A sludge age self-adaptive regulation method and system based on dissolved oxygen trend identification

Abstract

This invention discloses a sludge age adaptive control method and system based on dissolved oxygen trend identification. Addressing the problem of delayed shock response caused by difficulties in online COD measurement, this method, under stable influent and aeration conditions, filters and extracts features from the online DO signal using a sliding window to obtain the regression slope characterizing DO changes over time. Combined with the ORP persistence criterion, it reliably identifies COD surge shocks. A fusion model of mechanism and machine learning is used to determine the identification threshold and target sludge age (SRT). After a shock is triggered, the sludge discharge rate is rapidly adjusted to match the treatment capacity, and aeration is used to maintain DO stability. The system integrates modules for online monitoring, DO identification interval determination, feature calculation, load identification, SRT decision-making, and sludge discharge execution. This invention eliminates the need for online COD instruments, achieving rapid shock identification and closed-loop control, significantly improving the process's shock resistance and operational stability.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment process control and intelligent regulation technology. Specifically, in the field of wastewater treatment process control, it relates to a method for online identification of organic load shocks caused by COD increases during the operation of a high-load activated sludge process (HRAS) when chemical oxygen demand (COD) is difficult to measure online. The method and system achieve adaptive control by adjusting the sludge discharge rate to achieve linkage regulation of sludge age (SRT). Background Technology

[0002] High-load activated sludge process (HRAS) is a type of activated sludge process that operates under high organic loads, typically achieving rapid treatment of influent organic matter with a short hydraulic retention time (HRT). The stable operation of an HRAS largely depends on its ability to maintain sufficient treatment capacity during load fluctuations. In actual operation, chemical oxygen demand (COD) is not a stable, constant indicator; periodic increases or sudden spikes are common. Once COD rises rapidly, the oxygen demand of heterotrophic microorganisms increases accordingly, often leading to a significant drop in dissolved oxygen (DO), resulting in effluent fluctuations and operational risks.

[0003] Currently, key operating parameters commonly used in HRAS include HRT, DO, and SRT. SRT, by determining the biomass concentration and overall metabolic capacity within the system, directly affects the reactor's capacity to handle organic loads and its resistance to shock loads. In engineering practice, it's common to fix the SRT or perform sludge removal based on periodic experience. This method can maintain basic operation when the load is relatively stable. However, when a COD surge occurs, a fixed SRT often fails to synchronously match the increased treatment demand in the initial stages of the surge, easily leading to a mismatch where "the load has increased while the system's treatment capacity remains at the original level." Typical manifestations include a rapid drop in DO and increased effluent fluctuations. While increasing aeration or adjusting the DO setpoint can superficially alleviate the DO decline, it is essentially an oxygen supply-side compensation, failing to quickly match load changes at the biomass level, and may lead to increased energy consumption or control oscillations. Therefore, from a control perspective, the key lies in timely identification of COD surges and translating the identification results into coordinated adjustments to the SRT.

[0004] However, the timeliness of SRT control depends on the ability to obtain COD change information quickly and continuously. Under actual HRAS operating conditions, online continuous and reliable measurement of COD is usually difficult to achieve. In engineering practice, manual sampling and offline detection are still the main methods, with low detection frequency and large lag, which are difficult to support the need for rapid and timely adjustment when shocks occur.

[0005] On the other hand, dissolved oxygen (DO) is a commonly configured and continuously acquired online signal in wastewater treatment systems. Mechanistically, an increase in COD raises the oxygen consumption rate, and DO will generate a corresponding dynamic response. Therefore, using DO trends as proxy information for organic load shocks is feasible in engineering. However, it should be noted that DO is not only affected by COD, but also by factors such as aeration oxygen input, mixing state, hydraulic disturbances, and sensor noise. Judging solely based on instantaneous DO values ​​or simple thresholds can easily lead to misinterpreting DO fluctuations caused by changes in oxygen supply or disturbances as COD surges, resulting in false triggers and frequent adjustments, which weakens system stability. Furthermore, even if shock signals can be identified, current technology lacks an operable method for determining parameters that ensures consistency and interpretability of the identifiable operating range of DO, shock identification thresholds, and decision rules from shock intensity and target sludge age (SRT) to sludge discharge rate under different operating conditions, making it difficult to form a stable closed-loop control system.

