Slope farmland runoff collection and irrigation and ladder type purification method based on intelligent shunt

Abstract

The application discloses a slope farmland runoff collection and irrigation and ladder type purification method based on intelligent distribution, relates to the technical field of agricultural water conservancy purification, collects rainfall, monitoring, crop water requirement, pool position and water quality, generates a target mapping table and a control window, and unifies distribution instructions, water distribution instructions, water release instructions and a valve time sequence table; initial runoff is identified in the control window, contaminated incoming water is introduced into an initial temporary storage unit, and a distribution proportion trajectory is written into a valve time sequence segment according to the target mapping table. According to the distribution proportion and the incoming water characteristics, the hydraulic retention of a settling section, a filtering section and an ecological section in a purification unit sequence is set, multi-stage water storage pools are linked to store water at different times, and a pool position trajectory and a rolling estimation of available storage capacity are output. When operation deviates from the target mapping table, water release is rearranged and a water release time sequence is generated, an event is ended and template parameters are updated, and the synergy of water quality reaching the standard, irrigation guarantee, facility safety and ecological base flow guarantee is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of agricultural water purification technology, in particular to a slope farmland runoff collection and irrigation and ladder type purification method based on intelligent diversion. BACKGROUND

[0002] Under heavy rainfall, the small watershed of hilly and gully slope farmland produces runoff rapidly and carries sediment and nutrients. The initial stage of rainfall often produces "initial runoff" with significant pollution, and the quantitative criteria have not been unified for a long time. Engineering is often handled by empirical threshold or flow rejection device, which is difficult to integrate with subsequent scheduling. Graded purification and constructed wetlands are widely used in agricultural non-point source pollution control, but research and practice focus on unit removal and construction materials, and less on real-time scheduling and reservoir group management. Real-time control of urban rainwater system has gradually matured, but it mainly faces pipe networks and urban storage facilities, and it is difficult to be directly transferred to slope farmland reservoir group. The rigid requirements of irrigation water quality standards and ecological flow determination and protection specifications have been proposed, and existing slope farmland projects often rely on static threshold and manual switching. Monitoring is not the same, time period is mismatched, and rules conflict, resulting in initial pollution mixed into water storage body, purification chain under pressure, and downstream constraints difficult to coordinate. Slope farmland often uses a combination of interception ditches, terraces and ponds to collect rainwater and return it to irrigation, which can reduce confluence and erosion, but in events with varying intensity and duration, single-point structures are difficult to form a unified time sequence with dynamic diversion, purification and water release.

[0003] Within the same event time scale, a set of coordinated control mechanisms that can be calculated, transmitted and reused are constructed. Multi-source monitoring and crop water input are mapped into a target mapping table and a control window, so that initial runoff identification directly triggers diversion instructions and valve timing table fragments. The diversion ratio trajectory and the incoming water characteristics drive the hydraulic retention and inlet rhythm of the purification unit sequence. The incoming allocation vector, pool trajectory and available storage capacity are jointly determined to determine the time allocation and inter-pool transfer, and to rearrange the water release allocation and water release timing when it deviates. At the same time, the ecological base flow hard constraint is prepositioned in the allocation layer, and the template parameters are updated to sediment for the next round of callable reference at the end of the event.

[0004] Without this mechanism, the initial stage of heavy rain pollution is easily mixed into the water storage body, and the purification chain and pool group bear concurrent load. The subsequent incoming and water release are subject to static threshold and post-check, causing scheduling mismatch and constraint conflict and amplifying operation risk. In the multi-objective and multi-constrained scenario of slope farmland small watershed, the existing operation mode mainly based on manual switching and static rules is difficult to maintain semantic consistency and time sequence alignment within the event, and cannot concatenate identification, allocation, shaping, incoming, water release and experience update into a single chain, and continuously reproduce similar mismatches across events. SUMMARY

[0005] (I) Technical problems solved

[0006] To address the shortcomings of existing technologies, this invention provides a method for runoff collection, irrigation, and tiered purification of sloping farmland based on intelligent diversion. By aggregating rainfall, monitoring data, crop water requirements, reservoir locations, and water quality, a target mapping table and control window are generated, unifying diversion commands, water distribution commands, water release commands, and valve timing tables. Within the control window, initial runoff is identified, and contaminated water is introduced into an initial temporary storage unit. A diversion ratio trajectory is generated according to the target mapping table and written into the valve timing segment. Based on the diversion ratio and influent characteristics, hydraulic retention times in the sedimentation, filtration, and ecological sections of the purification unit sequence are set, linking multi-stage reservoirs for timed inflow, and outputting reservoir location trajectories and rolling estimates of available storage capacity. When the operation deviates from the target mapping table, water is re-discharged and a water release sequence is generated. The event ends, and template parameters are updated, achieving synergy between water quality compliance, irrigation supply assurance, facility safety, and ecological baseflow protection, thereby solving the technical problems described in the background art.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] The method of intelligent diversion for runoff collection, storage, irrigation and tiered purification in sloping farmland includes collecting rainfall, monitoring, crop water requirements, pool location and water quality status and ecological base flow constraints, generating a target mapping table and control window, and setting reference fields for diversion instructions, water distribution instructions, water release instructions and valve timing table and diversion ratio trajectory benchmark.

[0010] Within the control window, the initial runoff is determined according to the rainfall sequence and water quality status, triggering a diversion command to guide it into the initial temporary storage unit, and directing it to the purification unit sequence and multi-stage water storage tank according to the target mapping table, forming a valve timing segment.

[0011] Based on the target mapping table and the diversion ratio trajectory, water distribution instructions and hydraulic retention time settings are generated. The sequence of segments and retention allocation are determined in the purification unit sequence. The multi-stage water storage tanks are filled in time-sharingly, and the tank position trajectory and available storage capacity are output as rolling estimates.

[0012] When the operation deviates from the target mapping table, the discharge command, water distribution command and diversion command are re-issued with reference to the pool position trajectory and water quality status. The discharge sequence is generated and written into the valve sequence table. The target mapping table and valve sequence table template are updated afterwards.

[0013] Furthermore, the target mapping table includes fields for irrigation supply guarantee, water quality compliance, facility safety, ecological base flow guarantee, as well as reference fields for the diversion command, water distribution command, and water release command. The control window is generated by weighted rules of rainfall response, reservoir capacity shortage, and water quality vulnerability, and corresponds to the time index of the target mapping table.

[0014] Furthermore, the initial runoff determination integrates the characteristics of the initial rise of the comprehensive rainfall segment with the abrupt change characteristics of the water quality state to form a trigger signal. The diversion command determines the channel of the initial temporary storage unit and the purification unit sequence inlet according to the target mapping table, and sets mutually exclusive segments and allowed segments in the valve timing table.

[0015] Furthermore, the initial temporary storage unit includes a temporary storage pool connected to the upstream intercepting ditch, a controllable overflow weir connected to the inlet of the purification unit sequence, and a drain branch.

[0016] The diversion command, within the control window, first enters the initial temporary storage unit, then switches to the purification unit sequence according to the valve timing table, and restores the normal channel after the emptied branch is disposed of.

[0017] Furthermore, the purification unit sequence includes a settling section, a filtration section, and an ecological section. The hydraulic retention time is set according to the target mapping table and the pollution flux estimation. The retention allocation is given in the order of the sections, and the obtained passage order is aligned with the diversion ratio trajectory and written into the valve timing table for inlet rhythm control.

[0018] Furthermore, the flow distribution at the inlet of the purification unit sequence is determined by the updated diversion ratio trajectory within the control window and the available water volume.

