Carbonate wastewater treatment parameter matching method and system based on process model

Abstract

The application discloses a carbonic acid ester wastewater treatment parameter matching method and system based on a process model, and belongs to the field of industrial wastewater treatment process control and process parameter optimization. The method takes the cumulative water volume as the material coordinate to recursively propagate the proportion trajectory of the production batch in each treatment unit, pairs the discharge end component load with the same batch biochemical response, estimates the influent load state and the residual capacity state of the treatment unit, and solves the executable parameter combination under the effluent, resource consumption, equipment capacity, safety and parameter compatibility constraints. After execution, the node residual is formed according to the same batch trajectory, and the influent state or unit capacity is corrected, and the prediction interval is expanded when the evidence is insufficient. Therefore, the parameter mismatch, error attribution and process fluctuation caused by batch switching are reduced.

Description

Technical Field

[0001] This invention relates to the field of industrial wastewater treatment process control and process parameter optimization, and in particular to a method and system for matching parameters in carbonate wastewater treatment based on process models. Background Technology

[0002] The production, separation, and equipment cleaning processes of carbonates generate wastewater containing carbonate organic matter, alcohols, by-products, and salts. Existing treatment systems typically set reagent dosages, pH, reaction temperatures, residence times, aeration rates, and reflux ratios based on individual influent parameters, fixed thresholds, or operator experience, and can maintain basic operation when the influent composition and load are relatively stable.

[0003] When production batches and operating conditions change, wastewater flow rate, chemical oxygen demand, pH, salinity, temperature, and organic components will fluctuate in tandem. Measurements at a single moment or at a single node are insufficient to reflect the mixing and lag of different batches of materials in the conditioning, physicochemical, and biological treatment units, and it is also difficult to correlate the components at the discharge end with the respiration response of the same batch after arriving at the biological treatment unit. This can easily lead to distortions in the judgment of the biodegradability and inhibition status of the influent.

[0004] Furthermore, there is coupling between the remaining capacity of the treatment unit, equipment boundaries, and multiple control parameters. Adjusting a single parameter independently may lead to parameter conflicts, overdosing, or excessive aeration. When actual results deviate from predictions, existing methods struggle to determine whether the deviation originates from changes in influent or a decline in treatment unit performance. Incorrect attribution can cause control oscillations and effluent fluctuations.

[0005] Therefore, there is a need for a process model-based parameter matching method and system for carbonate wastewater treatment that can overcome the shortcomings of the existing technologies. Summary of the Invention

[0006] One objective of this invention is to propose a parameter matching method and system for carbonate wastewater treatment based on a process model. Addressing the issues of nonlinear coupling between influent components, pollution load, biodegradability, treatment capacity, and equipment boundaries during batch and operating condition switching, as well as the difficulty in distinguishing the actual sources of deviation, the method establishes a batch proportion trajectory based on cumulative flow volume. This completes the pairing of component loads and biochemical responses within the same batch and estimates the influent load state and the remaining capacity state of the treatment unit. The state is then input into the process model, and executable parameter combinations are solved in a rolling manner under constraints of effluent, resource consumption, equipment capacity, safety, and parameter compatibility. Finally, the corresponding state is corrected and constraints are updated through dual-channel routing of node residuals, thereby forming a closed-loop parameter matching system adapted to batch migration and unit capacity.

[0007] This invention provides a method for matching parameters in carbonate wastewater treatment based on a process model, comprising: S1: Collect water quality, flow rate, biochemical response and equipment operation data of carbonate wastewater discharge and treatment nodes, and establish a node status dataset with production batch labels using the cumulative water volume as material coordinates. S2, based on the effective volume, inflow and outflow, return flow and mixing or dwell characteristics of each processing unit, the node status dataset is recursively calculated to obtain the batch proportion trajectory of each production batch in each processing unit. S3. Based on the batch proportion trajectory, the load of the emission end component is matched with the biochemical response when the same production batch arrives at the biochemical unit, and the influent load status is obtained by combining material balance and reaction kinetics. The remaining capacity status of each treatment unit is obtained according to the load removal amount and equipment output of the treatment node. S4, take the influent load state and the remaining capacity state as the state input of the process model, and use the predicted effluent boundary, resource consumption target, equipment capacity boundary, safety boundary and compatibility relationship between treatment parameters as constraints, and solve the executable parameter combination of reagent dosage, pH, reaction temperature, residence time, aeration rate and reflux ratio in a rolling manner. S5, execute the executable parameter combination of the current control segment, align the measured results and predicted results of the processing node according to the batch proportion trajectory to form node residuals, and select to correct the water inlet load state, correct the remaining capacity state of the corresponding processing unit, or expand the prediction interval according to the propagation relationship of the node residuals along the material trajectory and the continuous deviation relationship of different production batches in the same processing unit. Then, update the constraints using the corrected state or prediction interval and re-execute step S4.

[0008] Optionally, S1 includes: collecting cumulative flow, chemical oxygen demand, pH, salinity or conductivity, temperature, characterization of carbonate organic matter, characterization of alcohols, characterization of by-reaction products, respiration rate and equipment operation data; The production batch label is determined based on the production emission switching signal and cumulative flow change, and the collection time of each data is mapped to the cumulative water volume to form the node status dataset; When any collected data exceeds the calibrated range of the corresponding sensor or the duration for which the continuously collected value remains unchanged exceeds the allowable duration in the equipment maintenance record, the collected data is marked as invalid and its data source record is retained, so that the invalid data does not participate in the update of the corresponding influent load status or remaining capacity status. When the collected data re-enters the calibrated range and the continuously collected values ​​change within the allowed duration, the collected data is restored to valid status.

[0009] Optionally, S2 includes: for each processing unit, calculating the cumulative material replacement amount based on the effective volume and the inflow and outflow rates, establishing the reflux material transfer relationship based on the reflux flow rate, and characterizing the mixing or residence characteristics using a complete mixing model, a series mixing model, or a residence time distribution determined by tracer test records; The proportion of each production batch entering the processing unit is combined with the cumulative material replacement amount, the return material transfer relationship and the mixing or residence characteristics to recursively deduce that the sum of the proportions of each production batch under the same material coordinate is one, thus obtaining the batch proportion trajectory. When the flow is interrupted, the cumulative material replacement amount is frozen, and it continues to be calculated from the cumulative flow at the time of flow recovery after the flow is restored. Furthermore, when traffic data is marked as invalid, the sum of batch proportions deviates from one, or the recursive result exceeds the dwell time distribution boundary of the processing unit, the corresponding batch proportion trajectory is stopped from being published to subsequent steps, the abnormal status record is saved, and a maintenance alarm is issued. After valid flow data is regained and the sum of batch percentages in two consecutive cumulative material replacement cycles is restored to one, the maintenance alarm is lifted and the restored batch percentage trajectory is published.

[0010] Optionally, S3 includes: using the batch proportion trajectory to perform weighted separation of the respiratory rate of the biochemical unit to obtain the respiratory response increment corresponding to each production batch; By inputting the respiratory response increment, emission component load, inlet and outlet loads of the treatment node, and reaction kinetic parameters into the material balance relationship, the influent load state, including readily biodegradable load, poorly biodegradable load, and inhibition coefficient, is obtained. The capacity utilization rate is determined by the ratio of the measured removal amount of the processing node to the available output of the corresponding processing unit under the equipment calibration conditions. The product of the available output and the capacity utilization rate is determined as the occupied output. The remaining capacity status is obtained by subtracting the occupied output from the available output.

[0011] Optionally, S4 includes: determining the predicted effluent boundary based on the effluent discharge limit and the prediction error range; determining the resource consumption target based on the reagent consumption and aeration energy consumption; determining the equipment capacity boundary based on the equipment rated output and maintenance derating records; and determining the safety boundary based on the permitted reagent dosing range, the permitted pH range, the permitted reaction temperature range, and the interlocking conditions of the aeration equipment. The compatibility relationship is determined based on the allowable operating condition records of each processing unit, equipment interlock records, and the feasible domain of parameters verified by historical processing results; Input the dosage of the reagent, pH, reaction temperature, residence time, aeration rate and reflux ratio into the process model, eliminate parameter combinations that violate the safety boundary, equipment capacity boundary or compatibility relationship, and select the parameter combination that satisfies the predicted effluent boundary and has the minimum resource consumption target value from the remaining parameter combinations as the executable parameter combination; When any data in the allowable operating condition record, equipment interlock record, or parameter feasible field is unavailable, the corresponding processing parameter will be restricted to the parameter field that was most recently verified by the historical processing result of the processing unit. Furthermore, the safety boundary and equipment capability boundary are set as strong constraints that do not allow the introduction of slack, and the resource consumption target is set as a soft constraint that allows the introduction of non-negative slack and applies a penalty weight to the non-negative slack. When no parameter combination satisfies the strong constraints and the predicted effluent boundary, the output includes a safety backoff instruction that includes at least one of reducing the influent flow rate to within the equipment capacity boundary, extending the residence time, and activating the standby treatment equipment, and prohibits the output of parameter combinations that increase the dosage of chemicals or the aeration rate beyond the safety boundary.

[0012] Optionally, S5 includes: forming a residual evidence set by combining the component residuals, breathing residuals, inlet and outlet load residuals of the processing unit, equipment output deviations and deviation duration periods corresponding to the same production batch; When the component residual and respiration residual appear sequentially at two or more downstream treatment nodes along the batch percentage trajectory, the correction amount will be allocated to the influent load state. When different production batches continuously exhibit the same load residual in the same processing unit and the equipment output deviation exceeds the output deviation boundary determined by the equipment calibration error, the correction amount will be allocated to the remaining capacity state of the processing unit. When both of the above allocation conditions are met, if the component residual and respiration residual continue to propagate to the downstream treatment node of the treatment unit, only the influent load state is corrected; otherwise, only the remaining capacity state of the treatment unit is corrected. When neither of the above two allocation conditions is met, the state is not changed, the version of the prediction interval before expansion is recorded, and the prediction interval is expanded according to the dispersion of the residual evidence set. The upper limit of the dispersion of the residual evidence set of historical production batches that use the version of the prediction interval before expansion and whose effluent meets the predicted effluent boundary is determined as the recovery threshold. When the dispersion of the residual evidence set of two consecutive subsequent production batches decreases to within the recovery threshold, or when one of the above two allocation conditions is met, the prediction interval is restored to the version of the prediction interval before expansion, and the safety boundary and equipment capability boundary are not relaxed. Furthermore, the production batch label, material coordinate range, residual evidence set, correction object, state before correction, state after correction, and constraint boundary version for each state correction are written into the verification record. The asymmetric tightening boundary is selected according to the correction object. When the correction object is the inlet load state, the upper boundary of the inlet load of the subsequent control section is tightened. When the correction object is the remaining capacity state, the upper boundary of the equipment output of the corresponding processing unit is lowered and the change rate boundary of the corresponding processing parameter is tightened. The state correction is verified based on the deviation between the subsequent measured results and the predicted results of the same batch, and the constraint boundary is restored to the original boundary corresponding to the version of the constraint boundary if the deviation does not decrease.

