Method and system for predicting and controlling cadmium absorption of rice

Abstract

The application relates to the technical field of agricultural environment monitoring and heavy metal pollution prevention and control, and discloses a rice cadmium absorption prediction prevention and control method and system. The method comprises the following steps: acquiring field soil environment data and rice root growth state data, and constructing a soil environment characteristic data set containing a cadmium solubility index and a cadmium bioavailability index; based on the data set, the influence of soil water and nutrient levels on cadmium absorption rate is analyzed, and a high absorption risk interval is determined; further, the cadmium concentration change of rice roots to stems and leaves is tracked, the migration characteristics of cadmium among different organizations are analyzed, and the bottleneck point in the transport channel and the cadmium accumulation key node are determined; combined with the interaction of soil characteristics and rice physiological mechanism, a whole-cycle transfer rule report is generated, and the basis for rice cadmium absorption prevention and control is obtained. Through multi-source data fusion and dynamic simulation, the application can accurately locate the cadmium accumulation key node, and provide the basis for rice cadmium pollution risk identification and accurate prevention and control.

Description

Technical Field

[0001] This application relates to the field of agricultural environmental monitoring and heavy metal pollution control technology, and in particular to a method and system for predicting and controlling cadmium absorption in rice. Background Technology

[0002] As a vital food crop, the safe production of rice is directly related to the quality and safety of agricultural products and human health. In some regions, due to cadmium enrichment in the soil environment, variations in farmland irrigation conditions, and differences in farming management methods, rice is prone to cadmium absorption, migration, and accumulation during its growth, leading to excessive cadmium content in rice. Cadmium is a typical heavy metal pollutant, characterized by strong mobility, high concealment, and significant bioaccumulation. Therefore, research on the prediction and control of cadmium absorption in rice has become an important technical direction in the fields of agricultural environmental safety and the prevention and control of heavy metal pollution in agricultural products.

[0003] The development of existing technologies for cadmium pollution control in rice has roughly followed a development path from end-point detection to process regulation, from single-factor analysis to multi-factor synergistic analysis, and from static evaluation to dynamic prediction. Early technologies mainly focused on detecting cadmium content in soil and rice, and determining exceedances, emphasizing the identification of pollution results. Subsequently, related research gradually expanded to reduce cadmium bioavailability by adjusting soil pH, moisture conditions, nutrient supply, and applying passivating materials to achieve agronomic regulation and field control. Furthermore, with the development of agricultural sensing monitoring, plant physiological analysis, and data processing technologies, joint analysis of soil environmental characteristics, rice root absorption characteristics, inter-tissue migration patterns, and key accumulation nodes, and based on this analysis, early identification and precise intervention of cadmium absorption risks in rice, has become an important development trend in this field.

[0004] However, existing technologies still have significant shortcomings. First, current methods primarily rely on total soil cadmium content or final cadmium content in rice as the main criteria for judgment, focusing on result detection and lacking continuous analysis of cadmium absorption and translocation behavior throughout the entire rice growth process, making it difficult to identify high-risk stages and areas in a timely manner. Second, existing technologies do not adequately utilize factors such as soil moisture, nutrient distribution, cadmium bioavailability, and rice root growth status. They typically rely on a single environmental factor or a single detection indicator, failing to accurately reflect the dynamic changes during cadmium absorption. Third, existing technologies lack effective identification of bottlenecks, hindering factors, and key accumulation nodes in the migration of cadmium between different rice tissues, making it difficult to establish a complete technical chain from soil environmental changes to plant absorption and translocation to key accumulation. Fourth, existing control measures are largely based on experience and lack universality, lacking targeted control measures based on monitoring data, migration patterns, and formation conditions, resulting in weak accuracy and adaptability.

[0005] Therefore, how to construct a rice cadmium absorption prediction and control method that can combine field soil environmental characteristics, rice root growth status and cadmium migration and accumulation patterns in rice to predict the risk of cadmium absorption in rice, identify high-risk absorption ranges, locate key accumulation nodes, and output targeted control basis has become a key technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the shortcomings of existing technologies, such as the lack of dynamic analysis of the entire cadmium absorption process in rice, inaccurate identification of high-risk absorption zones, insufficient location of bottlenecks and key nodes in cadmium migration and accumulation between different tissues, and weak targeting of control criteria, this application provides a method and system for predicting and controlling cadmium absorption in rice. This method aims to predict and identify the risk of cadmium absorption in rice, dynamically analyze the cadmium migration and accumulation process, and effectively extract control criteria.

[0007] In a first aspect, this application provides a method for predicting and controlling cadmium absorption in rice, the method comprising: S1. Obtain field soil environmental data and combine it with rice root growth status data to construct a soil environmental characteristic dataset that includes cadmium solubility and cadmium bioavailability indicators. The soil environmental data includes soil moisture and nutrient distribution. S2. Based on the soil environmental characteristics dataset, analyze the impact of soil moisture and nutrient levels on cadmium uptake rate and determine the high uptake risk zone. S3. When there is a high absorption risk zone, track the changes in cadmium concentration from rice roots to stems and leaves to obtain transfer dynamic sequence data; S4. Based on the transfer dynamic sequence data, analyze the migration speed and interaction characteristics of cadmium between different tissues to identify bottlenecks in the transport channels; S5. Based on the bottleneck point, construct a transfer path simulation map by integrating sample data of rice physiological mechanisms, and obtain a distribution map of hindering factors; S6. Based on the distribution map of hindering factors, analyze the interaction between soil properties and rice physiological mechanisms to determine the location and formation conditions of key nodes in cadmium accumulation. S7. Based on the location and formation conditions of key nodes in cadmium accumulation, generate a full-cycle transfer pattern report to obtain a basis for control.

[0008] Secondly, this application provides a rice cadmium absorption prediction and control system, the system comprising: The data acquisition module is used to acquire field soil environmental data and, in combination with rice root growth status data, construct a soil environmental characteristic dataset that includes cadmium solubility and cadmium bioavailability indicators. The soil environmental data includes soil moisture and nutrient distribution. The risk analysis module is used to analyze the impact of soil moisture and nutrient levels on cadmium uptake based on soil environmental characteristics datasets, and to determine high uptake risk zones. The dynamic monitoring module is used to track changes in cadmium concentration from rice roots to stems and leaves and obtain dynamic sequence data of cadmium transfer when there is a high absorption risk zone. The bottleneck analysis module is used to analyze the migration speed and interaction characteristics of cadmium between different tissues based on the transfer dynamic sequence data, and to identify bottleneck points in the transport channel; The path simulation module is used to construct a transfer path simulation map based on bottleneck points and integrate sample data of rice physiological mechanisms, and to obtain a distribution map of hindering factors. The node determination module is used to analyze the interaction between soil properties and rice physiological mechanisms based on the distribution map of hindering factors, and to determine the location and formation conditions of key nodes for cadmium accumulation. The report generation module is used to generate a full-cycle transfer pattern report based on the location and formation conditions of key nodes in cadmium accumulation, thereby obtaining a basis for control.

[0009] The beneficial effects of this application are as follows: By integrating field soil environmental data, rice root growth status data, and cadmium-related detection data, a soil environmental characteristic dataset is constructed. This dataset characterizes the rice cadmium absorption environment from multiple dimensions, including soil moisture, nutrient distribution, and cadmium bioavailability, improving the completeness of basic data and the accuracy of risk characterization. Analyzing the impact of soil moisture and nutrient levels on cadmium absorption rates identifies high-risk absorption zones, enabling early identification of potentially high-risk areas during rice growth and significantly improving the timeliness and relevance of cadmium absorption risk prediction. Tracking cadmium concentration changes from rice roots to stems and leaves, a dynamic sequence data of cadmium transfer is constructed, and the migration characteristics of cadmium between different tissues are analyzed. This allows for precise identification of bottlenecks in transport pathways, improving the dynamic analysis of cadmium migration within the plant. By integrating rice physiological mechanism sample data to construct a transfer path simulation map and generate a distribution map of hindering factors, the distribution area, scope of action, and hierarchical differences of hindering factors during cadmium migration are revealed, enhancing the ability to identify cadmium migration hindering mechanisms. By analyzing the interaction between soil characteristics and rice physiological mechanisms, the location and formation conditions of key nodes in cadmium accumulation can be determined. This establishes a correlation between changes in the field environment and accumulation within the plant, improving the accuracy of identifying key accumulation stages and sites. By generating a full-cycle transfer pattern report and extracting control evidence, a complete technical support chain is achieved, from risk identification and precise monitoring to targeted control, which is beneficial for improving the adaptability and accuracy of control measures. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart of the method for predicting and controlling cadmium absorption in rice according to this application. Figure 2 This is a scatter plot comparing the predicted and measured values ​​of the cadmium accumulation level index in this application embodiment; Figure 3 This is a heatmap showing the evolution of the tissue pathway and reproductive stages of key nodes in the full-cycle cadmium accumulation in this application embodiment. Figure 4 This is a schematic diagram of the rice cadmium absorption prediction and control system of this application. Detailed Implementation

[0012] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0013] For ease of understanding, the specific process of the embodiments of this application is described below. Figure 1 The diagram shows a flowchart of the method for predicting and controlling cadmium absorption in rice provided by this invention. The flowchart specifically includes the following steps: S1. Obtain field soil environmental data and combine it with rice root growth status data to construct a soil environmental characteristic dataset that includes cadmium solubility and cadmium bioavailability indicators. The soil environmental data includes soil moisture and nutrient distribution.

[0014] In one specific embodiment, the process of performing step S1 may specifically include the following steps: Raw data on soil moisture and nutrient distribution are collected through a sensor network. The raw data is then preprocessed to obtain cleaned soil environmental data. Soil samples were collected, and the total cadmium content, available cadmium content, and soil physicochemical parameters related to cadmium form transformation were detected as cadmium-related detection data. Rice root samples were collected, and quantitative characteristic data of root length, root density, and root activity were obtained as data on the growth status of rice roots. Based on data on rice root growth status, soil environment, and cadmium-related detection data, cadmium solubility and cadmium bioavailability indicators were determined. A soil environmental characteristics dataset was constructed by integrating cadmium solubility index, cadmium bioavailability index, soil environmental data, cadmium-related detection data, and rice root growth status data.

