Ecological cycle agricultural data collection and analysis method and system based on internet of things

Abstract

This invention discloses a method and system for data acquisition and analysis in ecological circular agriculture based on the Internet of Things (IoT), belonging to the field of IoT data processing technology in ecological agriculture. It acquires multi-point sensor signals from planting areas, breeding areas, and circular processing units, aligning them according to a unified time reference to form a continuous time-series acquisition sequence. The continuous time-series acquisition sequence is then subjected to segmented response analysis to construct a set of discontinuous transition features. A micro-circulation transmission chain is generated according to the direction of material flow, calculating the synchronization degree and phase difference of transition responses between adjacent units to obtain a set of coupling imbalance indicators. Subsequently, the discontinuous transition features are progressively corrected to form an evolution trajectory, and the synergistic abnormal intervals of nutrient pulse superposition and dissolved oxygen decline are identified, generating hierarchical control instructions. This invention can be used for the identification and control of local imbalances in ecological circular agriculture.

Description

Technical Field

[0001] This invention relates to the field of data processing technology in ecological agriculture Internet of Things (IoT), specifically to a method and system for data collection and analysis in ecological circular agriculture based on IoT. Background Technology

[0002] In hilly distributed ecological circular agriculture systems, there are non-uniform microcirculation channels between planting areas, breeding areas, and organic processing units. Affected by rainfall infiltration, matrix pore structure, and local microbial activity differences, nutrients and dissolved oxygen exhibit sudden "lag-jump" alternating transfer phenomena on a centimeter scale. Existing IoT monitoring mainly relies on steady-state thresholds or single-point trend judgments, making it difficult to capture such short-term discontinuous jump behavior. Especially under the transient high-concentration pulse conditions formed in the early stage of manure reflux, a hidden coupling imbalance of nutrient enrichment and rhizosphere hypoxia amplification can easily occur. Such problems occur in a small area, last for a short time, and have complex cross-unit transmission paths, making it difficult for traditional methods to identify and intervene in a timely manner. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for data collection and analysis in ecological circular agriculture based on the Internet of Things, so as to solve the shortcomings of the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for data collection and analysis in ecological circular agriculture based on the Internet of Things, comprising: Multiple sensor signals from the planting area, breeding area, and recycling unit are acquired, aligned according to a unified time reference, and formed into a continuous time-series acquisition sequence. The continuous time-series acquisition sequence is subjected to segmented response analysis to identify numerical abrupt changes and hysteresis recovery segments, extract the corresponding change intensity and duration, and construct a non-continuous transition feature set; Based on the discontinuous transition feature set, the units are sequentially connected according to the material flow direction to generate a micro-circulation transmission chain. Using the micro-circulation transmission chain, the synchronization degree and phase difference of the transition response between adjacent units are calculated to obtain the coupling imbalance indicator set; Based on the aforementioned set of coupling imbalance indicators, the discontinuous transition characteristics are progressively corrected to form an evolution trajectory that reflects the trend of local imbalance expansion. Based on the aforementioned evolution trajectory, the synergistic abnormal intervals of nutrient pulse superposition and dissolved oxygen decline are identified, and corresponding hierarchical regulation instructions are generated.

[0005] Preferably, the data is aligned according to a unified time base to form a continuous time-series acquisition sequence, including: A local incrementing time identifier is added to the original records of each acquisition node, and invalid records are removed based on the stability of the interval between adjacent records to obtain the initial time series segment; The sampling node located upstream of the material flow path and where the effective fluctuation first appears is selected as the reference node. The initial time series segment of the reference node is used as a reference to perform segment-by-segment offset correction on the initial time series segments of the remaining sampling nodes to form a relatively aligned sequence segment. Missing segments in the relatively aligned sequence segments are progressively filled in, and the mutation boundaries are processed before being spliced ​​together in chronological order to form a continuous time-series acquisition sequence.

[0006] Preferably, segment-by-segment offset correction includes: In each initial time series segment, identify positions where 2 to 5 consecutive sampling points change in the same direction and the change exceeds the dispersion by 2 to 4 times. The first position that exceeds this position is taken as the effective fluctuation moment. Candidate time offsets are set within the range of negative 5 times the reference sampling interval to positive 5 times the reference sampling interval. The segments of the acquisition nodes to be corrected are advanced or delayed respectively, and the consistency of change direction and the difference of change amplitude are compared with the segments of the reference node in the same time range. Candidate time offsets with consistent change direction, large number, and small cumulative difference in change magnitude are selected as segmented correction values, and the difference in correction values ​​between adjacent segments is progressively transitioned to no more than one reference sampling interval.

[0007] Preferably, a discontinuous transition feature set is constructed, including: A sliding window is set along the continuous time-series acquisition sequence. Based on the fluctuation of the change in adjacent sample values ​​within the window, the window is marked as a stable window or a transition window, and then merged to form an initial segmented sequence. The transition segments in the initial segmented sequence are refined by boundary refinement. By comparing the point-by-point change gradients of the transition segments with those of the preceding and following stable segments, the mutation start and end points are determined, forming mutation response fragments. In the stable region following the mutation response fragment, identify the intervals where the deviation decreases continuously, determine the recovery start position and the end position of the regression persistence interval, and form the hysteresis recovery fragment.

[0008] Preferably, the formation of the discontinuous transition feature set includes: The magnitude and duration of the changes in the mutation response segment and the hysteresis recovery segment are calculated separately, and the magnitude of the changes is compared with the median of the effective magnitudes of the corresponding acquisition nodes in the most recent 1 hour to 24 hours to obtain the intensity of the change. When a hysteresis recovery fragment appears within 1 to 20 reference sampling intervals after the mutation response fragment, the two are combined into a discontinuous transition unit; Multiple discontinuous transition units are sorted according to a unified time identifier, and the data acquisition node identifier, mutation start point, mutation end point, recovery start position, regression duration end position, change intensity, duration and change direction are recorded to construct a discontinuous transition feature set.

[0009] Preferably, generating a microcirculation transport chain includes: Based on the transition starting point of each acquisition node in the discontinuous transition feature set, and combined with the preset material flow path and the allowable transmission time range of adjacent nodes, the upstream node and the candidate downstream node are matched in time sequence to form an initial node connection sequence. The temporal overlap and sequential conflict in the initial node connection sequence are corrected, and the connection relationships that are directly adjacent in the material flow path, have the first appearance of the transition response and the relatively stable decay of the change intensity are retained first, so as to obtain an ordered transmission node sequence. Based on the ordered transmission node sequence, the temporal continuity, amplitude continuity and recovery continuity of adjacent nodes are screened, and node pairs that meet the transmission continuity are connected in series to form a micro-circulation transmission chain.

[0010] Preferably, the set of coupling imbalance indicators is obtained, including: The discontinuous transition characteristics of adjacent units are read along the micro-circulation transmission chain. The transition start point and recovery end point of the upstream unit are paired with the corresponding positions of the downstream unit to form an adjacent response pairing sequence. For each response pair in the adjacent response pairing sequence, calculate the time offset and determine the response overlap interval based on the common coverage of the upstream and downstream response intervals; By combining time offset, response overlap range, difference in change intensity, and difference in recovery duration, the synchronization level is classified into high, medium, and low, and response pairs with excessive time offset, insufficient response overlap, or low synchronization level are identified as coupling imbalance indicator sets.

[0011] Preferably, the evolutionary trajectory includes: Based on the time offset, synchronization and imbalance of abnormal nodes in the coupling imbalance indicator set, the correction priority value is determined, and the discontinuous transition units with a correction priority value of not less than 3 are identified as the set of critical transition segments. The key transition segment set is corrected step by step along the micro-circulation transmission chain. The transition starting point of downstream segments that appear early and recover late is adjusted. The amplitude of segments with deviation in change intensity is compensated or reduced. Delay marks are added to segments with abnormally long recovery duration to obtain the corrected transition sequence. By comparing the number of anomalous nodes, the total duration of anomalous events, and the coverage of change intensity in adjacent stages of the modified transition sequence, the direction and rate of expansion are determined, and the evolution trajectory is formed by continuously associating them in chronological order.

