Methods and related devices for coordinated control of wastewater treatment processes
By generating disturbance fingerprints and predicting process lag propagation, a time-sharing control strategy is collaboratively generated, solving the problems of response lag and control instability in the wastewater treatment system under influent shock loads, and achieving efficient and stable wastewater treatment.
Patent Information
- Application Number
- CN202611447284.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-17
- Publication Date
- 2026-11-17
AI Technical Summary
Existing wastewater treatment systems are slow to respond to influent shock loads and have uncoordinated control actions, resulting in risks to effluent quality, high energy and chemical consumption, and unstable control due to abnormal online data.
By acquiring multi-source operational data to generate disturbance fingerprints, predicting process section lag propagation, collaboratively generating time-sharing control strategies for multiple actuators, and considering the coupling effects between actuators, feedforward control is achieved.
It significantly improves the response speed and control stability of the wastewater treatment system to complex influent load fluctuations, reduces energy and chemical consumption, avoids process conflicts, and ensures that the effluent water quality meets the standards.
Smart Images

Figure CN123113572A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of smart water management and wastewater treatment process control technology, and in particular to a collaborative control method and related apparatus for wastewater treatment processes. Background Technology
[0002] The influent load of urban wastewater treatment plants exhibits significant time-varying and uncertainties. Factors such as peak residential water consumption, rainfall infiltration, intermittent industrial wastewater discharge, and seasonal water temperature changes can all cause drastic fluctuations in influent flow rate, pollutant concentration, and carbon-nitrogen ratio within a short period. Existing operation and control methods typically use blowers, return pumps, dosing pumps, and sludge discharge valves as the regulating objects, relying on real-time monitoring values from online instruments or manual experience for feedback regulation.
[0003] However, existing control strategies have significant technical shortcomings in practical applications. On the one hand, control actions often lag behind influent disturbances. Systems typically increase aeration or dosing only after ammonia nitrogen levels rise in the aerobic tank or total phosphorus levels in the effluent exceed standards. By this time, the shock load has already been transmitted to subsequent treatment units, easily causing short-term risks to effluent quality. On the other hand, some intelligent control schemes only optimize dissolved oxygen or single agent dosage independently, failing to consider aeration, recirculation, carbon sources, phosphorus removal agents, and sludge discharge as interconnected and coupled execution objects. This easily leads to process conflicts such as over-aeration inhibiting denitrification, mismatch between carbon source dosage and internal recirculation, and excessive phosphorus removal agent dosage affecting sludge settling. Furthermore, online instruments in wastewater treatment plants are susceptible to sludge adhesion, probe drift, or communication interruptions. Relying solely on measured values can easily lead to model misjudgments and control malfunctions, and fixed-parameter models are difficult to adapt to changes in hysteresis characteristics under different seasons and operating conditions.
[0004] Therefore, the existing technology lacks a control method that can identify the process impact of influent disturbances before they are transmitted to the effluent end, transform shock loads into interpretable process disturbance characteristics, and establish collaborative constraint relationships among multiple actuators to adapt to changes in data reliability. This results in wastewater treatment systems exhibiting slow response, poor control stability, and high energy and chemical consumption when dealing with complex influent fluctuations. Summary of the Invention
[0005] In view of the problems existing in the prior art, such as delayed response to influent shock loads, process conflicts easily caused by independent adjustment of multiple actuators, and control instability due to abnormal online data, this application provides a collaborative control method and related device for a wastewater treatment process to achieve early identification of influent disturbances and multi-variable collaborative feedforward control. The specific scheme is as follows: The first aspect of this application provides a method for coordinated control of a wastewater treatment process, comprising: Acquire multi-source operational data during the wastewater treatment process, including influent side data, process data, external disturbance data, and actuator status data; Based on the multi-source operational data, a perturbation fingerprint is generated, which characterizes the direction and intensity of the impact of influent shock on the subsequent biochemical treatment process. Based on the perturbation fingerprint, process segment hysteresis propagation prediction is performed to obtain the predicted state of the subsequent biochemical process segment. Based on the predicted state of the process section, and under the constraints of effluent water quality and process conflict constraints that take into account the coupling effects between various actuators, a time-sharing control strategy for multiple actuators is collaboratively generated.
[0006] In one possible implementation, generating the perturbation fingerprint based on the multi-source runtime data includes: Multidimensional feature components are calculated based on the multi-source operational data. These multidimensional feature components include any combination of organic load intensity, nitrogen load intensity, phosphorus load intensity, carbon-nitrogen ratio deviation, rainwater dilution, industrial shock probability, low-temperature nitrification pressure, load rise rate, duration, and data reliability. Obtain the credibility score of the data source corresponding to each of the aforementioned feature components; Based on the credibility score, the feature components are weighted and fused to generate an influent load disturbance feature vector. The disturbance type is labeled based on the preset category labels to obtain the disturbance fingerprint. The disturbance type labels include residential peak type, rainwater dilution type, industrial shock type, low temperature nitrification pressure type, carbon source deficiency type and composite disturbance type.
[0007] In one possible implementation, the step of performing process segment hysteresis propagation prediction based on the perturbation fingerprint to obtain the predicted process segment state for subsequent biochemical processing includes: Obtain current tank volume, hydraulic retention time, reflux ratio, sludge concentration, sludge age, water temperature, and historical response curves; Based on the disturbance fingerprint, the current tank volume, the hydraulic retention time, the reflux ratio, the sludge concentration, the sludge age, the water temperature, and the historical response curve, process segment hysteresis propagation prediction is performed to obtain the predicted state of the subsequent biochemical treatment process segment.
[0008] In one possible implementation, the subsequent biochemical treatment process includes an anaerobic zone, an anoxic zone, an aerobic zone, and a mud-water separation unit; The predicted state of the process segment includes: Anaerobic lag time, anoxic lag time, aerobic lag time, and mud-water separation lag time.
[0009] In one possible implementation, the collaborative generation of a time-sharing control strategy for the multiple actuators includes: Taking the effluent quality meeting the standard as a hard constraint, and taking aeration energy consumption, reagent consumption and equipment operation frequency as optimization objectives, the time-sharing control strategy for multiple actuators is generated by combining the safety boundary of the actuator and process conflict verification for collaborative constraint decision-making. The effluent water quality is used to indicate any one or more of the following: ammonia nitrogen, total nitrogen, total phosphorus, chemical oxygen demand (COD), and suspended solids (SS). The multiple execution structures include an aeration system, a return system, a chemical system, and a sludge removal system; The time-sharing control strategy for the multiple actuators includes setting the blower frequency, setting the aeration valve opening degree, setting the internal return pump speed, setting the external return pump speed, setting the carbon source dosing pump, setting the phosphorus removal agent dosing pump, setting the sludge discharge valve opening degree, and setting the sludge discharge cycle.
[0010] In one possible implementation, the actuator safety boundary includes adjustment boundaries, action intervals, and backoff conditions.
[0011] In one possible implementation, the process conflict verification includes at least one of the following: When a candidate control strategy reduces the risk of a certain pollutant but introduces process side effects, the priority of the candidate control strategy is reduced through a conflict penalty term. When increasing the internal recirculation flow leads to an increase in dissolved oxygen or an increased risk of insufficient carbon source in the anoxic zone, a conflict penalty between internal recirculation and denitrification is triggered. When increasing the aeration rate leads to an increased risk of internal backflow carrying oxygen or when energy consumption exceeds a preset threshold, a conflict penalty is triggered between aeration and energy consumption and backflow carrying oxygen. When increasing the dosage of phosphorus removal agent leads to an increased risk of deterioration in sludge settling performance, a conflict penalty between phosphorus removal agent and sludge settling is triggered.
[0012] In one possible implementation, the method further includes: The online monitoring data in the multi-source operation data is verified, and the data status is divided into four levels: reliable, repairable, low reliability, and unavailable. For trusted data, it directly participates in the generation of the perturbation fingerprint; For recoverable data, it is compensated by soft measurement or historical similar operating conditions before being used in the generation of the perturbation fingerprint; For low-confidence data, reduce the weight of the corresponding feature components in the perturbation fingerprint and limit the control amplitude of the time-division control strategy; For unavailable data, trigger the policy safety boundary and limit the single adjustment range of the actuator.
[0013] In one possible implementation, after the multiple actuators execute the time-sharing control strategy, the method further includes: The actual conditions are collected, and the deviation between the predicted conditions and the actual conditions is calculated. The deviation includes any combination of time deviation, concentration deviation, response deviation, energy consumption deviation, and chemical consumption deviation. The actual conditions include the actual effluent quality, actual aeration rate, return flow rate, chemical dosage, and sludge discharge rate. Based on the aforementioned deviation, the perturbation fingerprint weight, the predicted state of the process segment, the aeration response coefficient, the carbon source response coefficient, the phosphorus removal agent response coefficient, and the strategy safety boundary are corrected.
[0014] In one possible implementation, after the collaborative generation of time-sharing control strategies for the multiple actuators, the method further includes: The time-sharing control strategy is verified in two layers. The first layer is software verification to determine whether the target value exceeds the process boundary and equipment boundary. The second layer is programmable logic controller verification to determine the remote controllable status of the equipment and the start-stop interval. If any of the two-layer verifications fails, the prediction result is retained but automatic control is not executed, and a running prompt is generated.