[0015] Specifically, a sensor network is deployed within the target field according to preset monitoring units. Each monitoring unit corresponds to a spatial location and soil depth range in the field, continuously collecting raw data on soil moisture and nutrient distribution. Soil moisture data can be obtained using buried moisture sensors, while soil nutrient data can be obtained through ion detection probes, soil conductivity monitoring devices, or time-sampling detection methods. Nutrient data includes at least the supply status of nitrogen, phosphorus, and potassium.

[0016] When preprocessing the original monitoring data table, missing record identification, abnormal record removal, time synchronization, and spatial mapping are performed first. If necessary, smoothing filtering can be applied to the data after removing abnormalities to reduce the impact of instantaneous disturbances in the field. The washed moisture data are standardized to volumetric water content, and the nutrient content is standardized to mg / kg. Independent fields for the surface layer and different depth layers are retained, thus obtaining the washed soil environmental data.

[0017] When acquiring cadmium-related detection data, soil samples were collected at corresponding locations in each monitoring unit, with sampling depths corresponding to depth ranges. Samples were numbered, preserved, and sent for testing within the same batch. The total cadmium content and available cadmium content of the soil samples were measured, both in mg / kg. Simultaneously, soil physicochemical parameters related to cadmium speciation were measured, including pH, organic matter content, cation exchange capacity, redox state, and, where necessary, soil texture parameters. Total cadmium content characterizes the overall cadmium abundance in the soil, while available cadmium content and soil physicochemical parameters characterize the conditions for cadmium migration from the solid phase to the soil solution, thus forming cadmium-related detection data corresponding to the monitoring unit.

[0018] When acquiring data on rice root growth status, root samples were collected from the rice plant areas corresponding to each monitoring unit, ensuring a spatial correspondence between the root samples and soil samples. After washing the root samples, root images or measurement data were obtained, and root length, root density, and root activity were derived from these. Root activity, a quantitative characteristic of rice root absorption capacity, was obtained by physiological activity testing of the collected root samples, for example, using the triphenyltetrazolium chloride (TTC) reduction method to obtain the corresponding quantitative value of root activity. Root length, root density, and root activity together constitute the rice root growth status data, characterizing the scale and absorption capacity of the root absorption interface.

[0019] When determining the cadmium solubility and cadmium bioavailability indices, a calculation record for each monitoring unit is first constructed. The cadmium solubility index characterizes the degree of cadmium migration from the solid phase to the solution phase in soil. Its calculation inputs include at least the available cadmium content, soil moisture, pH, redox state, and organic matter content. For example, the above inputs are first normalized and homogenized, and then weights are assigned according to the direction of each factor's effect on cadmium release to obtain a dimensionless cadmium solubility index. The index value increases with increasing available cadmium content and soil moisture, and decreases with increasing pH. The cadmium bioavailability index characterizes the likelihood of releaseable cadmium in the soil reaching the rhizosphere and being absorbed by rice roots. Its calculation inputs include the cadmium solubility index, root length, root density, root activity, and root zone nutrient status. For example, based on the cadmium solubility index, root absorption interface correction terms and root activity correction terms are introduced, and the influence of nutrient status on the absorption process is combined to obtain a dimensionless cadmium bioavailability index.

[0020] When integrating the above data, the plot number, monitoring unit number, depth range, sampling time, and growth period identifier were used as unified primary keys to perform time alignment, spatial alignment, and field mapping on the multi-source data. The integrated dataset forms a structured soil environmental characteristics dataset, which includes at least the following fields: soil moisture, soil nutrients, total cadmium content, available cadmium content, soil physicochemical parameters, root length, root density, root activity, cadmium solubility index, and cadmium bioavailability index.

[0021] S2. Based on the soil environmental characteristics dataset, analyze the impact of soil moisture and nutrient levels on cadmium uptake rate and determine the high uptake risk zone.

[0022] In one specific embodiment, the process of performing step S2 may specifically include the following steps: Soil moisture distribution data and soil nutrient distribution data were extracted from the soil environmental characteristics dataset, and combined with cadmium absorption rate records to construct a comprehensive analysis base dataset. Based on the comprehensive analysis of the basic dataset, the correlation between soil moisture content, soil nutrient content and cadmium uptake rate was analyzed to determine the soil moisture range and soil nutrient range corresponding to high-risk areas. Acquire real-time environmental characteristics data of field soil and compare them with soil moisture range and soil nutrient range to determine whether there are abnormal data areas; If abnormal data is found, the data corresponding to the abnormal areas will be stratified according to spatial distribution and soil depth. Combined with the actual cadmium absorption rate records, the key monitoring areas and their corresponding soil moisture and soil nutrient ranges will be determined, and the key monitoring areas will be identified as high absorption risk zones.

[0023] Specifically, soil moisture distribution data, soil nutrient distribution data, and cadmium uptake rate records corresponding to each monitoring unit are extracted from the soil environmental characteristics dataset to construct a comprehensive analysis dataset. This dataset includes at least the plot number, monitoring unit number, depth range, sampling time, growth stage identifier, soil moisture field, soil nutrient field, cadmium solubility index field, cadmium bioavailability index field, and cadmium uptake rate field. The cadmium uptake rate is a quantitative indicator characterizing the degree of cadmium uptake by rice from the soil. In one embodiment, the cadmium uptake rate is expressed as the ratio of the cadmium content in the target part of the plant to the available cadmium content in the soil of the corresponding monitoring unit. To ensure comparability between different samples, the cadmium content in the target part of the plant and the available cadmium content in the soil are measured using a unified caliber and unit, resulting in a dimensionless ratio for the cadmium uptake rate. The target part can be selected as the root, the entire aboveground part, or a preset functional tissue, depending on the monitoring plan, and must remain consistent within the same analysis batch. Cadmium uptake rate records are obtained by collecting rice plant samples corresponding to the monitoring unit, detecting the cadmium content in the target area, and calculating the available cadmium content in the soil at the corresponding monitoring unit, depth range, and sampling time. These records are then linked to the monitoring unit number, sampling time, depth range, and growth stage identifier. When multiple cadmium uptake rate records exist within the same time window, duplicate records are averaged or processed using the median. If some monitoring units have missing environmental data, data from adjacent monitoring periods or data from the same layer of adjacent monitoring units can be used to supplement the data, and the supplemented records are marked. To avoid inconsistencies between historical and current sample data, when using historical field trial data, the historical data must at least meet the following requirements: consistent or identical variety, similar or identical planting conditions, consistent target area, and consistent detection method.

[0024] When determining the soil moisture and nutrient ranges corresponding to high-risk areas based on the comprehensive analysis dataset, samples are grouped according to monitoring units, depth intervals, and growth stages. This ensures that soil environmental data and cadmium uptake records correspond within the same spatial level and time period, avoiding the impact of sample mixing from different soil layers and growth stages on the analysis results. Within each group, soil moisture content, soil nutrient content, and cadmium uptake are jointly analyzed to obtain the response relationship of cadmium uptake to changes in soil moisture and nutrients. Joint analysis can employ piecewise statistical methods, regression fitting methods, or response surface methodology, where the inputs are soil moisture content and soil nutrient content, and the output is the response of cadmium uptake to environmental conditions. If necessary, a tree model can be used to train the relationship between soil moisture and nutrient combinations and cadmium uptake levels, and the corresponding high uptake condition range can be extracted based on feature contribution and node splitting results. In the above processing, soil moisture and nutrients are considered as influencing factors in the derivation of high uptake conditions, but do not directly replace the high-risk areas themselves.

[0025] Based on this, a high-absorption sample set is first determined according to cadmium absorption rate records. The high-absorption sample set is a set of samples whose cadmium absorption rate reaches the high-value judgment criteria under the same growth period, the same target location, and the same data caliber. In one embodiment, the high-value judgment criteria can be determined in any of the following ways: first, selecting samples whose cadmium absorption rate is within a pre-set upper quantile interval within the same growth period; second, selecting samples whose cadmium absorption rate is greater than or equal to a preset risk threshold; third, judging based on the statistical boundaries of historical high-absorption samples. For ease of implementation, unless otherwise specified, the preset risk threshold can be determined by the preset quantile of the cadmium absorption rate distribution of historical normal samples within the same growth period. After obtaining the high-absorption sample set, the soil moisture data and soil nutrient data corresponding to this sample set are extracted retrospectively, and their value distributions are statistically analyzed to extract the continuous value range or combined value range that keeps the cadmium absorption rate high; when nutrients include multiple sub-items, the value range of each sub-item can be determined separately, or the combined condition range of multiple sub-items can be determined.

[0026] To ensure the repeatability of subsequent assessments, the determination of soil moisture and soil nutrient ranges must at least meet the following conditions: the proportion of samples within the range to the high-absorption sample set reaches a preset threshold, and the statistical value of cadmium absorption rate of samples within the range is higher than the corresponding statistical value of samples outside the range; where the statistical value can be the mean, median, or upper quantile. In other words, high absorption risk conditions are not determined solely by a single moisture or nutrient value, but rather by environmental condition ranges that repeatedly occur in high-absorption samples and stably correspond to higher cadmium absorption rates.

[0027] When acquiring real-time environmental characteristic data of field soil, the soil moisture and soil nutrients of each monitoring unit within the current monitoring period are updated, and the real-time environmental characteristic data adopts the same field definitions, unit caliber, and depth layer caliber as the aforementioned high-risk condition range. The real-time soil moisture value of each monitoring unit within the corresponding depth interval is compared with the soil moisture range, and the corresponding nutrient value is compared with the soil nutrient range. When the real-time soil moisture value of a monitoring unit within the corresponding depth interval falls within the soil moisture range, and at least one preset nutrient combination simultaneously falls within the soil nutrient range, the monitoring unit is determined to be a suspected anomalous unit. To avoid misjudgment caused by transient disturbances, temporal continuity and spatial adjacency constraints are further applied to suspected anomalous units. When a suspected anomalous unit appears repeatedly within a continuous monitoring period, or appears simultaneously with an adjacent monitoring unit within the same monitoring period, its corresponding monitoring data is determined to be anomalous area data.