[0019] The settling section sets the effective water depth by combining the gate height and the bypass opening and closing; the filtration section sets the flow rate by throttling the inlet and outlet water; and the ecological section sets the residence scale by selecting branch channels and fine-tuning the water depth.

[0020] Furthermore, the timed inflow into the multi-stage water storage tank is determined by the inflow allocation vector, which determines the target tank type and proportion based on the tank position trajectory, the rolling estimate of available storage capacity, and the intermediate water quality status; when the tank position trajectory approaches the safety upper limit, the inter-tank transfer is performed according to the valve timing table and the inlet rhythm is simultaneously restricted.

[0021] Furthermore, the water release priority switching is based on a deviation index generated by weighting reservoir capacity shortage, water quality fragility and ecological gap to generate a water release allocation vector, and the time period is trimmed by the control window. In addition, a hard constraint of ecological base flow guarantee is set in the allocation layer. When the constraint value is insufficient, it is supplemented by deducting from non-ecological channels according to weight and then written into the valve timing table.

[0022] Furthermore, after generating the water release allocation vector, a quality-oriented short-term fine-tuning is performed on the hydraulic residence time setting. The fine-tuning is limited to a small monotonic gain on the filtration section and the ecological section, without changing the predetermined passage order, and is time-aligned and segmented with the control window and the valve timing table before generating the water release sequence.

[0023] Furthermore, the water release sequence is generated by aligning the water release allocation vector with the hydraulic residence time setting. Each channel is set with a minimum duration and a minimum interval between adjacent actions to limit the concurrency of channels in the same time slice. The water release sequence is then mapped to a unique valve position index of the valve sequence table for execution.

[0024] Furthermore, after the event ends, monitoring and execution records are collected, and an event performance function is constructed that includes ecological, engineering safety, water quality, and irrigation deviation. The weights are determined based on this function, and the template parameter vector is updated by convex combination to obtain the updated template parameter vector. This updated template parameter vector is then used to update the target mapping table and the valve timing table template for subsequent event calls.

[0025] (III) Beneficial Effects

[0026] This invention provides a method for runoff collection, irrigation, and tiered purification on sloping farmland based on intelligent diversion, which has the following beneficial effects:

[0027] Using the target mapping table and control window as a unified reference, the irrigation supply guarantee, water quality compliance, facility safety and ecological base flow guarantee are transformed into executable event target vectors, and are integrated and bound with the reference fields of diversion command, water distribution command and water release command, so that the valve timing table and the diversion ratio trajectory are consistent in time.

[0028] Initial runoff identification and diversion command implementation, through initial runoff probability discrimination and command priority sorting, directs high pollution loads to the initial temporary storage unit, while clearer inflows are directed into the purification unit sequence inlet and multi-stage reservoir inlet, so that the intermediate water quality state vector and purification load are peaked in the first half of the process, avoiding the initial pollution from mixing into the water body and reducing the instantaneous impact of the purification unit sequence.

[0029] In the joint regulation of water distribution and storage in the purification unit sequence, the updated diversion ratio vector, valve opening vector, and pollution flux estimate are used together to construct the inlet flow vector and hydraulic residence time vector of the purification unit sequence. This enables the sedimentation, filtration, and ecological sections to form a "pilot buffer, segment-by-segment balancing" carrying sequence. At the same time, the inflow allocation vector drives timed inflow and inter-pool transfer, outputting the pool position trajectory and rolling estimate of available storage capacity, thereby stabilizing the outflow and ensuring the amount of water available for irrigation.

[0030] The priority switching of water release commands is triggered by the deviation index to set the water release allocation vector and water release timing segment. During the short-term shaping phase, the water quality is smoothed by the adjusted hydraulic residence time vector, so that the ecological base flow guarantee and facility safety are realized in advance. After the risk is mitigated, it returns to the irrigation orientation and is aligned with the valve timing table to avoid execution failure caused by channel competition and time overlap.

[0031] After the event ends, the template update quantifies the deviations in ecology, safety, water quality and irrigation by using the event performance function, and uses the updated template parameter vector to precipitate a reusable target mapping table and valve timing table template. This makes the preliminary judgment, proportional generation and timing arrangement of the next event more realistic without the need for manual parameter adjustment, thereby continuously reducing strategy jitter and execution uncertainty across events. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the process for the runoff collection, irrigation, and tiered purification method for sloping farmland according to the present invention. Detailed Implementation

[0033] 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, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figure 1 This invention provides a method for runoff collection, irrigation, and tiered purification on sloping farmland based on intelligent diversion, including:

[0035] Step 1: Construct a target mapping table, focusing on data consistency and target computability. With control window Simultaneously, reference fields and templates are generated to make the event target vector... It can be reliably transmitted to the splitter command. Water distribution instructions Water release instructions The actual execution end.

[0036] Given the suddenness, temporal variability, and spatial heterogeneity of natural water inflows, threshold rules alone cannot simultaneously guarantee irrigation supply and water quality compliance, as well as facility safety and ecological baseflow assurance. To avoid local optima and temporal mismatches, it is necessary to incorporate rainfall input flow... On-site monitoring vector Crop water input Pool position status Water quality status Ecological base flow protection constraints The target intensity is elevated to a computable, comparable, and traceable quantity, and a set of corresponding actionable time periods is provided, thereby establishing a target mapping table. With control window .

[0037] Therefore, firstly, multi-source sensing is unified into the system state vector, and then the event target vector is obtained through nonlinear target mapping. Thus, control windows are screened out using risk gating. Then, both are written into the target mapping table. The primary key field ultimately binds the reference fields required for subsequent calls to a unified semantic space.

[0038] Without changing the physical measurement parameters, the rainfall data will be input into the flow. On-site monitoring vector Crop water input Pool position status With water quality status Concatenate to form the system state vector Furthermore, the intensity of the four types of targets is projected into event target vectors through an interpretable monotonic mapping. .

[0039] To avoid over-reliance on a single threshold, a smooth mapping is employed, allowing the priorities of four components—irrigation supply assurance, water quality compliance, facility safety, and ecological baseflow assurance—to be continuously adjusted according to state changes. This reduces false triggering and makes different objectives comparable on the same scale, thereby ensuring the accuracy of subsequent diversion instructions. Water distribution instructions Water release instructions The ordering and strength selection have a consistent origin, where:

[0040] In the formula: the target component of the event. : Represents the event target vector The Quantity, System state vector : Input flow from rainfall On-site monitoring vector Crop water input Pool position status Water quality status Composition; target response function Regarding the first Differentiable response of a target class, real number, representing the system state vector Projected onto the target domain, the type is a weighted differentiable combination function, used to uniformly compress multi-source monitoring quantities into a dimensionless index in the interval of 0 to 1; target benchmark value. : indicates the first A benchmark or reference for a class of targets; shape parameters : indicates the first The sensitivity coefficient of the target class is an open interval of positive real numbers.

[0041] Through event target components Continuous representation of the event target vector Balancing stability and sensitivity, the weights of irrigation supply assurance and water quality compliance, facility safety and ecological base flow assurance can be smoothly transitioned under state disturbances, reducing strategy jitter and increasing tolerance to sudden water inflows.