[0012] Preferably, generating corresponding graded control instructions includes: extracting nutrient concentration increase segments and dissolved oxygen decrease segments along the evolution trajectory, determining the common coverage portion or delayed synergistic portion of the response intervals of the two as synergistic response intervals, and forming a synergistic response interval sequence; determining the first, second, or third level of abnormal intensity based on the relative intensity of the nutrient concentration increase segment, the relative intensity of the dissolved oxygen decrease segment, the response overlap ratio, and the duration of the synergistic response interval, and forming a graded control strategy sequence in combination with the node positions in the microcirculation transmission chain, and converting the graded control strategy sequence into control signals for irrigation rhythm, circulation channels, and aeration intensity.

[0013] This invention provides an IoT-based ecological circular agriculture data acquisition and analysis system, comprising: Multi-source acquisition and alignment module: Acquires multi-point sensor signals from planting area, breeding area and recycling unit, aligns them according to a unified time reference, and forms a continuous time-series acquisition sequence; Segmented response analysis module: Performs segmented response analysis on the continuous time-series acquisition sequence, identifies numerical abrupt changes and hysteresis recovery segments, extracts the corresponding change intensity and duration, and constructs a non-continuous transition feature set; The transfer chain construction module: Based on the discontinuous transition feature set, the units are sequentially connected according to the material flow direction to generate a micro-circulation transfer chain; Coupling Imbalance Identification Module: Utilizing the micro-circulation transmission chain, the synchronization degree and phase difference of the transition responses between adjacent units are calculated to obtain a coupling imbalance indicator set; Progressive Correction Analysis Module: Based on the coupling imbalance indicator set, progressively correct the discontinuous transition characteristics to form an evolution trajectory reflecting the local imbalance expansion trend; Graded regulation output module: Based on the evolution trajectory, it identifies the synergistic abnormal intervals of nutrient pulse superposition and dissolved oxygen decline, and generates corresponding graded regulation instructions.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention differs from existing technologies that judge agricultural environmental anomalies based on single-point thresholds or general trends. Its core technical approach lies in: first, aligning dispersed data collection nodes to a unified time reference; then, identifying discontinuous transition characteristics of numerical mutations followed by delayed recovery from continuous time-series data collection. This processing does not simply summarize sensor data, but preserves transient fluctuations, recovery delays, and their durations. This allows short-duration pulses, sensor time differences, and local response lags in the initial stages of nutrient reflux to be compared on the same time scale, thereby improving the accuracy of identifying localized, hidden anomalies.

[0015] This invention further connects discontinuous transition characteristics into a micro-circulation transmission chain according to the direction of material flow, and performs progressive corrections based on the degree of synchronization, phase difference, and coupling imbalance indicators between adjacent units, forming a local imbalance expansion trend. This invention differs from existing technologies that analyze planting or breeding area data in isolation. It can link the transmission relationship between nutrient concentration increases and dissolved oxygen decreases between different units, identify synergistic abnormal intervals, and generate corresponding irrigation, circulation channel, and aeration graded control instructions, thereby achieving phased intervention in local imbalances of the ecological cycle chain. (See attached figures.) To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0016] Figure 1 This is a flowchart of the method of the present invention.

[0017] Figure 2 This is a flowchart of the system modules of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1, please refer to Figure 1 As shown in this embodiment, the data collection and analysis method for ecological circular agriculture based on the Internet of Things includes: Multiple sensor signals from the planting area, breeding area, and recycling unit are acquired and aligned according to a unified time reference to form a continuous time-series acquisition sequence.

[0020] In the ecological circular agriculture scenario, data acquisition nodes are deployed in the planting area, breeding area, and recycling unit. These nodes record at least one multi-point sensor signal from among temperature, humidity, pH, conductivity, dissolved oxygen, nutrient concentration, and circulation flow rate. After completing a single data acquisition, each node adds a locally incrementing time stamp to the original record.

[0021] The local incremental time identifier is based on the cumulative time since the data collection node started. The time identifier of the later record is greater than the time identifier of the previous record. When there is a time rollback, duplicate numbering, or a record with an interval of 0, the record is marked as an invalid record and will not participate in the subsequent alignment process.

[0022] After attaching a locally incrementing time identifier to the raw records of each acquisition node, the actual sampling interval between adjacent records is calculated. Based on multiple actual sampling intervals of the same acquisition node within a preset observation period, the median is used as the reference sampling interval for that acquisition node.

[0023] The preset observation period can be 10 to 60 minutes, or it can be the most recent 50 to 300 consecutive records. The stability of the interval between adjacent records is determined by the deviation between the actual sampling interval and the reference sampling interval. The deviation is the ratio of the absolute value of the difference between the two to the reference sampling interval. The acceptable range for interval stability is a deviation of no more than 0.05 to 0.20, preferably no more than 0.10. When there is wireless transmission jitter in the field, the acceptable range can also be set to no more than 3 times the median of the historical interval deviation of the acquisition node.

[0024] A valid time segment consists of records that continuously meet the interval stability requirements. The minimum length of a valid time segment is no less than 5 sampling points, and the duration is no less than 4 times the reference sampling interval, thereby avoiding the misidentification of a single sporadic record as the initial time series segment.

[0025] After obtaining the valid time segments, the sampled values, acquisition node identifiers, and locally incrementing time identifiers within each valid time segment are extracted in chronological order to obtain the initial time series segment. The initial time series segment retains the original sampling order and does not change the sampled values; only records with time backtracking, duplicate numbers, and unqualified interval stability are removed.

[0026] In cases where multiple valid time segments exist at the same acquisition node, the segments are arranged in chronological order of their start times, and the discontinuities between segments are preserved for subsequent identification of missing segments.

[0027] The reference node is determined from the acquisition node located upstream of the loop path where the fluctuation first occurs.

[0028] The upstream of the circulation path is determined according to the direction of material flow. For example, the discharge end of the aquaculture area, the outlet end of the fermentation liquid, or the outlet end of the circulation treatment unit can be used as the upstream location.

[0029] The first occurrence of fluctuations is determined by comparing the first valid fluctuation moment in each initial time series segment. The process for determining the valid fluctuation moment is as follows: first, calculate the median value of the changes in adjacent sampled values ​​during the stable period of the acquisition node; then calculate the dispersion of each change relative to the median value; when the changes in 2 to 5 consecutive sampled points all exceed 2 to 4 times the dispersion and the direction of change is consistent, the time marker corresponding to the first exceeding position is taken as the valid fluctuation moment. If multiple upstream acquisition nodes simultaneously meet the conditions, the acquisition node closest to the exit of the cyclic processing unit or with the most stable sampling interval is selected as the reference node.

[0030] After determining the reference node, the initial time series segment of the reference node is used as a reference to perform segment-by-segment offset correction on the initial time series segments of other acquisition nodes.

[0031] Segmented offset correction employs segmented processing, with each segment ranging from 30 to 300 seconds in length, or from 10 to 80 sampling points; adjacent segments can be set with an overlap length of 10% to 50% to preserve continuous variations at segment boundaries.

[0032] For any node to be corrected, multiple candidate time offsets are tried sequentially within a preset offset range. The preset offset range can be from -5 times the reference sampling interval to +5 times the reference sampling interval. When the on-site transmission delay is large, it can be extended to -30 seconds to +30 seconds.

[0033] Each candidate time offset is used, which means advancing or delaying the time identifier of the segment to be corrected, and then comparing it with the segment of the reference node within the same time range.