[0028] Anomaly regions are clustered based on the adjacency of monitoring units, grouping spatially continuous or neighboring anomaly monitoring units into the same candidate region. Within each candidate region, the data is stratified by depth interval to obtain anomaly data subsets for different soil layers. For each anomaly data subset, soil moisture, soil nutrients, cadmium solubility index, cadmium bioavailability index, and measured or historical cadmium absorption rate records matching the corresponding monitoring unit and growth stage are extracted, and the degree of matching with the historical high absorption sample set is calculated. The matching degree is a quantitative result characterizing the consistency between anomaly data subsets and historical high absorption characteristics. For example, it can be determined using a weighted scoring method: Matching Degree = a × Condition Hit Ratio + b × Absorption Rate High Value Degree + c × Continuous Occurrence Degree + d × Spatial Contiguousness Degree, where a, b, c, and d are preset weights, and a + b + c + d = 1. The condition hit ratio characterizes the proportion of samples in the anomaly data subset that meet the aforementioned high-risk condition range. The absorption rate high value degree characterizes the degree to which the mean, median, or upper quantile of the corresponding cadmium absorption rate measurement record is higher than the historical baseline. The continuous occurrence degree characterizes the frequency of repeated occurrences of the anomaly state within a continuous monitoring period. The spatial contiguousness degree characterizes the spatial clustering degree of the anomaly monitoring units. Each weight can be preset based on historical sample statistical results, validation set performance, or empirical calibration results. When a candidate region simultaneously meets the preset matching conditions within a specific depth layer, the depth layer corresponding to that candidate region is determined to be a key monitoring layer. The preset matching conditions include at least: the matching degree is not lower than the preset matching threshold, and the condition hit rate is not lower than the preset ratio threshold; if necessary, the corresponding cadmium absorption rate statistical value can also be required to be higher than the historical baseline for the same growth period. Through the above processing, the identification results of abnormal areas are not only limited to planar location, but can also distinguish the soil layers where the anomalies mainly occur, thus corresponding to the distribution of rice roots and the actual absorption path.

[0029] Historical cadmium uptake records from key monitoring layers were retrospectively matched to extract a set of historical high-uptake samples corresponding to each key monitoring layer. Within this sample set, the effective range of soil moisture and soil nutrients was recalculated. If multiple depth layers within the same candidate area meet the criteria, the priority layer is determined based on the frequency, duration, or statistical mean of high cadmium uptake records, and the spatial area corresponding to the priority layer is designated as the key monitoring area. The soil moisture and soil nutrient ranges corresponding to the key monitoring area serve as the basis for subsequent dynamic monitoring and risk warning in that area. When a key monitoring area is designated as a high-uptake risk zone, the high-uptake risk zone must include at least its spatial location range, depth range, and corresponding soil moisture and soil nutrient ranges, thus ensuring that the high-uptake risk zone simultaneously possesses spatial boundaries, depth boundaries, and environmental condition boundaries.

[0030] S3. When there is a high absorption risk zone, track the changes in cadmium concentration from rice roots to stems and leaves to obtain transfer dynamic sequence data.

[0031] In one specific embodiment, the process of performing step S3 may specifically include the following steps: On rice plants in the high absorption risk range, cadmium concentration change data from roots to stems and leaves were continuously collected using monitoring equipment to obtain initial transfer sequence data. Based on the initial transfer sequence data, the temporal correlation between cadmium uptake in roots and cadmium transfer in stems and leaves was analyzed, and the corresponding correlation characteristic parameters were determined. When the associated feature parameters exceed the preset parameter range, the initial transfer sequence data is segmented to determine the critical time window in which the rate of change of cadmium concentration changes abruptly. Extract cadmium concentration change data within key time windows and perform time-series feature extraction to generate transfer dynamic sequence data.

[0032] Specifically, the target plant set is determined based on the spatial location and depth range of the high absorption risk zone, and each target plant is assigned a plant number. Repeated sampling and testing of roots, stems, and leaves are performed according to the plant number within a continuous monitoring period. If necessary, the stem can be further subdivided into stem base and upper stem segments, and the leaves into functional leaf parts. Considering that cadmium concentration in plant tissues is difficult to measure directly and stably continuously using a single online device under field conditions, continuous collection is carried out using repeated sampling and testing at preset time intervals, or a combination of in-situ monitoring and timed sampling and testing. The cadmium concentration of samples from each part is measured after pretreatment, and the test results are linked to the plant number, sampling time, growth stage identifier, and spatial location identifier. The root, stem, and leaf cadmium concentration sequences obtained from the same plant within a continuous monitoring period are time-aligned, and outliers and missing values ​​are processed to obtain initial transfer sequence data.

[0033] Temporal correlation analysis was performed on the root cadmium concentration change sequences and the stem and leaf cadmium concentration change sequences of each target plant. Root cadmium uptake could be determined by the increase in root cadmium concentration between adjacent monitoring time points or the rate of change in root cadmium concentration within a unit monitoring period. Stem and leaf cadmium transfer could be determined by the increase in stem or leaf cadmium concentration between adjacent monitoring time points or the rate of change in cadmium concentration at the corresponding site within a unit monitoring period. To mitigate the influence of individual plant differences, cadmium concentrations at each site were uniformly converted to fresh or dry weight. Temporal correlation analysis could employ time-lag correlation analysis, sliding time window cross-correlation analysis, or dynamic time warping analysis. The inputs were the root cadmium concentration change sequences and the stem and leaf cadmium concentration change sequences; the outputs were the temporal correlation results from root to stem and from root to leaf. For example, when using time-lag correlation analysis, the correlation between root change sequences and stem or leaf change sequences is calculated within multiple candidate lag periods, and the lag period with the highest correlation is determined as the migration lag parameter. When using sliding time window cross-correlation analysis, the continuous monitoring period is divided into multiple sliding windows, and the cross-correlation between roots and stems / leaves is calculated within each window. The maximum correlation strength, corresponding lag time, correlation duration, and ratio of concentration change rates between parts are extracted as association feature parameters. The association feature parameters include at least the root-to-stem migration lag, the root-to-leaf migration lag, the corresponding correlation strength, and the ratio of concentration change rates between parts.

[0034] Based on historical normal migration sample sets or historical field trial data of the same variety and growth stage, preset parameter ranges for associated characteristic parameters are determined, and these ranges serve as the basis for judging abnormal migration trends. Normal intervals can be set for migration lag, correlation strength, and the ratio of concentration change rates between different parts. When any associated characteristic parameter of a target plant exceeds the corresponding normal interval, the target plant is judged to have an abnormal migration trend. Subsequently, the initial migration sequence data of the target plant is segmented chronologically to identify the time points where the rate of change of cadmium concentration changes significantly changes. The time point where the rate of change of cadmium concentration changes significantly changes is the time boundary where the difference in the rate of change of cadmium concentration between adjacent time periods exceeds a preset rate change threshold, and the change trends of the preceding and following time periods remain different. Segmentation processing can employ the sliding window difference method, change point detection method, or piecewise linear fitting method; the input is the cadmium concentration time series of roots, stems, and leaves, and the output is the cadmium concentration change rate of each part and its corresponding mutation time position. When using the sliding window difference method, the cadmium concentration sequence of a location within a continuous monitoring period is calculated using a rolling method according to a preset window length and window step size. The rate of change of cadmium concentration within each window is obtained, and the rate difference between adjacent windows is compared. When the rate difference between adjacent windows exceeds a preset rate change threshold, the corresponding window boundary time is identified as a candidate mutation point. If the persistence condition is met, this candidate mutation point is identified as a mutation point. When using the piecewise linear fitting method, the time series of cadmium concentrations at the location is piecewise fitted to obtain the slope of concentration change in each time period. When the slope difference between adjacent time periods exceeds a preset rate change threshold, the corresponding segment boundary time is identified as a mutation point. Using the time corresponding to the mutation point as the center, continuous time intervals of a preset duration are extracted forward and backward, or continuous time intervals covered by adjacent segments before and after the mutation point are extracted as key time windows. When the interval between adjacent key time windows is less than a preset merging threshold, they are merged.

[0035] The cadmium concentration sequences of roots, stems, and leaves of corresponding plants within key time windows, along with concentration changes at adjacent time points, were extracted. The extracted sequences were then reconstructed using a time index to ensure a consistent time scale for window data across different plants and monitoring periods. Temporal features were extracted from the cadmium concentration changes within the key time windows to obtain a dynamic feature set describing the cadmium migration process. This dynamic feature set includes at least the root cadmium concentration at the start of the window, the stem and leaf cadmium concentrations at the end of the window, the rate of change in root concentration, the rate of change in stem concentration, the rate of change in leaf concentration within the window, the root-to-stem migration lag, the root-to-leaf migration lag, the maximum concentration gradient within the window, the concentration ratio between different parts, and the magnitude of concentration changes before and after the mutation point. If necessary, the available cadmium content in the soil, soil moisture, and soil nutrients within the key time windows can be extracted simultaneously as additional environmental fields. However, the core of the transfer dynamic sequence data remains the temporal changes in cadmium concentration in roots, stems, and leaves, and their derived features. The dynamic feature set was organized by plant number, time window number, and spatial location number to form structured transfer dynamic sequence data.

[0036] S4. Based on the transfer dynamic sequence data, analyze the migration speed and interaction characteristics of cadmium between different tissues to identify bottlenecks in the transport channels.

[0037] In one specific embodiment, the process of performing step S4 may specifically include the following steps: Based on the transfer dynamic sequence data, the migration rate change parameters of cadmium in different tissues of rice were calculated to obtain the distribution dynamics of cadmium in each tissue. Based on the distribution dynamics, identify areas of high cadmium concentration accumulation over time and determine the limiting locations of key limiting points; For restricted locations, abrupt changes in migration rate parameters are extracted to identify the main obstruction zones in the transport path; Based on the main obstruction zones, a migration restriction analysis model was constructed to analyze the degree of migration restriction of cadmium between adjacent tissues and to determine the corresponding migration restriction range. By combining the location of key limiting points with the range of restricted migration, local features are extracted from the dynamic sequence data of the transfer to identify bottleneck points in the transportation channel.