[0006] In existing research, to achieve stable operation and optimized control of HRAS processes, one approach uses oxygen consumption-related indicators such as oxygen consumption rate (OUR) or specific oxygen consumption (SOC) as parameters for operational evaluation or oxygen supply management. The control objective primarily targets optimizing aeration intensity or reducing operational energy consumption. It should be noted that SOC is often used in existing studies as an indicator of oxygen consumption per unit change in organic matter. It typically needs to be correlated and normalized with organic matter indicators such as COD removal or changes in soluble COD. Some published studies have proposed using SOC related to soluble COD as an effective parameter for optimizing oxygen supply. Therefore, when using SOC as a control parameter, it is often necessary to obtain information on COD removal or changes in soluble COD. In engineering, this information usually relies on sampling and detection, or on soft sensor models that provide sufficient accuracy and long-term stability. Another approach is to achieve SRT control based on solids indicators such as MLSS to reduce COD removal fluctuations and improve operational stability. The control signals of this type of method focus on maintaining the concentration of solids or biomass. When faced with a sudden increase in influent COD, its control effect is easily affected by the update frequency of solid indicators, the representativeness of the measurement, and the inherent hysteresis of the system, making it difficult to achieve real-time identification and rapid response to the source of the shock.

[0007] In summary, existing technologies, under the operating scenario of the HRAS process itself, still lack a closed-loop method and system that can reliably characterize and promptly identify COD shocks by utilizing DO change trends under conditions where COD is difficult to measure online and there are COD rise shocks, and use the identification results to quickly generate the target sludge age SRT with the support of model-based rules and achieve adaptive control through sludge discharge, so as to reduce the lag in identification and adjustment and improve the operational stability and shock resistance of HRAS. Summary of the Invention

[0008] To overcome the aforementioned deficiencies of existing technologies, this invention aims to provide a sludge age adaptive control method and system based on dissolved oxygen trend identification for high-load activated sludge processes. It should be noted that this invention differs from existing aeration optimization control routes centered on SOC or OUR, and SRT maintenance control routes centered on solids indices such as MLSS. This invention does not require obtaining COD removal or soluble COD changes as a necessary prerequisite. Instead, it constructs a DO identifiable interval under boundary conditions of stable influent flow and controlled aeration oxygen supply. Within this interval, DO trend characteristics are extracted and combined with persistence criteria to trigger and identify organic load shocks caused by COD increases. The shock identification results are directly used to generate the target sludge age SRT and converted into sludge discharge volume for control, thus forming a closed-loop adaptive control system from identification to decision-making and final execution of control measures. The core objective of this method is to utilize the dynamic trend characteristics of DO as a characterizing indicator of organic load changes under operating conditions where continuous and reliable online measurement of chemical oxygen demand (COD) is difficult. This enables online identification of organic load shocks caused by rising COD, and upon triggering the identification, the sludge discharge rate is adjusted accordingly to achieve adaptive control of SRT (Self-Recovery Time), thereby reducing identification and adjustment lag and improving the operational stability and shock resistance of HRAS (High-Resistance System) under high load and fluctuating conditions. To achieve the above objective, the present invention employs the following technical solution:

[0009] An adaptive sludge age control method based on dissolved oxygen trend identification, the method comprising the following steps:

[0010] S1. Data Acquisition and Preprocessing: Real-time acquisition of operating data of the HRAS reactor, including at least dissolved oxygen (DO) signal, and optionally acquisition of oxidation-reduction potential, influent flow rate signal and aeration oxygen supply input signal, and preprocessing of the operating data to obtain a standardized signal sequence;

[0011] S2. Determination of DO Identifiable Range: Under the condition that the influent flow rate remains stable, the DO identifiable range is determined, wherein within the DO identifiable range, the aeration oxygen supply input remains stable, and the DO concentration is within the preset working range and continues to be no less than the preset time window.

[0012] S3. Trend feature extraction: Within the identifiable range of DO, feature extraction is performed on the DO signal, and trend feature parameters characterizing the change of DO over time are calculated;

[0013] S4. Organic load shock identification: Based on the trend characteristic parameters and combined with the persistence criterion of the oxidation-reduction potential (ORP) signal, determine whether an organic load shock event caused by an increase in chemical oxygen demand (COD) has occurred;

[0014] S5. SRT Decision and Sludge Discharge Execution: After identifying the organic load shock event, determine the target sludge age (SRT), convert the target sludge age SRT into sludge discharge volume Qw, and control the sludge discharge execution mechanism to output the sludge discharge volume Qw;

[0015] The DO identifiable interval parameter, the impact identification threshold of the organic load shock event, the relationship between the trend characteristic parameter and the load shock intensity, and the target sludge age SRT are at least determined by the control-oriented mechanism model and participate in online control. The DO identifiable interval parameter includes: the allowable fluctuation range for influent flow stability determination, the set value and allowable fluctuation range of aeration oxygen supply input, the preset working range of DO concentration, and the preset time window for determining whether DO is continuously within the working range.