[0034] When comparing, first count the number of sampling points with the same direction of change in the two segments, and then count the cumulative value of the difference in the magnitude of change of the sampling values ​​near the same time. Select the candidate time offset with a large number of points with the same direction of change and a small cumulative value of the difference in the magnitude of change as the correction amount for the segment.

[0035] When adjacent segments have different correction values, the segment boundaries are adjusted in a progressive manner to ensure a smooth transition between the correction value at the end of the previous segment and the correction value at the beginning of the next segment.

[0036] If the difference between the correction values ​​of adjacent segments is no greater than one reference sampling interval, the correction values ​​are gradually increased or decreased according to the time sequence; if the difference is greater than one reference sampling interval, the effective fluctuation time of the corresponding segment is re-examined, and the offset results caused by communication interruption or short-term sensor abnormality are eliminated.

[0037] After completing the segment-by-segment offset correction, each acquisition node forms a relatively aligned sequence segment. The time stamps in the relatively aligned sequence segment are all referenced to the time of the reference node, and the sampled values ​​still retain the original recorded values ​​of each acquisition node.

[0038] When missing segments exist in a relatively aligned sequence segment, progressive padding is performed based on the continuity of changes before and after the segment. Missing segments are determined by the time difference between adjacent records. When the time difference between adjacent records is greater than 1.5 to 3 times the reference sampling interval, the time range between them is considered a missing segment. If the duration of the missing segment does not exceed 10 reference sampling intervals, and the direction of change of the sample values ​​before and after the missing segment is consistent, padding values ​​are allocated point-by-point in chronological order according to the magnitude of change between the previous and next valid sample values. If the direction of change before and after the missing segment is inconsistent, the valid records closest to both ends of the missing segment are used as constraints to ensure a gradual transition of padding values ​​between the two ends, and the amount of change at a single point is limited to no more than twice the median value of the changes in adjacent valid records.

[0039] If the duration of the missing segment exceeds 10 reference sampling intervals, only 1 to 3 sampling points at each end of the missing segment are added, while the missing marker in the middle is retained to avoid introducing uncertain trends.

[0040] A smooth transition process is applied to the abrupt change boundaries of relatively aligned sequence segments. Abrupt change boundaries are identified by the change in adjacent sample values. A boundary is defined as a point where a single change exceeds 3 to 6 times the median of recent changes at that node, and there are at least 3 consecutive sample points before and after the abrupt change. The smooth transition process applies only to 1 to 3 sample points on either side of the abrupt change boundary, without altering its start and end points or the overall direction of change.

[0041] During processing, the effective sampled values ​​before and after the mutation boundary are used as endpoints. The sampled values ​​near the boundary are adjusted to transitional values ​​that gradually approach the endpoints, and the difference between the total change after adjustment and the original mutation amplitude does not exceed 10%. After missing fragment completion and mutation boundary smoothing processing, a continuous transition sequence segment is obtained.

[0042] Finally, the continuous transition sequence segments of each node are spliced ​​together in time order according to a unified time base.

[0043] During stitching, the reference node time identifier is used as the main axis, and records from each acquisition node that are near the same time position are grouped into the same acquisition time; the allowable grouping error is no more than 0.5 times the reference sampling interval. For overlapping records, records with higher interval stability and fewer missing markers are retained first; when two records have the same stability, the position closer to the unified time reference is taken. For short discontinuities that still exist, they are supplemented according to the supplementation rules in the continuous transition sequence segment; for long discontinuities, the discontinuity marker is retained and the start and end times of the discontinuity are recorded in the continuous time-series acquisition sequence. After overlap elimination, discontinuity processing, and time sequence arrangement, a continuous time-series acquisition sequence is formed. This continuous time-series acquisition sequence simultaneously contains acquisition node identifiers, unified time identifiers, sensor values, supplementation markers, and discontinuity markers, which can be used for subsequent segmented response analysis.

[0044] The continuous time-series acquisition sequence is subjected to segmented response analysis to identify numerical abrupt changes and hysteresis recovery segments, extract the corresponding change intensity and duration, and construct a non-continuous transition feature set.

[0045] When a continuous time-series acquisition sequence is entered into segmented response analysis, a unified time identifier is used as the arrangement basis, and the sensor values ​​of each acquisition node are processed separately. Sliding scanning uses a fixed-length window that moves point-by-point along the time sequence. The window length can be 5 to 30 sampling points, preferably the number of sampling points within 1 to 5 minutes corresponding to the reference sampling interval; the movement step size of adjacent windows can be 1 to 5 sampling points. When the sampling frequency is low, the window length should be no less than 5 sampling points to ensure continuous basis for segment determination.

[0046] Within each window, the change in adjacent sampled values ​​is calculated. The change is the absolute value of the difference between the previous and subsequent sampled values. Then, the median value of each change within the window is calculated as the baseline change level within the window.

[0047] The continuity of change within the window is determined by the degree of fluctuation between adjacent changes, which is the ratio of the absolute value of the difference between two adjacent changes to the baseline level of change.

[0048] When the baseline change level is 0, the median value of the historical effective change at that data collection node is used instead; if it is still 0, the fluctuation level is recorded as 0. If at least 70% of the fluctuation levels within the same window are less than 0.30 to 0.80, and the difference between the first and last sampled values ​​of the window does not exceed 2 to 4 times the baseline change level, then the window is marked as a stable window; if at least 30% of the fluctuation levels within the same window are greater than 0.80 to 2.50, or the difference between the first and last sampled values ​​of the window exceeds 4 to 8 times the baseline change level, then the window is marked as a transitional window.

[0049] Windows with consecutive identical markings are merged to form segments. Adjacent stable windows are merged into stable segments, and adjacent transitional windows are merged into transitional segments. When a single transitional window is sandwiched between two stable segments, and the direction of change of the transitional window is inconsistent with both the preceding and following stable segments, the transitional window is discarded as an isolated fluctuation and merged into a stable segment with a closer time interval.

[0050] The minimum length for a stable segment is 3 windows, and the minimum length for a transition segment is 2 windows. Segments shorter than the minimum length are merged according to the duration, direction of change, and boundary change of adjacent segments. After the above processing, an initial segmented sequence is formed, which includes stable segments and transition segments arranged by time, and retains the unified start time identifier, unified end time identifier, start sample value, and end sample value of each segment.

[0051] When refining the boundary of the transition segment in the initial segmented sequence, the gradient of change is first calculated in the transition segment and one stable segment before and after it.

[0052] The point-by-point gradient is the ratio of the change in adjacent sampled values ​​to the interval between adjacent unified time markers; the time interval is in seconds. The average gradient of the first segment is equal to the average of the point-by-point gradients of at least 3 sampled points before the transition segment, and the average gradient of the second segment is equal to the average of the point-by-point gradients of at least 3 sampled points after the transition segment. The point-by-point gradient within the transition segment is compared one by one with the average gradient of the first segment and the average gradient of the second segment.

[0053] The mutation starting point is searched from the front end of the transition section backward. If the point-by-point change gradient at a certain position is 3 to 10 times greater than the average change gradient of the preceding section, and the subsequent 2 to 5 positions continue to change in the same direction, then that position is considered a candidate mutation starting point; when there are multiple candidate mutation starting points, the position with the earliest time and the change direction consistent with the overall change direction of the transition section is selected.

[0054] The mutation endpoint is searched backward from the end of the transition segment. If the pointwise gradient after a certain position decreases to within 1 to 2 times the average gradient of the subsequent segment, and remains consistent for 2 to 5 positions, then that position is considered a candidate mutation endpoint. When multiple candidate mutation endpoints exist, the position with the shortest duration between it and the candidate mutation initiation point, and which covers the main amplitude changes, is selected. The sampling segment between the candidate mutation initiation point and the candidate mutation endpoint forms a mutation response segment. If the amplitude change between the initiation point and the endpoint is less than 3 times the baseline change level of the adjacent stable segment, then the mutation response segment is removed from the label to prevent ordinary disturbances from being included in the discontinuous transition feature set.