[0038] Specifically, the transfer dynamic sequence data is recombined according to the tissue pathway, forming directed migration segments from root to stem base, stem base to upper stem segment, and upper stem segment to leaf. When S3 only outputs data for three tissue types—root, stem, and leaf—two migration paths are formed: root to stem and stem to leaf. For each migration segment, the concentration difference change, concentration transfer ratio, and migration lag between adjacent tissues at continuous monitoring time points are calculated. The migration rate change parameter can be calculated based on the concentration difference and time interval between adjacent tissues at adjacent monitoring time points; the concentration transfer ratio is the ratio of the concentration increment of the later tissue to the concentration increment of the earlier tissue; the migration lag is the time difference between the peak concentration change of the earlier tissue and the peak concentration change of the later tissue. The first-order difference method can be used to calculate the rate of change of cadmium concentration sequence in each tissue between adjacent time points, and further obtain the rate change amplitude and rate fluctuation degree between adjacent tissues. After the above processing, the migration rate change sequence, concentration difference change sequence, transfer ratio sequence and time lag sequence of each migration segment within a continuous time window are obtained, and the distribution dynamics of cadmium in each tissue are constructed based on these.

[0039] For each migration segment, the concentration change trends of the preceding and following tissues within a continuous time window are compared. When the cadmium concentration of a preceding tissue continuously increases or remains high, while the concentration of the adjacent following tissue increases slowly, is interrupted, or the increase is lower than a preset transfer ratio threshold, it is determined that the preceding tissue has a cadmium retention trend. To avoid misjudging short-term fluctuations as high retention areas, persistence and cumulative conditions are further applied to the retention trend. When the same tissue simultaneously meets the conditions of maintaining a high concentration, a low transfer ratio to the following tissue, and an extended migration lag within multiple consecutive time windows, the corresponding tissue segment is identified as a high retention area, and the connection point between this tissue and the next tissue is identified as the restriction point of the critical constraint.

[0040] Around the restricted location, the migration rate change sequence, inter-tissue concentration difference change sequence, transfer ratio sequence, and migration lag sequence of the corresponding migration segment within each key time window are extracted. Change point detection or piecewise fitting is performed on these sequences to identify the time boundary point where the migration segment transitions from a normal transfer state to a restricted transfer state. Abrupt change characteristics include at least a significant decrease in migration rate, a continuous widening of the inter-tissue concentration difference, a continuous decrease in the transfer ratio, and a continuous increase in the migration lag. For example, a sliding window difference method can be used to continuously calculate the migration rate change parameters within a continuous time window and compare the rate difference between adjacent windows; when the rate difference between adjacent windows exceeds a preset rate change threshold, and this difference remains unchanged for at least one subsequent preset duration, the corresponding window boundary time is determined as the abrupt change point. Alternatively, a piecewise linear fitting method can be used to piecewise fit the migration rate change sequence to obtain the rate change slope for each time period; when the slope difference between adjacent time periods exceeds a preset slope difference threshold, the corresponding boundary time is determined as the abrupt change point. The continuous time interval before and after the mutation point is used as the obstacle identification period. Based on the degree of increase in inter-tissue concentration difference, decrease in transport ratio, and increase in migration lag within this period, the tissue connective tissue corresponding to the restricted location is determined to be restricted. When this tissue connective tissue continuously meets the conditions of decreased migration rate, low transport ratio, and increased inter-tissue concentration difference within the obstacle identification period, it is identified as the main obstacle interval in the transport path. The main obstacle interval includes the restricted tissue connective tissue segment and its corresponding continuous time range.

[0041] The model input consists of inter-tissue migration parameters within the main obstruction interval, and the model output consists of the migration restriction index and restriction level of adjacent tissue connection segments within the corresponding time window. To maintain consistency with field monitoring data, the migration restriction analysis model is implemented using a rule-based hierarchical analysis model or a threshold-based weighted judgment model. Specifically, the decrease in migration rate, the cumulative magnitude of inter-tissue concentration difference, the decrease in transport ratio, and the increase in migration lag of adjacent tissue connection segments within the local time window are normalized to obtain the corresponding rate-restricted component, concentration retention component, insufficient transport component, and lag anomaly component; then, these components are weighted and summed according to preset weights to obtain the migration restriction index of the adjacent tissue connection segment. The degree of migration restriction can be expressed as: R = w1 × V + w2 × C + w3 × T + w4 × L, where R is the degree of migration restriction, V is the normalized value of the decrease in migration rate, C is the normalized value of the cumulative magnitude of inter-tissue concentration difference, T is the normalized value of the decrease in transfer ratio, L is the normalized value of the increase in migration lag, and w1, w2, w3, and w4 are corresponding weights, with w1 + w2 + w3 + w4 = 1. Each weight can be pre-set based on the differentiation effect of historical normal migration samples and historical restricted migration samples, or calibrated based on the recognition accuracy in validation samples. Based on the numerical range of the degree of migration restriction index, the migration state is divided into corresponding restriction levels of mild, moderate, and severe restriction. The range of migration restriction consists of a time range and an organizational range. The time range is the period during which the degree of migration restriction index continuously reaches a preset level or higher, and the organizational range is the adjacent tissue connection segment where restriction occurs.

[0042] Within the migration-restricted range corresponding to each restricted location, parameters such as inter-tissue migration rate change, concentration difference accumulation, transport ratio, migration lag, and cadmium concentration change in the corresponding tissue are extracted within a local time window to form a local feature set. A joint judgment is then made based on the recurrence of restricted locations, the persistence of the restriction degree, and the cumulative degree of insufficient inter-tissue transport. When the same tissue segment falls within the migration-restricted range in multiple consecutive monitoring windows, and its migration restriction degree index is consistently higher than a preset judgment threshold, and simultaneously meets the conditions of high retention in the preceding tissue and insufficient response in the following tissue, the tissue segment is identified as a bottleneck in the transport channel. If multiple candidate bottlenecks exist within the same plant, they are sorted according to the migration restriction degree index, duration, and impact on cadmium accumulation in subsequent tissues. The tissue segment with the highest ranking is identified as the primary bottleneck, and the rest are identified as secondary bottlenecks. The output of the bottleneck point includes at least the plant number, bottleneck tissue segment, corresponding time range, migration-restricted range, and local feature parameter set.

[0043] S5. Based on the bottleneck point, a transfer path simulation diagram is constructed by integrating sample data of rice physiological mechanisms to obtain the distribution map of hindering factors.

[0044] In one specific embodiment, the process of performing step S5 may specifically include the following steps: Based on bottlenecks and combined with sample data on the physiological mechanisms of rice, we identified the factors that hinder the migration of cadmium between different tissues in rice. Based on the hindering factors and bottlenecks, a cadmium transfer path simulation map was constructed, and the distribution area of ​​the hindering factors in the transfer path simulation map was determined. Based on the distribution area, the relationship between the hindering factors and the cadmium migration between different tissues of rice was analyzed, and the area of ​​influence of the hindering factors was determined. Based on the region of action, stratified analysis was performed on the sample data of rice physiological mechanisms to determine the distribution differences of hindering factors in different tissue levels. Based on the distribution area, the area of ​​influence, and the distribution differences, a distribution map of hindering factors is generated.

[0045] Specifically, bottleneck point results should include at least the bottleneck tissue segment, corresponding growth stage, corresponding time window, range of migration restriction, and a set of local migration characteristic parameters. Rice physiological mechanism sample data should include at least tissue structure parameters, transmembrane transport parameters, and physiological metabolic parameters. Tissue structure parameters include cell wall lignin content, pectin content, cell wall thickness, tissue porosity, vascular bundle density, and vascular bundle cross-sectional area. Transmembrane transport parameters include the relative expression levels of heavy metal transporters, the relative expression levels of vacuolar septum-related transporters, and membrane permeability parameters. Physiological metabolic parameters include transpiration intensity, root pressure characterization parameters, antioxidant enzyme activity, and chelation-related metabolite content. Relative expression levels can be characterized by transcriptional quantification or protein quantification, while membrane permeability parameters can be characterized by tracer experiment results or membrane electrophysiological test results. To reduce field implementation costs and maintain sample representativeness, physiological mechanism sample data can be generated from representative sample batches corresponding to the bottleneck tissue segment. Representative sample batches should at least meet the requirements of consistent tissue location, consistent sampling time window, consistent growth stage, and corresponding spatial location, without requiring full molecular testing on every monitored plant. This ensures that the bottleneck results correspond to the physiological mechanism sample data at both the tissue and time levels, avoiding object misalignment.

[0046] Physiological mechanism sample records were extracted from the upstream and downstream tissues corresponding to the bottleneck tissue segment. Missing values ​​were removed, outliers were truncated, dimensions were standardized, and intervals were normalized for each parameter, forming a physiological parameter matrix that corresponds one-to-one with the bottleneck tissue segment. Using the S4 migration restriction index as the target value and the physiological parameter matrix as input, a hindering factor identification model was constructed. The hindering factor identification model employed an interpretable supervised learning model or a weighted decision model. When using a supervised learning model, a gradient boosting tree model was adopted. The model input consisted of tissue structure parameters, transmembrane transport parameters, and physiological metabolic parameters, and the model output was the contribution value of each parameter to the migration restriction index. The training samples consisted of historical field trial data and current monitoring data. Historical data and current data were required to be at least identical in variety, tissue location, growth period, and detection caliber. The training and validation sets were stratified by plant number, and the squared error loss function was used to ensure a continuous correspondence between the model output and the migration restriction index.