[0042] Only event target vector This is still insufficient for implementing rhythm control; a control window is necessary. By limiting the permitted execution time periods, high-risk periods are excluded. To avoid one-sided reliance on a single factor, the risk parameters of rainfall intensity response, reservoir capacity constraints, and water quality vulnerability are integrated into a single risk measure, and the control window is defined using a hyper-level set approach. ;

[0043] Therefore, while maintaining operability, it is possible to strictly exclude non-operational periods; thus providing a safety enclosure for valve timing and proportional allocation, ultimately controlling the window. Write to target mapping table Time period field: In the formula: control window : Represents the set of times when control is allowed to be executed, a subset of time indices; Rainfall input stream : Represents the rainfall intensity or its proxy at any given time, a non-negative real number, characterizing the inflow-driven force; reservoir capacity tension function : Status of pool position The monotonic transformation of the non-negative real number represents the stress level of reservoir capacity reserve; the water quality vulnerability function Regarding water quality status The dimensionality reduction quantization, non-negative real numbers, characterize the stress resistance of the purification chain; rainfall intensity response function : Input flow of rainfall Normalized mapping, nonnegative real numbers, equilibrium scale differences; reservoir capacity tension function To smoothly assess how close the water level is to the safe upper limit, a monotonically differentiable Logistic-Sigmoid curve is chosen: ; When the current water level Approximately the total storage capacity With available storage capacity When the difference is less than 1, the function value rises rapidly to 1; when it is far below the design water level, it approaches 0. The Sigmoid curve is continuous and strictly increasing, and it exhibits Lipschitz behavior across the entire range, which is beneficial for real-time control. For reservoir capacity Sigmoid slope control parameters; water quality vulnerability function For multiple water quality indicators, deviations are normalized using weighted Euclidean distance, and then the sensitivity is enhanced using Sigmoid. ;in Let i be the instantaneous value of the i-th water quality indicator. As a reference value, The water quality weight for the i-th item is usually determined according to a water quality index system. The greater the distance, the more severe the pollution. Approaching 1; when the indicator is close to the baseline Approaching 0 This is a Sigmoid slope control parameter for water quality. This is a weighted diagonal matrix;

[0044] Rainfall Intensity Response Function Used to determine the current rainfall intensity This is converted into a dimensionless index between 0 and 1, used to generate the control window. A piecewise Sigmoid approximation with linearity within the threshold and saturation outside the threshold is recommended, as follows:

[0045] In the formula; Real-time rainfall intensity (after initial damage / conversion); Triggering the lower limit (light rain threshold). The following assumes that the rainfall intensity is insufficient to affect the scheduling, so the output is 0; the middle section uses a standard Sigmoid function to make the response smooth and differentiable; its inflection point is located at... ; , saturation value (rainstorm threshold). The above is directly considered as extreme rainstorm, and the function saturates to 1. Sigmoid slope parameter;

[0046] Weighting coefficients : Represents the relative importance of the three factors, a non-negative real number; threshold parameter : Indicates the boundary between executable and non-executable, a non-negative real number.

[0047] Control Window Cut the time axis according to the risk level to enable subsequent splitting instructions. Water distribution instructions Water release instructions The execution only occurs within the acceptable range, thereby reducing the probability of concurrent overflows and water quality exceeding limits, and improving the availability and interpretability of the strategy.

[0048] Input rainfall data On-site monitoring vector Water quality status Pool position trajectory Timestamps are unified to event clocks, valve timing tables Fragment generation progresses in a clockwise manner; when any critical quantity is missing, short-term interpolation is performed using the previous valid value and physical feasibility constraints, only within the control window. It takes effect within the current period and is covered by new observations in the next sampling period.

[0049] Target mapping table With control window While specifying what to do and when to do it, it's still necessary to specify the order of priority and quantity of tasks at the reference field level, thereby forming a valve timing table. Trajectory of Diversion Ratio The template. To avoid conflicts caused by hard-coded rules, the event target vector needs to be... The projection is a sortable instruction score, which is then further translated into a proportional template so that subsequent execution can directly read it without ambiguity. Therefore, instruction priority scores are first generated, and then a probabilistic template for weight allocation is given based on these scores. Finally, the two types of output are integrated with the control window. After field alignment, write back to the target mapping table. .

[0050] To ensure the routing instructions Water distribution instructions Water release instructions The ordering is interpretable and monotonically consistent, which relates to the event target vector. Linear projection scores instructions by priority and determines the execution order by sorting; therefore, it can maintain style stability across different events; thus ensuring a consistent target mapping. Different control windows A consensus was reached internally; the final decision was the valve timing table. The construction provides a direct sequence, where:

[0051] In the formula: instruction priority score : indicates the first Instruction precedence, a real number, is used as the sorting criterion; weight vector. : For the event target vector Linear weights, real vectors; event target vector. : An intensity vector composed of four types of target components, in the component domain.

[0052] Scoring by instruction priority The monotonic mapping can be used with different event intensities and different control windows. To maintain consistency in the order of commands, prevent command preemption and interruption, and enhance the valve timing table. Stability.

[0053] The order has been determined, but the proportions still need to be determined. To balance robustness and feasibility, the event target vector... The splitting ratio trajectory is generated through structure mapping. The template is configured and its concentration is adjusted using temperature parameters; therefore, the scaling ratio can be automatically adjusted under different event intensities; thus ensuring that high-priority targets receive greater allocation when resources are scarce, while all targets can still be considered when resources are plentiful; finally, the resulting template is integrated with the control window. Alignment, generating valve timing table The occupied fragment:

[0054] In the formula: the diversion ratio template : Indicates the diversion ratio trajectory Single-time vector, And the sum of the components is 1; temperature parameter : A positive real number that adjusts the concentration of the distribution;

[0055] Structure mapping matrix : Event target vector The matrix projected onto the allocation channel, a real matrix, reflects the facility topology and reachability; the event target vector. Same as before, component domain, providing strength information for each target.

[0056] Traffic splitting ratio template The probabilistic representation of the valve timing table makes Trajectory of Diversion Ratio In the control window It features scalable bias, enabling stable operation under both resource-constrained and resource-abundant conditions, significantly reducing the frequency of manual parameter tuning, and maintaining semantic consistency and easy auditability.

[0057] By jointly identifying the time period and composition of the initial runoff, high pollution loads are first isolated to the inlet of the initial temporary storage unit, and then the clearer water is directed to the inlet of the purification unit sequence and the inlet of the multi-stage reservoir. This compresses the identification-command-action chain into an event time scale, reduces mixing and impact, and provides predictable load and time-series-executable input for subsequent water distribution and water storage coordination of the purification unit sequence.

[0058] Step 2: In the control window Internally, using the event target vector Based on the ranking criteria, the occurrence and dissipation of initial runoff are identified online and translated into diversion instructions in real time. Simultaneously output valve timing table Trajectory of Diversion Ratio The executable segments allow high-pollution loads to enter the initial temporary storage unit inlet, while clearer incoming water enters the purification unit sequence inlet and the multi-stage reservoir inlet, ensuring that water quality meets standards and irrigation supply is guaranteed in the same direction under the constraints of engineering safety and ecological base flow protection.

[0059] Initial runoff carries a high proportion of suspended solids and soluble nutrients during the initial stages of rainfall. If it is not identified and diverted in time, it will push a high pollution load into multi-stage reservoirs in a short period of time, causing the quality of the incoming water to deviate from the target mapping table. The water quality meets the standards, thereby reducing the hydraulic retention and treatment capacity of the purification unit sequence.

[0060] To avoid the passive situation of contamination followed by dilution, it is necessary to [implement measures] in the control window. The internal structure determines the initial runoff and maps the discriminant directly to the diversion command. The valve position action and time slice allocation are determined. Therefore, a feature vector representing the initial runoff is extracted from the observable rainfall characteristics and water quality conditions. The initial runoff probability is obtained through a smooth and interpretable function, and a binary initial runoff indicator is generated based on this. This indicator is then compared with the event target vector. The sorting and combination of these parameters form a valve opening vector with priority and proportion, which is then written into the valve timing table. The current segment is updated and the splitting ratio trajectory is refreshed simultaneously. .