[0055] The hysteresis recovery segment is extended from the stable region for identification. For the stable region located after the mutation response segment, the deviation of the sampled values ​​within the stable region from the baseline value before the mutation is first calculated.

[0056] The baseline value before the mutation is the median value of the sampled values ​​in the most recent stable segment before the start of the mutation response segment. The deviation is the absolute value of the difference between the current sampled value and the baseline value before the mutation, divided by the amplitude of the change in the mutation response segment.

[0057] If the deviation at the beginning of the stable segment is not less than 0.20 to 0.60, and then decreases continuously in chronological order, then a delayed regression characteristic is identified.

[0058] The recovery starting position is the position where the deviation first continuously decreases; the regression duration is the position from the recovery starting position until the deviation drops to within 0.05 to 0.20 and remains within 3 to 10 sampling points.

[0059] If the deviation increases briefly during the decrease, and the increase does not exceed 10% to 25% of the amplitude of the change in the mutation response fragment, the same regression duration interval is retained. If it exceeds this range, the position of the increase is taken as the regression interruption position, and subsequent extended identification is performed again. The hysteresis recovery fragment is formed from the recovery start position to the end position of the regression duration interval.

[0060] The magnitude and duration of the change were extracted from the mutation response fragment and the hysteresis recovery fragment, respectively.

[0061] The change amplitude of the mutation response fragment is the absolute value of the difference between the sampled value at the end of the fragment and the sampled value at the beginning of the fragment, and the duration is the time length obtained by subtracting the unified time identifier at the beginning of the fragment from the unified time identifier at the end of the fragment.

[0062] The change amplitude of the lag recovery segment is the absolute value of the difference between the sampled value at the recovery start position and the sampled value at the end position of the regression duration interval, and the duration is the time length obtained by subtracting the unified time identifier of the recovery start position from the unified time identifier of the regression duration interval end position.

[0063] To avoid the influence of different sensor dimensions on subsequent comparisons, the change amplitude can be divided by the median of the effective amplitude of the corresponding acquisition node in the most recent 1 to 24 hours to obtain the normalized change intensity. When the median of the effective amplitude is 0, the range of at least 50 effective sample values ​​of the acquisition node is used as a substitute. When the range is still 0, the segment does not participate in the intensity normalization process and only the original change amplitude is retained.

[0064] The discontinuous transition feature set is combined and labeled according to the temporal correlation order. For the same acquisition node, if a hysteresis recovery segment appears within 1 to 20 reference sampling intervals after the mutation response segment, the two form a discontinuous transition unit; if multiple hysteresis recovery segments correspond to the same mutation response segment, the segment with the recovery start position closest to the mutation end point and the longest regression duration interval is selected first.

[0065] Each discontinuous transition unit records the acquisition node identifier, mutation start point, mutation end point, recovery start position, regression duration end position, change intensity, mutation duration, recovery duration, change direction, and fragment source segment. Multiple discontinuous transition units are sorted according to a unified time identifier to form a discontinuous transition feature set, which is used for subsequent generation of microcirculation transmission chains according to the material flow direction.

[0066] Based on the discontinuous transition feature set, the units are sequentially connected according to the material flow direction to generate a micro-circulation transmission chain.

[0067] When generating a micro-circulation transmission chain based on a discontinuous transition feature set, the mutation start point, mutation end point, recovery start position, regression duration end position, change intensity, mutation duration, recovery duration, and change direction of each acquisition node in the discontinuous transition feature set are first read.

[0068] The preset material flow path includes the positional order of each sampling node in the planting area, breeding area, and circulating treatment unit, as well as the time range within which transitions are allowed between adjacent nodes. The time range within which transitions are allowed can be set based on the residence time of the circulating liquid between adjacent nodes, or based on the median value of the confirmed transition time difference between adjacent nodes within the same operating day; the value range is generally 1 to 60 reference sampling intervals, with 1 to 15 reference sampling intervals for shorter circulating branches, and 10 to 60 reference sampling intervals for longer branches or branches with packing material retention.

[0069] Temporal matching is performed from upstream nodes to downstream nodes according to the direction of material flow. For any discontinuous transition unit of an upstream node, the time difference between the transition start point of the candidate downstream node and the transition start point of the upstream node is calculated; if the time difference is positive and falls within the time range during which transitions can occur between adjacent nodes, it is retained as a candidate connection.

[0070] If there are multiple discontinuous transition units in the candidate downstream node, the direction and intensity of their changes are also compared.

[0071] The directional condition is met when the changes are in the same direction, or when the increase in nutrient concentration and the decrease in dissolved oxygen are related by a pre-defined synergistic relationship. The intensity condition is met when the ratio of the change in intensity falls between 0.20 and 5.00. Candidate connections that meet both the directional and intensity conditions are sorted according to how close their time difference is to the median of the allowable time range, and the top-ranked candidate connections are written into the initial node connection sequence. A maximum of two candidate connections are retained for an upstream discontinuous transition unit to handle diversion or bypass backflow scenarios.

[0072] After the initial node connection sequence is formed, the connections that have time overlap or sequence conflict are corrected.

[0073] Time overlap refers to the simultaneous connection of the same downstream discontinuous transition unit to two or more upstream discontinuous transition units, and the time difference corresponding to each connection relationship falls within the allowable range.

[0074] During processing, the positions of upstream nodes in the material flow path are first compared, prioritizing connections directly adjacent to downstream nodes. If all are directly adjacent, the transition start times are compared, retaining the earlier-appearing connection with a more stable attenuation of change intensity. The method for determining a more stable attenuation of change intensity is as follows: the intensity difference is obtained by subtracting the downstream change intensity from the upstream change intensity; the smaller the absolute value of the intensity difference, the more stable the attenuation. When the absolute values ​​of the intensity differences are the same, the connection with the smaller duration difference is selected for restoration.

[0075] Sequence conflict refers to a connection relationship where the transition start point of a downstream node is earlier than that of an upstream node, or the transition start point of a later-level node is earlier than the corresponding transition end point of a previous-level node.

[0076] For sequence conflicts, first check if the connection originates from a continuous transition sequence segment with a large number of completion markers. If the number of completion markers exceeds 30% of the number of sampling points in the segment, delete the connection. If the number of completion markers does not exceed 30%, compare the leading and lagging aspects of the transition response. Leading is determined based on the order of the mutation start points, and lagging is determined based on the order of the recovery start positions. When a node's mutation start point is earlier than its adjacent upstream node, but its recovery start position is significantly later than the upstream node, and the time difference does not exceed three reference sampling intervals, the node is considered a boundary early response node, the connection is retained, and its ordering position is postponed to after the upstream node. After correction, the connections are rearranged according to the direction of material flow, the order of mutation start points, and the order of recovery start positions to obtain an ordered sequence of transmission nodes.

[0077] Based on the ordered sequence of nodes, the continuity of transmission consistency of transition characteristics between adjacent nodes is calculated. Continuous transmission consistency is determined by temporal continuity, amplitude continuity, and recovery continuity. Temporal continuity is obtained by the closeness of the time difference between the transition start points of adjacent nodes and the median value of the allowable time range; the closer the difference is to the median value, the higher the temporal continuity. Amplitude continuity is obtained by the difference between the intensity of upstream and downstream changes; the amplitude continuity requirement is met when the difference does not exceed 20% to 80% of the upstream change intensity. Recovery continuity is obtained by the time difference between the upstream recovery start position and the downstream recovery start position; the recovery continuity requirement is met when this time difference is positive and does not exceed 1.5 times the upper limit of the corresponding allowable time range. For the synergistic relationship between the increase in nutrient concentration and the decrease in dissolved oxygen, amplitude continuity does not require the same direction, but it requires that the trends of the increase and decrease in the intensity of the two changes are consistent, that is, when the intensity of upstream nutrient change increases, the intensity of downstream dissolved oxygen decrease is not lower than the median level of the previous adjacent period.