[0047] After model training, hindering factors are identified based on the contribution values ​​and directions of each parameter. Directional constraints are not abstract judgments, but rules used to filter and correct model outputs: only when a parameter's contribution to the migration restriction index reaches a preset contribution threshold, and the parameter's direction of change aligns with the direction of enhanced migration restriction, is the corresponding physiological mechanism identified as a hindering factor. If the contribution value is high but the direction is inconsistent with the direction of enhanced migration restriction, it is marked as a non-hindering factor or removed. The direction of enhanced migration restriction includes at least the following: reduced effective vascular bundle conduction area leading to decreased long-distance transport capacity; increased cell wall fixation leading to increased tissue retention; enhanced vacuolar partitioning leading to obstructed trans-tissue transport; and decreased membrane permeability leading to weakened transmembrane transport. This avoids directly identifying hindering factors based solely on model numerical values.

[0048] To facilitate subsequent structured expression, the contribution value range of hindering factors is divided into preset level ranges, which are then mapped to hindering factor types. In one embodiment, the contribution value range can be divided into high-contribution hindering factors, medium-contribution hindering factors, and low-contribution hindering factors; alternatively, the hindering factors can be further classified into tissue-fixed, transmembrane transport-restricted, vascular conduction-restricted, and metabolic regulation-restricted types based on the parameter category. Specifically, when the contribution value corresponding to cell wall lignin content, pectin content, or cell wall thickness reaches a preset range, it is determined to be a tissue-fixed hindering factor; when the contribution value corresponding to the relative expression level of heavy metal transporters, the relative expression level of vacuolar septum-related transporters, or membrane permeability parameters reaches a preset range, it is determined to be a transmembrane transport-restricted hindering factor; when the contribution value corresponding to vascular bundle density, vascular bundle cross-sectional area, or transpiration / root pressure-related parameters reaches a preset range, it is determined to be a vascular conduction-restricted hindering factor; and when the contribution value corresponding to antioxidant enzyme activity or chelation-related metabolite content reaches a preset range, it is determined to be a metabolic regulation-restricted hindering factor. This yields a list of hindering factors, their types, and their contribution ranges corresponding to each bottleneck segment.

[0049] The cadmium migration pathway within rice is abstracted as a directed tissue graph. The nodes of this graph represent migration units such as root epidermis, cortex, endodermis, stele, xylem, vascular bundles at the stem base, vascular bundles at stem nodes, vascular bundles at leaf sheaths, and leaf tissues. Edges represent the mass transfer relationships between adjacent migration units. Edge weights are not directly replaced by obstruction factors, but rather by deriving migration capacity parameters based on the migration rate, migration lag, concentration transfer ratio, and obstruction factors identified in this step, as output from S4. The migration capacity parameter characterizes the actual migration passage capacity between adjacent tissue units and can be determined by the baseline migration capacity of the corresponding tissue segment and an obstruction correction term. The baseline migration capacity is obtained statistically from historical normal samples, while the obstruction correction term is formed by superimposing the contribution values ​​of obstruction factors according to preset weights. To ensure physical rationality, the migration capacity parameter is limited to a preset effective range to avoid negative values ​​or conflicts with tissue connectivity.

[0050] A cadmium transport path simulation map is generated based on a directed tissue graph. Each node retains its tissue identity, time window, and bottleneck association markers, while each edge retains its migration capacity parameters and restriction level. Then, hindering factors are mapped to corresponding regions. If a hindering factor primarily affects fixation or partitioning processes within a tissue, it is mapped to a node region; if it primarily affects transmembrane transport, loading, unloading, or vascular bundle conduction processes between adjacent tissues, it is mapped to an edge region. This determines the distribution area of ​​hindering factors in the transport path simulation map.

[0051] A record table of interactions was established for each hindering factor. This table included at least the type of hindering factor, its corresponding tissue level, corresponding migration edge, affected migration indicators, and direction of action. For each hindering factor, the differences in migration capacity parameters, migration lag, and concentration transfer ratios between its occurrence and non-occurrence areas were compared. Combined with the contribution value output from the hindering factor identification model, the primary effects of this hindering factor on intra-tissue fixation, inter-tissue loading, long-distance vascular transport, or inter-tissue unloading were determined. During this process, available cadmium in the soil, soil moisture, soil nutrients, and growth period were used as control variables to eliminate confounding effects and not directly replace the determination of the area of ​​action. When a hindering factor appears consecutively in multiple adjacent tissue segments and its influence on migration capacity parameters is consistent, that consecutive tissue segment is identified as the area of ​​action of the hindering factor. When a hindering factor only causes an increase in migration lag at a single tissue interface and has no significant effect on other tissue segments, that tissue interface is identified as the local area of ​​action.

[0052] Using the region of action as an index, the sample data is divided into the root absorption layer, root-stem connection layer, stem-node transport layer, and leaf accumulation layer. Within each sub-layer, further subgroups are formed based on growth stage and bottleneck severity. Within each group, the distribution center, dispersion, and frequency of consecutive occurrence of parameters corresponding to hindering factors are statistically analyzed to form hierarchical distribution characteristics. When multiple hindering factors coexist, Gaussian mixture clustering or hierarchical clustering is used to identify the combination patterns of hindering factors. The clustering input includes the contribution vector of the hindering factor, its tissue level, and the length of the region of action; the clustering output is the category of hindering pattern under different tissue levels. The number of clusters can be determined based on the silhouette coefficient, Bayesian information criterion, or a preset upper limit for the number of clusters, thus avoiding arbitrary classification of hindering pattern categories. If necessary, analysis of variance is performed on the abundance differences of key hindering factors in each tissue level to test whether the distribution differences between different tissue levels meet the preset significance condition. The distribution difference results include at least the hierarchical location, corresponding hindering pattern, frequency of occurrence, and duration.

[0053] Using a directed organizational graph as the base map, the distribution area, area of ​​effect, and hierarchical distribution differences are written into the attribute fields of nodes and edges to form a structured organizational path map. Node fields must include at least the organization name, type of hindering factor, hierarchical category, frequency of occurrence, and duration. Edge fields must include at least the migration restriction level, main hindering pattern, and corresponding time window. The hindering factor distribution map is output in the order of the organizational path. In the two-dimensional graph, the organizational path is the horizontal axis and the organizational hierarchy is the vertical axis, with different identifiers distinguishing the distribution area and area of ​​effect of different hindering factors. When time information needs to be retained, reproductive period and time window markers are overlaid in the node and edge attributes to express the spatial and temporal distribution of hindering factors. If used for system implementation, the hindering factor distribution map is saved as a graph database record or a matrix layer, and the hindering factor category, area of ​​effect, distribution differences, and corresponding time window for each key organizational segment are output.

[0054] S6. Based on the distribution map of hindering factors, analyze the interaction between soil characteristics and rice physiological mechanisms to determine the location and formation conditions of key nodes in cadmium accumulation.

[0055] In one specific embodiment, the process of performing step S6 may specifically include the following steps: Based on the distribution map of hindering factors, combined with soil property data and rice physiological mechanism sample data, the interaction characteristics of soil properties and rice physiological mechanisms on cadmium accumulation were analyzed. Based on the interaction characteristics, the distribution areas of key nodes in cadmium accumulation were identified, and the locations of key nodes in cadmium accumulation were determined. Based on the location of key nodes in cadmium accumulation, we analyze the characteristics of the interaction effects on cadmium accumulation and determine the main areas of action corresponding to the key nodes in cadmium accumulation. Based on the main areas of action, soil property data and rice physiological mechanism sample data were stratified and analyzed according to environmental variables to obtain the variation characteristics of cadmium accumulation under different environmental variables. Based on the characteristics of the changes, the formation conditions of key nodes in cadmium accumulation were determined.

[0056] Specifically, the obstacle factor distribution map should include at least the obstacle factor type, corresponding tissue location, corresponding tissue level, corresponding time window, area of ​​action, direction of action, duration, and limiting markers corresponding to the tissue migration path. Soil characteristic data are extracted from the soil environmental characteristic dataset formed in S1 and a correspondence is established with the obstacle factor distribution map in terms of plot number, monitoring unit number, sampling time, soil layer interval, and growth stage identifier. The soil characteristic data should include at least the available cadmium content, total cadmium content, soil moisture, soil nutrients, pH, organic matter content, cation exchange capacity, and redox state. The rice physiological mechanism sample data uses the parameters from S5 and is consistent with the corresponding tissue location, corresponding time window, and corresponding plant number. At the same time, tissue cadmium accumulation data corresponding to the obstacle factor distribution map is extracted. The tissue cadmium accumulation data should include at least the tissue cadmium content, the cumulative concentration difference between tissues, and the duration of retention.

[0057] Using the area of ​​effect output by S5 as an index, soil characteristic data, rice physiological mechanism sample data of corresponding tissue parts, and tissue cadmium accumulation data within the same area of ​​effect were extracted. These data were then aligned according to monitoring unit number, plant number, tissue part, and time window to form an interactive analysis sample set. Missing value removal, outlier truncation, dimensional unification, and interval normalization were performed on the interactive analysis sample set. Based on this, interactive feature terms were generated. These interactive feature terms include at least the combination of soil available cadmium content and heavy metal transporter protein expression level, pH and cell wall fixation capacity, soil moisture and membrane permeability, and soil nutrients and transpiration intensity. These interactive feature terms are only used to characterize the coupling effect between soil environmental factors and physiological mechanism factors and do not directly replace the cadmium accumulation results themselves.

[0058] A multi-factor coupled interactive analysis model is constructed, implemented using an interpretable supervised learning model, such as a gradient boosting tree model. The model input includes soil characteristic data, rice physiological mechanism sample data, and interactive feature terms. The model output is an index of cadmium accumulation in corresponding tissue parts within a given time window. This index characterizes the strength of local cadmium accumulation in the target tissue within the current time window and is obtained by weighted summation of tissue cadmium content, cumulative inter-tissue concentration difference, and duration of residence after normalization. Specifically, tissue cadmium content represents the actual accumulation level in the tissue, cumulative inter-tissue concentration difference represents the enrichment gradient of cadmium between adjacent tissues, and duration of residence represents the persistence of this accumulation state. Weights can be determined based on historical normal sample statistical boundaries, validation set performance, or preset calibration rules. For example, the training samples consist of historical field trial data and current continuous monitoring data. The training and validation sets are stratified according to monitoring unit number and plant number. The number of trees is set to be no less than one hundred, and the maximum tree depth is set to control overfitting. Five-fold cross-validation is used for training, and the squared error loss function is employed. After training, the contribution values ​​and directions of each soil characteristic, physiological characteristic and interaction characteristic to the cadmium accumulation index are output. Furthermore, the interaction intensity statistics can be used to quantify the main two-factor interactions, thereby obtaining the interaction characteristic results.