[0061] In the rain input stream The initial rising phase, water quality status The turbidity and particulate matter indices rise synchronously. To condense this complex feature of rapid rise-high turbidity-short duration into a single discriminant, a smooth mapping based on feature vectors is constructed to obtain the initial runoff probability and generate a binary initial runoff indicator, so that subsequent instructions can be directly referenced.

[0062] Therefore, features from rainfall time series and water quality composition are concatenated into a feature vector, and the probability is output as a monotonically increasing function. This maintains stable pre-threshold behavior under different event intensities, thereby reducing false shunting.

[0063] Where: Initial runoff probability : Quantifying the likelihood of whether a point in time is in the initial runoff phase; initial runoff indicator : A binary set, serving as the switching variable for valve position action; 1 indicates the initial stage, and 0 indicates exit; feature weight vector. : Real vectors, which are weighted feature vectors and applied to the discriminant space;

[0064] Feature vector Real vector, carrying the rainfall input stream On-site monitoring vector With water quality status Extracted time period and component characteristics; threshold shift parameter Real numbers; adjusting the entry point of the probability curve; decision threshold. Typically, 0.5 is used;

[0065] Indicator Function Binary output, translating probability into binary control signals; probability threshold. : Define the trigger boundary for initial runoff; S-shaped function : This ensures the continuous differentiability of the output probability.

[0066] Based on initial runoff probability With initial runoff indicator The combined output converges the heterogeneity of multi-source features into a single control variable, significantly reducing false and false identifications caused by the instability of a single threshold, thus improving the subsequent splitting instructions. It exhibits predictable consistency and auditability. Continuous probability supports binary actions, and smooth judgment is used to handle valve position switching, reducing oscillations and ensuring repeatable triggering.

[0067] After completing the judgment, it is necessary to check the control window. The internal process translates the initial conclusion into which valve to open, how much to open, and when to open it. To avoid conflicts between different valve positions within the same time slice, the event target vector... Under the sorting, construct a flow path from the initial runoff indicator. From the flow ratio template to the valve opening vector A linear combination, and forced with the control window. Time slice alignment, thereby aligning the valve timing table Trajectory of Diversion Ratio A consistent update is generated, where:

[0068] In the formula: valve opening vector : Values ​​are taken in a closed interval vector domain, representing the opening ratio of each valve position at any given time; time slice indication function Binary output, the output only takes the value 0 or 1, and is only displayed in the control window. Allowed actions; initial clear water mapping matrix : Real matrix, mapping the diversion ratio vector to the clear water channel valve position; Initial wastewater mapping matrix : Real matrix, mapping a fixed initial wastewater ratio to the inlet valve position of the initial temporary storage unit; Diversion ratio vector : And the sum of the components is 1, which comes from the diversion ratio trajectory. The current segment; initial sewage basis vector Unit basis vector, pointing to the inlet channel of the initial temporary storage unit; initial runoff indicator. Choose the route opening option.

[0069] In use, the valve opening vector is used. Time slice alignment, splitting instructions In the control window The system operates stably, with mutually exclusive and switchable valve positions at the initial temporary storage unit inlet and purification unit sequence inlet, significantly reducing the probability of contamination and minimizing instantaneous impact on the purification chain. The identification results are transformed into executable vectors, and matrix mapping is used to fix channels and proportions, ensuring conflict-free valve position actions.

[0070] Relying solely on initial binary conclusions is insufficient to control subsequent loads and proportions. In the early stages of an event, pollution flux may peak within a short period. Without load quantification and pre-allocation, this could lead to exceedances at the inlet of the purification unit sequence and fluctuations in water quality within the pool. Therefore, based on initial runoff identification, pollution flux estimation is introduced and fed back into the diversion proportion trajectory. In terms of template updates, the recognition-mapping-update process is promoted in a single chain.

[0071] Therefore, we should first consider the water quality. Component weights and rainfall input flow Pool position status Hydrological response synthetic pollution flux estimation, then using event target vector Sequencing and initial runoff indicators The states together determine the proportional bias, and finally output a new diversion proportional vector and the corresponding valve timing table fragment, ensuring the carrying order of the purification unit sequence inlet and the inlet quality of the multi-stage water storage tank.

[0072] In the initial stage, pollutants are significantly contributed by rapid flushing, with suspended solids and particulate nutrients being the dominant components. To quantify this type of load before it enters the purification unit sequence, a pollutant flux estimate determined by both water quality status and influent volume is constructed and used as the driving force for proportional updates. Therefore, purification pressure can be predicted in the first half of the event, thus providing a proportional strategy of first peak shaving and then balancing, where:

[0073] Where: pollution flux estimation : Non-negative real number, representing the potential pollution load intensity entering the processing chain at a given time; component weight vector : A non-negative real vector representing the water quality state Multi-component polymerization is equivalent to pollution intensity; water quality status : A non-negative real vector describing the water composition index at a given time; hydrological response function Non-negative real numbers, inputting rainfall data into the stream. Pool position status The mapping is to the equivalent water inflow, which is mathematically continuous and differentiable, such as the Nash-Gamma cascade hydrological response function and the exponentially decaying linear reservoir hydrological response function.

[0074] When used, pollution flux estimation The introduction of this feature enables quantifiable pressure input for proportional updates, avoiding coarse switching based solely on binary judgments, thus achieving a more robust peak-shaving strategy during initial high-load periods. It allows both composition and water volume to provide pressure measurements, quantitatively driving subsequent adjustments to the proportional and valve positions.

[0075] In obtaining pollution flux estimates Then, this pressure metric needs to be correlated with the event target vector. Ranking, initial runoff indicator The states are collectively transcribed into a shunt proportional vector, allowing the bias to adjust with pressure and vary with stage. To balance stability and directivity, a normalized mapping with a temperature parameter is used, and a proportional bias is applied to the inlet of the initial temporary storage unit when an initial marker is present, subsequently synchronized with the control window. Align time slices and refresh valve timing table The corresponding fragment, in which:

[0076] In the formula: the updated splitting ratio vector : And the sum of the components is 1, which serves as the diversion ratio trajectory. The current replacement fragment; temperature parameters Positive real numbers, controlling the degree of concentration in the proportional distribution; structural mapping matrix. : Real matrix, representing the event target vector Projected onto the executable channel; bias gain : A non-negative real number that adjusts the strength of the initial bias; the initial bias vector : Real vector, which pulls the scale closer to the initial temporary storage unit entry direction; Event target vector As before, it provides a ranking of the strength of four types of targets;

[0077] When using it, update the split ratio vector. The temperature- and bias-based design allows the proportional controller to focus more precisely on the initial temporary storage unit entrance during high-pressure periods, while automatically returning to a multi-channel approach after pressure relief, achieving a smooth transition from biased peak shaving to balanced storage. By incorporating pressure signals and target sorting into the proportional controller, and using temperature and bias to control concentration and directionality, the execution end becomes both stable and responsive.

[0078] Valve timing table The segments are used as time instructions to establish a peak-shaving and then balanced flow sequence in the sedimentation, filtration and ecological sections, and realize timed entry and inter-pool transfer at the inlet of the multi-stage reservoir, so that the water quality and irrigation volume of the reservoir can be coordinated to meet the goals of water quality compliance and irrigation guarantee under the constraints of facility safety and ecological base flow guarantee.