[0078] When screening valid transitive connection pairs, the connection relationship between adjacent nodes needs to simultaneously satisfy temporal continuity and restorative continuity, and satisfy either amplitude continuity or cooperative change relationship.

[0079] If a single node connects to multiple downstream nodes, they are sorted from highest to lowest according to their continuous transmission consistency. If the continuous transmission consistency is the same, the downstream node closest in the material flow path is prioritized. Continuous transmission consistency can be defined in three levels: high if time continuity, amplitude continuity, and recovery continuity are all satisfied; medium if any two of these are satisfied; and low if fewer than two are satisfied. High and medium-level connections are included in the effective transmission connection set, while low-level connections are not included but are retained in the process record for subsequent verification.

[0080] Effective transfer links are connected in series according to the direction of material flow. When connecting in series, start from the upstream effective transfer link and compare the upstream node of the next effective transfer link with the downstream node of the previous effective transfer link. If they are the same and the transition start order satisfies the condition that the upstream node is earlier than the downstream node, then they are connected as the same transfer link segment.

[0081] If branches occur, they are formed into parallel chain segments, and the branch start point, downstream node of the branch, and the continuity consistency level of each branch are recorded. If a chain segment lacks a valid transmission connection pair in the middle, but the duration of the gap does not exceed the upper limit of two allowable time ranges, and the direction of change or cooperative change relationship before and after the gap remains consistent, the chain segments at both ends of the gap can be connected, and the discontinuity position is marked in the chain segment; if it exceeds this range, the chain segment terminates.

[0082] After the above sequential connection, conflict correction, consistency screening, and chain segment concatenation, a micro-loop transmission chain is formed. The micro-loop transmission chain records the node sequence, the connection relationship between adjacent nodes, the corresponding discontinuous transition units, the transition start time difference, the recovery start time difference, the change intensity relationship, the continuous transmission consistency level, and the possible branch or discontinuity positions.

[0083] In this embodiment, the micro-circulation transfer chain is used to characterize the actual transfer path and correlation of discontinuous transitions along the material flow direction between the planting area, the breeding area and the recycling unit, and serves as the data basis for subsequent calculation of the synchronization degree and phase difference between adjacent units.

[0084] Using the micro-circulation transmission chain, the synchronization degree and phase difference of the transition response between adjacent units are calculated to obtain the coupling imbalance indicator set.

[0085] The transition response analysis between adjacent units takes the micro-circulation transmission chain as the input object and processes it segment by segment according to the node order recorded in the chain.

[0086] Adjacent units are upstream and downstream units directly connected in the microcirculation transmission chain. The discontinuous transition characteristics corresponding to each unit include the transition start point, mutation end point, recovery start position, regression duration end position, change intensity, change direction, mutation duration, and recovery duration.

[0087] When extracting discontinuous transition features corresponding to adjacent units along the micro-circulation transmission chain, the connection relationship between adjacent nodes is used as an index to read the discontinuous transition units within the same transmission chain segment of the upstream and downstream nodes. The transition start point is used to indicate the start time of the mutation, and the recovery endpoint is represented by the end position of the regression persistence interval.

[0088] During pairing, the transition start point of the upstream unit is used as the reference, and the transition start point in the downstream unit is searched within the allowable transmission time range. When the downstream transition start point is later than the upstream transition start point, and the time difference between the two does not exceed the upper limit of the allowable time range of the corresponding connection relationship, a candidate response pairing is formed.

[0089] If multiple candidate objects exist for the same downstream unit, objects with time differences close to the median of historical time differences for that connection relationship are prioritized; if the degree of closeness is the same, objects with the same direction of change or conforming to a cooperative change relationship are selected. After pairing, they are arranged in the order of the upstream transition starting points to form an adjacent response pairing sequence.

[0090] For each response pair in an adjacent response pairing sequence, an alignment comparison of the transition times is performed. The time offset is determined by the delay between the downstream unit's transition start point and the upstream unit's transition start point, and the unit can be seconds or minutes.

[0091] To facilitate comparison at different sampling frequencies, the delay time can be divided by the duration of the upstream unit from the transition start point to the recovery end point to obtain the relative offset ratio.

[0092] A relative offset ratio less than 0 indicates that the downstream response occurs earlier; a ratio greater than 1 indicates that the downstream response occurs after the upstream recovery endpoint. The normal range of the relative offset ratio can be determined based on historical stable operating days, with a limit value set at twice the dispersion of the median historical relative offset ratio above and below it; if historical data is unavailable, it can be set to 0 to 1.5.

[0093] The response overlap interval is obtained by comparing the response intervals of the upstream and downstream units. The response interval starts from the transition start point and ends at the recovery end point.

[0094] When two response intervals share common coverage under a unified time reference, the portion of the common coverage is designated as the response overlap interval; when there is no common coverage, the response overlap interval is designated as empty. The duration of the overlap is the time difference between the start and end of the portion of common coverage.

[0095] The overlap ratio is determined by the ratio of the overlap duration to the shorter duration of the two response intervals. An overlap ratio between 0.40 and 1.00 indicates that adjacent cells have a basis for synchronous response in time; an overlap ratio below 0.20 indicates significant separation in the responses of adjacent cells; and a ratio between 0.20 and 0.40 requires further evaluation in conjunction with the time offset. The time offset, relative offset ratio, response overlap interval, and overlap ratio together constitute the time series offset characteristic set.

[0096] After obtaining the time-series offset feature set, the degree of response coordination is graded and labeled based on the consistency of transition amplitude changes. The consistency of transition amplitude changes is determined from three aspects: change direction, difference in change intensity, and difference in recovery rhythm. Consistent change direction means that adjacent units have the same change direction; for the combination of nutrient concentration increase and dissolved oxygen decrease, it is treated as a coordinated change relationship. Difference in change intensity is determined by comparing the upstream and downstream change intensities. When the downstream change intensity is between 0.30 and 3.00 times the upstream change intensity, it is considered comparable in amplitude transmission; when it exceeds this range, it is considered a deviation in amplitude. Difference in recovery rhythm is determined by comparing the upstream and downstream recovery durations. When the difference between the two does not exceed 0.50 to 2.00 times the shorter recovery duration, the recovery rhythms are considered close.

[0097] The level of response coordination can be categorized into three levels: high, medium, and low. A high level indicates that the time offset is within the normal range, the overlap ratio is not less than 0.40, the direction of change is consistent or conforms to a coordinated change relationship, and the difference in change intensity meets the amplitude propagation comparability condition. A medium level indicates that the time offset slightly exceeds the normal range but does not exceed 1.5 times the upper limit of the normal range, or the overlap ratio is between 0.20 and 0.40, and the direction of change or coordinated change relationship is established.

[0098] When the time offset exceeds 1.5 times the upper limit of the normal range, the overlap ratio is less than 0.20, the direction of change is inconsistent and does not belong to a cooperative change relationship, or the change intensity exceeds the comparable range of amplitude propagation, it is marked as low. The cooperative level, corresponding time offset, and amplitude consistency results of each response pair are summarized in chronological order to form a synchronization degree characterization set.

[0099] The coupling imbalance indicator set is obtained by screening response pairs with significant deviations. Significant deviations are determined by meeting at least one of the following conditions: the time offset exceeds 1.5 times the upper limit of the normal range; the relative offset ratio is greater than 1.5 or less than 0; the overlap ratio is less than 0.20; the coordination level in the synchronization characterization set is low; the downstream change intensity exceeds the upstream change intensity by 3.00 times and the recovery duration is extended by more than 2.00 times; nutrient concentration increase and dissolved oxygen decrease occur simultaneously, and the overlap ratio of their responses is not less than 0.30. For response pairs that meet the conditions, the upstream unit, downstream unit, transition start point, recovery end point, time offset, response overlap interval, coordination level, and reason for deviation are recorded.