[0013] On the other hand, the present invention also provides a carbonate wastewater treatment parameter matching system based on a process model, comprising: The data acquisition module is used to execute step S1; the trajectory recursion module is used to execute step S2; the state estimation module is used to execute step S3; the parameter matching module is used to execute step S4; and the feedback correction module is used to execute step S5.

[0014] The beneficial effects of this invention are: 1. Using the cumulative water volume as the material coordinate and combining the treatment unit volume, flow rate, reflux, and mixing or residence characteristics to recursively deduce the batch proportion trajectory, so that the discharge load, biochemical response and treatment results can be aligned according to the same production batch, reducing state mismatch caused by hydraulic lag and batch mixing.

[0015] 2. Simultaneously estimate the easily biochemical load, the difficult biochemical load, the inhibition state, and the remaining capacity of the treatment unit, and solve the parameter combination under the common constraints of predicted effluent, resource consumption, equipment capacity, safety, and parameter compatibility, so as to coordinate the dosing of chemicals, pH, temperature, residence time, aeration, and reflux control.

[0016] 3. Based on the propagation relationship of node residuals along the material trajectory and the correction object of deviation routing across batches and units, when the evidence is insufficient, the prediction interval is expanded and the recovery mechanism is retained, which can reduce excessive dosing, excessive aeration, control oscillation and effluent fluctuations caused by incorrect attribution. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the method for matching parameters in carbonate wastewater treatment based on a process model, as described in this invention.

[0018] Figure 2This is a feedback control flowchart for S5 of the present invention, which corrects the state or expands the prediction interval based on the node residual. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-2 A method for matching parameters in carbonate wastewater treatment based on a process model, including: S1: Collect water quality, flow rate, biochemical response and equipment operation data of carbonate wastewater discharge and treatment nodes, and establish a node status dataset with production batch labels using the cumulative water volume as material coordinates. S2, based on the effective volume, inflow and outflow, return flow and mixing or dwell characteristics of each processing unit, the node status dataset is recursively calculated to obtain the batch proportion trajectory of each production batch in each processing unit. S3. Based on the batch proportion trajectory, the load of the emission end component is matched with the biochemical response when the same production batch arrives at the biochemical unit, and the influent load status is obtained by combining material balance and reaction kinetics. The remaining capacity status of each treatment unit is obtained according to the load removal amount and equipment output of the treatment node. S4, take the influent load state and the remaining capacity state as the state input of the process model, and use the predicted effluent boundary, resource consumption target, equipment capacity boundary, safety boundary and compatibility relationship between treatment parameters as constraints, and solve the executable parameter combination of reagent dosage, pH, reaction temperature, residence time, aeration rate and reflux ratio in a rolling manner. S5, execute the executable parameter combination of the current control segment, align the measured results and predicted results of the processing node according to the batch proportion trajectory to form node residuals, and select to correct the water inlet load state, correct the remaining capacity state of the corresponding processing unit, or expand the prediction interval according to the propagation relationship of the node residuals along the material trajectory and the continuous deviation relationship of different production batches in the same processing unit. Then, update the constraints using the corrected state or prediction interval and re-execute step S4.

[0021] In this specific embodiment, S1 includes: This specific implementation takes a continuously operating carbonate production wastewater treatment line as the example. The treatment line sequentially includes a discharge collection end, a regulating unit, a physicochemical reaction unit, a biochemical unit, and an end monitoring node. The biochemical unit is divided into a front-end biochemical subunit and a back-end biochemical subunit along the material trajectory, and each is equipped with a respiration response observation node. The physicochemical reaction unit and the biochemical unit are respectively equipped with temperature sensors, heat exchange jackets, hot and cold medium regulating valves, and temperature controller interfaces. The temperature controller stores the allowable heating and cooling rates, heat exchange capacity boundaries, hot and cold medium pressure interlocks, and setpoint reception confirmation. The data acquisition module writes the node number, production batch label, acquisition time, and cumulative data for each record. Flow rate, instantaneous flow rate, chemical oxygen demand, pH, salinity characterized by conductivity, temperature, characterization of carbonate organic matter, characterization of alcohols, characterization of by-reaction products, respiration rate, equipment current, valve position, pump frequency, aeration volume, and data validity flags are all included. The data at each node is timestamped using a unified clock, and the corrected records are written to the node status data cache. If the target processing line is not equipped with an active temperature control loop, the reaction temperature is used as the measured disturbance quantity and parameter range index. When the S4 candidate is generated, it is fixed as the predicted value of the current measured temperature. No temperature setpoint is issued. When the temperature exceeds the feasible range, only flow reduction, extended residence time, or manual handling are allowed. The discharge flow meter outputs instantaneous and cumulative flow rates at 1-minute intervals; the water quality analyzer outputs chemical oxygen demand (COD), pH, conductivity, and temperature at 5-minute intervals; the online chromatograph outputs characterization values ​​of three organic components at 15-minute intervals; and the online respirators in the upstream and downstream biochemical subunits output the oxygen consumption rate per unit effective tank volume at 1-minute intervals. When any online respirator is under maintenance, samples are taken at the corresponding node at 30-minute intervals, and offline respiration measurements are performed under constant temperature, same dissolved oxygen setting, and same sample volume conditions. The offline results are recorded with the node number, sampling time, detection time, and measurement method flag. The equipment controller reads the operating fields of the pump, agitator, dosing device, aeration device, and reflux device at 1-minute intervals. The data acquisition module holds the low-frequency water quality value until the next effective sampling time, but retains the original sampling time and hold flag in the record so that subsequent state estimation can distinguish between measured values ​​and held values. The data acquisition module uses the cumulative water volume as the material coordinate, according to... Update the current material coordinates, where, Indicates the first The cumulative water volume at the end of each collection cycle, in cubic meters. 3 This is derived from cross-validation of current cumulative traffic records and recursive results. This represents the cumulative water volume at the end of the previous data collection cycle, in cubic meters (m³). 3 , Indicates the first The effective instantaneous flow rate for each collection cycle, in m³ / s. 3 / h, sourced from the flow meter at the emission end. This indicates the data collection period, expressed in hours (h). In this specific implementation, it is set to 1 / 60 h. This represents the initial cumulative water volume at the flow meter reset point, in cubic meters (m³). 3 The value originates from reset messages and manual verification records. During a normal reset, it is set to 0. During a reset with a reference cumulative value, the verified reference cumulative value is used. The first acquisition cycle starts with this initial value. This represents the sampling period index from the current cumulative flow meter reset point. It is a positive integer, and the unit is not applicable because it is a discrete period index. When the difference between the cumulative value and the recursive value given by the flow meter does not exceed the flow meter calibration error, the cumulative value of the flow meter is used; otherwise, the recursive value is retained and a verification flag is generated. When the production control system switches the production grade, raw material ratio, or reactor discharge path, it outputs a discharge switching signal. The data acquisition module establishes a new production batch label based on the rising edge of the signal and sets the material coordinates corresponding to the rising edge as the batch start coordinates and the material coordinates corresponding to the next rising edge as the batch end coordinates. If the switching signal is missing, a candidate batch boundary is established based on the combined conditions that the cumulative water volume increment reaches the single reactor calibration drainage volume set in this specific embodiment and the characterization amount of at least one component in the online chromatography changes by more than 20% relative to the previous stable window. The candidate boundary enters the set to be checked. For the set to be verified, the data acquisition module first deletes candidate boundaries that are less than 80% of the single reactor calibration drainage volume or whose component changes only last for one chromatographic sampling cycle. Then, it sorts the candidate boundaries in ascending order according to the absolute time difference between the candidate boundaries and the reactor number switching time of the production execution system. If the time differences are the same, the candidate boundary with the larger peak value of the cumulative volume change rate is selected. If they are still the same, the candidate boundary with the earlier material coordinates is selected. The selected boundary takes effect after being verified by at least two of the three records of reactor number, production grade and discharge path being consistent. Candidates that fail the verification are retained in the verification log but no production batch label is generated. Sensor calibration records are managed uniformly by the equipment maintenance database. Flow meters are calibrated using a standard volumetric tank at three points. Chemical oxygen demand and component analyzers are calibrated using certified standard solutions covering the low, medium, and high ranges. pH, conductivity, and temperature sensors are calibrated using standard buffer solutions, standard conductivity solutions, and traceable thermometers, respectively. The calibration version is valid only when the relative errors at the three points do not exceed the maximum permissible error given in the instruction manual of each device. Calibration versions that fail verification shall not be used as the source of range, resolution, or permissible invariance duration. After each production batch, the data acquisition module performs data integrity verification, and counts the number of records to be collected, the number of valid records, the monotonicity of timestamps, and the monotonicity of material coordinates node by node. When the ratio of the number of valid records to the number of records to be collected is less than 95%, the batch dataset is marked as incomplete. When timestamps or material coordinates are in reverse order, they are rearranged according to the original message sequence number and the rearrangement log is retained. Conflict records that cannot be eliminated by the message sequence number retain the original values ​​of the online instrument according to the source priority, while manually entered values ​​are marked as pending review. For asynchronous indicators that need to be mapped to the same material coordinates, the module performs linear interpolation only when the interval between two consecutive valid sampling points does not exceed the instrument's allowed holding time. The interpolation record saves the left and right valid record pointers and interpolation weights. When the interval exceeds the allowed holding time, no interpolation value is generated and a missing value is written. When reading data in subsequent steps, the original or interpolation type is read at the same time, so that the state estimation can increase the uncertainty of the interpolation data and will not treat the missing value as zero. For the data acquired at each processing node at the acquisition time, the data acquisition module performs piecewise linear integration using two adjacent valid flow sampling points, mapping the acquisition time to the node's local cumulative material coordinates. The discharge collection end record is written into the production batch label according to the confirmed batch start and end coordinates. Downstream processing nodes, due to the existence of effective volume, mixing, backflow, and residence lag, are first written into the unresolved batch label set. This set is generated by the initial inventory batch, the node flow meter coordinate origin, and the previous valid trajectory version, and is not directly assigned a single batch label based on the discharge end start and end coordinates. Each record in the node status dataset includes at least the node number, material coordinates, batch label or unresolved batch label set, original acquisition time, calibration acquisition time, indicator name, indicator value, unit, sensor number, calibration version, valid flag, and source record pointer. The record is indexed by the node number and material coordinates. Different indicators under the same index are retained according to the original sampling time, and the interpolated value is not overwritten with the original value. The equipment maintenance database saves the calibrated range, resolution, allowable invariance duration, and most recent calibration date for each sensor. When the collected value is lower than the lower limit of the range or higher than the upper limit of the range, it is immediately marked as invalid. When the difference between consecutive collected values ​​is less than the corresponding resolution and the duration exceeds the allowable invariance duration, it is also marked as invalid. Invalid records still save the sensor number, original value, diagnostic code, and source message, but their validity flag is set to zero and they are filtered out when the influent load status or remaining capacity status is updated in subsequent updates. When an invalid indicator re-enters the calibration range and changes greater than the corresponding resolution at least twice within the allowed constant time, the data acquisition module restores the records after the last change time to be valid. Invalid records before restoration are not filled back. If chemical oxygen demand or component characterization is temporarily invalid, only the state components dependent on that indicator are frozen. If the flow data is invalid, the flow invalid status is written to S2 without publishing new material coordinate increments to avoid invalid data participating in batch proportion recursion. Before the node status dataset is released, batch label closure verification is performed. The module confirms that each completed batch has unique start coordinates, end coordinates and production execution system batch number. When the boundaries of two batches overlap, one boundary is retained according to the rule of priority of verified switching signals and secondary to candidate boundaries, and the conflict segment is marked as pending review. When the conflict segment accounts for more than 2% of the corresponding batch volume, the release of the batch is suspended. After the operator checks the production records, he can only add confirmation records and cannot delete the original conflict records. At the end of each collection cycle, the data collection module submits the incremental node status dataset in ascending order of material coordinates, and writes the dataset version, start and end material coordinates, set of production batch labels or pending batch labels, effective indicator bitmap, and source record hash into the release list. The release list and node status dataset are stored together in the process database, serving as the sole data entry point for S2 to recursively predict the batch percentage trajectory and S3 to estimate the load and capacity status. After S2 completes the trajectory recursion, it writes back the determined batch labels and percentages of downstream records with the percentage of each batch and the coordinates of the arriving materials. When starting for the first time and there is no effective trajectory, the initial inventory composition verified by manual verification is used and the pending mark is maintained. During the trajectory stoppage period, pending records are not converted into determined labels.