[0015] A second aspect of this application provides a coordinated control device for a wastewater treatment process, comprising: A multi-source operation data acquisition unit is used to acquire multi-source operation data during the wastewater treatment process. The multi-source operation data includes influent side data, process data, external disturbance data, and actuator status data. The perturbation fingerprint generation unit is used to generate a perturbation fingerprint based on the multi-source operating data. The perturbation fingerprint characterizes the direction and intensity of the impact of influent impact on the subsequent biochemical treatment process. The prediction unit is used to predict the process segment hysteresis propagation based on the perturbation fingerprint, so as to obtain the predicted state of the process segment for subsequent biochemical treatment. The strategy generation unit is used to collaboratively generate a time-sharing control strategy for multiple actuators based on the predicted state of the process section, under the constraints of effluent water quality and process conflict constraints that take into account the coupling effects between the actuators.
[0016] A third aspect of this application provides a computer program product including computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the coordinated control method for a wastewater treatment process as described in the first aspect or any implementation thereof.
[0017] A fourth aspect of this application provides an electronic device, including at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the electronic device can implement the coordinated control method of the wastewater treatment process according to the first aspect or any implementation thereof.
[0018] The fifth aspect of this application provides a computer-readable storage medium carrying one or more computer programs that, when executed by an electronic device, enable the electronic device to implement the coordinated control method for the wastewater treatment process described in the first aspect or any implementation thereof.
[0019] By employing the aforementioned technical solutions, the wastewater treatment process coordinated control method and related apparatus provided in this application generate disturbance fingerprints characterizing the direction and intensity of influent impact by acquiring multi-source operational data. This transforms the originally discrete and random fluctuations in influent water quality and quantity into composite feature vectors with clear process meanings, solving the semantic ambiguity problem caused by traditional methods relying solely on feedback from a single effluent index. Based on this, the disturbance fingerprint is used for process segment lag propagation prediction, quantifying the spatiotemporal transmission patterns of disturbances in tandem units such as anaerobic, anoxic, aerobic, and sludge-water separation. This allows the control system to detect risk states before disturbances reach sensitive process segments, thus overcoming the deficiency in existing technologies where control actions lag significantly behind influent disturbances. Furthermore, when generating the time-sharing control strategy, process conflict constraints considering the coupling effects between actuators are introduced. Aeration, recirculation, chemical dosing, and sludge discharge are incorporated into a unified collaborative decision-making framework. This avoids process side effects caused by optimizing a single actuator, such as over-aeration inhibiting denitrification, mismatch between carbon source and recirculation, or excessive chemical dosing worsening sludge properties. It achieves optimal global energy efficiency and chemical consumption while ensuring effluent quality. This method constructs a complete feedforward control closed loop from disturbance nature identification and spatiotemporal propagation location to global collaborative execution, significantly improving the response speed and control stability of the wastewater treatment system in the face of complex influent load fluctuations. Attached Figure Description
[0020] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.
[0021] Figure 1 A schematic flowchart illustrating a collaborative control method for a wastewater treatment process provided in an embodiment of this application; Figure 2 A schematic diagram of the arrangement of A2 / O process measurement points and actuators provided in the embodiments of this application; Figure 3 This is a schematic diagram of the perturbation fingerprint data structure provided in an embodiment of this application; Figure 4 This is a schematic diagram of the time-space propagation of process segment delay provided in the embodiments of this application; Figure 5 This application provides a closed-loop diagram for the coordinated control of aeration return agent and sludge discharge in an embodiment of the present application. Figure 6 The online credibility grading and degradation control state transition diagram provided in the embodiments of this application; Figure 7 A schematic diagram of the prediction bias and parameter feedback correction matrix provided in the embodiments of this application; Figure 8 A schematic diagram of the structure of a coordinated control device for a wastewater treatment process provided in an embodiment of this application; Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0022] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.
[0023] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.
[0024] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.
[0025] The collaborative control method for the wastewater treatment process according to embodiments of this application will be described in detail below with reference to the accompanying drawings.
[0026] Example 1
[0027] Reference Figure 1 , Figure 1This application provides a flowchart illustrating a collaborative control method for a wastewater treatment process. This method establishes a complete control framework from disturbance identification and propagation prediction to collaborative decision-making, applicable to A2 / O (Anaerobic-Anoxic-Oxic), oxidation ditch, SBR (Sequencing Batch Reactor Activated Sludge Process), MBR (Membrane Bioreactor), and their combined processes. For ease of understanding, this application provides an example of the A2 / O process to illustrate the arrangement of measuring points and actuators. It should be understood that although... Figure 2 The arrangement of measuring points and actuators is illustrated using the A2 / O process as an example. However, in other embodiments, this method can also be applied to other wastewater treatment processes with series biochemical treatment units, as long as the functional requirements of multi-source data acquisition and multi-actuator collaborative control are met.
[0028] like Figure 1 As shown in the embodiments of this application, a method for coordinated control of a wastewater treatment process may include the following steps, which are described in detail below.
[0029] Step S100: Obtain multi-source operational data during the wastewater treatment process. The multi-source operational data includes influent side data, process data, external disturbance data, and actuator status data.
[0030] Specifically, combined Figure 2 The diagram showing the layout of measuring points and actuators in the A2 / O process demonstrates that the collection of multi-source operational data covers key nodes throughout the entire wastewater treatment process, ensuring a sufficient information foundation for subsequent control.
[0031] The data on the influent side is mainly obtained through online instruments above the influent channel, including influent flow rate, COD (Chemical Oxygen Demand) concentration, ammonia nitrogen concentration, total nitrogen concentration, total phosphorus concentration, pH value, water temperature and conductivity, etc., which are used to sense the absolute amount and changing trend of the influent load in real time.
[0032] The process data covers the internal state of each biochemical reaction zone and sludge-water separation unit, such as the ORP value and phosphorus release status in the anaerobic tank, the nitrate nitrogen concentration in the anoxic tank, the dissolved oxygen (DO) and ammonia nitrogen concentration in the aerobic tank, the mixed liquor suspended solids (MLSS) concentration, and the sludge level in the secondary sedimentation tank or the pressure difference in the membrane tank. These data reflect the real-time progress of the pollutant removal process.
[0033] External disturbance data includes water inflow from off-site pumping stations, rainfall and intensity, operational status of key discharge outlets in industrial parks, holiday types and time period labels, etc., which are used to capture environmental and scheduling factors that have a significant impact on the process but are not water quality itself.
[0034] The status data of the actuators are directly collected from the field control cabinet and frequency converter, including blower frequency, aeration valve opening, internal and external return pump speed, set value and actual flow rate of carbon source addition pump and phosphorus removal agent addition pump, sludge discharge valve opening and sludge discharge cycle, etc.
[0035] By integrating the above four types of heterogeneous data, the system can construct a dynamic digital image that reflects the overall picture of the wastewater treatment plant, avoiding information loss and response lag caused by relying on a single effluent indicator.
[0036] Step S200: Based on multi-source operational data, a perturbation fingerprint is generated. The perturbation fingerprint characterizes the direction and intensity of the impact of influent impact on the subsequent biochemical treatment process.
[0037] Specifically, the perturbation fingerprint is not a traditional single water quality prediction value or a simple load statistic, but a composite feature vector. Its core function is to transform the discrete, random, and multidimensional fluctuation signals on the influent side into interpretable features with clear process implications. This feature vector integrates multi-dimensional information such as organic load, nitrogen and phosphorus load, carbon-nitrogen ratio matching degree, hydraulic dilution effect, industrial shock risk, and low-temperature nitrification pressure. Through data credibility weighted fusion, it forms a unified quantitative expression of the direction of influence (such as promoting nitrification, inhibiting denitrification, increasing phosphorus release demand, etc.) and the intensity of influence (such as mild fluctuations, moderate shocks, and severe overload) on subsequent biological treatment units.
[0038] For example, when the influent flow rate increases sharply but the pollutant concentration decreases simultaneously and is accompanied by heavy rainfall, the disturbance fingerprint will be characterized as a "hydraulic dilution type" shock, indicating that the main risks faced by subsequent process stages are a shortened hydraulic retention time and an increase in the solid load of the secondary sedimentation tank, rather than an increase in the absolute amount of organic matter or ammonia nitrogen; conversely, if COD and conductivity rise sharply without rainfall, the fingerprint will be characterized as an "industrial shock type", indicating a surge in oxygen demand in the aerobic zone and a possible suppression of sludge activity.
[0039] By transforming raw data into perturbation fingerprints, the system achieves a shift from "data-driven" to "process semantic-driven" approaches, providing a standardized and noise-resistant input benchmark for subsequent hysteresis propagation prediction.
[0040] Step S300: Based on the perturbation fingerprint, perform process segment hysteresis propagation prediction to obtain the predicted state of the process segment for subsequent biochemical treatment.
[0041] Specifically, process-stage delay propagation prediction aims to address the widespread spatial and temporal delay issues in wastewater treatment. As wastewater flows sequentially through the anaerobic, anoxic, aerobic, and sludge-water separation units, disturbances at the inlet end do not immediately manifest at the outlet end. Instead, they propagate segment by segment along the process flow, and the residence time and response characteristics in each segment dynamically change due to various factors such as tank volume, reflux ratio, sludge concentration, and water temperature.