[0100] The degree of imbalance can be categorized as mild, moderate, and severe. A mild imbalance is defined as a single deviation with downstream recovery duration not exceeding 1.50 times the previous value. A moderate imbalance is defined as two deviations, or a low synergy level with downstream change intensity exceeding upstream change intensity by 2.00 times. A severe imbalance is defined as three or more deviations, or an overlap between nutrient concentration increase and dissolved oxygen decrease that persists for more than 5 to 20 reference sampling intervals.

[0101] After screening and labeling, a set of coupling imbalance indicators is obtained, which can be used to progressively correct the discontinuous transition characteristics and to identify the key nodes of abnormal transmission in the microcirculation transmission chain.

[0102] Based on the aforementioned set of coupling imbalance indicators, the discontinuous transition characteristics are progressively corrected to form an evolution trajectory that reflects the trend of local imbalance expansion.

[0103] When progressively correcting discontinuous transition characteristics based on the coupling imbalance indicator set, the abnormal nodes are first processed. The time offset, relative offset ratio, response overlap interval, synchronization level marker, imbalance level marker, and associated discontinuous transition units are read for each abnormal node. Each abnormal node maintains the same position as the node in the micro-circulation transmission chain, without changing the existing node order, thus preventing the correction process from deviating from the material flow direction.

[0104] During the tiered screening process, a correction priority value is calculated for abnormal nodes. This correction priority value is obtained by combining the time offset contribution, synchronization contribution, and imbalance contribution. The time offset contribution is determined by the degree to which the time offset exceeds the normal range: 0 for not exceeding the normal range, 1 for exceeding the upper limit of the normal range but not more than 1.5 times, 2 for exceeding 1.5 times but not more than 2.5 times, and 3 for exceeding 2.5 times. The synchronization contribution is determined by the degree of synchronization: 0 for high, 1 for medium, and 2 for low. The imbalance contribution is determined by the degree of imbalance: 1 for mild, 2 for moderate, and 3 for severe. The correction priority value is obtained by adding these three values ​​together.

[0105] Abnormal nodes with a correction priority value of 5 or higher are moved to the first correction level, abnormal nodes with a correction priority value of 3 to 4 are moved to the second correction level, and abnormal nodes with a correction priority value of 1 to 2 are moved to the third correction level. The discontinuous transition units corresponding to the first and second correction levels constitute the critical transition segment set. When multiple abnormal nodes exist at the same acquisition node in adjacent time periods, the segment with the earliest time and the highest degree of imbalance is retained, and the remaining segments are processed as auxiliary segments.

[0106] Once the set of critical transition segments is determined, stepwise correction is performed along the micro-circulation transmission chain. The correction direction starts from the upstream anomalous node and proceeds sequentially to the downstream adjacent units.

[0107] For critical transition segments of adjacent units, the order of transition start points is first compared. If the downstream transition start point is later than the upstream transition start point but falls within the allowable propagation time range, it is processed normally. If the downstream transition start point appears earlier but the advance does not exceed 3 reference sampling intervals, and the recovery start position is later than the upstream recovery start position, the transition start point of the downstream segment is adjusted to a position no earlier than the upstream mutation endpoint; the adjustment amount is half to all of the advance, preferably two-thirds of the advance. If the advance exceeds 3 reference sampling intervals, the segment's start point is not directly adjusted, but only marked as an anomalous leader segment in the corrected transition sequence.

[0108] The magnitude correction is based on the relationship between the magnitudes of changes in adjacent units. First, the ratio of the downstream magnitude change to the upstream magnitude change is calculated. When the ratio is between 0.30 and 3.00, the original downstream magnitude change is retained. When the ratio is less than 0.30, the corrected magnitude change is obtained by adding a differential compensation amount to the original downstream magnitude change, where the differential compensation amount is 30% to 70% of the difference between 0.30 times the upstream magnitude change and the original downstream magnitude change. When the ratio is greater than 3.00, the corrected magnitude change is obtained by subtracting an excess reduction amount from the original downstream magnitude change, where the excess reduction amount is 30% to 70% of the difference between the original downstream magnitude change and 3.00 times the upstream magnitude change. For the synergistic relationship between the increase in nutrient concentration and the decrease in dissolved oxygen, the magnitude correction does not require the two to be in the same direction; it only requires that the magnitude change after correction maintains a sequential trend of either increasing or decreasing in the same direction.

[0109] Duration correction is performed separately based on mutation duration and recovery duration. When the difference between the mutation duration and the adjacent upstream segment exceeds 2.00 times the shorter mutation duration, the mutation endpoint of the excessively long segment is shrunk towards the recovery start position, with a reduction of 20% to 60% of the excess portion. When the recovery duration exceeds 2.00 times the adjacent upstream segment, the recovery endpoint is retained as a delay risk endpoint, and a delay marker is added to the corrected transition sequence, without directly shortening the recovery duration.

[0110] After processing the starting point, amplitude of change, and duration, the segments are reordered according to a unified time identifier to obtain the corrected transition sequence. The corrected transition sequence records the original segment position, the start and end times after correction, the corrected amplitude of change, the corrected duration, the correction level, and whether there is an anomalous leading segment.

[0111] To address the diffusion trend in the corrected transition sequence, the range of change extension is compared between adjacent stages. Stages are divided according to adjacent connection pairs in the micro-circulation transmission chain, and the range of change extension in each stage is determined by the number of nodes involved in the anomaly, the total duration of the anomaly, and the coverage area of ​​the change intensity.

[0112] The total duration of anomalies is the time between the earliest transition point and the latest recovery point within the same phase; the range of change intensity is the difference between the maximum and minimum change intensity within that phase. An increase in the number of anomalous nodes, or an increase in the total duration of anomalies of 20% to 100%, or an increase in the range of change intensity of 20% to 100% in the subsequent phase, indicates an expansion; a decrease of more than 20% in the above indicators in the subsequent phase indicates a decline; a change between -20% and +20% indicates maintenance.

[0113] The direction of imbalance expansion is determined by the position of the expansion stage in the microcirculation transmission chain. When expansion stages occur consecutively from upstream to downstream, it is marked as downstream expansion; when expansion stages occur simultaneously at branch nodes, it is marked as branch expansion; when expansion stages return to adjacent upstream nodes, it is marked as reflux expansion. The expansion rate is obtained by the expansion distance and time difference between adjacent stages. The calculation process is as follows: first, determine the distance between the central nodes of adjacent stages on the material flow path, then divide by the time difference between the transition start point of the subsequent stage and the transition start point of the previous stage. When measured distances are unavailable, they are counted at intervals according to the node sequence, with each crossing of an adjacent node counted as one unit distance. An expansion rate below 1.5 times the median value of historical stable operating days is marked as low, 1.5 to 3.0 times is marked as medium, and above 3.0 times is marked as high. The trend extension characteristic group consists of the expansion direction, expansion rate, range of change, and expansion state.

[0114] Trend extension feature groups are continuously correlated according to a unified time identifier to form an evolution trajectory. In continuous correlation, if the time interval between the start time of a subsequent trend extension feature group and the end time of the previous trend extension feature group does not exceed the upper limit of two allowable transmission time ranges, and their expansion directions are the same or they have adjacent node succession relationships, they are grouped into the same evolution trajectory; if this range is exceeded, a new evolution trajectory is formed. The evolution trajectory records the initial trigger node, the abnormal nodes passed through, the magnitude of correction changes at each stage, the duration of correction, the expansion direction, the expansion rate, and the degree of imbalance changes, used for subsequent identification of synergistic abnormal intervals of nutrient pulse superposition and dissolved oxygen decline.