[0022] In this specific embodiment, S2 includes: The trajectory recursion module reads the node status dataset, volume ledger of each processing unit, inflow and outflow records, return flow records and mixing characteristic configurations published by S1, and establishes a unit status table according to the adjustment unit, physicochemical reaction unit and biochemical unit. The unit status table shall at least save the unit number, effective volume, mixing model type, current material coordinates, proportion of each production batch, return source unit, residence time distribution version, last effective flow time and trajectory publication status. The effective volume is the working volume obtained by shutdown liquid level calibration and online liquid level conversion. In this specific implementation, it is reviewed once a month. For unit The Each recursive cycle, the trajectory recursion module according to... Calculate the cumulative increase in material replacement volume, where, Representation unit In the Net inflow volume per cycle, in m³ 3 , This indicates the external inflow rate of the unit, in meters per second (m³). 3 / h, This indicates the flow rate returning to this unit, in meters per second (m³). 3 / h, This represents the flow rate extracted from this unit and returned upstream, in meters (m). 3 / h, all three traffic items originate from valid traffic records in S1. This indicates the recursion period, in hours (h), which in this specific implementation is 1 / 60 h. This represents the processing unit index; units are not applicable because it is an index for discrete objects. This represents a recursive periodic index, taking positive integers; units are not applicable because it is a discrete periodic index; it only applies when... When the value is not less than 0, the recursion is executed. When a negative value occurs, the flow status of the current cycle is marked as invalid, the trajectory is stopped and the previous valid status is used. At the same time, the three flow items, the instrument version and the time of the abnormality are recorded. The material coordinates are not advanced with negative increments. For units configured as fully hybrid, the trajectory recursion module follows... Update batch percentage, of which, Representation unit Domestic production batch The volume percentage in the next cycle is a dimensionless quantity ranging from 0 to 1. Representation unit The percentage of this batch saved in the previous iteration of the status table is a dimensionless quantity ranging from 0 to 1. This represents the proportion of this batch in the external influent read from the synchronous output trajectory of the upstream main process, and is a dimensionless quantity ranging from 0 to 1. This indicates the proportion of this batch of refluxed material read from the synchronous output trajectory of the reflux source unit, a dimensionless quantity ranging from 0 to 1, and is calculated as follows: Calculate the external influent replacement coefficient. The external influent replacement coefficient is a dimensionless quantity ranging from 0 to 1, calculated as follows: Calculate the reflux displacement coefficient. This represents the reflux displacement coefficient, a dimensionless quantity ranging from 0 to 1. This represents the external influent flow rate, expressed in m³, following the complete explanation of the formula for the net influent volume of the material from the previous step. 3 / h, This represents the reflux inflow rate, expressed in m³, following the complete explanation of the formula for the net inflow material volume from the previous step. 3 / h, This indicates the recursive period using the complete explanation of the formula for the net incoming material volume from the previous step, in hours. This represents the effective volume of the unit, in meters (m). 3 This information originates from the unit volume ledger; only in... Greater than 0, volume version valid, and The recursion is executed when both are not less than 0 and their sum is not greater than 1. If the sum is greater than 1, the current recursion period is divided into a sufficient number of sub-periods, such that the sum of the two permutation coefficients in each sub-period is not greater than 1, and then the same recursion is executed for each sub-period. If a valid traffic or sub-period configuration that satisfies the joint upper bound cannot be obtained, the recursion for this period is stopped, the previous valid trajectory is used, and the traffic version, volume version, and reason for the anomaly are recorded. Updating the proportion with negative old state coefficients is prohibited. This represents the processing unit index; units are not applicable because it is an index for discrete objects. This indicates a production batch index; units are not applicable because it is a discrete batch index. This represents a recursive periodic index, taking positive integers. Units are not applicable because it is a discrete periodic index. For tandem mixing units, the trajectory recursion module divides the effective volume into equal-volume sub-pools according to the order of the tracer test fitting, and performs complete mixing recursion for each sub-pool level by level. The batch ratio of the previous sub-pool is used as the external influent ratio of the next sub-pool, and the ratio of the last sub-pool is used as the output trajectory of the treatment unit. For units using residence time distribution, the module first uses the external influent flow rate and effective return flow rate as weights to synthesize the external influent batch ratio and the batch ratio of the return source unit's output at the same time as a combined inlet ratio. Then, this combined inlet ratio is written into the directed unit graph for iteration, and then... Calculate the output percentage, where, Representation unit Output production batch In the The proportion of each cycle is a dimensionless quantity ranging from 0 to 1. Representation unit In the Production batches in the combined entry of each cycle The proportion, a dimensionless quantity ranging from 0 to 1, is derived from the flow-weighted result of the external inflow trajectory and the effective return trajectory. The table showing the number of stay times indicates the number of stay times. The dimensionless weights for each time delay interval are derived from tracer experiments at the same flow rate and return ratio level, and the sum of each weight is 1. This represents the maximum delay index covering 99% of the tracer response. It should be a non-negative integer, and units are not applicable because it is an upper bound for discrete delay. Indicates the delay interval index, ranging from 0 to... Integers are not applicable because they are discrete time-delay indices. This represents the processing unit index; units are not applicable because it is an index for discrete objects. This indicates a production batch index; units are not applicable because it is a discrete batch index. This represents the output period index, taken as a positive integer. Units are not applicable because it is a discrete period index; when The initial inventory batch composition is manually verified before the first effective trajectory period. If the composition is missing, the weight of the existing time delay interval is re-normalized and marked as low coverage. When the coverage weight is less than 0.95, the release is stopped. When the backflow topology, backflow composition source or backflow ratio level changes to outside the table, the table is disabled and switched to serial hybrid recursion. When switching, the current proportion vector is inherited and the conservation check that the proportion sum is 1 is performed. The residence time distribution table uses unit number, influent flow rate level, and return ratio level as query keys. Value fields include time delay interval, interval weight, weight reset confidence interval, test date, and version number. The construction process involves adding a conserved tracer to the stable operating unit, collecting outlet concentration data at 1-minute intervals, subtracting the baseline, and calculating the concentration-time integral values ​​for each time delay interval and the total concentration-time integral value for the entire time period. Both are in mg·h / L. The dimensionless interval weight (0-1) is obtained by dividing the integral value of each interval by the total integral value for the entire time period. The query key for a biochemical unit record is a flow rate of 40-50 m³ / h. 3 / h and a return ratio of 0.8 to 1.0, with a delay coverage range of 6 to 14 h. The table records are stored in the process parameter library and recalibrated after the effective volume changes by more than 5% or after equipment modification. When no matching gear is found, the residence time distribution with a larger delay coverage upper limit among the adjacent gears with a flow rate not higher than the query value is adopted. The transfer relationship of refluxed materials is represented by directed connection records. Each record includes the reflux source unit, reflux target unit, reflux pump number, reflux flow rate, reflux ratio, start / stop status and effective time. The trajectory recursion module only reads the reflux batch ratio when the reflux pump operation confirmation and the flow meter valid flag are established at the same time. When the reflux ratio exceeds the valid level of the corresponding residence time distribution table, the table is stopped and switched to series hybrid recursion. When switching, the current batch ratio vector is used as the initial value of each sub-pool. To ensure the cross-unit recursion order, the module establishes a directed unit graph based on pipeline connections and return connections. First, it updates the external water inlet item along the main process topology order, then updates the return item with the return ratio of the previous round in the same cycle and iterates until the maximum difference between the proportions of each batch in two adjacent rounds does not exceed 0.001. If the iteration count reaches 20 and still does not converge, the trajectory of the previous valid cycle is retained and written into the return recursion exception, and the unconverged trajectory is not published to S3. After each update, the trajectory recursion module compares the sum of the percentages of all active batches under the same material coordinates with 1. When the absolute deviation does not exceed 0.002, it normalizes by dividing each percentage by the sum of the percentages. Batch items with a normalization value less than 0.001 are not discarded directly, but their percentages are incorporated into the numerical residual items of other active batches with the source batch set. This residual item participates in the next cycle recursion as an item of the percentage vector, so that the sum of all percentage items and numerical residual items of the published trajectory remains 1. When the cumulative incorporation exceeds 0.002, the truncated trajectory is not published, but it reverts to the untruncated percentage vector and increases the numerical uncertainty. When the deviation exceeds 0.002, any percentage exceeds the range of 0 to 1, or the output delay exceeds the coverage boundary of the dwell time distribution, normalization is not performed, but the unit number, material coordinates, original percentage vector, flow version, and dwell time distribution version are saved. When the external inflow and return flow are interrupted at the same time or the flow data is marked as invalid by S1, the trajectory recursion module freezes the cumulative material replacement amount and the proportion of all batches, and does not advance the material coordinates with zero flow. After the flow is restored, the cumulative flow at the time of restoration is used as the new recursion starting point, and the last valid state before the interruption is used as the initial value. If the cumulative flow value is reset, the volume increment before and after the reset is spliced ​​using the reset flag in the S1 release list without changing the batch label. When the flow rate is invalid, the proportion is 1, or the recursive result exceeds the residence time distribution boundary, the trajectory recursion module sets the trajectory publishing status to stop and sends a maintenance alarm to the maintenance system, including the anomaly type, unit number, sensor number, material coordinates, and suggested verification items. S3 can only read the last valid trajectory version before the stop. When valid flow rate data is obtained again and the batch proportion of each sampling point in two consecutive cumulative material replacement cycles is between 0.998 and 1.002, the module removes the alarm and publishes the complete trajectory segment after recovery. Finally, a batch proportion trajectory is formed by unit number, material coordinates, and production batch label index, which is used by S3 to match the load and biochemical response of the same batch.