[0042] This step utilizes disturbance fingerprints as excitation inputs, combined with physicochemical parameters under current operating conditions, to predict the timing of disturbances reaching each series process stage within a future period and the resulting intermediate state changes, such as the peak time of anaerobic phosphorus release pressure, the size of the carbon source gap in the anoxic zone's denitrification, the increase in ammonia nitrogen oxidation load in the aerobic zone, and the solids flux risk in the sludge-water separation unit. This predictive mechanism enables the control system to detect potential risks in each process stage before disturbances cause effluent exceedances, thus transforming traditional post-effect remediation into pre-risk intervention, significantly improving the foresight and proactiveness of control.
[0043] Step S400: Based on the predicted state of the process section, and under the constraints of process conflict that meet the effluent quality constraints and consider the coupling effects between various actuators, a time-sharing control strategy for multiple actuators is collaboratively generated.
[0044] Specifically, the collaborative generation control strategy is the core element for achieving global optimization in this embodiment. The system takes the compliance of effluent ammonia nitrogen, total nitrogen, total phosphorus, COD, and SS as insurmountable hard constraints, while taking aeration energy consumption, reagent consumption, and equipment operation frequency as optimization targets. When generating control commands, the system does not calculate each actuator independently, but comprehensively considers the coupling effects between them under a unified decision-making framework.
[0045] For example, while increasing the internal recirculation flow rate helps nitrate nitrogen return to promote denitrification, excessive internal recirculation may bring a large amount of dissolved oxygen into the anoxic zone, damaging the denitrification environment or exacerbating carbon source deficiency; while increasing the aeration rate can enhance nitrification, excessive aeration not only increases energy consumption, but may also interfere with the function of the anoxic zone through oxygen carried by internal recirculation, and may even lead to sludge disintegration; while increasing the dosage of phosphorus removal agent can reduce total phosphorus in the effluent, excessive dosage may deteriorate sludge settling performance, thereby affecting the SS concentration in the secondary sedimentation tank effluent and the concentration of returned sludge.
[0046] Therefore, when generating time-sharing control strategies for the blower, return pump, dosing pump, and sludge discharge valve, the system introduces a process conflict verification mechanism to globally score and filter candidate strategies. This ensures that the final output control sequence is the optimal solution for energy efficiency and chemical consumption while guaranteeing effluent safety, rather than an overall imbalance caused by local optima. This collaborative decision-making model effectively overcomes the process side effects of traditional single-loop control or multivariate independent optimization, achieving stable and efficient operation of the wastewater treatment process.
[0047] Through steps S100 to S400, this embodiment constructs a complete technical closed loop of "multi-source data perception - disturbance feature extraction - spatiotemporal risk prediction - multivariate collaborative decision-making". This closed loop transforms the physical fluctuations of influent load into calculable process semantics, transforms delayed effluent feedback into forward-looking process prediction, and transforms decentralized equipment adjustment into global collaborative optimization. It fundamentally solves the technical defects of existing technologies, such as delayed response to influent impacts, mutual constraints on control actions, and over-reliance on single indicator feedback. It provides reliable methodological support for wastewater treatment plants to achieve stable compliance and energy conservation under complex influent conditions.
[0048] Example 2
[0049] In some embodiments, the process of generating perturbation fingerprints based on multi-source operational data is not a simple numerical statistical process, but is achieved through a structured data processing workflow.
[0050] Specifically, the process includes: Step S201: Calculate multidimensional feature components based on multi-source operational data. The multidimensional feature components include any combination of organic load intensity, nitrogen load intensity, phosphorus load intensity, carbon-nitrogen ratio deviation, rainwater dilution, industrial shock probability, low-temperature nitrification pressure, load rise rate, duration, and data reliability. Step S202: Obtain the credibility score of the data source corresponding to each feature component; Step S203: Based on the credibility score, perform weighted fusion of each feature component to generate the influent load disturbance feature vector; Step S204: Based on the preset category labels, the influent load disturbance feature vector is labeled with disturbance type to obtain disturbance fingerprint. The disturbance type labels include residential peak type, rainwater dilution type, industrial shock type, low temperature nitrification pressure type, carbon source deficiency type and composite disturbance type.
[0051] like Figure 3As shown, when generating fingerprints, the system first uses a multi-scale sliding time window to perform temporal alignment and feature extraction on the input data. The input sequences cover influent load sequences (such as flow rate Q, COD, ammonia nitrogen NH3-N, total nitrogen TN, total phosphorus TP, and suspended solids SS), external disturbance sequences (such as rainfall intensity, pump station inflow, industrial discharge outlet status, and holiday type), and confidence sequences (such as communication status, instrument drift markers, and maintenance records). This synchronous input of multi-source heterogeneous data provides a complete information foundation for the subsequent generation of composite features with process interpretability.
[0052] In step S201, the system calculates multidimensional characteristic components based on multi-source operational data. These components are not isolated water quality indicators, but composite parameters after process semantic transformation. For example, organic load intensity combines influent flow rate and COD or BOD concentration to characterize the instantaneous demand of biodegradable organic matter on the aeration system; nitrogen load intensity combines flow rate, ammonia nitrogen concentration, and water temperature to reflect the metabolic pressure on nitrifying bacteria; carbon-nitrogen ratio deviation quantifies the matching degree between internal recirculation and external carbon sources by comparing the actual C / N ratio with the theoretical denitrification demand; rainwater dilution is correlated with rainfall intensity, flow rate change rate, and conductivity decrease to identify the potential risk of hydraulic shock to the solid flux of the secondary sedimentation tank; industrial shock suspicion is used to predict the risk of toxic and harmful substances inhibiting sludge activity by analyzing synchronous abnormal changes in COD, conductivity, and pH, as well as the status of key discharge outlets; and low-temperature nitrification pressure is specifically quantified for specific operating conditions where the water temperature is below the threshold and the ammonia nitrogen load increases.
[0053] It should be understood that the above component list is only an illustrative example. In actual deployment, the system can select any combination of the above components according to the specific process configuration of the wastewater treatment plant (such as whether there is a primary sedimentation tank, whether an MBR is used, etc.), or add other derived variables that can characterize the influent shock characteristics, as long as they can reflect the direction and intensity of the impact of influent fluctuations on the subsequent biological treatment units.
[0054] In step S202, the system obtains the reliability score of the data source corresponding to each feature component. Due to the harsh environment at wastewater treatment sites, online instruments are highly susceptible to sludge adhesion, probe aging, electromagnetic interference, or unstable communication links, leading to drift, stagnation, or jumps in the collected raw data. Directly using data without quality assessment to calculate feature components can easily generate erroneous perturbation fingerprints, thus misleading subsequent control decisions. Therefore, this embodiment introduces an independent reliability evaluation mechanism, performing range verification, slope change verification, consistency verification of adjacent measuring points, and comparison with historical similar operating conditions for each data source, outputting a continuous reliability score or discrete quality level between 0 and 1. This step explicitly embeds data quality information into the feature generation chain, ensuring the physical authenticity of the perturbation fingerprint.
[0055] In step S203, the system performs weighted fusion of each feature component based on the reliability score to generate an influent load disturbance feature vector. Specifically, the fusion process does not involve a simple arithmetic average or fixed coefficient weighting of each component, but rather uses the reliability score of the corresponding data source as a dynamic weighting factor. When the reliability score of a certain data source (such as the influent ammonia nitrogen meter) is high, the weight of its corresponding nitrogen load intensity component in the feature vector increases accordingly; conversely, when the meter malfunctions or data anomalies cause the reliability score to decrease, the system automatically reduces the weight of that component, or even sets it to zero, while increasing the weight of soft measurement components derived from other relevant measuring points (such as oxygen consumption in the aerobic zone and the trend of effluent nitrate nitrogen).
[0056] This credibility-based adaptive fusion mechanism enables the generated influent load disturbance feature vector to effectively suppress interference from low-quality data and maintain the robustness of feature representation even when some sensors fail. This avoids the problem of the entire control system being paralyzed or malfunctioning due to single-point instrument failure, which is a problem in traditional methods.
[0057] In step S204, the system labels the influent load disturbance feature vector with disturbance type based on preset category labels, obtaining the final disturbance fingerprint. Although the feature vector itself already contains rich quantitative information, discrete semantic labels are more operable for operators and upper-level control strategy libraries. The system maps high-dimensional feature vectors to predefined operating condition categories through pattern recognition or classification algorithms.
[0058] For example, when the feature vector shows a surge in flow rate, dilution of pollutant concentration, and heavy rainfall, it is labeled as rainwater dilution type, indicating that the control system should focus on the shortened hydraulic retention time and the risk of sludge runoff in the secondary sedimentation tank; when the feature vector shows a simultaneous surge in COD and conductivity, pH shift, and no rainfall, it is labeled as industrial shock type, indicating that priority should be given to ensuring aeration and limiting sludge discharge to maintain sludge age; when the feature vector shows low water temperature and a decrease in ammonia nitrogen oxidation rate, it is labeled as low-temperature nitrification pressure type, guiding the system to extend the aeration time in the aerobic zone in advance.