[0079] Step 3: In the control window Internally, using pollution flux estimation With the updated split ratio vector This constitutes load shaping at the inlet of the purification unit sequence and time-sharing inlet to the multi-stage water storage tanks, enabling water distribution commands. Defined as a hydraulic residence time vector With the inbound allocation vector Based on this, the pool position trajectory is jointly updated. Rolling estimate of available storage capacity .

[0080] If the high pollution load in the initial stage is directly introduced into the purification unit sequence without being shaped, the sedimentation section, filtration section and ecological section will be under pressure simultaneously within a short period of time, the hydraulic residence time will be passively shortened and the effluent fluctuations will be triggered.

[0081] To further translate the identification-triage results into a sustainable entry order, the updated triage ratio vector needs to be updated. With valve opening vector The actual flow rate vector corresponds to the inlet of the purification unit sequence, and the hydraulic retention time vector is set based on this. This ensures that each section prioritizes the absorption of suspended solids and particulate loads during the first half of the event and gradually restores balanced flow during the second half. Therefore, the inlet flow vector of the purification unit sequence is synthesized first, and then the pollution flux is estimated. The hydraulic residence time vector is proportionally gained and distributed to ultimately control the water distribution command. Write it as an executable payload sequence - pass sequence - dwell time three-element information.

[0082] To ensure that the purification unit sequence inlet only accepts non-initial water during permitted periods and maintains consistency with the phased bias of the updated diversion ratio vector, the valve opening vector needs to be adjusted. With the updated split ratio vector Perform dual alignment of the time domain and channel domain within the control window;

[0083] Therefore, a selection matrix is ​​introduced to extract the proportional components related to the purification unit sequence, and these components are expressed as hydrological response functions. Input rainfall data Pool position status The flow rate is converted into an equivalent inflow volume, thus obtaining the inlet flow vector of the purification unit sequence. This inlet flow vector is then used as the direct input for calculating the hydraulic retention and load-bearing sequence, avoiding fluctuations caused by high-frequency rule switching.

[0084] Where: the inlet flow vector of the purification unit sequence. : Non-negative real vector, indicating the distribution flow entering the settling section, filtration section and ecological section at any given time;

[0085] Time slice indicator function A binary set that restricts actions to occur only within the control window. Inside;

[0086] Initial runoff indicator : Binary set, shielding the initial stage of water inflow to the purification unit sequence; hydrological response function Non-negative real numbers, inputting rainfall data into the stream. Pool position status Mapped to equivalent water inflow;

[0087] Rainfall input stream : Non-negative real number, representing the rainfall driving force at that moment; pool position status : Non-negative real numbers, representing the water storage status; selection matrix : Real matrix, derived from the updated splitting ratio vector Select the proportional component pointing to the inlet of the purification unit sequence; update the diversion proportional vector. : Furthermore, the sum of the components is 1, providing a reference for the proportion of each channel at the current stage.

[0088] During use, the inlet flow vector of the purification unit sequence The alignment and synthesis ensures that the purification unit sequence only accepts non-initial inflow water during controllable periods and maintains consistency with the stage bias, thereby achieving load shaving at the source and significantly reducing the fluctuation range of subsequent hydraulic retention settings. By unifying the timing, direction, and flow rate onto a single vector, the calculation of subsequent retention and load capacity has a precise entry point.

[0089] The three segments of the purification unit sequence differ in their removal mechanisms and sensitivity to load, and the hydraulic retention time vector... It needs to be estimated based on pollution flux. Directional gain is applied to ensure that the settling section preferentially absorbs the peak value and allows for a stable transition between the filtration and ecological sections.

[0090] Therefore, based on the baseline hydraulic residence time vector, the pollution flux estimate is translated into a gain term according to stage weights and applied multiplicatively to each stage; thus, it tilts towards the pilot stage when pressure rises and restores equilibrium when pressure falls, ultimately yielding a water distribution command. Directly invoked hydraulic residence time vector:

[0091] Where: hydraulic residence time vector : Positive real vectors, representing the current residence time settings for the settling section, filtration section, and ecological section, respectively; Reference hydraulic residence time vector. : Positive real vector, representing the reference dwell time without pressure gain;

[0092] Hadamard product symbol : Superposition gain by component multiplication; unit vector : A single vector that maintains the basic unit of measurement;

[0093] Gain coefficient Non-negative real numbers, controlling the intensity of the pressure response; stage weight matrix. A real matrix that distributes a uniform pressure metric across three segments;

[0094] Normalized mapping vector Non-negative real vectors are used to estimate pollution flux. The conversion to a dimensionless stage gain involves transforming the original diversion proportion trajectory (which may be any real number or an intermediate quantity containing negative or missing values) into a vector within the 0–1 interval, with the sum of its elements equal to 1. This facilitates subsequent one-to-one field alignment with hydraulic residence time, inlet rhythm, and valve timing; pollution flux estimation : A non-negative real number representing the load pressure entering the processing chain.

[0095] When in use, the hydraulic residence time vector The pressure response setting enables the purification unit sequence to have an adaptive capability of lead buffering and step-by-step equalization for short-duration high loads, reducing the risk of overflow and short circuits and smoothing the effluent quality. Mapping pressure to a directional gain of residence time allows each of the three stages to perform its function while maintaining a recoverable equilibrium state. If the shaping output of the purification unit sequence cannot be aligned with the inflow time and destination of the multi-stage reservoirs, it will result in disordered stacking within the reservoir group, compressing the rolling estimate of available storage capacity. This also causes uneven fluctuations in the water quality within the pool.

[0096] Therefore, it is necessary to base it on the pool position trajectory. Based on the real-time state of the rolling estimate of available storage capacity, assign inbound vectors. Selective cross-pool selection prioritizes high-quality effluent to the pool with the highest capacity, irrigation capability, and lowest risk; therefore, within the control window... The internal inbound tendency is constructed using state vectors, and the inbound allocation vector is obtained by temperature-normalized mapping. This vector is then compared with the inlet flow vector of the purification unit sequence. With hydraulic residence time vector The system then updates the pool position trajectory in a manner that conserves mass. Rolling estimate of available storage capacity Finally, the water distribution order was issued. The pool group is implemented as an executable inbound sequence for a given time period.

[0097] Different reservoir types exhibit temporal variations in their capacity and water quality compatibility, leading to variations in the inflow allocation vector. These differences need to be directly reflected in the proportions; therefore, a state vector containing reservoir capacity and water quality suitability is constructed, and aggregated into a comparable inflow tendency using a linear mapping. Then, an inflow allocation vector is obtained using a temperature-normalized function. This strengthens the focus on the safest and most suitable pool type when resources are scarce, while maintaining a balanced approach when resources are plentiful.

[0098] In the formula: warehousing allocation vector : And the sum of the components is 1, giving the inbound ratio of each pool category; concentration parameter Positive real numbers, used to regulate the concentration of distribution; mapping matrix. : Real matrix, mapping the state vector to the inbound tendency domain; state vector Real vectors, jointly represented, can be used for rolling estimation of storage capacity. Decomposition amount and intermediate water quality state vector The degree of fit; can be estimated using inventory capacity rolling. : A non-negative real number representing a dynamic estimate of the remaining capacity currently available for storage;

[0099] Intermediate water quality state vector : A non-negative real vector representing the instantaneous water quality index at the outlet of the purification unit sequence. Within the three segments of the purification unit sequence, it is represented by the inlet flow vector. With hydraulic residence time vector As the primary control quantity, the removal and transformation are achieved in stages based on the mechanistic responses of sedimentation, filtration, and ecological segments, forming an intermediate water quality state vector at the outlet side. .

