellular enzyme activity proxy signal continuously collected during the disease-free period of the same bed segment. ORP deviation terms, impedance distance terms, conductivity perturbation terms, water content deviation terms, temperature stress terms, dissolved oxygen deficit terms, acidification anomaly terms, enzyme proxy terms, nucleic acid confidence terms, and reachability terms are written into the shared situation vector RZS to form electrochemical deviation term groups and environmental perturbation term groups.

3. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The fusion update includes time updates and measurement updates. The time update decays the historical situation according to the forgetting factor to maintain trend continuity. The measurement update writes new observations into the shared situation vector RZS and adjusts the fusion weights according to the conductivity perturbation term and the temperature stress term. At the same time, it performs consistency checks on the ORP deviation term, impedance distance term and water content deviation term and writes the inconsistency window into the environmental perturbation flag for the suppression determination of the consistency criterion.

4. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The missing fingerprint MDF is written into the shared situation vector RZS and includes the missing location, missing duration, missing phase, and missing correlation, and generates a missing mechanism identifier. The missing location is set for root zone ORP data, rhizosphere impedance spectrum data, root zone water content data, root zone temperature data, conductivity data, leachate acidification signal, extracellular enzyme activity proxy signal, and water flow pressure difference signal, and distinguishes between continuous and intermittent missing. The missing duration includes the longest continuous missing duration, the cumulative missing duration, and the missing duty cycle. The missing phase includes intraday phase, irrigation phase, agricultural phase, and fertility phase, and records the phase peak and repeatability. The missing correlation includes the set of co-missing pairs, the number of co-missing times, and the co-missing delay, and is used to determine the link co-missing structure. The missing domain and the control domain are jointly determined by the missing duration and the missing correlation.

5. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The phenotypic event items are written into the shared situation vector RZS and include phenotypic indicator sub-items corresponding to water-soaked brown necrotic spots on the root or root stem surface, softening of root or root stem tissue, mold layer on the root stem surface, reduced pulling resistance, reduced growth potential, and chlorosis and yellowing of lower leaves, respectively. The phenotypic event items are generated by root area sampling records, root stem surface image discrimination output, and pulling resistance discrimination output and are written into the shared situation vector RZS with timestamps.

6. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The consistency criteria include combined drift gating and phenotypic gating. The combined drift gating is determined by the co-drift of the ORP deviation term and the impedance distance term satisfying the consistency check and the conductivity perturbation term not exceeding the perturbation threshold. The phenotypic gating is determined by the necrotic spot indication, softening indication, and pull-out resistance indication in the phenotypic event term satisfying the positive criteria. The confirmatory gating condition is determined by the missing fingerprint MDF restricting the update of the ORP deviation term or the impedance distance term and the acidification anomaly term and the enzyme proxy term showing a co-abnormal trend.

7. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The nucleic acid confirmation unit performs nucleic acid release processing on the shunt filtrate within the detection branch and performs nucleic acid recognition and signal amplification reactions to output nucleic acid confirmation results. When the nucleic acid confirmation result is negative, under the condition that the phenotypic event item is negative, the corresponding combination drift sample is written into the non-disease drift library and the threshold library and consistency criteria are updated. When the nucleic acid confirmation result is invalid, the timing and number of retests are determined based on the missing phase, missing duration, missing correlation and reachability, and the invalidity reason is written into the missing mechanism identifier to update the retest strategy.

8. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The control permission domain includes nucleic acid confidence terms reaching the positive threshold, reachability terms meeting the control permission conditions, and missing fingerprint MDF being in the control verifiable domain. The target window, pulse duration, and interval of the oxygen-regulating pulse root irrigation are calculated from the ORP deviation term, dissolved oxygen deficit term, water content deviation term, and reachability term. The regression rates of the ORP deviation term and dissolved oxygen deficit term are written back to the shared situation vector RZS as the gating condition for entering microcapsule control. When the missing fingerprint MDF sets the water flow and pressure difference signal or the reachability term does not meet the conditions, oxygen-regulating pulse root irrigation is prohibited, and the reason for prohibition is written into the missing mechanism identifier to trigger maintenance writeback.

9. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The microcapsule injection is performed after the root zone enters a controllable window, the acidification anomaly and enzyme proxy are in an observable state, and the missing fingerprint MDF has not entered a degraded state. The microcapsule root irrigation agent used by the microcapsule control unit is unlocked and released when the acidification environment and the extracellular enzyme enrichment environment are simultaneously satisfied, so as to form a local high gradient in the lesion microenvironment and inhibit non-target release in the healthy root zone. The dosage, concentration and injection rate of the microcapsule injection are determined by the impedance distance term, acidification anomaly term, enzyme proxy term and reachability term, and are dynamically corrected during the injection process based on the reachability term and missing mechanism identifier.

10. The method for early identification and root drenching control of root rot in Polygonatum sibiricum according to claim 1, characterized in that, The write-back object includes a threshold library object, a diversion and confirmation triggering strategy table object, an oxygenation parameter object, a microcapsule dose mapping object, a bed segment risk map object, a phase blacklist library object, a non-disease drift library object, and a waterway maintenance object. Among them, the threshold library object is used to update the early screening threshold, disturbance exclusion threshold, missing domain boundary, and control domain boundary; the diversion and confirmation triggering strategy table object is used to update the diversion parameters, retest interval, and retesting strategy; the oxygenation parameter object is used to update the target window, pulse parameters, and phase avoidance strategy; the microcapsule dose mapping object is used to update the situation and dose-response mapping relationship and use the missing fingerprint MDF and phenotypic event items as dose correction factors; the phase blacklist library object is used to solidify the missing phase contamination window and sampling scheduling avoidance rules; and the waterway maintenance object generates a maintenance mark based on the missing mechanism identifier and, after maintenance, uses the missing measurement recovery boundary event to write back, remove the degradation mode, and relearn the baseline window and drift rate upper limit.

Images