[0059] By transforming continuous mathematical features into explicit process semantic labels, this embodiment not only enhances the interpretability of the technical solution, but also lays the foundation for accurate matching of perturbation fingerprints with differentiated collaborative control strategies in subsequent embodiments.
[0060] Example 3
[0061] In some embodiments, in order to transform the abstract perturbation fingerprint into a control basis with explicit temporal attributes, process hysteresis propagation prediction is performed based on the perturbation fingerprint to obtain the predicted state of the process segment for subsequent biochemical treatment, specifically including: Step S301: Obtain the current tank volume, hydraulic retention time, reflux ratio, sludge concentration, sludge age, water temperature, and historical response curves; Step S302: Based on the disturbance fingerprint, current tank volume, hydraulic retention time, reflux ratio, sludge concentration, sludge age, water temperature and historical response curves, perform process section hysteresis propagation prediction to obtain the predicted state of the subsequent biochemical treatment process section.
[0062] Specifically, the process section delay propagation prediction in this embodiment does not use fixed empirical delay parameters, but instead constructs a multi-parameter coupled prediction mechanism that integrates hydraulic transport and biochemical reaction kinetics. For example... Figure 4 The diagram shows the time-space propagation of the disturbance in the process section. The disturbance loading time at the inlet is t0, and it then passes through the anaerobic zone, anoxic zone, aerobic zone, and mud-water separation unit in sequence, generating corresponding risk response curves in each section. In the figure, τ1 to τ5 represent the dynamic lag time of the disturbance between each series unit. These lag times are not constant values, but are calculated in real time from the above-mentioned multiple physicochemical parameters.
[0063] In step S301, the system acquires the current tank volume, hydraulic retention time, reflux ratio, sludge concentration, sludge age, water temperature, and historical response curves in real time. These parameters constitute the physical baseline of the prediction model. Among these factors, the current tank volume and hydraulic retention time determine the basic hydraulic transmission time of wastewater in the treatment unit, serving as the geometric boundary for calculating lag time. The recirculation ratio (including internal and external recirculation) directly alters the actual flow rate of each process section. When the recirculation ratio increases, the actual retention time per unit volume of wastewater is compressed, leading to accelerated disturbance transmission, and vice versa. Sludge concentration and sludge age reflect the biomass stock and metabolic activity of the biological system. High sludge concentration or long sludge age usually indicates stronger pollutant buffering and degradation capabilities, which can smooth out peaks and valleys, altering the duration and peak shape of risk states. Water temperature is a key environmental factor affecting the rate of biochemical reactions. According to the laws of microbial growth kinetics, low temperatures significantly reduce the metabolic rate of sensitive bacteria such as nitrifying bacteria, leading to a prolonged lag time in processes such as ammonia nitrogen oxidation. Historical response curves record the actual input-output relationship of the wastewater treatment plant under similar operating conditions in the past, used to compensate for deviations between theoretical models and non-ideal factors such as aging equipment and dead zones in the flow regime.
[0064] In step S302, the system uses the perturbation fingerprint as an excitation input and combines it with the real-time acquired physicochemical parameters to predict the lag propagation of the process segment. The prediction process essentially solves a state transition problem for a time-varying nonlinear system. For example, when the perturbation fingerprint indicates "low-temperature nitrification pressure type" and the current water temperature is below 12 degrees Celsius, the model automatically increases the biochemical reaction time constant in the aerobic zone, calculating a longer ammonia nitrogen oxidation lag time and a higher DO demand peak than under normal temperature conditions. When the perturbation fingerprint indicates "rainwater dilution type" and the internal recirculation ratio is high, the model comprehensively considers the hydraulic scouring effect and the recirculation carrying effect, re-estimating the arrival time of the denitrification carbon source gap in the anoxic zone. The final output of the predicted state of the subsequent biological treatment process segment not only includes the risk intensity of each segment but also accurately provides the time window for the risk to arrive.
[0065] This embodiment introduces multi-dimensional physical parameters to anchor the prediction model, achieving a leap from "static table lookup" to "dynamic mechanism deduction" in technical solutions. Compared to traditional methods that rely solely on the statistical average of historical data, this prediction mechanism based on multiple physical parameters can keenly perceive and adapt to the frequent operational drift that occurs during wastewater treatment. For example, in scenarios such as drastic water temperature changes due to seasonal transitions, fluctuations in sludge age due to adjustments in sludge discharge strategies, or adjustments to the return ratio due to energy-saving requirements, the prediction model can automatically correct the lag time parameter, ensuring that the control system always makes decisions based on accurate spatiotemporal coordinates. This avoids insufficient lead time or over-adjustment caused by model mismatch, significantly enhancing the robustness and accuracy of the collaborative control method under complex operating conditions throughout the year.
[0066] Example 4
[0067] In some embodiments, in order to further improve the accuracy and targeting of regulation, the subsequent biochemical treatment process includes an anaerobic zone, an anoxic zone, an aerobic zone, and a mud-water separation unit; then, the predicted state of the process includes: anaerobic lag time, anoxic lag time, aerobic lag time, and mud-water separation lag time.
[0068] Specifically, combined Figure 4 The diagram illustrates the time-space propagation of the process segment. Disturbances at the inlet do not affect the effluent quality instantaneously as a whole, but rather propagate sequentially along the physical flow of wastewater treatment, exhibiting different response characteristics in each functional unit. This embodiment decouples the abstract propagation process into four specific time-space nodes, each corresponding to an independent time lag parameter.
[0069] The anaerobic lag time characterizes the time required for influent disturbance to enter the system and trigger a change in phosphorus release pressure in the anaerobic zone. This timeframe is primarily influenced by influent flow rate, anaerobic tank volume, and sludge return ratio, and is a key indicator for determining whether sludge discharge or external return is necessary to maintain the anaerobic environment. The anoxic lag time characterizes the time it takes for the disturbance to reach the anoxic zone and cause a denitrification carbon source deficit to appear. Since the anoxic zone is typically located after the anaerobic zone and receives internal return liquid, its lag time depends not only on hydraulic transmission but also on the coupling effect of the internal return ratio and the concentration of nitrified liquid in the aerobic zone, directly determining the lead time for adding external carbon sources. The aerobic lag time characterizes the time it takes for the disturbance to reach the aerobic zone and cause ammonia nitrogen oxidation load or dissolved oxygen demand peak. This is the longest and most significant link in the entire biochemical reaction chain, and its length is determined by the aerobic tank volume, sludge age, water temperature, and microbial activity. The sludge-water separation lag time characterizes the time it takes for a disturbance to ultimately reach the sludge-water separation unit, such as the secondary sedimentation tank or membrane tank, and trigger a risk of solid flux or pressure difference changes. This parameter is crucial for preventing sludge runoff, controlling the concentration of returned sludge, and ensuring that the suspended solids in the effluent meet the standards.
[0070] The technical value of this segmented prediction mechanism lies in its transformation of the vague "effluent exceeding the standard risk" in traditional control into a clear "process unit risk," enabling the control system to prepare differentiated control strategies in advance for specific problems in different process stages. For example, when the prediction shows a short anoxic lag time and a large carbon source deficit, the system can prioritize adjusting the internal reflux pump speed or starting the carbon source dosing pump without waiting for the ammonia nitrogen in the aerobic zone to rise before reacting; when the prediction shows that the sludge-water separation lag time is approaching and the solids load is too high, the system can limit the sludge discharge or adjust the flocculant addition in advance to avoid sludge backflow in the secondary sedimentation tank. Compared to a single mode that only predicts the final effluent indicators, this embodiment achieves refined perception and graded response to risks throughout the entire wastewater treatment process by constructing a multi-dimensional process state prediction system. This effectively avoids global effluent accidents caused by the failure to detect local deterioration in a timely manner, and also provides precise time window constraints to prevent over-adjustment.
[0071] It should be understood that although this embodiment uses the four typical functional areas of the A2 / O process as an example for illustration, in other embodiments, this segmentation logic is also applicable to wastewater treatment processes with similar functional zoning, such as oxidation ditch, SBR, and MBR. It is only necessary to adjust the process segment division method and the corresponding lag time definition according to the actual process flow, which does not depart from the protection scope of this application.
[0072] Example 5
[0073] In some embodiments, in order to achieve global optimization and control of the wastewater treatment process, a time-sharing control strategy for multiple actuators is generated collaboratively, including: taking effluent quality compliance as a hard constraint, taking aeration energy consumption, reagent consumption and equipment operation frequency as optimization objectives, and combining actuator safety boundaries and process conflict verification to perform collaborative constraint decision-making to generate a time-sharing control strategy for multiple actuators.