[0023] In this specific embodiment, S3 includes: The state estimation module reads the batch proportion trajectory published by S2 and the node state dataset of S1, and writes the emission component characterization, inlet and outlet chemical oxygen demand of each treatment node, respiration rate of the biochemical unit, temperature, pH, effective volume, and equipment output records into the batch state working area, using the production batch label and material coordinates as the joint key. With production batch The paired records, the module is arranged by Calculate the material coordinate alignment error, and then... Calculate the allowed upper limit, where, This indicates the material coordinate alignment error, in meters (m). 3 , This represents the locally accumulated material coordinates at the sample time node read from S1, in meters. 3 , This represents the expected arrival coordinates of the material given by the S2 batch trajectory, in meters. 3 , This indicates the upper limit of the allowable alignment error, in meters (m). 3 , Represents a node The current calibration version of the flow meter provides the upper limit of the dimensionless relative error. This represents the effective flow rate corresponding to the sample time, in meters (m). 3 / h, This indicates the sampling period, in hours (h). This represents the standard deviation of the expected arriving material coordinates obtained from the conversion of the S2 trajectory state covariance, in meters. 3 The constant 1.645 represents the dimensionless 95% coverage factor on one side. This indicates the processing node index; units are not applicable because it is a discrete node index. This represents the production batch index; units are not applicable because it is a discrete batch index. Only if all the above inputs are valid and... The alignment status is marked as qualified. If any input, flowmeter calibration version, or trajectory covariance is invalid, it is marked as undecidable and cannot be marked as qualified. Each pairing record in the work area saves the trajectory version, flowmeter calibration version, allowable upper limit version, batch percentage, observation value, and valid flag. , The alignment status will only be updated when the trajectory publishing status is valid, the observation record is valid, and the alignment status is qualified. The state estimation module first subtracts the blank respiratory baseline under the same temperature and dissolved oxygen conditions from the respiratory rates obtained from the respiratory response observation nodes of the S1 front and rear segments, and then calculates the data within a sliding observation window containing 10 consecutive effective cycles. Establish a system of batch-unit percentage respiratory response equations with non-negativity constraints, where, Indicates the first observation window The measured respiratory rate of the ventilator for each cycle, in mgO2 / (L·h), is derived from valid records of the respiratory response observation node corresponding to S1. This represents the respiration baseline from the most recent blank water calibration under the same node, temperature, and dissolved oxygen conditions, expressed in mgO2 / (L·h). This indicates the production batch number given by S2. In the The dimensionless proportion of biochemical units in each cycle Indicates the first Production batches to be estimated within each period observation window The percentage of respiratory response, expressed in mgO2 / (L·h). Indicates the first The residual of respiratory observation over one cycle, in mgO2 / (L·h), Indicates the first The number of active production batches within each periodic observation window is taken as a positive integer; units are not applicable because it is a discrete count. Indicates the periodic index within the observation window, taken from... to Integers are not applicable because they are discrete periodic indices. This indicates a production batch index; units are not applicable because it is a discrete batch index. This represents the current estimated period index, taken as a positive integer. Units are not applicable because it is a discrete period index; the module solves using the non-negative least squares method. Only if the column rank of the proportion matrix is ​​equal to And when the condition number does not exceed 100, the estimate is published, and according to Calculate the batch respiratory response increment; Indicates production batch In the The increase in respiratory response per cycle, expressed in mgO2 / (L·h), is derived from... and The result is obtained by multiplication. This indicates that S2 indicates the batch number in the specified order. The dimensionless proportion of each period; in the latter formula This indicates a production batch index; units are not applicable because it is a discrete batch index. This represents the current estimated period index, taken as a positive integer. Units are not applicable because it is a discrete period index. If the rank is insufficient, the condition number exceeds the limit, or neither of the two respiratory response observation nodes has a valid record, the batch response will not be issued. Instead, the previous valid state will be used and the corresponding state covariance will be expanded. For components Modules are Calculate the batch load at the emission end, where, Indicates components Belongs to production batch Mass load, in kg / h This indicates that the flow meter at the S1 discharge collection end is at the first... The effective instantaneous flow record for each period reads the flow rate, in milliseconds (m). 3 / h, This represents the concentration read from the effective measured concentration record of the component analyzer corresponding to the S1 emission end under the same material coordinate, in mg / L. Records with a valid value of zero are not counted. And keep the missing measurement mark, Indicates the production batch at the material coordinates corresponding to the emission end. The dimensionless proportion is derived from the proportion of the confirmed batches written back in S2, with the divisor 1000 used to divide m 3 Convert the product of / h and mg / L to kg / h. This index represents components such as chemical oxygen demand, carbonate organics, alcohols, or by-reaction products. Units are not applicable because it is a discrete index. This indicates a production batch index; units are not applicable because it is a discrete batch index. This represents the load calculation cycle index, which is a positive integer and the unit is not applicable because it is a discrete cycle index. The module pairs the load of the emission end component with the biochemical response when the same production batch arrives at the biochemical unit based on the batch proportion trajectory. The pairing key is the production batch label, the trajectory arrival material coordinates, and the biochemical node number. The reaction kinetic parameter library stores easily biodegradable first-order reaction coefficients, difficult-to-biodegrade first-order reaction coefficients, respiration conversion coefficients, and inhibition benchmarks according to temperature, pH, conductivity levels, and sludge age in the biological treatment tank. These parameters are obtained by fitting data from intermittent respiration tests and chemical oxygen demand (COD) degradation tests of the same batch of wastewater. Each record includes a temperature of 28–32 °C, pH of 7.0–7.5, conductivity not exceeding 18 mS / cm, and an easily biodegradable reaction coefficient of 0.18 h. -1 The coefficient of the recalcitrant biochemical reaction is 0.025 h. -1 Version number and validity date; the parameter library is updated quarterly or when the sludge age changes by more than 20%; when the setting is not matched, the adjacent parameter that is more unfavorable to the reaction rate is used. The module performs joint estimation with non-negative constraints using the total load at the emission end of the same batch, the respiratory response increment, and the inlet and outlet loads of the nodes: within the range where both readily biochemical and difficult biochemical loads are non-negative and their sum does not exceed the effective total load at the emission end, with a candidate step size of 0.1 kg / h, it calculates the material balance residual, the respiratory prediction residual obtained by multiplying the respiratory conversion coefficient from the parameter library by the two types of candidate loads and summing them, and the node removal residual; the three types of residuals are squared after being divided by the corresponding instrument calibration error, and the weights are obtained by normalizing the inverse of the variance of the three types of residuals of the most recent 20 qualified batches, and the candidate with the smallest weighted sum of squares is selected. When the target values ​​are the same, the module selects the one with the smaller absolute value of the node removal residual, the smaller absolute value of the respiratory residual, and the smaller difficult biochemical load in that order to ensure the uniqueness of the output; the module is set according to Calculate the predicted removal load of the biochemical unit, where, Indicates production batch The predicted removal load, in kg / h. This represents the estimated biochemical load, expressed in kg / h. This represents the estimated recalcitrant biochemical load, expressed in kg / h. This represents the first-order biochemical reaction coefficient read from the parameter library, in units of h. -1 , This represents the coefficient of a first-order recurrent biochemical reaction read from the parameter library, in units of h. -1 , This represents the trajectory-weighted residence time of this batch in the biochemical unit, in hours. This represents the natural constant, which is a dimensionless constant and its unit is not applicable. This indicates the production batch index, and the unit is not applicable because it is a discrete batch index. When the respiratory conversion coefficient is missing, the observation category is missing, the rank of the candidate design matrix is ​​insufficient, or the condition number exceeds 100, the easy and difficult biochemical loads are not published. The previous verification status is used, the covariance is expanded, and the manual calibration task is output. Module by Calculate the inhibition coefficient, where, Indicates production batch The dimensionless inhibition coefficient; a larger value indicates stronger respiratory inhibition. This represents the average increase in respiratory response within the stable percentage range for this batch, expressed in mgO2 / (L·h). This represents the non-negative reference respiratory increment effectively calibrated from a historically uninhibited batch under the same biochemical load, with the same units. This represents the positive protective dose in units equivalent to the increase in respiratory response; in this specific embodiment, it is defined as 0.2 mgO2 / (L·h). This function represents a cutoff function that restricts the result to the range of 0 to 1. Units are not applicable because it is a mathematical operation function. This indicates a production batch index; units are not applicable because it is a discrete batch index. Only when... When the value is strictly greater than 0, the inhibition coefficient is calculated. When the reference respiratory increment is missing, the calibration is invalid, or the denominator is not positive, the previous effective inhibition coefficient is used. The corresponding state covariance is expanded and a reference respiratory increment calibration task is generated. The state is not updated with an invalid denominator. The influent load state is saved in batches for easy biochemical load, difficult biochemical load, inhibition coefficient, state covariance, and kinetic parameter versions. For each processing unit The module calculates the measured removal amount based on the effective inlet and outlet load difference; when and When not less than 0, press Determine the capacity utilization rate and truncate it to 0 to 1. Multiply the available output by the capacity utilization rate. Once it is confirmed that the output has been occupied, press [button]. The available output is minus the already occupied output; when When Set to 0, if Then Set to 1 and record unavailable output and overload. Then Set to 0, if Then mark the measured removal amount for this period as abnormal and do not update. And the capacity status, using the previous effective occupancy rate for diagnosis but maintaining The value is 0, and the inlet and outlet loads, instrument version, and abnormal time are recorded simultaneously; if the total available output is positive but the measured removal amount is negative, the capacity status is not updated, the previous effective status is used, and the corresponding covariance is increased; negative residual capacity is prohibited in all branches; among them, Representation unit Dimensionless capacity utilization rate This represents the measured load removal after trajectory alignment, in kg / h. This represents the total available output of the equipment after maintenance derating under its rated operating conditions, expressed in kg / h, and is derived from the equipment capacity register. This indicates the current operating capacity, expressed in kg / h. This indicates the incremental remaining capacity that can be allocated on top of the currently occupied output baseline, expressed in kg / h. This represents the processing unit index; units are not applicable because it is an index for discrete objects. The state estimation module writes the influent load status of each batch, the remaining capacity status of each unit, the trajectory version used, the parameter version, the valid data bitmap, and the estimated time into the process state cache. If the load or breathing response of the key components is invalid, the previous valid state is retained and the corresponding state covariance is increased without using invalid values ​​for updating. The cache is encapsulated as the process model state input in units of the current control segment for S4 to process parameter combinations in a rolling solution.