[0100] When using it, the vector is allocated into the database. The state-driven approach binds reservoir capacity, water quality, and inflow direction into an interpretable proportional quantity, prioritizing the placement of high-quality effluent into the most suitable reservoir type, reducing subsequent rebalancing costs and improving the availability of irrigation water. By making the destination of inflows a function of the state rather than a fixed rule, the behavior of the reservoir group becomes adaptive in real time.

[0101] Inbound allocation vector With the inlet flow vector of the purification unit sequence Having determined the quantity and direction of the incoming goods, this information still needs to be written into the pool position trajectory using the principles of mass conservation and inertial decay. Based on this, update the rolling estimate of available storage capacity. Therefore, a linear update method with storage inertia and release effects is used to characterize the short-term evolution of the pool position trajectory, and the available storage capacity is estimated by subtracting the current pool position from the total storage capacity.

[0102] The rolling estimate then flows back to the state vector at the next time step. With control window During the time-period screening, a unidirectional update chain is formed, ultimately stabilizing the relationship between inbound, pool location, and storage capacity within an interpretable range, where:

[0103] ; In the formula: pool position trajectory : A non-negative real number representing the water level or equivalent storage capacity in the pool at a given time; The predicted reservoir water level for the next time period is used to determine the safety of water supply during operation; the storage inertia coefficient... : This reflects the time inertia of the pool's response; the inbound conversion coefficient. Positive real numbers, converting the inbound quantity into pool position increments; all-one vectors. : For a single vector, sum the components; Hadamard product. Assign vectors to the inbound data based on components. With the inlet flow vector of the purification unit sequence Multiplication; Input allocation vector Each reservoir or inter-reservoir passage during a certain period The inbound ratio, with non-negative and normalized elements; inbound flow vector. Instantaneous flow rates of multiple branches from the outlet of the purification unit sequence;

[0104] Predicted water release : A non-negative real number representing a water release instruction. Expected release amount at the current moment; release impact coefficient Positive real number, converting water release into reservoir level drop; can be estimated using reservoir capacity rolling. Same as before, indicating the remaining capacity currently available for inbound storage; available storage capacity for the next period. The scheduling algorithm uses real-time constraints to determine whether the load has been reached or a transfer is needed; total storage capacity. : A positive real number that gives the upper limit of the capacity of the engineering design.

[0105] The definition of available outflow capacity is given as follows: Among them, all-one vectors Used for summation; function The structural outflow upper limit, which varies with the pool position trajectory, is a three-segment function of a polygonal line: first maintaining zero outflow (dead reservoir zone), then increasing linearly (regular scheduling zone), and finally increasing with a larger slope (flood control and discharge zone); constant. Indicates the rated capacity of the mechanical-hydraulic system in that time slice; time slice indication function Control the outflow outside the window to be zero.

[0106] During use, the pool position trajectory Rolling estimate of available storage capacity The joint update couples the purification side and the pool group side into a calculable evolution curve, enabling rapid determination of whether to reserve or release storage capacity during priority switching, thereby ensuring facility safety and ecological base flow protection. It completes the dynamic characterization of inflow-storage-outflow with minimal state variables, making the next switchover data reliable and computationally manageable.

[0107] By linking deviation detection with water release allocation, short-term shaping with time sequence alignment, and event assessment with template parameter updates, the system completes the sequential progression from deviation identification to ecological priority release, engineering safety protection, and simultaneous irrigation and water quality management. Ultimately, the event knowledge is precipitated into a reusable target mapping table and valve timing table template.

[0108] Step 4: In the control window Internally, when the running state is relative to the target mapping table When deviations occur, the water discharge instructions will be repeated in a prioritized manner. Water distribution instructions Flow splitting instructions The execution order and intensity are determined, and the monitoring and execution records are extracted into reusable template parameters after the event to ensure that the ecological base flow protection and facility safety are not compromised, while maintaining the consistent progress towards the goals of water quality compliance and irrigation supply guarantee.

[0109] In sloping farmland incidents, the constraints on water level, water quality, and ecological discharge often take effect simultaneously. If water release is switched based solely on static thresholds, it will lead to a dilemma of either excessive discharge resulting in the loss of irrigable water or insufficient discharge touching the safety boundary.

[0110] Therefore, it is necessary to analyze the pool position trajectory. Available storage capacity rolling estimation Intermediate water quality state vector With event target vector A deviation index is generated as the trigger; therefore, when the deviation index increases, the water allocation vector should automatically tilt towards the ecological and safety channels, and then return to the irrigation orientation after the risk is under control; thus, the water allocation vector and control window are derived accordingly. Alignment, then writing to the valve timing table. The executable fragments ultimately make identification-assignment-execution a single chain of actions rather than isolated, fragmented decisions.

[0111] Deviations may arise from reservoir capacity constraints, water quality fluctuations, or insufficient ecological discharge. In order to synthesize the three types of pressures into a single comparable quantity and maintain a continuous closeness to the boundary, a smooth additive synthesis and positive vectorization are required to drive subsequent allocations.

[0112] To address this, a deviation index is synthesized by weighting the reservoir capacity tension function, water quality vulnerability function, and ecological gap function. This index is used to trigger priority switching and characterize its intensity, thereby avoiding nonlinear jumps to a single threshold.

[0113] Where: Deviation index : Non-negative real numbers, comprehensively depicting the intensity of three types of pressure: reservoir capacity, safety and water quality, and ecology;

[0114] Weighting coefficient : Non-negative real numbers, representing strategy preferences and relative importance;

[0115] Storage capacity tension function : Non-negative real numbers, reflecting the degree of tension between the pool location trajectory and the rolling estimate of available reservoir capacity; the form can be found in the previous text; water quality vulnerability function : A non-negative real number that reflects the vulnerability of the intermediate water quality state vector relative to the water quality compliance target. The form can be referred to in the previous text.

[0116] Ecological gap function : Non-negative real numbers, representing the gap in ecological outflow demand based on the strength of the event target vector on the ecological component;

[0117] Ecological gap function Real-time ecological traffic With base current constraint The difference between them can be expressed as a piecewise linear equation:

[0118]

[0119] If the immediate water discharge does not meet the base flow requirement, then a proportional amount will be provided. For interval gaps, the function returns 0 once the flow rate is satisfied. The function is continuous and monotonically decreasing, facilitating the enforcement of hard constraints.

[0120] These three types of functions should satisfy the properties of monotonically increasing and piecewise differentiable, and should be smoothly extended at extreme values ​​to avoid frequent oscillations of the window boundary near the threshold. This requirement is fully compatible with existing formulas, only specifying the engineering selection of the function family;

[0121] When using, deviation index Expressing multi-source pressure on the same scale while maintaining fine-tunability ensures continuity between triggering and intensity, reduces operational oscillations caused by abrupt threshold switching, and facilitates control within the control window. An executable priority switching sequence is generated internally. Three types of pressure are aggregated by a single indicator, providing a unique driving force for subsequent allocation and timing.

[0122] Once the deviation is identified, the water allocation vector needs to be directed to the ecological and safety channels, and then gradually returned to irrigation and water quality optimization after the risk is mitigated.

[0123] To this end, a tradeoff that monotonically changes with the deviation index is constructed, which weights the ecological base vector and the preference allocation based on the event target vector, and uses a control window to prune the time slice, thereby obtaining a water allocation vector that can protect ecology and safety while taking into account irrigation and water quality.

[0124] In the formula: water distribution vector : A non-negative real vector that represents the water distribution for ecological channels, safety channels, and irrigation channels;

[0125] In the water distribution vector Add hard projection during generation: to ensure relationships.