[0074] Specifically, such as Figure 5 The diagram shown illustrates a closed-loop control scheme for aeration, recirculation, and sludge removal. This embodiment constructs a unified decision-making framework with multiple objectives and constraints. Instead of implementing independent single-loop control for each actuator, this framework uses disturbance fingerprints, predicted process conditions, effluent quality constraints, and equipment status constraints as common inputs to a collaborative constraint decision-maker for overall optimization. In this decision-making system, achieving effluent quality standards is set as an insurmountable hard constraint. Regardless of energy or chemical consumption optimization, the generated control strategy must first ensure that any one or more of the following indicators in the effluent—ammonia nitrogen, total nitrogen, total phosphorus, chemical oxygen demand (COD), and suspended solids (SS)—meet emission standards or internal control thresholds. This hard constraint mechanism ensures environmental safety and prevents water quality risks caused by excessive pursuit of economic benefits. Based on this, the system uses aeration energy consumption, chemical consumption, and equipment operation frequency as optimization objectives to seek the lowest overall cost solution while meeting the hard water quality constraints. Among them, the frequency of equipment operation is one of the optimization objectives, which aims to reduce the frequent start-stop and adjustment wear of electromechanical equipment such as blowers, water pumps and valves, extend the service life of equipment and maintain the stability of process operation.
[0075] Furthermore, the core of collaborative constraint decision-making lies in combining the safety boundaries of the actuators with process conflict verification. Due to the highly nonlinear and strongly coupled nature of biochemical reactions in wastewater treatment, adjustments to a single actuator often trigger chain reactions. For example, while increasing the speed of the internal recirculation pump helps increase the nitrate nitrogen return flow to promote denitrification, it may also introduce excessive dissolved oxygen into the anoxic zone, disrupting the anaerobic environment required for denitrification or exacerbating the risk of insufficient carbon sources. Increasing aeration can enhance nitrification, but excessive aeration not only directly increases power consumption but may also interfere with the function of the anoxic zone through oxygen carried by the internal recirculation, even leading to sludge disintegration and affecting the settling performance of the secondary sedimentation tank. Therefore, this embodiment forcibly introduces a process conflict verification mechanism when generating control strategies, performing global scoring and screening of candidate strategies to eliminate those strategies that, while reducing the risk of a certain pollutant, introduce serious process side effects, ensuring that the final output control command is the globally optimal solution after multi-variable coordination, rather than an overall imbalance caused by local optima.
[0076] In this embodiment, the controlled multiple actuators cover the key control measures of the entire wastewater treatment process, specifically including the aeration system, recirculation system, chemical system, and sludge discharge system. The time-sharing control strategies generated for these actuators include settings for blower frequency, aeration valve opening, internal recirculation pump speed, external recirculation pump speed, carbon source dosing pump, phosphorus removal agent dosing pump, sludge discharge valve opening, and sludge discharge cycle. These control quantities constitute a multi-dimensional time-sharing control vector, whose values are not calculated in isolation but are generated synchronously under the aforementioned unified constraint framework. For example, when the system predicts an increase in ammonia nitrogen oxidation load in the aerobic zone and insufficient carbon source in the anoxic zone within the next two hours, the collaborative decision-maker will not simply increase aeration and carbon source addition simultaneously. Instead, it will comprehensively consider the current internal recirculation ratio, sludge concentration, and equipment safety boundaries, and may generate a combined strategy: moderately increasing the blower frequency to ensure nitrification requirements, while limiting the increase in internal recirculation pump speed to avoid excessive dissolved oxygen in the anoxic zone, and accurately calculating the setpoint of the carbon source addition pump based on the carbon-nitrogen ratio deviation. If necessary, it will adjust the sludge discharge valve opening to maintain an appropriate sludge age. This time-sharing control strategy reflects foresight in the time dimension and synergy in the spatial dimension, enabling all actuators to operate in coordination under unified command, like an orchestra. This effectively overcomes the problems of mutual restraint and adjustment oscillation among subsystems in the traditional decentralized control mode, significantly improving the stability and energy efficiency of the wastewater treatment plant in dealing with complex influent load fluctuations.
[0077] Example 6
[0078] In some embodiments, to ensure the safe implementation of the collaboratively generated control strategy in actual engineering, the actuator safety boundary includes adjustment boundaries, action intervals, and backoff conditions. Specifically, this safety boundary constitutes a pre-filter layer of the collaborative constraint decision-maker, defining an absolutely insurmountable physical and technological safety envelope before or during algorithm optimization. Regardless of the theoretical optimal solution calculated by the collaborative decision-making model, the final output time-sharing control strategy must be verified and pruned by this safety boundary to prevent equipment damage or process collapse due to model distortion, data anomalies, or extreme operating conditions, thereby compensating for the robustness shortcomings of pure algorithm control in complex industrial environments.
[0079] The adjustment boundary refers to the numerical range within which the actuator is allowed to operate under current conditions. This range is determined by both the equipment's rated capacity and the process's permissible range, with the intersection of the two taken as the effective boundary. For example, for a blower in an aeration system, its frequency adjustment upper limit is not only limited by the rated power of the frequency converter and motor, but also by the maximum oxygen demand at the current sludge concentration in the aerobic tank and the shear force threshold to prevent sludge disintegration. For a sludge discharge valve in a sludge discharge system, its opening setting depends not only on the valve's mechanical stroke but also on the strict constraints of the real-time sludge level in the secondary sedimentation tank, sludge settling performance, and the sludge age maintenance requirements of the biological treatment tank. This dual constraint mechanism ensures that control commands neither exceed the hardware's limits nor violate the fundamental laws of biochemical reactions.
[0080] The action interval refers to the minimum time interval or frequency limit between two consecutive adjustment operations of an actuator. It aims to prevent mechanical wear and electrical shocks caused by frequent start-ups and shutdowns of equipment, while also avoiding process oscillations due to excessively frequent adjustments. For example, high-power blowers typically have minute-level cooling and stabilization windows for frequency adjustment; the start-up and shutdown operations of carbon source dosing pumps and phosphorus removal agent dosing pumps must meet specific protection cycles; and the opening and closing of sludge discharge valves must avoid periods of drastic fluctuations in influent flow. By solidifying these timing constraints within safety boundaries, the system can automatically smooth the control sequence, extend the service life of electromechanical equipment, maintain the relative stability of the internal environment of the biochemical system, and avoid unnecessary process disturbances caused by high-frequency noise in control signals.
[0081] The fallback condition refers to the logical rule that the system automatically triggers conservative control settings when a specific abnormal state or insufficient confidence level is detected. This is the most defensive element in the safety boundary. Different actuators are configured with differentiated fallback strategies. For example, when the dissolved oxygen (DO) meter is judged to be of low confidence or malfunctions, the aeration system's control strategy no longer relies on real-time DO feedback, but immediately falls back to a conservative aeration rate setting based on historical concurrent loads or feedforward models to prevent over-aeration or under-aeration due to erroneous readings. When dissolved oxygen in the anoxic zone rises abnormally or the risk of insufficient carbon source increases significantly, the internal recirculation system will limit further increases in the recirculation ratio, or even actively reduce the recirculation flow rate to preserve the denitrification environment, rather than blindly pursuing nitrate nitrogen recirculation efficiency. When encountering heavy rainfall that causes a surge in influent hydraulic load or abnormal fluctuations in the sludge level in the secondary sedimentation tank, the sludge discharge system will suspend large-scale sludge discharge operations and switch to low-flow sludge discharge to maintain sludge layer stability or temporarily close the sludge discharge valve to prevent sludge loss. Furthermore, when the confidence level of key predictive indicators (such as total nitrogen and total phosphorus in effluent) falls below a preset threshold, the reagent dosing system will limit the magnitude of a single adjustment and instead adopt a more gradual adjustment strategy. Through these multi-dimensional fallback conditions, this embodiment ensures that the collaborative control method remains safe and controllable even in the face of uncertainties such as sensor failure, communication interruptions, or extreme weather, effectively avoiding the "black box" risks common in intelligent control systems and significantly improving the operational safety and reliability of wastewater treatment plants.
[0082] Example 7
[0083] In some embodiments, to further reflect the inherent advantages of multi-actuator coordinated control, process conflict verification includes at least one of the following: when a candidate control strategy reduces the risk of a certain pollutant but introduces process side effects, the priority of the candidate control strategy is reduced through a conflict penalty item; when increasing the internal recirculation flow leads to an increase in dissolved oxygen in the anoxic zone or an increase in the risk of insufficient carbon source, a conflict penalty is triggered between internal recirculation and denitrification; when increasing the aeration rate leads to an increase in the risk of oxygen carrying by internal recirculation or energy consumption exceeding a preset threshold, a conflict penalty is triggered between aeration and energy consumption and oxygen carrying by recirculation; when increasing the dosage of phosphorus removal agent leads to an increase in the risk of deterioration of sludge settling performance, a conflict penalty is triggered between phosphorus removal agent and sludge settling.
[0084] Specifically, this embodiment is a further refinement of the collaborative constraint decision-making process in the aforementioned embodiments. Unlike the actuator safety boundary described in Embodiment 6, which serves as a hard filtration condition, process conflict verification is a soft optimization mechanism that operates during the scoring and ranking of candidate strategies. In wastewater treatment biochemical reaction systems, the regulation of each actuator often has a dual effect: while positively improving a certain water quality indicator, it may negatively interfere with another process step. If only traditional methods such as single-objective optimization or multi-objective weighted summation are used, it is easy to fall into local optimum traps, causing the system to frequently oscillate between solving one problem and creating another. This embodiment introduces a conflict penalty term into the collaborative constraint decision-maker, explicitly quantifying this coupling side effect as a negative score. This forces the optimization algorithm to actively avoid strategies that, while theoretically high in returns, have high actual process costs when searching the solution space, thereby ensuring that the final time-sharing control strategy is a globally optimal solution that balances water quality safety, operational stability, and economic costs.