[0024] In this specific embodiment, S4 includes: The parameter matching module reads the S3 process status cache and assembles the process model status by combining the biochemical load, non-biochemical load, inhibition coefficient, total available output of each treatment unit, currently occupied output, incremental remaining capacity, and status covariance of the current active production batch. It also assembles the control parameter vector by combining the reagent dosage, pH, reaction temperature, residence time, aeration rate, and reflux ratio. The model is set to 30... The rolling time domain of the min control segment and six consecutive control segments predicts the chemical oxygen demand, carbonate organics, alcohols, by-reaction products, respiration rate, pH, dissolved oxygen, total equipment output, new predicted load relative to the current occupied output baseline, incremental remaining capacity, and end-of-line effluent indicators for each treatment node according to production batch label, treatment node, material coordinate range, and trajectory version. The reaction temperature is only used as a controllable quantity when the S1 temperature control loop is available, the heat exchange capacity boundary is valid, and the temperature controller interlock self-test passes. It is mapped to the opening degree of the cold and hot medium regulating valve and the frequency of the heat exchange circulation pump. The field controller returns the temperature setpoint, actual temperature, valve position, and confirmation. When there is no active temperature control capability, the temperature is fixed as a predicted disturbance and no temperature setpoint is generated. The model parameters are fitted from historical batches in the most recent 90 days that meet the discharge limits, and incremental updates are made based on the state correction records verified in S5. For production batches The process model is based on and Applying the S3 inhibition coefficient to biochemical reactions, where, Indicates production batch The first-order reaction coefficient of biochemically efficient reaction after inhibition correction, in h. -1 , This indicates the first-order reaction coefficient of the recalcitrant biochemically effective batch after inhibition correction, in h. -1 , This indicates the dimensionless inhibition coefficient for this batch released by S3. This represents the biochemical first-order reaction coefficient in the S3 kinetic parameter version, in units of h. -1 , This represents the coefficient of first-order recalcitrant biochemical reactions in the same version, in units of h. -1 , This represents the production batch index; units are not applicable because it is a discrete batch index. Subsequent load removal, respiratory demand, and oxygen demand predictions will all use only this index. and Instead of using the uncorrected coefficients; when the suppression coefficient is effective but has low confidence, its upper bound of the state confidence interval is taken and truncated to 0 to 1 as a conservative input, while the corresponding prediction standard deviation is expanded; when the suppression coefficient has no effective value and no usable upper bound, 1 is taken as a conservative input and marked as low confidence. If no candidate satisfies the strong constraints as a result, a safe backoff is entered; the model prediction record saves the original suppression coefficient, conservative substitution value, effective response coefficient, state version and substitution reason; The process model employs a coupled structure of material balance and first-order reaction kinetics. The physicochemical unit calculates the sedimentation removal coefficient based on reagent dosage, pH, and temperature. The biochemical unit calculates the load removal based on the aforementioned effective first-order reaction coefficient, aeration rate, reflux ratio, and residence time. The equipment sub-model correlates pump frequency, aeration rate, and available output after maintenance derating. Each control segment of the model outputs the predicted mean and predicted standard deviation, and... This constitutes the predicted water outflow boundary, in which, Indicators In the predictive control segment The predicted mean, with units consistent with the indicator. This represents the predicted standard deviation for the same indicator, with the same unit, derived from the state covariance and historical model residuals. This represents the 95% coverage factor on one side, which is taken as 1.645 in this specific implementation and is dimensionless. This indicates the corresponding effluent discharge limit, with the unit consistent with the indicator. This represents an index of effluent indicators; units are not applicable because it is a discrete index. This represents the predictive control segment index, which is an integer from 1 to 6. Units are not applicable because it is a discrete control segment index. For carbonate organic compounds, alcohols, or by-product components The model only exists at nodes No. Predicted flow rates for each control segment When the value is a valid positive value, according to Calculate production batch The predicted concentrations of node components, among which, Indicates components At the node Belongs to production batch The predicted concentration, in mg / L. This represents the predicted outflow load for this component node obtained from material balance calculations, in kg / h. This represents the predicted flow rate of the node, in meters (m). 3 / h, the constant 1000 is used to express kg / m 3 Converted to mg / L Indicates component index, Indicates the node index to be processed. Indicates the production batch index. The predictive control segment index is indicated by these four indices; none of their units are applicable because they are all discrete indices. This applies to nodes configured for respiratory response monitoring. Model according to Calculate the predicted respiratory rate for the same batch, where, Represents a node In the This control section belongs to the production batch. The predicted respiratory rate, in mgO2 / (L·h), This indicates the dimensionless proportion of this batch at this node and control segment given by trajectory S2. This indicates the percentage of respiratory response for this batch released by S3, expressed in mgO2 / (L·h). This indicates the dimensionless inhibition coefficient for this batch. , and These represent discrete node, batch, and control segment indices, respectively, with no applicable units. Each predicted value is saved along with the prediction standard deviation, production batch label, node number, material coordinate range, trajectory version, and model version. When the predicted flow rate, batch percentage, unit percentage, breathing response, or any required state is invalid, the corresponding predicted value is not generated and is marked as unavailable. Based on this, S5 marks the corresponding residual condition as undeterminable and enters the low-confidence prediction interval expansion or supplementary measurement path. Zero values ​​must not be used to replace missing predictions. The sedimentation removal coefficient of the physicochemical unit adopts a three-dimensional piecewise linear lookup table structure of reagent dosage, pH, and temperature. The parameter table fields include three independent variable nodes, node removal coefficient, sample size, residual standard deviation, and version number. Trilinear interpolation is performed when the lookup point is between adjacent nodes, and extrapolation is prohibited when it exceeds the validated envelope. The equipment sub-model adopts a first-order piecewise regression with intercept. Pump frequency corresponds to flow rate, aerator frequency corresponds to aeration volume and power, and hot and cold medium valve position corresponds to heat exchange. Each segment saves intercept, slope, applicable equipment number, operating mode, maintenance derating factor, and residual standard deviation. Biochemical reaction parameters are adopted from the S3 kinetic parameter library and calculated within the validated combination domain of aeration, reflux, and residence time. Model identification samples are organized by control segment. Each sample must include at least the entry status, control action, ambient temperature, equipment mode, subsequent measured output, and calibration version. If there are no fewer than 30 valid qualified samples within the last 90 days and the number of design matrix conditions does not exceed 100, the first 80% of the samples are subjected to weighted least squares fitting with parameterized physical boundaries in chronological order, with the weights being the reciprocal of the measurement variance. The remaining 20% ​​of the samples are reserved for verification. A new version is only activated when the median absolute relative error of each key effluent indicator in the reserved set does not exceed 10%, the prediction interval coverage is not less than 95%, and the number of misjudgments at the safety boundary is 0. Otherwise, the update is rejected and the previous approved version is used. When deploying for the first time or when there are insufficient valid samples, a conservative initial model and equipment ledger parameters verified by offline experiments and historical playback are used and marked as low-confidence versions. If the initial model, the previous version, and key equipment parameters are all unavailable, automatic parameter matching is stopped, the most recent safe parameters are maintained, and manual calibration and safety rollback tasks are output until at least 30 valid samples and the reserved verification threshold are met simultaneously. Resource consumption targets by Calculate and apply in the solution and Two soft constraints on resources, among which, This represents the dimensionless resource consumption target value to be minimized. This represents the total amount of reagent consumed within the rolling time domain, in kg, and is obtained by summing the reagent dosage of each control segment by the duration of that control segment. This represents the baseline chemical consumption for historically qualified batches under the same influent load, expressed in kg. This represents the total energy consumption of aeration within the rolling time domain, expressed in kWh. It is obtained by summing the products of the aeration equipment power and the duration of each control segment. This represents the corresponding historical energy consumption benchmark, in kWh. and All values ​​must be positive. If any benchmark is zero or missing, the corresponding normalization target will not be called. Instead, the most recent valid positive benchmark will be read. If there is still no valid positive benchmark, a safe rollback will be initiated and the benchmark version will be recorded. This represents the upper limit of dimensionless drug consumption given in the resource budget table; in this specific implementation, it is taken as 1.05. This represents the dimensionless aeration energy consumption limit given in the resource budget table. In this specific implementation, it is taken as 1.05. The budget table can be queried according to the influent load level and temperature level. This represents the dimensionless weight determined by the monthly drug cost ratio; in this