[0115] Based on the aforementioned evolution trajectory, the synergistic abnormal intervals of nutrient pulse superposition and dissolved oxygen decline are identified, and corresponding hierarchical regulation instructions are generated.

[0116] When identifying collaborative anomaly intervals based on evolutionary trajectories, the variation extension characteristics of each stage are first read according to a unified time identifier, and then the corrected transition sequences related to the increase in nutrient concentration and the decrease in dissolved oxygen are extracted respectively. The criteria for the nutrient concentration increase segment are that the change direction is upward and the corrected change amplitude is greater than 2 to 6 times the median of the effective change amplitude in the past 1 to 24 hours of that node; the criteria for the dissolved oxygen decrease segment are that the change direction is downward and the corrected change amplitude is greater than 2 to 6 times the median of the effective change amplitude in the past 1 to 24 hours of that node.

[0117] When historical valid data is lacking, the range of the most recent 50 valid sample values ​​should be used as a reference, and the determination factor can be between 0.20 and 0.60.

[0118] When performing temporal overlap comparisons of nutrient concentration increase segments and dissolved oxygen decrease segments, the transition start point to recovery end point of each segment is used as the response interval. When two response intervals share common coverage, the common coverage portion is considered a candidate coordinated response interval. If the common coverage time is less than 0.20 times the duration of the shorter response interval, it is not included in the coordinated response interval sequence; if the common coverage time is 0.20 to 1.00 times the duration of the shorter response interval, it is retained as a coordinated response interval. If the nutrient concentration increase segment appears first, and the dissolved oxygen decrease segment appears within 1 to 10 reference sampling intervals after its recovery end point, and both are located at adjacent nodes in the microcirculation transport chain or downstream nodes of the same branch, this is also recorded as a delayed coordinated interval. The coordinated response interval sequence is arranged by start time, and the corresponding node position, response interval start and end times, overlap time, delay time, and segment source are recorded.

[0119] The intensity level of anomalies is determined by combining the degree of superposition of change amplitudes with the duration. The calculation process for the degree of superposition of change amplitudes is as follows: first, the corrected change amplitude of the nutrient concentration increase segment is divided by its reference amplitude to obtain the relative nutrient intensity; then, the corrected change amplitude of the dissolved oxygen decrease segment is divided by its reference amplitude to obtain the relative dissolved oxygen intensity; finally, the average of the two is taken and multiplied by the response overlap ratio to obtain the superposition intensity.

[0120] The response overlap ratio is the overlap time divided by the shorter duration of the two response intervals. For delayed coordination intervals, the response overlap ratio is determined after delay decay; the closer the delay time is to 10 reference sampling intervals, the lower the value, with a minimum of not less than 0.10.

[0121] Anomaly intensity levels are divided into three levels. Level 1 is defined as a superimposed intensity less than 1.0, or a duration of the coordinated response interval less than 3 reference sampling intervals; Level 2 is defined as a superimposed intensity of 1.0 to 2.5, with a duration of 3 to 10 reference sampling intervals; and Level 3 is defined as a superimposed intensity greater than 2.5, or a duration exceeding 10 reference sampling intervals, or two or more downstream nodes experiencing consecutive dissolved oxygen decreases within the same interval. If the coordinated response interval is located in the direction of backflow expansion, the level is increased by one level, but not exceeding Level 3.

[0122] The levels, superposition intensity, duration, and node locations of each collaborative response interval are summarized to form an anomaly classification marker set.

[0123] When forming a hierarchical control strategy sequence, the abnormal hierarchical marker set is mapped to the node positions in the microcirculation transmission chain.

[0124] For abnormal intervals located upstream of nodes, the control priority is higher than that for downstream nodes; for abnormal intervals before branch convergence, the control priority is higher than that for a single branch; when multiple abnormal intervals exist at the same node, the intervals with higher levels and longer durations are treated first. Level 1 abnormal intervals correspond to low-intensity intervention, which involves reducing irrigation frequency or extending irrigation intervals while maintaining the original circulation channel; Level 2 abnormal intervals correspond to medium-intensity intervention, which involves reducing the proportion of nutrient-containing return liquid entering the planting area, switching some circulation channels to buffer paths, and increasing aeration intensity; Level 3 abnormal intervals correspond to priority intervention, which involves suspending nutrient return to the corresponding branch, switching circulation channels to treatment units or bypass buffer zones, and increasing aeration intensity to the preset upper limit.

[0125] The control signals are switched according to a hierarchical regulation strategy sequence. The irrigation rhythm control signals include irrigation intervals and the duration of a single irrigation. Level 1 anomalies extend the irrigation interval by 10% to 30%, Level 2 anomalies extend it by 30% to 60%, and Level 3 anomalies suspend irrigation for 1 to 3 irrigation cycles. The circulation channel control signals include the channel opening, closing, and switching sequence. Level 2 anomalies switch 20% to 60% of the return flow to the buffer path, and Level 3 anomalies switch 60% to 100% of the return flow to the treatment unit or bypass buffer. The aeration intensity control signals include aeration start time and aeration duty cycle. Level 1 anomalies increase it by 10% to 20%, Level 2 anomalies increase it by 20% to 50%, and Level 3 anomalies increase it to 80% to 100% of the preset upper limit.

[0126] Understandably, after time sorting and node correspondence, hierarchical control instructions are formed. These hierarchical control instructions retain the anomaly level, trigger interval, target node, execution order, control signal type, and execution duration, and are used for phased intervention and recovery of local imbalance states.

[0127] Example 2, please refer to Figure 2 As shown in this embodiment, the IoT-based ecological circular agriculture data acquisition and analysis system includes: Multi-source acquisition and alignment module: Acquires multi-point sensor signals from planting area, breeding area and recycling unit, aligns them according to a unified time reference, and forms a continuous time-series acquisition sequence; Segmented response analysis module: Performs segmented response analysis on the continuous time-series acquisition sequence, identifies numerical abrupt changes and hysteresis recovery segments, extracts the corresponding change intensity and duration, and constructs a non-continuous transition feature set; The transfer chain construction module: Based on the discontinuous transition feature set, the units are sequentially connected according to the material flow direction to generate a micro-circulation transfer chain; Coupling Imbalance Identification Module: Utilizing the micro-circulation transmission chain, the synchronization degree and phase difference of the transition responses between adjacent units are calculated to obtain a coupling imbalance indicator set; Progressive Correction Analysis Module: Based on the coupling imbalance indicator set, progressively correct the discontinuous transition characteristics to form an evolution trajectory reflecting the local imbalance expansion trend; Graded regulation output module: Based on the evolution trajectory, it identifies the synergistic abnormal intervals of nutrient pulse superposition and dissolved oxygen decline, and generates corresponding graded regulation instructions.