[0085] In scenarios involving conflict between internal recirculation and denitrification, the system assesses the denitrification potential of the anoxic zone in real time during collaborative decision-making. While increasing the internal recirculation pump speed can theoretically increase the total amount of nitrate nitrogen recirculated from the aerobic zone to the anoxic zone, thus benefiting total nitrogen removal, in practice, an excessively high internal recirculation ratio can carry a large amount of the high-dissolved oxygen mixture from the aerobic zone into the anoxic zone, leading to an increase in dissolved oxygen concentration and disrupting the facultative anaerobic environment required by denitrifying bacteria. Simultaneously, the effective carbon source allocated per unit volume of wastewater is diluted, exacerbating the risk of carbon source deficiency. When the system predicts that increasing the internal recirculation flow rate will cause dissolved oxygen in the anoxic zone to exceed the denitrification inhibition threshold or the carbon-to-nitrogen ratio to fall below the critical value, even if the strategy significantly reduces the predicted total nitrogen in the effluent, the conflict penalty will significantly reduce its overall score, lowering the priority of this candidate strategy. This avoids the process pitfall of blindly pursuing nitrate nitrogen recirculation efficiency, which could lead to substantial failure of the denitrification process.
[0086] To address the conflict between aeration, energy consumption, and oxygen carried by the recirculation, the system comprehensively considers nitrification requirements, system energy efficiency, and environmental stability. While increasing blower frequency or aeration valve opening can directly improve dissolved oxygen levels in the aerobic zone and enhance ammonia nitrogen oxidation, its side effects are equally significant. On one hand, aeration energy consumption typically accounts for more than 50% of the total energy consumption of a wastewater treatment plant; excessive aeration can lead to a sharp increase in electricity consumption per ton of water. On the other hand, excessively high dissolved oxygen in the aerobic zone will transport excess oxygen to the anoxic zone through the internal recirculation system, creating a cumulative effect with the aforementioned internal recirculation conflict and further deteriorating the denitrification environment. Therefore, when the energy consumption increase caused by a candidate aeration strategy exceeds the preset economic threshold, or when the predicted oxygen carried by the internal recirculation will damage the function of the anoxic zone, a conflict penalty will intervene to adjust the system. The system may tend to choose a suboptimal aeration strategy with moderate aeration volume, but combined with other methods (such as optimizing sludge age or adding synergistic bacteria), rather than simply relying on an aggressive strategy that increases aeration volume, thereby achieving a dual guarantee of energy saving and process stability.
[0087] In scenarios involving conflicts between phosphorus removal agents and sludge settling, the system focuses on the compatibility of chemical phosphorus removal and biological treatment systems. While increasing the dosage of phosphorus removal agents can rapidly reduce the total phosphorus concentration in the effluent through chemical precipitation, excessive dosage generates a large amount of chemical sludge, altering the floc structure and density of activated sludge, thereby deteriorating sludge settling performance, leading to increased sludge levels in the secondary settling tank or even sludge overflow, ultimately causing excessive suspended solids and total phosphorus in the effluent. In this embodiment, when the collaborative decision-maker evaluates a phosphorus removal agent dosage strategy, it simultaneously predicts its impact on the sludge volume index (SVI) or the solids flux in the secondary settling tank. If the prediction results show a significantly increased risk of deterioration in sludge settling performance, the conflict penalty term will reduce the priority of this strategy, guiding the system to find alternative solutions, such as appropriately increasing the sludge discharge to renew the sludge age, adjusting anaerobic phosphorus release conditions to enhance biological phosphorus removal capacity, or precise micro-dosing while ensuring settling performance, rather than simply sacrificing sludge-water separation efficiency for instantaneous total phosphorus compliance.
[0088] Through the three typical process conflict verification mechanisms described above, this embodiment transforms the complex biochemical coupling relationships in wastewater treatment into calculable constraints, enabling the collaborative control method to possess a global judgment capability similar to that of a senior process engineer. This mechanism ensures that when the system copes with fluctuations in influent load, the long-term stability of the overall process will not be sacrificed for short-term optimization of a single indicator. It effectively solves the common problem of neglecting one aspect for another in traditional automatic control, and significantly improves the operational resilience and overall efficiency of wastewater treatment plants under complex operating conditions.
[0089] Example 8
[0090] In some embodiments, to improve the robustness of the collaborative control method in complex industrial environments, the method further includes: verifying the online monitoring data in the multi-source operating data and classifying the data status into four levels: reliable, repairable, low-reliability, and unavailable; for reliable data, it directly participates in the generation of disturbance fingerprints; for repairable data, it participates in the generation of disturbance fingerprints after compensation through soft measurement or historical similar operating conditions; for low-reliability data, it reduces the weight of the corresponding feature components in the disturbance fingerprint and limits the control amplitude of the time-sharing control strategy; for unavailable data, it triggers the strategy safety boundary and limits the single adjustment amplitude of the actuator.
[0091] Specifically, this embodiment addresses the pain point of data distortion caused by sludge adhesion, probe aging, electromagnetic interference, or unstable communication links in online instruments of wastewater treatment plants. It constructs a hierarchical defense mechanism that links data quality with control behavior. For example... Figure 6 The online reliability grading and degradation control state transition diagram shown illustrates that the system does not employ a simple "normal / abnormal" binary judgment. Instead, it uses multi-dimensional rules, including range verification, slope change verification, adjacent measurement point consistency verification, and communication status monitoring, to map the real-time collected online monitoring data into four refined state levels. The diagram illustrates the dynamic transition relationships between each state, as well as the control action amplitude coefficient γ corresponding to different states. This coefficient directly affects the optimal control quantity output by the collaborative decision-maker. This ensures that the instructions ultimately issued to the implementing agency are... satisfy This hierarchical linkage mechanism ensures that when the system detects a decline in data quality, it can automatically and smoothly reduce control permissions, rather than abruptly cutting off automatic control or blindly implementing aggressive strategies based on erroneous data.
[0092] For data deemed reliable, the control amplitude coefficient γ is 1.0, indicating that the data is completely reliable. The system directly uses this data for perturbation fingerprint calculation and subsequent collaborative decision-making, fully leveraging the model's optimization potential. This is the system's standard operating mode when instruments are healthy and operating conditions are stable.
[0093] Data deemed recoverable typically refers to data exhibiting slight drift, short-term communication interruptions, or localized anomalies that can be inferred from other information sources. In such cases, the system does not discard the data directly but instead activates a compensation mechanism. For example, if the online COD instrument reading remains constant, but the influent flow rate, conductivity, and historical COD for the same period all show significant trends, the system determines that the COD data is at risk of stagnation. It then invokes a soft-sensor model based on the correlation between flow load and conductivity, or retrieves typical COD curves from similar historical operating conditions for replacement compensation. The compensated data is considered valid information for perturbation fingerprint generation, but its corresponding control amplitude coefficient γ is adjusted to between 0.6 and 0.8 to moderately tighten the control and prevent over-adjustment due to compensation errors. This approach effectively avoids the problem of the entire feedforward control link failing due to a momentary failure of a single instrument.
[0094] For data deemed low-confidence, this typically refers to data that, while not completely invalid, has significantly reduced confidence, such as frequent numerical jumps, increased contradictions with other related measurement points, or being at the edge of the sensor's measurement range. For this type of data, the system employs a dual suppression strategy of "reduced control amplitude." On one hand, when generating perturbation fingerprints, the weight of corresponding feature components is significantly reduced, even set to zero, making the perturbation fingerprint more reliant on other high-confidence data sources, thereby suppressing the contamination of process semantics by noise signals. On the other hand, the control amplitude coefficient γ is further reduced to the 0.2 to 0.4 range, strictly limiting the adjustment step size of the time-sharing control strategy. For example, when the dissolved oxygen (DO) meter in the aerobic zone shows suspected drift but no confirmed fault, the system will not drastically adjust the blower frequency based on the abnormal DO value, but will only allow fine-tuning within a very small range, while increasing reliance on other indirect indicators (such as nitrate nitrogen trends and aeration current). This strategy allows the system to maintain basic safe operation even with faulty data, neither completely halting control due to minor issues nor recklessly causing process oscillations.
[0095] For data deemed unusable, meaning it has completely lost its reference value (e.g., sensor damage, prolonged communication interruption, or multiple serious failures in verification rules), the system immediately triggers a preset strategy safety boundary and limits the control amplitude coefficient γ to between 0 and 0.1, essentially transferring control to a conservative safety fallback logic. In this state, the actuator's single adjustment amplitude is locked at an extremely low level or remains unchanged. The system then maintains operation based on the previous effective state, fixed safety setpoints, or manually confirmed emergency parameters, and sends high-level alarms and verification prompts to operators. For example, when a critical influent ammonia nitrogen meter completely fails, the system will suspend aeration feedforward regulation based on ammonia nitrogen load and switch to maintaining a baseline aeration rate to ensure nitrification safety until the meter is repaired or manual intervention confirms it. This differs from the rigid approach of traditional control systems that directly report errors and shut down or switch to pure manual mode when data is lost, ensuring the continuity and safety of the wastewater treatment process under extreme conditions of sensor loss.