[0126]

[0127] If the conditions are not met, ecological channels are supplemented by deducting from non-ecological channels according to their weights. Here, the ecological basis vector... Selecting ecological pathways, Hadamard product Used for component extraction. Time slice indication function. Binary sets, with allocation enabled within the control window; S-shaped functions. With shape factor Function output The shape coefficient is a positive real number, mapping the deviation exponent to ecological tendency weights; ecological basis vectors : Unit basis vectors, pointing to ecological baseflow protection channels; mapping matrix and : Real matrices, serving as structural mappings of channel reachability and target vector to preference domain, respectively; concentration parameter Positive real numbers, used to regulate the concentration of preference allocation; available outflow capacity. : A non-negative real number that limits the maximum amount of water that can be released within this time frame.

[0128] When in use, the water distribution vector The ecological tendency is monotonically adjusted according to the deviation index, and then smoothly reverts to the preference allocation after the risk decreases, and the entire process is consistent with the control window. Alignment ensures that ecological base flow and facility safety are prioritized without sacrificing the long-term stability of irrigable water volume. Continuing the deviation-bias-allocation process guarantees ecological priority and a return to a more sustainable irrigation orientation.

[0129] Simply distributing water is insufficient to guarantee the immediate stability of effluent quality and the downstream process's capacity to handle it, especially when the deviation index has just begun to decline. If the effluent is released directly without a short period of adjustment, it may cause transient fluctuations in water quality and instability in downstream processes. Therefore, it is necessary to vectorize the hydraulic residence time. Make minor directional adjustments to reinforce the filtration or ecological sections during critical periods;

[0130] Therefore, the intermediate water quality state vector and the deviation index are combined to form a quality correction vector, and a multiplicative gain is applied to the residence time. Then, the finely adjusted flow order and the water discharge distribution vector are time-aligned to form a water discharge timing segment, which is then executed by the valve timing table. Finally, it is ensured that the water is shaped first and then released, and that the same time axis is followed to prevent mutual interruption.

[0131] During the period when the risk has just eased, the filtration and ecological sections need to assume a higher responsibility for smoothing the flow. In order to avoid drastically changing the existing order, the quality correction vector is used as a dimensionless gain to make a small directional adjustment to the hydraulic residence time vector, so as to enhance its ability to suppress volatile indicators in a short period of time, while keeping the overall flux from changing abruptly.

[0132] In the formula: the adjusted hydraulic residence time vector. : Positive real vector, used as the dwell time setting for the short-term shaping phase; hydraulic dwell time vector The meaning is the same as before; the original dwell time setting; unit vector. : All-uniform vector, maintaining consistent dimensions; quality-corrected gain Non-negative real numbers control the fine-tuning intensity; allocation matrix : Real matrix, which projects the quality correction vector onto each segment; quality correction vector : A non-negative real vector, a dimensionless indicator synthesized from the intermediate water quality state vector and the deviation index, reflecting the indicators and segments that need to be strengthened.

[0133] When using, adjust the hydraulic residence time vector. Directional gains are applied only to key sections, without disrupting the existing order and suppressing transient fluctuations, resulting in more stable effluent quality before water release, which is beneficial to downstream ecology and water supply continuity. This minimal adjustment achieves short-term shaping, avoiding secondary disturbances caused by major changes to the existing order.

[0134] After short-term shaping, it is necessary to ensure that the water discharge execution is consistent with the valve timing table and to avoid competition between different channels in the same time slice. To this end, an alignment operator is used to link the water discharge allocation vector, the adjusted hydraulic residence time, and the control window into a water discharge timing segment, and this segment is mapped to the valve position index to ensure uniqueness on the execution side.

[0135] In the formula: water release sequence segment : A discrete sequence that describes the order and duration of water release in several future time slices;

[0136] Alignment Operator : Sequence generation mapping, which transcribes the water release allocation vector and dwell settings into a timing sequence within the control window; water release allocation vector The meaning is the same as before; it refers to the allocation of channels and proportions; the adjusted hydraulic residence time vector. The meaning is the same as before, providing plastic surgery support during critical periods; control window. : Same meaning as before, limits the allowed execution time range.

[0137] When using it, the water release sequence segment This ensures a unique sequence from valve placement to water release, preventing channel conflicts and time overlaps, and giving the valve timing schedule an interpretable and auditable linear progression. It transforms proportions and dwell times into specific timing sequences, ensuring that actions at the execution end do not conflict and have a clear order.

[0138] If the temporal relationship between deviation, water release, and water quality is not integrated, evaluated, and settled into template parameters after the event, the same trial-and-error path will be repeated in the next event, reducing the overall irrigation capacity and the stability of water quality compliance.

[0139] Therefore, an event performance function needs to be defined to aggregate deviations in ecology, engineering safety, water quality, and irrigation into a single quantitative score. After obtaining the score, the template parameter vector is updated using a convex combination method, enabling the target mapping table and valve timing table templates to possess interpretable progressive learning capabilities. This update directly benefits subsequent preliminary analysis and timing generation. Finally, without forming a closed loop, the knowledge is stored in a template library for later event calls. It is necessary to simultaneously measure the ecological release fulfillment rate, pool safety margin, water quality compliance stability, and irrigation supply satisfaction. To avoid conflicting indicators making decision-making difficult, the four types of deviations are aggregated into an event performance function in a weighted manner and ranked according to the principle of smaller is better. This ranking is used as the update weight for the template parameters, constructing the event performance function. :

[0140] ; Where: event performance function : Non-negative real number, representing scores for four categories of deviations: ecology, engineering, water quality, and irrigation / supply.

[0141] Weighting coefficient : Non-negative real numbers, representing the relative importance of the four types of biases in the evaluation; ecological bias function : A non-negative real number that measures the constraint of the water release allocation vector on the protection of ecological baseflow. The deviation in fulfillment rate;

[0142] ; The actual water discharge through the ecological channel during this period. Ecological base flows defined in planning or guidelines;

[0143] When the actual water release is lower than the baseflow, the proportional difference is directly used as the deviation; if the baseflow is met, the output is 0. This proportional gap method is widely used in ecological scheduling literature and facilitates supervision and verification.

[0144] Safety Deviation Function : A non-negative real number that measures the relative safety upper limit of the pool position trajectory. margin deviation;

[0145] In the formula: Real-time pool position (or equivalent storage capacity). Safety control water level / flood limit water level Sigmoid slope (steepness coefficient)

[0146] The Sigmoid function causes a rapid rise in the function near the safe water level, and is both differentiable and easily adjustable in sensitivity. Similar forms are often used for flood control constraints in reservoir MPC (Mean Control Limitation).

[0147] Water quality deviation function : A non-negative real number used to measure the intermediate water quality state vector relative to the water quality reference vector. Deviation;

[0148] In the formula: The overall water quality vector at the outlet (COD, SS, TN, TP...); , Irrigation or emission limit vector; Indicator weight matrix ; Slope, controlling sensitivity; Weighting of water quality indicators;

[0149] The Euclidean distance measures the degree of exceedance of multiple indicators, which is then normalized by Sigmoid; it is suitable for docking with WQI classification results and can highlight key pollutants through weighting.

[0150] Irrigation deviation function : A non-negative real number that measures the actual available irrigation flow relative to the crop's water requirement. Satisfaction deviation; let Total inbound flow at different times , Let be the evapotranspiration water requirement of the crop at the current time.

[0151] In the formula: , Input and allocate vector (elements normalized); Instantaneous flow vectors for each channel Water requirement for crop evapotranspiration;

[0152] When the inflow is less than the evapotranspiration demand, the deviation increases proportionally; if it is greater than or equal to the demand, the deviation is 0. This irrigation gap function is consistent with agricultural deficit-irrigation research and is easily coupled with crop models or DSSAT outputs.