specific implementation, it is taken as 0.45. This represents the dimensionless weight determined by the proportion of monthly aeration energy consumption costs; in this specific implementation, it is taken as 0.55. and The sum is 1. This represents a non-negative dimensionless slack that simultaneously relaxes the budget limits for the two aforementioned resources; it is taken as 0 if the budget is not exceeded. This represents the dimensionless relaxation penalty weight, which is set to 10 in this specific implementation. The equipment capacity boundary is obtained by multiplying the rated output of the equipment by the maintenance derating factor, which is read from the maintenance record and limited to 0 to 1. The safety boundary stores the permitted dosing range of chemicals, the permitted pH range, the permitted temperature range, the minimum start-stop interval of the aerator, the maximum aeration rate, and the permitted frequency of the return pump. Both the safety boundary and the equipment capacity boundary are set as strong constraints that do not allow the introduction of slack. The resource consumption target allows the use of non-negative slack, but the target value will be increased according to the aforementioned penalty weight. During candidate screening, the newly predicted load relative to the currently occupied output baseline of S3 is compared with... The incremental remaining capacity is compared, and the total predicted output, formed by the sum of the currently occupied output and the newly predicted load, is compared with... The comparison of the total available output after maintenance derating is indicated. Any candidate parameter combination will be deleted from the candidate set if the new predicted load exceeds the incremental remaining capacity, the predicted total output exceeds the available total output, the chemical dosing range is exceeded, or the aeration equipment interlock is triggered. It is forbidden to directly compare the predicted total output with the incremental remaining capacity. The compatibility database uses the treatment unit number, influent load level, and equipment combination as query keys. Value fields include the joint feasible region of chemical dosage and pH, the joint feasible region of temperature and residence time, the joint feasible region of aeration rate and reflux ratio, the upper limit of parameter change rate, interlock prohibited combinations, the number of verification batches, and the version number. Records are constructed from permitted operating condition records, equipment interlock records, and historical treatment results. A physical and chemical unit record stipulates that when the influent chemical oxygen demand load is 120 to 160 kg / h, the combination of chemical dosage and pH must be within the polygonal feasible region verified by 12 qualified historical batches. The compatibility relationship is rebuilt when 10 new qualified batches are added or when the equipment interlock logic is changed. During the rolling solution process, the parameter matching module first generates candidate combinations within the permissible range of each parameter according to the executable resolution of the equipment. Then, it performs Boolean filtering using the compatibility relation library. For the retained combinations, it calls the process model to obtain the predicted effluent, equipment output, and resource consumption. Subsequently, it refines the continuous parameters in the neighborhood of the candidate combinations using sequential quadratic programming. During the refinement, the discrete interlocking state is fixed, and the feasible region of the continuous polygon, the upper limit of the parameter change rate, the interlocking prohibited combinations, as well as the safety boundary and equipment capacity boundary in the compatibility relation library are continuously written as solution constraints. The continuous solution is then quantized according to the resolution of the field equipment, and the full verification of the compatibility relation, the predicted effluent boundary, the safety boundary, and the equipment capacity boundary is re-executed. If any verification fails after quantization, the corresponding discrete candidate is returned or a safety backoff is entered. The solver sorts the parameters in ascending order of strong constraint feasibility, predicted effluent boundary feasibility, and target value. When the difference in target value does not exceed 0.001, it selects the combination with the smallest sum of squares of parameter changes in adjacent control segments. The dosage of the first control segment, pH, temperature, residence time, aeration rate, and reflux ratio, as well as their effective time, are written into the executable parameter combination. The predicted effluent boundary table can be queried by effluent index, emission permit version, and control segment date. The value fields include emission limit, unit, predicted coverage factor, monitoring method, and effective time. The equipment capacity table can be queried by equipment number and operating mode, including rated output, maintenance derating factor, minimum stable output, and maintenance status. The safety boundary table can be queried by treatment unit and reagent type, including permitted range, interlocking conditions, and upper limit of change rate. All three types of boundaries save version hashes, and the solution task can only reference the same version that is effective throughout the entire rolling time domain. The baselines for chemical consumption and aeration energy consumption are established based on the median of historical control periods with the same influent load and temperature and qualified effluent. When the monthly resource cost changes by more than 10%, the weights are recalculated. The new baselines and weights are first verified on the most recent 10 qualified historical batches. The new version is only used when the effluent boundary feasibility rate is not lower than the original version and the average resource consumption target value decreases. Otherwise, the original version is maintained and the reason for the verification failure is recorded. The upper and lower bounds of the continuous variables in the sequential quadratic programming are derived from the current safety boundary. The initial value is the lowest target value in the discrete candidate set. The termination condition is that the change between two consecutive target values ​​does not exceed 0.0001 and all strong constraint margins are non-negative. The maximum number of iterations is set to 50. If the maximum number of iterations is reached but the termination condition is not met, the refinement result is discarded and the discrete candidate is returned. If the discrete candidate is also empty, the safe backoff branch is entered immediately. The selection of the safety backoff instruction is based on the predicted overload source. When the newly predicted load exceeds the unit's incremental remaining capacity or the predicted total output exceeds the available total output, the maximum allowable inflow rate that satisfies both capacity boundaries is calculated first, and the flow rate is reduced accordingly. The hydraulic retention time is a derived control quantity determined by both the effective volume and the total excess flow rate, and is only applied when... When the value is a valid positive value, the parameter matching module proceeds according to... Conversion, among which, Indication processing unit The hydraulic residence time, measured in hours (h). This indicates the effective volume of the unit, read from the S2 volume register and corrected for the current liquid level, in cubic meters (m³). 3 , This indicates the influent flow rate read from the effective flow record of S1 by this unit, in meters (m³). 3 / h, This indicates the return flow rate read from the valid record of the return flow meter, in units of m³. 3 / h, This represents the processing unit index; units are not applicable because it is a discrete object index. It is not calculated when the total throughput is 0 or any throughput is invalid. Maintain the previously confirmed dwell time status and mark it as uncalculable, then proceed to the traffic recovery or safe shutdown branch; the module does not directly issue... Within the allowable liquid level range of the regulating tank, the storage capacity of the two future control sections, and the constraints of the reflux linkage, it is mapped to the set values ​​of the inlet pump frequency, outlet valve position, and reflux pump frequency. The field controller calculates the actual residence time based on the measured flow rate and liquid level and returns confirmation. When the hydraulic residence time is insufficient and the available volume of the regulating tank meets the storage capacity of the two future control sections, the residence time is extended according to this mapping. When the standby equipment is in the state of completed maintenance and interlock self-test passed, the standby equipment is activated. When multiple commands are feasible at the same time, the command combination is determined in the order of not exceeding the upper limit of the regulating tank liquid level, the lowest resource increment, and the fewest equipment start-stop times. When any data in the feasible domain of the allowed operating condition record, equipment interlock record, or parameter verified by historical results is unavailable, the module will restrict the affected parameter to the parameter domain that was most recently verified by historical processing results and has a complete version, and freeze the version of that parameter domain until the missing data is recovered. When there is no combination that simultaneously satisfies the strong safety constraint, the strong equipment capacity constraint, and the predicted effluent boundary, the module will not output the over-limit dosing or over-limit aeration parameters, but will generate a safety backoff command. The backoff command will be given in sequence according to the predicted overload, such as reducing the influent flow rate to within the equipment capacity boundary, extending the equivalent residence time, and activating one or more of the standby treatment equipment that has passed the interlock self-test. The parameter matching module writes the selected combination, predicted trajectory, target value, strong constraint margin, soft constraint relaxation amount, compatibility relationship version, model version, and safety rollback status into the control segment list. The control segment list saves the predicted mean, predicted standard deviation, available flag, production batch label, node number, material coordinate range, trajectory version, model version, inhibition coefficient input, and effective reaction coefficient for carbonate organics, alcohols, by-reaction products, and respiration rate batch by batch and node by node. Only the parameters of the current control segment are sent to the field controller, and the parameters of the remaining control segments are retained as the prediction plan. After the field controller returns a confirmation of receipt, it enters S5 for execution and residual verification. If no confirmation is received, the most recent safety parameters are retained and the unexecuted status is recorded.