[0059] Based on the tissue path map output by S5, interaction features are mapped to corresponding nodes and edges, and a node accumulation risk value is calculated for each organizational node. The node accumulation risk value characterizes the risk intensity of a given organizational node developing into a critical accumulation node within a corresponding time window. It can be obtained by a weighted sum of the organizational cadmium accumulation level index, interaction contribution value, duration of hindering factors, and migration restriction level after normalization. Specifically, the organizational cadmium accumulation level index characterizes the intensity of the outcome, the interaction contribution value characterizes the driving force, the duration of hindering factors characterizes the stability of the effect, and the migration restriction level characterizes the degree of conduction obstruction. A high node accumulation risk value indicates that the organizational node not only has a high level of cadmium accumulation, but its formation causes and persistence conditions are also relatively stable, making it more likely to develop into a critical cadmium accumulation node.

[0060] For each tissue node, continuity determination and adjacency merging are performed along the time window sequence. If the node accumulation risk value of the same tissue node is consistently higher than a preset risk threshold within a continuous preset time window, and the corresponding tissue cadmium content is higher than that of adjacent nodes or the historical baseline at the same level, while simultaneously showing an increased level of migration restriction, the region where this tissue node is located is identified as a candidate region for key cadmium accumulation nodes. The candidate regions are then further screened by combining the local peak location of tissue cadmium content, the cumulative location of inter-tissue concentration differences, and the main area of ​​influence of hindering factors in S5. When a candidate region simultaneously meets the local peak condition, the duration condition, and the significant interaction condition, it is identified as the distribution region of key cadmium accumulation nodes. The local peak condition indicates that the cadmium content of the node's tissue forms a local high value in the adjacent tissue path; the duration condition indicates that the high accumulation state recurs within the continuous monitoring window; and the significant interaction condition indicates that the corresponding interaction feature contribution value or interaction intensity statistic reaches a preset significance standard. The tissue name, spatial location, corresponding time window, and reproductive period identifier of the key node are then output to determine the location of the key cadmium accumulation node.

[0061] Centered on key nodes of cadmium accumulation, tissue pathway sub-maps within a preset range are extracted for upstream and downstream tissue segments. The interaction contribution value, migration capacity parameters, concentration transfer ratio, migration lag, and tissue cadmium accumulation increment within each sub-map are extracted. Local interaction analysis is performed on the pathway sub-maps, comparing the differences between key node regions and adjacent regions in interaction contribution value, migration capacity reduction, and local cadmium accumulation increment. Based on these differences, it is determined that the interaction mainly occurs in fixed areas within tissues, loading areas at tissue interfaces, long-distance vascular bundle transport areas, or unloading areas at tissue interfaces. For example, partial correlation analysis combined with sliding time window local regression analysis is used. Partial correlation analysis is used to eliminate the confounding effects of differences in total soil cadmium content, growth stage, and background varieties, while sliding time window local regression analysis is used to identify the stable range of interaction effects within continuous monitoring periods. The local interaction intensity characterizes the driving force of interaction effects on cadmium accumulation changes within local tissue segments and can be formed by weighting the interaction contribution value, migration capacity reduction, and local cadmium accumulation increment after normalization. When the local interaction intensity within a continuous tissue segment is consistently higher than a preset interaction threshold, and the corresponding cadmium accumulation increment in the tissue continuously increases, the continuous tissue segment is identified as the main interaction area corresponding to the key node of cadmium accumulation; when the interaction only causes a rapid expansion of the local concentration difference at a single tissue interface and does not continue to extend to adjacent tissues, the tissue interface is identified as the local main interaction area.

[0062] Using the primary area of ​​action as an index, the samples were divided into different soil environmental layers and different physiological state layers, and further subdivided within each layer according to growth stage and tissue level. The soil environmental layer included at least different soil moisture ranges, different pH ranges, different available cadmium ranges, and different nutrient supply ranges; the physiological state layer included at least different ranges of transporter protein expression levels, different cell wall fixation capabilities, different membrane permeability ranges, and different transpiration intensities. Stratification was based on the gradient of field measured values, ensuring that each stratified group had a statistically significant sample size. Within each stratified group, the distribution center, variation range, duration, and growth rate of cadmium content in key tissue nodes were statistically analyzed to establish cadmium accumulation and variation characteristics. When identifying the combined effects of multiple environmental variables, conditional inference tree models or stratified response surface methodology models were used. The conditional inference tree model takes environmental variable stratification values, physiological state stratification values, and interaction characteristics as inputs and outputs the cadmium accumulation level in key node tissues. When using the stratified response surface methodology model, soil moisture, pH, and available cadmium are taken as soil-side inputs, while transporter protein expression levels and cell wall fixation capacity are taken as physiological-side inputs. The degree of cadmium accumulation in key node tissues is taken as the output. The range of variable combinations corresponding to high accumulation areas is identified through response surface methodology, thereby outputting the cadmium accumulation change characteristics under different combinations of environmental variables, including the environmental intervals where high accumulation occurs, the corresponding physiological states, the duration, and the tissue level.

[0063] Conditional retrospective analysis was conducted on the cadmium accumulation characteristics under different combinations of environmental variables to identify variable combinations that consistently corresponded to high accumulation states at key nodes. These formation conditions are not abstract concepts, but rather variable combinations that repeatedly corresponded to high accumulation states at key nodes and passed stability tests. Formation conditions were determined according to three categories of constraints: soil supply conditions, tissue transport conditions, and local fixation conditions. Soil supply conditions consist of available cadmium levels, soil moisture status, pH status, and nutrient status, used to limit external cadmium supply and rhizosphere activation. Tissue transport conditions consist of decreased migration capacity, increased migration lag, and transport protein expression status, used to limit the degree of cadmium stagnation after transport to the target tissue. Local fixation conditions consist of cell wall fixation capacity, vacuolar compartmentalization capacity, and tissue metabolic status, used to limit the retention of cadmium within the target tissue.

[0064] Based on the consistency of the key node accumulation results when the formation conditions occur in historical samples and current monitoring samples, a stability test is performed on each condition item. When a certain condition combination repeatedly occurs within a continuous preset monitoring period, and the corresponding key node accumulation enhancement results remain consistent, this condition combination is determined as the formation condition of the key node for cadmium accumulation. If multiple condition combinations can lead to the accumulation of the same key node, they are sorted according to frequency of occurrence, duration, and accumulation intensity, with the top-ranked combination being the primary formation condition and the remaining combinations being supplementary formation conditions.

[0065] To verify the predictive effectiveness of the multi-factor coupled interaction analysis model constructed in this step for cadmium accumulation indicators, validation samples not used in training were employed for model evaluation. Validation samples were derived from monitoring unit data within the same field monitoring period but not used for model training. Soil characteristic data, rice physiological mechanism sample data, and interaction feature terms from the validation samples were input into the trained model to obtain the predicted cadmium accumulation indicators. Simultaneously, the measured cadmium accumulation indicators were calculated based on the measured tissue cadmium content, cumulative concentration difference between tissues, and duration of residence. Figure 2 The scatter distribution of predicted and measured values ​​of cadmium accumulation indicators in the validation sample is shown. Figure 2 As shown, the overall distribution of the validation samples is near the diagonal (y=x), and the predicted and measured values ​​show high consistency. The coefficient of determination R² reaches 0.944, and the root mean square error (RMSE) is 0.049, indicating that the constructed model has a good fitting effect and prediction accuracy for the cadmium accumulation index, and can well characterize the comprehensive influence of soil properties and rice physiological mechanisms on the cadmium accumulation index. This validation result provides quantitative support for the identification of key nodes of cadmium accumulation and the analysis of formation conditions in this step.

[0066] S7. Based on the location and formation conditions of key nodes in cadmium accumulation, generate a full-cycle transfer pattern report to obtain a basis for control.

[0067] In one specific embodiment, the process of performing step S7 may specifically include the following steps: Based on the location and formation conditions of key nodes in cadmium accumulation, cadmium accumulation data were processed in stages according to the rice growth period to obtain the accumulation status at different growth stages throughout the entire cycle and to determine the distribution of key nodes in cadmium accumulation at each stage. Based on the distribution, a dynamic change model of cadmium transfer pathways was constructed to obtain the evolution trajectory of transfer pathways throughout the entire cycle. By combining environmental variable fluctuation data, correlation analysis was conducted on the evolution trajectory of the transfer path to obtain the path change characteristics of cadmium accumulation under different conditions; Based on the characteristics of path changes, the absorption and transport patterns of cadmium throughout the entire cycle are determined, and a report on the transfer patterns throughout the entire cycle is generated. Based on the report on the full-cycle transfer pattern, the basis for controlling cadmium absorption in rice was extracted.

[0068] Specifically, the location of key nodes in cadmium accumulation includes at least the corresponding tissue part, corresponding tissue level, corresponding tissue pathway location, corresponding time window, and corresponding growth stage marker. The formation conditions include at least soil supply conditions, tissue transport conditions, and local fixation conditions. Simultaneously, dynamic sequence data on cadmium transfer, bottleneck point data, distribution map data of hindering factors, and data on key node locations and formation conditions are retrieved. Furthermore, environmental variable fluctuation data corresponding to each time window are retrieved, including at least soil moisture, soil nutrients, pH, available cadmium content, temperature, and irrigation status. Multi-source data are aligned according to plot number, monitoring unit number, plant number, growth stage marker, tissue part, and sampling time to form a full-cycle analysis dataset.