[0153] Ecological base flow protection constraints Safety Limit Water quality reference vector : The values ​​are non-negative real numbers, non-negative real numbers, and non-negative real vectors, respectively, serving as a reference scale for ecology, safety, and water quality;

[0154] In the water distribution vector Before execution, a prediction is made using a pool-level recursive method to verify the safety limit. When forecasts exceed the limits, the weighting of ecological and safety considerations is increased, and the direction of irrigation and water inflow is adjusted downwards until the forecasts do not exceed the limits.

[0155] Among them, the storage inertia coefficient The average effect of evaporation and leakage can be absorbed, and the inbound conversion coefficient is [missing information]. With release effect coefficient It reflects the structural and hydraulic conditions.

[0156] When using event performance functions Expressing the four types of biases with the same dimension facilitates cross-event comparison and ranking, directly providing an interpretable weight basis for template parameter updates, avoiding local bias caused by single-index optimization, and turning multi-objective biases into comparable numbers with a single score to serve subsequent template updates.

[0157] With the event performance function, template parameters can be updated intuitively by prioritizing smaller values. To maintain stability and interpretability, a convex combination is used to update the template parameter vector. The representative summary vector of the current event is mixed with existing parameters according to weights. The updated parameters are directly read in the next event and do not form a feedback loop in the current event.

[0158] In the formula: template parameter vector With the updated template parameter vector : Real vector, carrying adjustable parameters of the target mapping table and valve timing table template; updates weights : According to the event performance function Convex combination weights set by relative performance; summary mapping Vector-generating operators average or represent the state of events. The mapping is a vector of the same type as the template parameters. It compresses the time series, valve actions and monitoring curves during an event into a fixed-length vector with the same dimension and field order as the template parameters, so that it can be directly written back to the target mapping table template or valve timing table template.

[0159] Event target vector average state Average state of pool position trajectory , intermediate water quality state vector average state The average state can be estimated using the rolling storage capacity estimate. Valve timing table average state : The values ​​can be either vectors or scalars, representing the statistical summary of the template update.

[0160] In practice, the convex combination update of the template parameter vectors preserves event knowledge in an interpretable, parameterized way. This avoids overly following a single event while gradually refining the target mapping table and valve timing table templates, making the preliminary analysis and execution of the next event closer to real-world operational characteristics. Learning remains between events, rather than within events, ensuring stable evolution and avoiding oscillations caused by immediate feedback.

[0161] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0162] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0163] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0164] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0165] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations 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. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for slope farmland runoff collection, irrigation and cascade purification based on intelligent diversion, characterized in that: comprises, collecting rainfall, monitoring, crop water requirement, pool position and water quality state and ecological base flow constraints, generating target mapping table and control window, setting diversion instruction, water distribution instruction, water release instruction reference field and valve timing table and diversion proportion trajectory benchmark; In the control window, according to the rainfall sequence and the water quality state, the initial runoff is determined, the diversion instruction is triggered to guide it into the initial temporary storage unit, and according to the target mapping table, it is directed to the purification unit sequence and the multi-stage water storage pool to form the valve timing segment; According to the target mapping table and the diversion proportion trajectory, the water distribution instruction and the hydraulic retention time setting are formed, the sequence and the residence distribution in the purification unit sequence are determined, and the multi-stage water storage pool is divided into time storage, and the pool position trajectory and the rolling estimation of available storage capacity are output; When the operation deviates from the target mapping table, referring to the pool position trajectory and the water quality state, rearrange the water release instruction, the water distribution instruction and the diversion instruction, generate the water release timing sequence and write it into the valve timing table, and update the target mapping table and the valve timing table template.

2. The slope field runoff collection irrigation and step-by-step purification method according to claim 1, characterized in that: The target mapping table includes a guarantee field for irrigation, a water quality standard field, a facility safety field, an ecological base flow guarantee field, and reference fields for the diversion instruction, the water distribution instruction, and the water release instruction. The control window is generated by weighted rules of rainfall response, storage capacity tension, and water quality vulnerability, and corresponds to the time index of the target mapping table.

3. The slope field runoff collection irrigation and step-by-step purification method according to claim 2, characterized in that: The initial runoff determination integrates the initial rising characteristics of the rainfall segment and the mutation characteristics of the water quality state to form a trigger signal. The diversion instruction determines the passage of the initial temporary storage unit and the inlet of the purification unit sequence according to the target mapping table, and sets mutually exclusive segments and allowed segments in the valve timing table.

4. The slope field runoff collection irrigation and step-by-step purification method according to claim 3, characterized in that: The initial temporary storage unit includes a temporary storage pool connected with an upstream intercepting ditch, a controllable overflow weir connected with the inlet of the purification unit sequence, and an emptying branch; The diversion instruction limits the initial entry into the initial temporary storage unit, then switches to the purification unit sequence according to the valve timing table, and restores the normal passage after disposal at the emptying branch.

5. The slope field runoff collection irrigation and step-by-step purification method according to claim 4, characterized in that: The purification unit sequence includes a sedimentation segment, a filtration segment, and an ecological segment. The hydraulic retention time setting is given according to the residence distribution in the target mapping table corresponding to the pollution flux estimation, and the obtained through sequence is aligned with the diversion proportion trajectory to write into the valve timing table for inlet rhythm control.

6. The slope field runoff collection irrigation and step-by-step purification method according to claim 5, characterized in that: The flow distribution at the inlet of the purification unit sequence is determined by the updated diversion proportion trajectory and the available water quantity in the control window. The settling section sets the effective water depth by the gate height and bypass combination, the filtering section sets the passing rate by the inlet and outlet throttling, and the ecological section sets the residence scale by the branch selection and water depth fine-tuning.

7. The method according to claim 6, wherein: The time-sharing storage of the multi-stage reservoir is determined by the storage allocation vector, the target pool and proportion are determined according to the pool trajectory, the rolling estimation of available storage and the intermediate water quality state; when the pool trajectory approaches the safety upper limit, the inter-pool transfer is performed according to the valve timing table and the inlet rhythm is simultaneously limited.

8. The method according to claim 7, wherein: The water discharge priority switching generates the water discharge allocation vector based on the deviation index weighted by the storage tension, water quality vulnerability and ecological gap, and the time period is clipped by the control window, and the hard constraint of ecological base flow guarantee is set in the allocation layer, and when the constraint value is insufficient, the non-ecological channel is deducted by weight to make up, and then written into the valve timing table.

9. The method according to claim 8, wherein: After generating the water discharge allocation vector, the quality-oriented short-time fine-tuning is performed on the hydraulic residence time setting, the fine-tuning is limited to small and monotonous gain of the filtering section and the ecological section, the passing order is not changed, and the timing alignment and segment splicing are performed before generating the water discharge timing and the control window and the valve timing table.

10. The method according to claim 9, wherein: The water discharge timing is generated by the timing alignment of the water discharge allocation vector and the hydraulic residence time setting, each channel is set with minimum duration and minimum interval of adjacent actions to limit concurrent channels in the same time slice, and the water discharge timing is mapped to the unique valve index of the valve timing table for execution.

11. The method according to claim 10, wherein: After the event ends, the monitoring records and execution records are collected, the event performance function including ecology, engineering safety, water quality and irrigation deviation is constructed, the weight is determined according to the function, the convex combination update is performed on the template parameter vector, the updated template parameter vector is obtained, and is used to update the target mapping table and the valve timing table template for subsequent event calling.

Images