[0096] Through the aforementioned four-level classification and degradation control mechanism, this embodiment elevates data quality evaluation from a background monitoring function to a core constraint element in front-end control decision-making. It not only solves the industry problem of poor reliability of online instruments hindering the implementation of intelligent control, but more importantly, it establishes a dynamic balance capability between "perceived uncertainty" and "control certainty." Regardless of the health status of the field instruments, the collaborative control method can output safety control commands that match the current information confidence level, achieving a seamless transition from ideal operating conditions to fault conditions, significantly enhancing the system's survivability and practical value in real industrial environments.
[0097] Example 9
[0098] In some embodiments, to ensure the accuracy and adaptability of the coordinated control method in long-term operation, after multiple actuators execute the time-sharing control strategy, the method further includes: collecting the actual state, calculating the deviation between the predicted state and the actual state, the deviation including any combination of time deviation, concentration deviation, response deviation, energy consumption deviation and chemical consumption deviation; the actual state including the actual effluent quality and the actual aeration rate, return flow rate, chemical dosage and sludge discharge rate; based on the deviation, correcting the disturbance fingerprint weight, the predicted state of the process section, the aeration response coefficient, the carbon source response coefficient, the phosphorus removal agent response coefficient and the strategy safety boundary.
[0099] Specifically, this embodiment constructs a refined closed-loop feedback correction mechanism to address the model parameter drift problem caused by seasonal changes, sludge activity evolution, equipment aging, or structural changes in influent components during wastewater treatment. For example... Figure 7The diagram illustrating the prediction bias and parameter feedback correction matrix shows that the system does not use a single global error to update all model parameters indiscriminately. Instead, it establishes a multi-dimensional matrix mapping relationship. The vertical axis of the diagram lists five different types of bias sources, while the horizontal axis lists the key model parameters to be corrected, including perturbation weights, anaerobic / hypoxic / aerobic lag times, various response coefficients, and policy safety boundaries. Specific adjustment instructions are defined within the matrix cells, such as increasing, decreasing, maintaining, or limiting, thus achieving precise matching between bias types and correction targets. This matrix-based correction logic avoids the parameter coupling interference that affects the entire system in traditional adaptive algorithms, ensuring that the model can independently calibrate for mismatches in specific aspects.
[0100] In the specific execution of feedback correction, the system first collects actual effluent water quality indicators and the actual operating volume of each actuator after executing the control strategy and undergoing the corresponding process lag time. Subsequently, the system compares the actual values with the previous predicted values, separating deviation information in different dimensions. For time deviation, i.e., the difference between the actual arrival time of the risk and the predicted time, the system mainly corrects the lag time parameters of each segment in the predicted state of the process section. For example, if the actual ammonia nitrogen peak occurs later than the predicted time, it indicates that the hydraulic transmission or biochemical reaction rate under the current operating conditions is slower than the model expects; the system will correspondingly extend the aerobic lag time parameter to make subsequent predictions more consistent with the actual flow pattern. For concentration deviation, i.e., the difference between the actual pollutant concentration change range and the predicted range, the system mainly corrects the perturbation fingerprint weights. For example, if a certain type of influent perturbation feature is identified but the actual effluent does not show the expected water quality fluctuation, it indicates that the feature has a weak substantial impact on the process under the current operating conditions; the system will reduce its weight in perturbation fingerprint generation to reduce interference from invalid information.
[0101] To address response bias—the difference between the actual removal effect after the actuator's action and the theoretically calculated effect—the system adjusts the corresponding response coefficients. This is the core part of matrix correction. For example, if the actual decrease in total nitrogen is lower than the predicted value after increasing the carbon source dosage, it indicates a decrease in the carbon source utilization efficiency of the current denitrifying bacteria or the existence of other limiting factors. The system will lower the carbon source response coefficient to prevent over-dosing or effluent exceeding standards due to overestimating carbon source efficiency in subsequent control. Conversely, if the total phosphorus removal effect is better than expected after adding the phosphorus removal agent, the phosphorus removal agent response coefficient will be appropriately increased to improve control sensitivity. Regarding energy consumption bias and chemical consumption bias—the difference between actual resource consumption and the optimization target—the system mainly adjusts the strategy safety boundary and conflict penalty weights. For example, if aeration energy consumption continues to exceed the optimization target for a certain period of time and the effluent quality still has a surplus, the system will tighten the upper limit of the aeration safety boundary or increase the penalty weight of the high-energy-consumption strategy, forcing the collaborative decision-maker to search for a more energy-efficient control solution space in the next cycle; if the chemical consumption deviation shows that the phosphorus removal agent consumption is abnormally high and accompanied by a deterioration trend in sludge settling performance, the system will adjust the threshold for the conflict verification between phosphorus removal agent and sludge settling, and strengthen the inhibition of excessive addition.
[0102] Through the aforementioned matrix feedback correction mechanism, this embodiment enables the collaborative control system to possess continuous online learning and adaptive evolution capabilities. Unlike offline training models that are calibrated only once before deployment, the method in this application can continuously absorb actual operating data during daily operation, automatically compensating for model mismatch caused by factors such as changes in ambient temperature, sludge age adjustments, and fluctuations in the proportion of industrial wastewater. This refined correction not only ensures the control accuracy during long-term operation but also effectively extends the model's maintenance-free cycle, reduces the reliance of wastewater treatment plants on on-site debugging by high-level process experts, and significantly enhances the practical value and robustness of the intelligent control system in complex and ever-changing real-world engineering environments.
[0103] Example 10
[0104] In some embodiments, to further ensure the safety and reliability of the collaborative control method in the engineering field, after collaboratively generating time-sharing control strategies for multiple actuators, the method further includes: performing two-layer verification on the time-sharing control strategy, wherein the first layer of software verification determines whether the target value exceeds the process boundary and equipment boundary, and the second layer of programmable logic controller verification determines the remote controllable status and start-stop interval of the equipment; when any two-layer verification fails, the prediction result is retained but automatic control is not executed, and an operation prompt is generated.
[0105] Specifically, this embodiment constructs a defense-in-depth system located between the collaborative decision-making output and the field execution mechanism. The execution mechanism safety boundary described in Embodiment 6 above mainly serves as an optimization constraint condition within the collaborative constraint decision-maker, used to eliminate obviously unreasonable candidate solutions during the strategy generation stage. The dual-layer verification in this embodiment is an independent verification step after strategy generation and before actual instruction issuance, designed to prevent unsafe instructions from being erroneously executed due to instantaneous model anomalies, communication data delays, or sudden changes in the state of field equipment. This design, which separates the logical verification at the algorithm layer from the physical verification at the device layer, constitutes the final security line for the practical application of the intelligent control system.
[0106] In the first layer of software verification, the system conducts a comprehensive compliance review of the generated time-sharing control strategy target values on the algorithm server or edge computing gateway side. Process boundary verification primarily relies on the biochemical reaction limits and environmental emission requirements under current operating conditions. For example, it determines whether the calculated blower frequency setting will cause dissolved oxygen in the aerobic tank to exceed the sludge disintegration threshold, or whether the carbon source dosage will cause the effluent COD to exceed the standard. Equipment boundary verification is based on equipment ledger parameters and real-time operating capabilities. For example, it determines whether the internal return pump speed setting exceeds the inverter's rated upper limit, or whether the sludge discharge valve opening adjustment rate exceeds the maximum allowable rate of change of the mechanical structure. This layer of verification only passes when the target value simultaneously meets both process safety and equipment capability constraints. This process ensures that the instructions output from the algorithm level are theoretically and design-wise safe and feasible.
[0107] In the second-level programmable logic controller (PLC) verification, after receiving control commands from the upper-level system, the field PLC or dedicated control cabinet performs a secondary verification based on the real-time status of the underlying hardware. This layer of verification focuses on the actual executability of the equipment, including determining whether the target equipment is in remote control mode rather than local manual maintenance mode, whether there are fault alarm signals such as overheating or overload, and whether the minimum start-stop interval required for motor protection is met. For example, if the collaborative decision-maker suggests adjusting the frequency of a high-power blower twice within one minute, but the motor protection logic set internally by the PLC requires an interval of at least three minutes between the two frequency conversion operations, then the second-level verification will determine that the command is invalid. This verification mechanism based on the underlying hardware logic effectively shields the upper-level software from potential timing errors or lag in transient perception of field equipment, ensuring that the control commands are strictly matched with the actual capacity of the physical equipment.
[0108] When either layer of the aforementioned two-layer verification fails, the system adopts a strategy of retaining the prediction result but not executing automatic control, and generating operational prompts. This mechanism embodies the core design concept of this application in human-machine collaboration. Retaining the prediction result means that the system does not negate its assessment of operational risks due to execution failure; the relevant disturbance fingerprint recognition, hysteresis propagation prediction, and collaborative optimization suggestions are still fully recorded and displayed to the operators as valuable trend warnings and decision-making references. Not executing automatic control means that when it is confirmed that the execution conditions are not met or there are safety hazards, the system proactively relinquishes the authority to intervene in the machine, avoiding production accidents caused by forcibly executing flawed instructions. At the same time, the system will generate differentiated operational prompts based on the specific reasons for the verification failure. For example, when the first-layer verification fails, it will prompt details of process parameter out-of-bounds errors and suggested checkpoints; when the second-layer verification fails, it will prompt the specific equipment number, fault type, and remote status anomaly information. This approach allows operators to safely take over decision-making when the automatic control system is uncertain or limited. It avoids the rigidity of traditional fully automated systems where a single failure can cause a complete shutdown, and eliminates operators' concerns about the black box operation of intelligent algorithms. This significantly improves the credibility and practical value of collaborative control methods in complex industrial environments.