[0025] In this specific embodiment, S5 includes: The feedback correction module receives the executable parameter combination of the current control segment S4 and the confirmation from the field controller. It records the actual executed values ​​of the reagent dosage, pH setpoint, residence time corresponding actuator setpoint, aeration volume, and reflux ratio according to the parameter effective time. When the active temperature control loop of S1 is available, it also records the reaction temperature setpoint, actual temperature, hot and cold medium valve position, and heat exchange circulation pump frequency. When there is no active temperature control capability, it only records the measured temperature disturbance and its predicted level, but does not record the temperature setpoint execution value. Using the batch proportion trajectory of S2 as an index, it maps the measured results of each processing node after execution to the same production batch and material coordinate range used in the process model prediction. Control segments that have not received execution confirmation are not used for state correction. For processing nodes Production batch and indicators Modules are Forming nodal residuals, where, Represents nodal residuals, units and indicators Consistent, This represents the measured value weighted and aligned according to the batch proportion, with units and indicators. Consistent, This indicates the predicted values ​​of the same batch stored in S4, with units and indicators. Consistent, This indicates the processing node index; units are not applicable because it is a discrete node index. This indicates a production batch index; units are not applicable because it is a discrete batch index. This module represents the index of components, respiratory response, inlet and outlet loads, or equipment output indicators. Units are not applicable because it is a discrete index. The module also saves the trajectory version used for alignment and the material coordinate range. The module uses the residuals of carbonate organics, alcohols and by-reaction products in the same batch as component residuals, the residual of respiration rate as respiration residuals, the difference of chemical oxygen demand load between the inlet and outlet of each unit as load residuals, and the difference between actual equipment output and equipment sub-model predicted output as equipment output deviation. All residuals are normalized according to their respective equipment calibration errors. Together with the number of control segments with continuous same deviation, data valid bitmap and timestamp, they form a residual evidence set. When the calibration error is zero or missing, the indicator does not participate in the attribution. The evidence set also saves the S4 process model version, operating status input, actual control parameters and corresponding predicted outputs, so that each predicted value can be traced back to the operating status, static parameters, execution actions and calibration version. When the component residual and the breathing residual appear sequentially at two or more downstream processing nodes in the trajectory sequence, in the same direction as the upstream, and each of their absolute values ​​exceeds twice the corresponding calibration error, the module determines that the water inlet propagation condition is met, and allocates the residual correction amount to the water inlet load state of the same batch according to the gain matrix of the state estimator. When different production batches have the same load residual in three consecutive control sections after the same treatment unit, and the absolute value of the equipment output deviation exceeds the output deviation boundary determined by the equipment calibration error, the module determines that the unit continuous deviation condition is met, and allocates the correction amount to the remaining capacity state of the unit. When both allocation conditions are met simultaneously, the module performs a three-state determination on the batch component residuals and respiration residuals of the downstream nodes of the processing unit: only when the downstream effective observations cover at least two respiration response observation nodes ordered by trajectory, the effective bitmap is complete, the batch ratio is not less than 0.2, and the material coordinate alignment status recorded in S3 is qualified, can it be determined to continue propagation or truly terminate; if the two types of residuals continue to appear in trajectory order and exceed their respective boundaries, it is determined to continue propagation and only the influent load status is corrected; if the two types of residuals return to within the boundary in the downstream nodes that meet the above effective observation conditions, it is determined to truly terminate and only the remaining capacity status of the processing unit is corrected; if the downstream observations are missing, invalid, insufficiently covered, the required S4 prediction value is unavailable, or the alignment error and its allowable upper limit are undeterminable, it is determined to be insufficient evidence, no status is corrected, the current verified status is maintained, the corresponding low confidence prediction interval is expanded, and manual review or subsequent supplementary testing is initiated; the corrected status is truncated by the same non-negative range and capacity upper limit as S3 and written into the unverified status version to avoid deducting the influent status and processing capacity repeatedly with one piece of evidence; When both allocation conditions are determined to be invalid based on complete and valid evidence, or when any allocation condition is undeterminable due to missing data, the module does not change the influent load state and remaining capacity state; when at least one valid standardized residual exists in the residual evidence set, it is processed according to... Calculate the discreteness of the effective subset, where, Indicates production batch The dimensionless discreteness, Indicates the first The dimensionless standardized residual is obtained by dividing each effective residual by its corresponding calibration error. This represents the number of valid standardized residuals in the residual evidence set for this batch, expressed as a positive integer. Units are not applicable because it is a discrete count. Indicates a valid residual index, ranging from 1 to... Integers are not applicable because they are discrete residual indices. This indicates a production batch index; units are not applicable because it is a discrete batch index. Under complete evidence, the standard deviations of each prediction are multiplied by... When the evidence is incomplete but there is at least one valid residual, the same valid subset is used for calculation and the magnification factor shall not be less than 1.5; when there is no valid residual, the most recently verified upper bound of the dispersion is used, and when there is still no usable upper bound, the preset maximum magnification factor of 2 is directly used; the module records the version of the prediction interval before magnification, the evidence integrity status and the source of low confidence, without changing the prediction mean, and simultaneously initiates supplementary testing or manual review; The recovery threshold is the maximum value of the dispersion of the most recent 20 historical production batches that use the pre-expansion prediction interval version and whose effluent meets the predicted effluent boundary. If there are 5 to 19 qualified historical batches, all existing qualified batches are used. If there are fewer than 5 qualified historical batches, the cold start path is entered. The conservative initial threshold or the upper limit of the equipment calibration residual verified offline is used first and marked as a low confidence version. Each qualified production batch is updated according to the versioning rules. After 5 qualified batches are obtained, the maximum value of the sample is switched. If there is no verifiable initial threshold and fewer than 5 qualified batches, the prediction interval expansion state is maintained and a manual review or calibration task and a clear exit condition are output. Silent recovery is not allowed. When the dispersion of two consecutive subsequent production batches does not exceed the effective recovery threshold, or when any of the conditions of water inlet propagation and unit continuous deviation are met, the module recovers the pre-expansion prediction interval version. The recovery process only cancels the prediction standard deviation amplification and does not relax the safety boundary and equipment capability boundary in S4. Before performing attribution assignment, the module filters records from the effective residual evidence set that meet the trajectory arrival order, have a batch ratio of not less than 0.2, and have a qualified S3 alignment status. These records are then sorted in descending order of the absolute value of the standardized residuals, with those having the same absolute value being retained first. Smaller records, if still identical, prioritize retaining the original measured records over interpolated records; modules record whether the allocation conditions are true, false, or indeterminate, and only determine false when the evidence is complete and the value does not reach the corresponding threshold, such as component residuals or respiration residuals less than two downstream nodes, equipment output deviation lacking effective calibration boundaries, insufficient continuity between different batches (less than three control segments), or the required predicted value for S4 being unavailable. and If any condition is invalid, the corresponding allocation condition will be marked as undeterminable due to missing data and shall not be confused with conditions that are invalid based on complete evidence. The state estimator reads an initial conservative gain matrix verified through offline testing and historical playback during its first run. This matrix stores the input residual type, output state type, element limit, version number, and applicable operating conditions. The correction increment must not exceed 10% of the corresponding state's physical range. When fewer than 10 records have passed verification, only the initial matrix is ​​used, and no online updates are performed. After reaching 10 records, a recursive least squares update is performed every 5 new records, reserving the 3 most recent verifications. After reaching 30 records, the state correction is fitted using the 30 most recent verified state correction records, reserving the 5 most recent verifications. In the first stage, a new matrix is ​​only activated if the median of the sum of the absolute values ​​of the updated residuals decreases and there are no state out-of-bounds errors; otherwise, it is rejected and the previous valid version is used. If there is no valid initial matrix, state correction is not allowed. The current verification state is maintained, the prediction interval is expanded, or the temporary boundary is tightened, and the manual calibration task is output until the initial matrix passes offline playback. The input of the gain matrix is ​​various standardized residuals and batch proportions, and the output is the correction increment of easily biochemical load, difficult biochemical load, suppression coefficient, and cell residual capacity. The correction increment is truncated according to the physical upper and lower limits of the corresponding state before application. For the expansion branch of the prediction interval, if the residual evidence set contains effective component residuals, effective respiration residuals, and effective treatment unit inlet and outlet load residuals, it is marked as complete evidence and expanded according to the complete evidence rule; if any type is missing but there is still at least one effective standardized residual, it is marked as low confidence evidence and expanded according to the dispersion of the effective subset and an amplification factor of not less than 1.5; if there are no effective standardized residuals, it is marked as no evidence, and the most recently verified upper limit of dispersion or twice the preset maximum amplification factor is used. The original interval must not be maintained. Supplementary testing or manual review is initiated for all three branches, and clear exit conditions are saved. The expanded version of the interval saves the original half-width, new half-width, dispersion or alternative upper limit, evidence integrity status, recovery threshold source batch and effective time. Subsequent recovery operations are based on this to accurately return to the version before expansion. Each status correction generates a verification record. Record fields include production batch label, material coordinate range, residual evidence set identifier, correction object, pre-correction status, post-correction status, correction gain, total available output, currently occupied output, incremental remaining capacity, trajectory version, basic boundary version, temporary tightening layer version, and effective control segment. These three types of capacity quantities are separately identified and cannot use fields with the same name. The basic safety boundary and rated equipment capacity boundary originate from S4 and cannot be relaxed or overridden by S5. S5 can only establish versioned temporary upper operating boundaries; the effective boundary is based on the baseline. The stricter of the basic boundary and the temporary operating upper boundary shall apply; when the correction object is the influent load state, the temporary influent load upper boundary of the subsequent control section shall be tightened to the lower limit of both the corrected estimated value and the original upper boundary; when the correction object is the incremental remaining capacity state, the currently confirmed occupied output shall be added to the corrected incremental remaining capacity to obtain the temporary total output upper boundary, and the available total output after the S4 maintenance derating shall be used as the cap. The incremental remaining capacity shall not be directly used as the upper boundary of the total equipment output. At the same time, the upper limit of the single control section change rate of relevant chemicals, aeration and return parameters shall be tightened by 20%. After the verification record enters the verification queue, the module sorts the records according to the expected arrival time of the material coordinates. If the expected arrival times are the same, the records that will trigger the tightening of the equipment capacity boundary will be verified first. Each record is only compared with the subsequent actual test results of the same batch, the same node, the same index, and the same trajectory version. If the subsequent data is invalid, the control segment will be postponed. If valid verification data cannot be obtained in three consecutive control segments, the current status correction will be cancelled and the boundary before the correction will be restored to avoid long-term control parameter matching of unverified corrections. Before re-executing S4, the feedback correction module assembles the available unverified status versions and prediction interval versions into an update candidate set. First, it deletes candidates with incomplete valid bitmaps, revoked trajectory versions, or expired boundary versions. Then, it selects update objects in order of priority based on verified status, descending order of coordinate coverage of materials in the same batch, and reverse order of effective time. If the order is the same, it selects the object with stricter constraints. If the set is empty, it maintains the current verified status and the original prediction interval and writes a no-update flag to S4. No update object may relax or cover the basic safety boundary and rated equipment capacity boundary. Only the versioned temporary tightening layer is allowed to be retained based on the verification record. After verification, the layer is retained. After verification failure or timeout, the layer is revoked and restored to the previous valid boundary version it points to. After the materials of the same batch arrive at the corresponding node, the feedback correction module compares the sum of the absolute values ​​of the same set of residuals before and after correction. If the sum after correction is less than the sum before correction, the version of the state to be verified is marked as passed and the tightened boundary is retained. If it does not decrease, the state and constraint boundary are restored to the version before correction pointed to by the verification record. All pass or rollback results are written to the model verification library. Then, the S4 constraint is updated with the corrected state or the restored prediction interval and the rolling parameter matching is re-executed to form the executable parameter combination of the next control segment.