[0069] Based on the growth cycle records corresponding to rice varieties, the entire growth cycle is divided into seedling stage, tillering stage, jointing and booting stage, heading and flowering stage, and grain filling and ripening stage, or divided into several continuous stage intervals according to the actual monitoring scheme. Within each growth stage, the corresponding tissue cadmium content, cumulative inter-tissue concentration difference, duration of retention, location of key nodes, frequency of occurrence of formation conditions, and migration rate and transmission ratio for the corresponding time period are extracted. Repeated observations of the same tissue part within the same stage are time-merged using a sliding time window mean or median. Based on the merged stage data, stage accumulation status indicators are calculated. These indicators include at least the stage average accumulation level, stage maximum accumulation level, stage accumulation growth rate, mean inter-tissue migration rate, and duration of key nodes. The stage accumulation growth rate is derived by dividing the increase in tissue cadmium content between adjacent time windows by the time interval. When a tissue part meets the following conditions simultaneously within the same growth stage: the accumulation level is higher than the stage baseline, the growth rate continues to rise, and the duration of the critical node reaches the preset conditions, the tissue part is identified as the cadmium accumulation critical node of that stage. If there are multiple critical nodes in the same stage, they are sorted according to the stage average accumulation level, duration, and frequency of formation conditions, thereby obtaining the spatial and temporal distribution of cadmium accumulation critical nodes in each stage.

[0070] The tissue path map constructed in S5 serves as the path skeleton, and the dynamic sequence data of cadmium transfer in S3, the bottleneck data in S4, and the distribution of key nodes at each stage are mapped to corresponding nodes and edges. The dynamic change model of cadmium transfer path can be implemented using a state transition graph model or a time series graph network model. When using a state transition graph model, the model input includes the cadmium content of each tissue node at each growth stage, the proportion of concentration transfer between tissues, the migration lag, migration capacity parameters, the bottleneck restriction level, the abundance of hindering factors, and the state label of key nodes. The model output is the state change results of each tissue node and each tissue connection edge between adjacent stages. The node state includes at least a low accumulation state, an increasing accumulation state, a stable stagnation state, and a high accumulation state, and the edge state includes at least a normal transmission state, a weakly restricted state, and a strongly restricted state. The state transition order is set to first order, and the inter-stage transfer probability is obtained by statistical analysis of historical field monitoring data and current full-cycle monitoring data, and the state transition parameters are solved using maximum likelihood estimation.

[0071] When an artificial intelligence model is required, a temporal graph network model can be used. This model includes a node feature input layer, an edge feature input layer, a stage temporal encoding layer, a graph convolutional propagation layer, and a state output layer. The node feature input layer receives tissue cadmium content, stage accumulation state indicators, key node state markers, and distribution characteristics of hindering factors. The edge feature input layer receives concentration transfer ratios, migration lags, migration capacity parameters, and restriction levels. The stage temporal encoding layer encodes temporal relationships according to the order of reproductive stages to maintain temporal continuity between adjacent stages. The graph convolutional propagation layer propagates node and edge state information along the tissue path adjacency relationships to characterize state transfer across tissue paths. The state output layer outputs the node states, edge states, and path state changes between stages. The training samples consist of historical continuous monitoring period data. The training and validation sets are stratified by plot number and year. The optimization objective is to minimize the mean square error between the predicted and actual observed path states. After model training, the state change sequences of each tissue node and each tissue connection edge are output in stage order. This yields the evolution trajectory of the transfer path throughout the entire cycle, used to reconstruct the migration and retention process of cadmium in different reproductive stages.

[0072] Environmental variable fluctuation data within each reproductive stage are resampled at a time granularity consistent with the path state to form a stage environmental feature sequence. This stage environmental feature sequence is then mapped to node state changes, edge state changes, and key node transfers in the evolution trajectory of the transfer path, constructing a path association analysis sample set. A path state association model is used to analyze the correspondence between environmental variable fluctuations and path changes. In one implementation, a conditional inference tree model or a piecewise regression model is employed. The inputs of the conditional inference tree model include stage environmental features, formation condition markers, and the path state of the previous stage; the output is the path change category of the current stage. When using a piecewise regression model, the amplitude of environmental variable fluctuations is used as the independent variable, and the stage path state index is used as the dependent variable to identify the intervals where changes in environmental variables cause abrupt changes in path state.

[0073] Pathway changes are categorized into at least four types: upward migration, local stagnation, cross-node restriction, and terminal accumulation. The following rules apply: Upward migration is defined as a path change where key accumulation nodes or high-risk path segments shift from downstream to upstream tissues, or where the accumulation level and restriction grade of upstream tissue segments significantly increase compared to the previous stage. Local stagnation is defined as a path change where key nodes maintain high accumulation levels continuously within the same tissue or adjacent short path regions, and the local migration lag continues to increase while the proportion of transfer to subsequent tissues does not increase synchronously. Cross-node restriction is defined as a path change where the migration lag increases, migration capacity decreases, and the concentration transfer ratio between tissues before and after the interface increases at stem nodes, root-stem junctions, or other cross-node tissue interfaces. Terminal accumulation is defined as a path change where the cadmium content in leaves, panicles, or other terminal tissues continuously increases, and the corresponding local fixation conditions strengthen or prolong their duration. If necessary, continuity verification can be performed in conjunction with the path status of the previous stage to avoid misclassifying a single disturbance as a path type change. This yields the path change characteristics under different environmental conditions, including at least the corresponding combination of environmental variables, the corresponding path change category, the corresponding reproductive stage, and the corresponding organizational path range.

[0074] The pathway changes throughout all growth stages were sequentially analyzed to construct a temporal chain of "soil supply changes—root absorption changes—tissue transport changes—local accumulation changes." High-risk transition stages, high-risk tissue pathway segments, and key node migration directions were identified within this chain. High-risk transition stages are those where the pathway state changes from normal transport to strongly restricted or high-accumulation states. High-risk tissue pathway segments are tissue connection segments where the level of restriction repeatedly increases within consecutive stages. Key node migration directions are those where key nodes move from upstream to downstream tissues or from vegetative organs to reproductive organs. Based on these identification results, the full-cycle cadmium absorption and translocation patterns were extracted. These patterns include at least the key absorption stage, key retention stage, key translocation restriction stage, key accumulation amplification stage, and the corresponding dominant environmental and formation conditions for each stage. Subsequently, a full-cycle transfer pattern report is generated. The report includes at least the stage division information, the accumulation status table for each stage, the distribution table of key nodes for each stage, the trajectory diagram of the transfer path evolution, the correspondence table of environmental variables and path changes, and the pattern extraction result table. The tables are linked by plot number, plant number, growth period identifier, and tissue path location to ensure that the report results can be traced back to the original monitoring data and intermediate analysis data.

[0075] Based on the key absorption stage, key translocation restriction stage, key accumulation amplification stage, corresponding environmental variable combinations, and corresponding formation condition combinations reported in the report, control rules are generated according to three levels: external supply regulation, tissue conduction mitigation, and local accumulation weakening. External supply regulation is used to extract field management basis for key absorption stages and corresponding soil supply conditions. Field management basis includes at least the water regulation range, pH regulation direction, and nutrient supply adjustment direction. Tissue conduction mitigation is used to extract conduction regulation basis for key translocation restriction stages and corresponding tissue conduction conditions. Conduction regulation basis includes at least the management period for reducing cross-node restriction risk and the scope of key tissue pathways. Local accumulation weakening is used to extract key node intervention basis for key accumulation amplification stages and corresponding local fixed conditions. Key node intervention basis includes at least the key monitored tissue parts, key intervention time windows, and the order of lifting formation conditions. The aforementioned control rules are matched with the current monitoring data of each plot. When the current monitoring data falls within the condition combination range corresponding to a certain rule, that rule is determined as the control basis for the current plot. If multiple rules are met simultaneously, they are prioritized according to the risk level of the corresponding stage, the duration of key nodes, and the cumulative growth rate. The rule with the highest priority is used as the primary control basis, and the remaining rules are used as supplementary control basis. The control basis includes at least the target reproductive stage, the target organizational path segment, the target key nodes, the corresponding environmental conditions, and the corresponding management action type, so that the control output can not only characterize the source of risk but also correspond to specific management directions.

[0076] To visually demonstrate the dynamic evolution of key cadmium accumulation nodes in different tissues and growth stages throughout the entire rice growth period, this embodiment generates a heat map of the spatiotemporal evolution of tissue pathways and growth stages based on the locations of key cadmium accumulation nodes identified at each stage and the corresponding cadmium accumulation levels in the tissues. Figure 3 An example of this heatmap is shown. Figure 3 As shown, the horizontal axis represents the rice growth stages, including the seedling stage, tillering stage, jointing and booting stage, heading and flowering stage, and grain-filling and ripening stage; the vertical axis represents the tissue pathways, including roots, stem base, upper stem segments, leaf sheaths, and leaves. The intensity of the thermal color indicates the degree of cadmium accumulation in the corresponding tissue at the corresponding growth stage, with darker colors indicating higher accumulation levels. White asterisks indicate key nodes of cadmium accumulation identified through comprehensive evaluation. These key nodes are determined by combining the degree of cadmium accumulation, migration restriction characteristics, duration, and interaction intensity, and do not necessarily coincide completely with the tissue with the highest thermal value in a single stage. Figure 3 It can be seen that the key nodes of cadmium accumulation generally show a trend of gradual evolution from lower to upper tissues: during the seedling stage, it is mainly concentrated near the roots; during the tillering stage, it is still mainly located along the root-stem base path; after entering the jointing and booting stage, it shifts upward to the upper stem segments; towards the later stages of growth, the degree of cadmium accumulation in the leaf sheath and leaf blades increases significantly, with the highest accumulation in the leaves during the grain-filling and ripening stage. This evolutionary trend is generally consistent with the temporal chain of "key absorption stage → key retention stage → key translocation restriction stage → key accumulation amplification stage" extracted from S7, indicating that this invention can better characterize the dynamic changes of cadmium migration and accumulation throughout the entire cycle. Figure 3 Planting managers can intuitively identify key risk tissues at different growth stages, thus providing a basis for developing precise monitoring and control strategies by stage and part.