[0128] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An ecological recycling agricultural data collection and analysis method based on the Internet of Things, characterized in that, include: Multiple sensor signals from the planting area, breeding area, and recycling unit are acquired, aligned according to a unified time reference, and formed into a continuous time-series acquisition sequence. The continuous time-series acquisition sequence is subjected to segmented response analysis to identify numerical abrupt changes and hysteresis recovery segments, extract the corresponding change intensity and duration, and construct a non-continuous transition feature set; Based on the discontinuous transition feature set, the units are sequentially connected according to the material flow direction to generate a micro-circulation transmission chain. Using the micro-circulation transmission chain, the synchronization degree and phase difference of the transition response between adjacent units are calculated to obtain the coupling imbalance indicator set; Based on the aforementioned set of coupling imbalance indicators, the discontinuous transition characteristics are progressively corrected to form an evolution trajectory that reflects the trend of local imbalance expansion. Based on the aforementioned evolution trajectory, the synergistic abnormal intervals of nutrient pulse superposition and dissolved oxygen decline are identified, and corresponding hierarchical regulation instructions are generated. 2.The Internet of Things based ecological cycle agricultural data collection and analysis method according to claim 1, characterized in that, Aligned according to a unified time base, forming a continuous time-series acquisition sequence, including: A local incrementing time identifier is added to the original records of each acquisition node, and invalid records are removed based on the stability of the interval between adjacent records to obtain the initial time series segment; The sampling node located upstream of the material flow path and where the effective fluctuation first appears is selected as the reference node. The initial time series segment of the reference node is used as a reference to perform segment-by-segment offset correction on the initial time series segments of the remaining sampling nodes to form a relatively aligned sequence segment. Missing segments in the relatively aligned sequence segments are progressively filled in, and the mutation boundaries are processed before being spliced ​​together in chronological order to form a continuous time-series acquisition sequence. 3.The Internet of Things based ecological cycle agricultural data collection and analysis method according to claim 2, characterized in that, Segment-by-segment offset correction, including: In each initial time series segment, identify positions where 2 to 5 consecutive sampling points change in the same direction and the change exceeds the dispersion by 2 to 4 times. The first position that exceeds this position is taken as the effective fluctuation moment. Candidate time offsets are set within the range of negative 5 times the reference sampling interval to positive 5 times the reference sampling interval. The segments of the acquisition nodes to be corrected are advanced or delayed respectively, and the consistency of change direction and the difference of change amplitude are compared with the segments of the reference node in the same time range. Candidate time offsets with consistent change direction, large number, and small cumulative difference in change magnitude are selected as segmented correction values, and the difference in correction values ​​between adjacent segments is progressively transitioned to no more than one reference sampling interval.

4. The method for data collection and analysis of ecological circular agriculture based on the Internet of Things according to claim 1, characterized in that, Construct a feature set of discontinuous transitions, including: A sliding window is set along the continuous time-series acquisition sequence. Based on the fluctuation of the change in adjacent sample values ​​within the window, the window is marked as a stable window or a transition window, and then merged to form an initial segmented sequence. The transition segments in the initial segmented sequence are refined by boundary refinement. By comparing the point-by-point change gradients of the transition segments with those of the preceding and following stable segments, the mutation start and end points are determined, forming mutation response fragments. In the stable region following the mutation response fragment, identify the intervals where the deviation decreases continuously, determine the recovery start position and the end position of the regression persistence interval, and form the hysteresis recovery fragment. 5.The Internet of Things based ecological cycle agricultural data collection and analysis method according to claim 4, characterized in that, The formation of the feature set of discontinuous transitions includes: The magnitude and duration of the changes in the mutation response segment and the hysteresis recovery segment are calculated separately, and the magnitude of the changes is compared with the median of the effective magnitudes of the corresponding acquisition nodes in the most recent 1 hour to 24 hours to obtain the intensity of the change. When a hysteresis recovery fragment appears within 1 to 20 reference sampling intervals after the mutation response fragment, the two are combined into a discontinuous transition unit; Multiple discontinuous transition units are sorted according to a unified time identifier, and the data acquisition node identifier, mutation start point, mutation end point, recovery start position, regression duration end position, change intensity, duration and change direction are recorded to construct a discontinuous transition feature set. 6.The Internet of Things based ecological cycle agricultural data collection and analysis method according to claim 1, characterized in that, Generate a microcirculation transport chain, including: Based on the transition starting point of each acquisition node in the discontinuous transition feature set, and combined with the preset material flow path and the allowable transmission time range of adjacent nodes, the upstream node and the candidate downstream node are matched in time sequence to form an initial node connection sequence. The temporal overlap and sequential conflict in the initial node connection sequence are corrected, and the connection relationships that are directly adjacent in the material flow path, have the first appearance of the transition response and the relatively stable decay of the change intensity are retained first, so as to obtain an ordered transmission node sequence. Based on the ordered transmission node sequence, the temporal continuity, amplitude continuity and recovery continuity of adjacent nodes are screened, and node pairs that meet the transmission continuity are connected in series to form a micro-circulation transmission chain. 7.The Internet of Things based ecological cycle agricultural data collection and analysis method according to claim 6, characterized in that, The set of coupling imbalance indicators is obtained, including: The discontinuous transition characteristics of adjacent units are read along the micro-circulation transmission chain. The transition start point and recovery end point of the upstream unit are paired with the corresponding positions of the downstream unit to form an adjacent response pairing sequence. For each response pair in the adjacent response pairing sequence, calculate the time offset and determine the response overlap interval based on the common coverage of the upstream and downstream response intervals; By combining time offset, response overlap range, difference in change intensity, and difference in recovery duration, the synchronization level is classified into high, medium, and low, and response pairs with excessive time offset, insufficient response overlap, or low synchronization level are identified as coupling imbalance indicator sets.

8. The method for data collection and analysis of ecological circular agriculture based on the Internet of Things according to claim 7, characterized in that, The evolutionary trajectory includes: Based on the time offset, synchronization and imbalance of abnormal nodes in the coupling imbalance indicator set, the correction priority value is determined, and the discontinuous transition units with a correction priority value of not less than 3 are identified as the set of critical transition segments. The key transition segment set is corrected step by step along the micro-circulation transmission chain. The transition starting point of downstream segments that appear early and recover late is adjusted. The amplitude of segments with deviation in change intensity is compensated or reduced. Delay marks are added to segments with abnormally long recovery duration to obtain the corrected transition sequence. By comparing the number of anomalous nodes, the total duration of anomalous events, and the coverage of change intensity in adjacent stages of the modified transition sequence, the direction and rate of expansion are determined, and the evolution trajectory is formed by continuously associating them in chronological order. 9.The Internet of Things based ecological cycle agricultural data collection and analysis method according to claim 8, characterized in that, The corresponding hierarchical control instructions are generated, including: extracting nutrient concentration increase segments and dissolved oxygen decrease segments along the evolution trajectory, determining the common coverage part or delayed synergistic part of the response intervals of the two as the synergistic response interval, and forming a synergistic response interval sequence; determining the first, second or third level of abnormal intensity based on the relative intensity of the nutrient concentration increase segment, the relative intensity of the dissolved oxygen decrease segment, the response overlap ratio and the duration of the synergistic response interval, and forming a hierarchical control strategy sequence in combination with the node positions in the microcirculation transmission chain, and converting the hierarchical control strategy sequence into control signals for irrigation rhythm, circulation channels and aeration intensity.

10. The system for ecological cycle agriculture data collection and analysis based on Internet of Things, which is used for realizing the method for ecological cycle agriculture data collection and analysis based on Internet of Things as claimed in any one of claims 1-9, characterized in that, include: Multi-source acquisition and alignment module: Acquires multi-point sensor signals from planting area, breeding area and recycling unit, aligns them according to a unified time reference, and forms a continuous time-series acquisition sequence; Segmented response analysis module: Performs segmented response analysis on the continuous time-series acquisition sequence, identifies numerical abrupt changes and hysteresis recovery segments, extracts the corresponding change intensity and duration, and constructs a non-continuous transition feature set; The transfer chain construction module: Based on the discontinuous transition feature set, the units are sequentially connected according to the material flow direction to generate a micro-circulation transfer chain; Coupling Imbalance Identification Module: Utilizing the micro-circulation transmission chain, the synchronization degree and phase difference of the transition responses between adjacent units are calculated to obtain a coupling imbalance indicator set; Progressive Correction Analysis Module: Based on the coupling imbalance indicator set, progressively correct the discontinuous transition characteristics to form an evolution trajectory reflecting the local imbalance expansion trend; Graded regulation output module: Based on the evolution trajectory, it identifies the synergistic abnormal intervals of nutrient pulse superposition and dissolved oxygen decline, and generates corresponding graded regulation instructions.

Images