[0026] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0027] This invention places the emission component load, treatment node status, and biochemical response in a unified material coordinate system through batch proportion trajectory, and converts the influent load status and treatment unit remaining capacity status into executable parameter combinations subject to multiple boundary constraints through process model, so that parameter matching can be updated in a rolling manner as batch migration and equipment capacity changes.

[0028] This invention sets up a batch ratio and biochemical response pairing structure at the state estimation inlet, sets up a node residual dual-channel routing structure at the feedback end, and asymmetrically tightens the influent load boundary or equipment output and parameter change rate boundary based on the attribution results; when the evidence is insufficient, it keeps the state unchanged and expands the prediction interval, thus avoiding erroneous attribution between influent changes and unit performance degradation, and enhancing the stability of the control process.

Claims

1. A method for matching parameters in carbonate wastewater treatment based on a process model, characterized in that, include: S1. Collect water quality, flow rate, biochemical response and equipment operation data of carbonate wastewater discharge end and treatment node, and establish a node status dataset with production batch label using the cumulative water volume as material coordinate. S2. Based on the effective volume, inflow and outflow, return flow and mixing or dwell characteristics of each processing unit, the node status dataset is recursively calculated to obtain the batch proportion trajectory of each production batch in each processing unit. S3. Based on the batch proportion trajectory, the load of the emission end component is matched with the biochemical response when the same production batch arrives at the biochemical unit, and the influent load status is obtained by combining material balance and reaction kinetics. The remaining capacity status of each treatment unit is obtained based on the load removal amount and equipment output of the treatment node. S4. Using the influent load state and remaining capacity state as the state input of the process model, and taking the predicted effluent boundary, resource consumption target, equipment capacity boundary, safety boundary and compatibility relationship between treatment parameters as constraints, the executable parameter combination of reagent dosage, pH, reaction temperature, residence time, aeration rate and reflux ratio is solved in a rolling manner. S5. Execute the executable parameter combination of the current control segment, align the measured results and predicted results of the processing nodes according to the batch proportion trajectory to form node residuals, and select to correct the water inlet load status, correct the remaining capacity status of the corresponding processing unit, or expand the prediction interval based on the propagation relationship of the node residuals along the material trajectory and the continuous deviation relationship of different production batches in the same processing unit. Then, update the constraints using the corrected status or prediction interval and re-execute step S4.

2. The method for matching parameters for carbonate wastewater treatment based on a process model according to claim 1, characterized in that, S1 includes: Collect cumulative flow, chemical oxygen demand, pH, salinity or conductivity, temperature, characterization of carbonate organic matter, characterization of alcohols, characterization of by-reaction products, respiration rate and equipment operation data; The production batch label is determined based on the production emission switching signal and cumulative flow change, and the collection time of each data is mapped to the cumulative water volume to form the node status dataset; When any collected data exceeds the calibrated range of the corresponding sensor or the duration for which the continuously collected value remains unchanged exceeds the allowable duration in the equipment maintenance record, the collected data is marked as invalid and its data source record is retained, so that invalid data does not participate in the updating of the water inflow load status or the remaining capacity status. When the collected data re-enters the calibrated range and the continuously collected values ​​change within the allowed duration, the collected data is restored to valid status.

3. The method for matching parameters for carbonate wastewater treatment based on a process model according to claim 1, characterized in that, S2 include: For each processing unit, the cumulative material replacement amount is calculated based on the effective volume and inflow / outflow rate, the reflux material transfer relationship is established based on the reflux flow rate, and the mixing or residence characteristics are characterized by a complete mixing model, a series mixing model, or the residence time distribution determined by tracer test records. The proportion of each production batch entering the processing unit is combined with the cumulative material replacement amount, the return material transfer relationship and the mixing or residence characteristics to recursively deduce that the sum of the proportions of each production batch under the same material coordinate is one, thus obtaining the batch proportion trajectory. When the flow is interrupted, the cumulative material replacement amount is frozen, and it continues to be calculated from the cumulative flow at the time of flow recovery.

4. The method for matching parameters for carbonate wastewater treatment based on a process model according to claim 1, characterized in that, S3 include: The respiratory rate of the biochemical unit is weighted and separated using the batch proportion trajectory to obtain the respiratory response increment corresponding to each production batch; By inputting the respiratory response increment, emission component load, inlet and outlet loads of the treatment node, and reaction kinetic parameters into the material balance relationship, the influent load state, including readily biodegradable load, poorly biodegradable load, and inhibition coefficient, is obtained. The capacity utilization rate is determined by the ratio of the measured removal amount of the processing node to the available output of the corresponding processing unit under the equipment calibration conditions. The product of the available output and the capacity utilization rate is determined as the occupied output. The remaining capacity status is obtained by subtracting the occupied output from the available output.

5. The method for matching parameters for carbonate wastewater treatment based on a process model according to claim 1, characterized in that, S4 include: The predicted effluent boundary is determined based on the effluent discharge limit and the prediction error range; the resource consumption target is determined based on the reagent consumption and aeration energy consumption; the equipment capacity boundary is determined based on the equipment rated output and maintenance derating records; and the safety boundary is determined based on the permitted reagent dosing range, the permitted pH range, the permitted reaction temperature range, and the interlocking conditions of the aeration equipment. The compatibility relationship is determined based on the allowable operating condition records of each processing unit, equipment interlock records, and the feasible domain of parameters verified by historical processing results; Input the dosage of the reagent, pH, reaction temperature, residence time, aeration rate and reflux ratio into the process model, eliminate parameter combinations that violate the safety boundary, equipment capacity boundary or compatibility relationship, and select the parameter combination that satisfies the predicted effluent boundary and has the minimum resource consumption target value from the remaining parameter combinations as the executable parameter combination; When any data in the allowed operating condition record, equipment interlock record, or parameter feasible field is unavailable, the corresponding processing parameter is restricted to the parameter field that was most recently verified by the historical processing result of the processing unit.

6. The method for matching parameters for carbonate wastewater treatment based on a process model according to claim 1, characterized in that, S5 include: The residual evidence set is composed of the component residuals, breathing residuals, inlet and outlet load residuals of the processing unit, equipment output deviations and deviation duration periods corresponding to the same production batch. When the component residual and the respiration residual appear sequentially at two or more downstream processing nodes along the batch proportion trajectory, the influent propagation condition is determined to be met, and the correction amount is allocated to the influent load state. When different production batches continuously exhibit the same load residual in the same processing unit and the equipment output deviation exceeds the output deviation boundary determined by the equipment calibration error, the unit's continuous deviation condition is determined, and the correction amount is allocated to the remaining capacity state of the processing unit. When the influent propagation condition and the unit continuous deviation condition are met simultaneously, if the component residual and respiration residual continue to propagate to the downstream treatment node of the treatment unit, only the influent load state is corrected; otherwise, only the remaining capacity state of the treatment unit is corrected. When neither the water inflow propagation condition nor the unit continuous deviation condition is met, the state is not changed, the prediction interval version before expansion is recorded, and the prediction interval is expanded according to the dispersion of the residual evidence set. The upper limit of the dispersion of the residual evidence set of historical production batches that use the pre-expansion prediction interval version and whose effluent meets the predicted effluent boundary is determined as the recovery threshold. When the dispersion of the residual evidence set of two consecutive subsequent production batches decreases to within the recovery threshold, or when either the influent propagation condition or the unit continuous deviation condition is met, the prediction interval is restored to the pre-expansion prediction interval version, and the safety boundary and equipment capability boundary are not relaxed.

7. The method for matching parameters for carbonate wastewater treatment based on a process model according to claim 3, characterized in that, When traffic data is marked as invalid, the sum of batch proportions deviates from one, or the recursive result exceeds the dwell time distribution boundary of the processing unit, stop publishing the corresponding batch proportion trajectory to subsequent steps, save the abnormal status record, and issue a maintenance alarm. After valid flow data is regained and the sum of batch percentages in two consecutive cumulative material replacement cycles is restored to one, the maintenance alarm is lifted and the restored batch percentage trajectory is published.

8. The method for matching parameters for carbonate wastewater treatment based on a process model according to claim 5, characterized in that, The safety boundary and equipment capability boundary are set as strong constraints that do not allow the introduction of slack, and the resource consumption target is set as a soft constraint that allows the introduction of non-negative slack and applies a penalty weight to the non-negative slack. When no parameter combination satisfies the strong constraints and the predicted effluent boundary, the output includes a safety backoff instruction that includes at least one of reducing the influent flow rate to within the equipment capacity boundary, extending the residence time, and activating the standby treatment equipment, and prohibits the output of parameter combinations that increase the dosage of chemicals or the aeration rate beyond the safety boundary.

9. The method for matching parameters for carbonate wastewater treatment based on a process model according to claim 6, characterized in that, Write the production batch label, material coordinate range, residual evidence set, correction object, state before correction, state after correction, and constraint boundary version for each state correction into the verification record. The asymmetric tightening boundary is selected according to the correction object. When the correction object is the inlet load state, the upper boundary of the inlet load of the subsequent control section is tightened. When the correction object is the remaining capacity state, the upper boundary of the equipment output of the corresponding processing unit is lowered and the change rate boundary of the corresponding processing parameter is tightened. The state correction is verified based on the deviation between the subsequent measured results and the predicted results of the same batch, and the constraint boundary is restored to the original boundary corresponding to the version of the constraint boundary if the deviation does not decrease.

10. A process model-based carbonate wastewater treatment parameter matching system, used to execute the process model-based carbonate wastewater treatment parameter matching method according to any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to execute step S1; The trajectory recursion module is used to execute step S2; The state estimation module is used to execute step S3; The parameter matching module is used to execute step S4; The feedback correction module is used to execute step S5.

Images