[0077] The above describes the method for predicting and controlling cadmium absorption in rice in the embodiments of this application. The following describes the system for predicting and controlling cadmium absorption in rice in the embodiments of this application. Please refer to [link to relevant documentation]. Figure 4 The present application provides a schematic diagram of the structure of a rice cadmium absorption prediction and control system, which includes: The data acquisition module 10 is used to acquire field soil environmental data and, in combination with rice root growth status data, construct a soil environmental characteristic dataset containing cadmium solubility and cadmium bioavailability indicators. The soil environmental data includes soil moisture and nutrient distribution.

[0078] Risk analysis module 20 is used to analyze the impact of soil moisture and nutrient levels on cadmium uptake rate based on soil environmental characteristics dataset, and to determine high uptake risk ranges.

[0079] The dynamic monitoring module 30 is used to track changes in cadmium concentration from rice roots to stems and leaves and obtain dynamic sequence data of cadmium transfer when there is a high absorption risk zone.

[0080] The bottleneck analysis module 40 is used to analyze the movement speed and interaction characteristics of cadmium between different tissues based on the transfer dynamic sequence data, and to identify bottleneck points in the transport channel.

[0081] The path simulation module 50 is used to construct a transfer path simulation map based on bottleneck points and integrate sample data of rice physiological mechanisms, and to obtain a distribution map of hindering factors.

[0082] The node determination module 60 is used to analyze the interaction between soil characteristics and rice physiological mechanisms based on the distribution map of hindering factors, and to determine the location and formation conditions of key nodes for cadmium accumulation.

[0083] The report generation module 70 is used to generate a full-cycle transfer pattern report based on the location and formation conditions of key nodes in cadmium accumulation, thereby obtaining a basis for control.

[0084] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for predicting and controlling cadmium uptake in rice, characterized by, The method includes: S1. Obtain field soil environmental data and combine it with rice root growth status data to construct a soil environmental characteristic dataset containing cadmium solubility and cadmium bioavailability indicators. The soil environmental data includes soil moisture and nutrient distribution. S2. Based on the soil environmental characteristics dataset, analyze the impact of soil moisture and nutrient levels on cadmium uptake rate and determine the high uptake risk range. S3. When there is a high absorption risk zone, track the changes in cadmium concentration from rice roots to stems and leaves to obtain transfer dynamic sequence data; S4. Based on the aforementioned transfer dynamic sequence data, analyze the movement speed and interaction characteristics of cadmium between different tissues to determine the bottleneck points in the transport channel; S5. Based on the bottleneck point, construct a transfer path simulation diagram by integrating sample data of rice physiological mechanisms, and obtain a distribution map of hindering factors; S6. Based on the distribution map of the hindering factors, analyze the interaction between soil characteristics and rice physiological mechanisms to determine the location and formation conditions of key nodes in cadmium accumulation. S7. Based on the location and formation conditions of the key nodes of cadmium accumulation, generate a full-cycle transfer law report to obtain the basis for control.

2. The method of claim 1, wherein, S1 includes: Raw data on soil moisture and nutrient distribution are collected through a sensor network. The raw data is then preprocessed to obtain cleaned soil environmental data. Soil samples were collected, and the total cadmium content, available cadmium content, and soil physicochemical parameters related to cadmium form transformation were detected as cadmium-related detection data. Rice root samples were collected, and quantitative characteristic data of root length, root density, and root activity were obtained as data on the growth status of rice roots. Based on the rice root growth status data, the soil environment data, and the cadmium-related detection data, cadmium solubility index and cadmium bioavailability index were determined. The soil environmental characteristics dataset is constructed by integrating the cadmium solubility index, the cadmium bioavailability index, the soil environmental data, the cadmium-related detection data, and the rice root growth status data.

3. The method of claim 1, wherein S2 include: Soil moisture distribution data and soil nutrient distribution data were extracted from the soil environmental characteristics dataset, and combined with cadmium absorption rate records to construct a comprehensive analysis base dataset. Based on the comprehensive analysis dataset, the correlation between soil moisture content, soil nutrient content and cadmium uptake rate was analyzed to determine the soil moisture range and soil nutrient range corresponding to high-risk areas. Real-time environmental characteristic data of field soil are obtained and compared with the soil moisture range and the soil nutrient range to determine whether there are abnormal areas of data. If abnormal data is found, the data corresponding to the abnormal area will be stratified according to spatial distribution and soil depth. Combined with the actual cadmium absorption rate records, the key monitoring area and its corresponding soil moisture range and soil nutrient range will be determined, and the key monitoring area will be identified as the high absorption risk zone.

4. The method of claim 1, wherein S3 include: On rice plants corresponding to the high absorption risk range, cadmium concentration change data from roots to stems and leaves were continuously collected using monitoring equipment to obtain initial transfer sequence data; Based on the initial transfer sequence data, the temporal correlation between cadmium absorption in roots and cadmium transfer in stems and leaves was analyzed, and the corresponding correlation characteristic parameters were determined. When the associated feature parameters exceed the preset parameter range, the initial transfer sequence data is segmented to determine the critical time window in which the cadmium concentration change rate changes abruptly. Extract cadmium concentration change data within the key time window and perform time-series feature extraction to generate the transfer dynamic sequence data.

5. The method according to claim 1, characterized in that, S4 includes: Based on the aforementioned transfer dynamic sequence data, the migration rate change parameters of cadmium in different tissues of rice were calculated to obtain the distribution dynamics of cadmium in each tissue. Based on the aforementioned distribution dynamics, high-value retention areas of cadmium concentration accumulated over time are identified, and the limiting locations of key limiting points are determined. For the restricted locations, extract the abrupt change features of the migration rate change parameters to determine the main obstruction intervals in the transportation path; Based on the main obstruction intervals, a migration restriction analysis model was constructed to analyze the degree of migration restriction of cadmium between adjacent tissues and to determine the corresponding migration restriction range. By combining the location of the key limiting points with the range of migration restrictions, local feature extraction is performed on the dynamic sequence data of the transfer to identify the bottleneck points in the transportation channel.

6. The method according to claim 1, characterized in that, S5 include: Based on the aforementioned bottlenecks, and combined with sample data on rice physiological mechanisms, the factors hindering the migration of cadmium between different tissues in rice were identified. Based on the aforementioned hindering factors and bottlenecks, a cadmium transfer path simulation map is constructed, and the distribution area of ​​the hindering factors in the transfer path simulation map is determined. Based on the distribution area, analyze the relationship between the hindering factors and cadmium migration between different tissues of rice, and determine the area of ​​action of the hindering factors; Based on the said area of ​​action, a stratified analysis was performed on the sample data of rice physiological mechanisms to determine the distribution differences of the hindering factors in different tissue levels. Based on the distribution area, the area of ​​effect, and the distribution differences, the distribution map of the hindering factors is generated.

7. The method according to claim 1, characterized in that, S6 include: Based on the aforementioned distribution map of hindering factors, combined with soil characteristic data and rice physiological mechanism sample data, the interaction characteristics of soil characteristics and rice physiological mechanisms on cadmium accumulation were analyzed. Based on the interaction characteristics, the distribution areas of key nodes for cadmium accumulation are identified, and the locations of the key nodes for cadmium accumulation are determined. Based on the locations of the key nodes in cadmium accumulation, the characteristics of the interaction on cadmium accumulation are analyzed, and the main areas of action corresponding to the key nodes in cadmium accumulation are determined. Based on the main areas of action, the soil characteristic data and rice physiological mechanism sample data were stratified and analyzed according to environmental variables to obtain the variation characteristics of cadmium accumulation under different environmental variables. Based on the aforementioned change characteristics, the formation conditions of the key nodes in cadmium accumulation were determined.

8. The method according to claim 1, characterized in that, S7 includes: Based on the location and formation conditions of the key nodes for cadmium accumulation, the cadmium accumulation data is processed in stages according to the rice growth period to obtain the accumulation status at different growth stages throughout the entire cycle, and to determine the distribution of key nodes for cadmium accumulation at each stage. Based on the distribution, a dynamic change model of cadmium transfer pathways is constructed to obtain the evolution trajectory of transfer pathways throughout the entire cycle. By combining environmental variable fluctuation data, correlation analysis was performed on the evolution trajectory of the transfer path to obtain the path change characteristics of cadmium accumulation under different conditions; Based on the path change characteristics, the cadmium absorption and transport patterns throughout the entire cycle are determined, and a full-cycle transfer pattern report is generated. Based on the report on the full-cycle transfer pattern, the basis for controlling cadmium absorption in rice was extracted.

9. A rice cadmium absorption prediction and control system, used to implement the method as described in any one of claims 1 to 8, characterized in that, The system includes: The data acquisition module is used to acquire field soil environmental data and, in combination with rice root growth status data, construct a soil environmental characteristic dataset containing cadmium solubility and cadmium bioavailability indicators. The soil environmental data includes soil moisture and nutrient distribution. The risk analysis module is used to analyze the impact of soil moisture and nutrient levels on cadmium uptake rate based on the soil environmental characteristics dataset, and to determine the high uptake risk range. The dynamic monitoring module is used to track changes in cadmium concentration from rice roots to stems and leaves and obtain dynamic sequence data of cadmium transfer when there is a high absorption risk zone. The bottleneck analysis module is used to analyze the movement speed and interaction characteristics of cadmium between different tissues based on the transfer dynamic sequence data, and to determine the bottleneck points in the transport channel; The path simulation module is used to construct a transfer path simulation map based on the bottleneck point and integrate sample data of rice physiological mechanisms, and to obtain a distribution map of hindering factors. The node determination module is used to analyze the interaction between soil characteristics and rice physiological mechanisms based on the distribution map of the hindering factors, and to determine the location and formation conditions of key nodes for cadmium accumulation. The report generation module is used to generate a full-cycle transfer pattern report based on the location and formation conditions of the key nodes of cadmium accumulation, thereby obtaining a basis for control.

Images