[0109] The above describes a method for coordinated control of a wastewater treatment process provided by the embodiments of this application. The following will describe the apparatus for implementing the above-described method for coordinated control of a wastewater treatment process.
[0110] Please see Figure 8 , Figure 8 This is a schematic diagram of a coordinated control device for a wastewater treatment process, provided as an embodiment of this application. Figure 8 As shown, the coordinated control device for the wastewater treatment process includes: The multi-source operation data acquisition unit 11 is used to acquire multi-source operation data during the wastewater treatment process. The multi-source operation data includes influent side data, process data, external disturbance data, and actuator status data. The disturbance fingerprint generation unit 12 is used to generate a disturbance fingerprint based on the multi-source operating data. The disturbance fingerprint characterizes the direction and intensity of the impact of water ingress on the subsequent biochemical treatment process. Prediction unit 13 is used to predict the process segment hysteresis propagation based on the perturbation fingerprint, and obtain the predicted state of the process segment for subsequent biochemical treatment. The strategy generation unit 14 is used to collaboratively generate a time-sharing control strategy for multiple actuators based on the predicted state of the process section, under the constraints of effluent water quality and process conflict constraints that take into account the coupling effect between each actuator.
[0111] Each unit in the aforementioned coordinated control device for wastewater treatment processes can be implemented entirely or partially through software, hardware, or a combination thereof. These units can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each unit.
[0112] This application also provides an electronic device in its embodiments. (See reference...) Figure 9 The diagram illustrates a structural schematic suitable for implementing the electronic device in the embodiments of this application. The electronic device in the embodiments of this application may include, but is not limited to, fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 9 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.
[0113] like Figure 9 As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. When the electronic device is powered on, the RAM 603 also stores various programs and data required for the operation of the electronic device. The processing unit 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.
[0114] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 608 including, for example, memory cards, hard drives, etc.; and communication devices 609. Communication device 609 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 9 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.
[0115] This application also provides a computer program product including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the wastewater treatment process coordinated control methods provided in this application.
[0116] This application also provides a computer-readable storage medium that carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the wastewater treatment process coordination and control methods provided in this application.
[0117] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.
[0118] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.
[0119] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.
[0120] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).
Claims
1. A method for coordinated control of a wastewater treatment process, characterized in that, include: Acquire multi-source operational data during the wastewater treatment process, including influent side data, process data, external disturbance data, and actuator status data; Based on the multi-source operational data, a perturbation fingerprint is generated, which characterizes the direction and intensity of the impact of influent shock on the subsequent biochemical treatment process. Based on the perturbation fingerprint, process segment hysteresis propagation prediction is performed to obtain the predicted state of the subsequent biochemical process segment. Based on the predicted state of the process section, and under the constraints of effluent water quality and process conflict constraints that take into account the coupling effects between various actuators, a time-sharing control strategy for multiple actuators is collaboratively generated.
2. The method according to claim 1, characterized in that, The generation of the perturbation fingerprint based on the multi-source operational data includes: Multidimensional feature components are calculated based on the multi-source operational data. These multidimensional feature components include any combination of organic load intensity, nitrogen load intensity, phosphorus load intensity, carbon-nitrogen ratio deviation, rainwater dilution, industrial shock probability, low-temperature nitrification pressure, load rise rate, duration, and data reliability. Obtain the credibility score of the data source corresponding to each of the aforementioned feature components; Based on the credibility score, the feature components are weighted and fused to generate an influent load disturbance feature vector. The disturbance type is labeled based on the preset category labels to obtain the disturbance fingerprint. The disturbance type labels include residential peak type, rainwater dilution type, industrial shock type, low temperature nitrification pressure type, carbon source deficiency type and composite disturbance type.
3. The method according to claim 1, characterized in that, The step of predicting the process segment hysteresis propagation based on the perturbation fingerprint to obtain the predicted state of the subsequent biochemical treatment process segment includes: Obtain current tank volume, hydraulic retention time, reflux ratio, sludge concentration, sludge age, water temperature, and historical response curves; Based on the disturbance fingerprint, the current tank volume, the hydraulic retention time, the reflux ratio, the sludge concentration, the sludge age, the water temperature, and the historical response curve, process segment hysteresis propagation prediction is performed to obtain the predicted state of the subsequent biochemical treatment process segment.
4. The method according to claim 3, characterized in that, The subsequent biochemical treatment process includes an anaerobic zone, an anoxic zone, an aerobic zone, and a mud-water separation unit. The predicted state of the process segment includes: Anaerobic lag time, anoxic lag time, aerobic lag time, and mud-water separation lag time.
5. The method according to claim 1, characterized in that, The collaborative generation of time-sharing control strategies for multiple actuators includes: Taking the effluent quality meeting the standard as a hard constraint, and taking aeration energy consumption, reagent consumption and equipment operation frequency as optimization objectives, the time-sharing control strategy for multiple actuators is generated by combining the safety boundary of the actuator and process conflict verification for collaborative constraint decision-making. The effluent water quality is used to indicate any one or more of the following: ammonia nitrogen, total nitrogen, total phosphorus, chemical oxygen demand (COD), and suspended solids (SS). The multiple execution structures include an aeration system, a return system, a chemical system, and a sludge removal system; The time-sharing control strategy for the multiple actuators includes setting the blower frequency, setting the aeration valve opening degree, setting the internal return pump speed, setting the external return pump speed, setting the carbon source dosing pump, setting the phosphorus removal agent dosing pump, setting the sludge discharge valve opening degree, and setting the sludge discharge cycle.
6. The method according to claim 5, characterized in that, The safety boundaries of the actuator include adjustment boundaries, action intervals, and backoff conditions.
7. The method according to claim 5, characterized in that, The process conflict verification includes at least one of the following: When a candidate control strategy reduces the risk of a certain pollutant but introduces process side effects, the priority of the candidate control strategy is reduced through a conflict penalty term. When increasing the internal recirculation flow leads to an increase in dissolved oxygen or an increased risk of insufficient carbon source in the anoxic zone, a conflict penalty between internal recirculation and denitrification is triggered. When increasing the aeration rate leads to an increased risk of internal backflow carrying oxygen or when energy consumption exceeds a preset threshold, a conflict penalty is triggered between aeration and energy consumption and backflow carrying oxygen. When increasing the dosage of phosphorus removal agent leads to an increased risk of deterioration in sludge settling performance, a conflict penalty between phosphorus removal agent and sludge settling is triggered.
8. The method according to claim 1, characterized in that, The method further includes: The online monitoring data in the multi-source operation data is verified, and the data status is divided into four levels: reliable, repairable, low reliability, and unavailable. For trusted data, it directly participates in the generation of the perturbation fingerprint; For recoverable data, it is compensated by soft measurement or historical similar operating conditions before being used in the generation of the perturbation fingerprint; For low-confidence data, reduce the weight of the corresponding feature components in the perturbation fingerprint and limit the control amplitude of the time-division control strategy; For unavailable data, trigger the policy safety boundary and limit the single adjustment range of the actuator.
9. The method according to claim 1, characterized in that, After the multiple actuators execute the time-sharing control strategy, the method further includes: The actual conditions are collected, and the deviation between the predicted conditions and the actual conditions is calculated. The deviation includes any combination of time deviation, concentration deviation, response deviation, energy consumption deviation, and chemical consumption deviation. The actual conditions include the actual effluent quality, actual aeration rate, return flow rate, chemical dosage, and sludge discharge rate. Based on the aforementioned deviation, the perturbation fingerprint weight, the predicted state of the process segment, the aeration response coefficient, the carbon source response coefficient, the phosphorus removal agent response coefficient, and the strategy safety boundary are corrected.
10. The method according to claim 1, characterized in that, After the collaborative generation of time-sharing control strategies for multiple actuators, the method further includes: The time-sharing control strategy is verified in two layers. The first layer is software verification to determine whether the target value exceeds the process boundary and equipment boundary. The second layer is programmable logic controller verification to determine the remote controllable status of the equipment and the start-stop interval. If any of the two-layer verifications fails, the prediction result is retained but automatic control is not executed, and a running prompt is generated.
11. A computer program product, characterized in that, It includes computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the coordinated control method for the wastewater treatment process as described in any one of claims 1 to 10.
12. An electronic device, characterized in that, It includes at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program to enable the electronic device to implement the coordinated control method of the wastewater treatment process as described in any one of claims 1 to 10.
13. A computer-readable storage medium, characterized in that, The storage medium carries one or more computer programs, which, when executed by an electronic device, enable the electronic device to implement the coordinated control method for the wastewater treatment process as described in any one of claims 1 to 10.