[0016] It should be noted that, in this invention, the DO trend feature refers to the set of feature parameters obtained by extracting the trend of the online DO signal within a preset DO identifiable range using a sliding window. This set includes at least the linear regression slope dDO / dt of the relationship between DO and time within the window. The DO identifiable range is used to ensure that DO changes primarily reflect changes in oxygen consumption rather than oxygen supply disturbances. The determination is based on the following conditions: the influent flow rate Q is within the allowable fluctuation range; the aeration oxygen supply input remains at the set value or its variation is limited; and the DO is within the preset operating range and remains for at least a preset time window. The DO trend feature is used to characterize the dynamic response of DO caused by changes in oxygen consumption under controlled oxygen supply boundaries. It is also combined with the persistence criterion of the oxidation-reduction potential (ORP) signal to verify the impact triggering caused by COD, thereby reducing the risk of misjudgment and frequent adjustments.

[0017] Furthermore, the method of the present invention includes periodic updates: the above-mentioned process of "DO change feature extraction - impact identification - SRT decision - sludge discharge execution" is repeatedly executed with a fixed control cycle to achieve dynamic adaptive control.

[0018] Furthermore, in step S1, the running data undergoes preprocessing such as filtering and time alignment to obtain a standardized signal sequence that can be used for feature calculation.

[0019] Further, in step S2, the DO identifiable range is determined as follows: under the operating condition that the influent flow rate remains stable, that is, the hydraulic retention time (HRT) remains unchanged or is within the allowable fluctuation range, the DO identifiable range is determined. Within the DO identifiable range, the aeration oxygen supply input is maintained at the preset set value or within the allowable fluctuation range, and the DO is kept within the preset working range and continuously not less than the preset time window, so as to reduce the interference of oxygen supply disturbance and short-term noise on the DO signal and ensure the identifiability of trend characteristics.

[0020] Furthermore, the trend characteristic parameters include, but are not limited to, the regression slope dDO / dt of DO changing over time, wherein dDO / dt is obtained by performing linear regression fitting on the DO data within the sliding window.

[0021] Within the identifiable range of DO, a sliding window feature extraction is performed on the DO signal, and trend feature parameters characterizing the change of DO over time are calculated. These trend feature parameters include, but are not limited to, the regression slope dDO / dt of DO change obtained by linear regression fitting.

[0022] Furthermore, the determination of the DO identifiable range must simultaneously meet the following conditions: the deviation of the aeration oxygen supply input signal from the set value does not exceed a preset threshold; the DO is within a preset working range and the duration is not less than a preset time window; and the influent flow rate is within a preset allowable fluctuation range.

[0023] Further, in step S4, the persistence criterion for the redox potential (ORP) signal includes: when the trend characteristic parameter meets a preset impact triggering condition, and the ORP signal meets a preset unidirectional change condition and lasts for no less than a preset holding time, then it is determined that an organic load impact event caused by an increase in COD has occurred. At this time, the impact identification signal and the characterization of the impact intensity are output.

[0024] In step S5, after identifying an organic load shock event, the target sludge age SRT is determined based on the control-oriented mechanism model, and the target sludge age SRT is converted into sludge discharge volume Qw. The sludge discharge actuator is controlled to output the sludge discharge volume Qw ​​to adjust the biomass concentration of the reactor, so that the treatment capacity is synchronously matched with the organic load shock. Preferably, aeration is used as an auxiliary adjustment means to maintain DO within the preset working range to ensure the reliability of identification and the stability of operation.

[0025] Furthermore, in step S5, the determination of the target sludge age SRT is based on a control-oriented mechanism model, which includes at least an oxygen transfer term determined based on aeration oxygen input and an oxygen consumption term determined based on DO change trends.

[0026] Furthermore, the control-guided mechanism model is configured to perform the following function: based on the equivalent oxygen transfer coefficient kLa and the saturated dissolved oxygen concentration SO determined by the aeration oxygen supply input. * Oxygen transfer relationships are constructed, and the oxygen consumption rate (OUR) or the change in oxygen consumption rate is calculated based on the DO change trend to determine the impact identification threshold and SRT decision.

[0027] Based on the aeration oxygen supply input, the equivalent oxygen transfer coefficient kLa and the saturated dissolved oxygen concentration SO* are determined to establish the oxygen transfer relationship. The oxygen consumption rate (OUR) or the change in oxygen consumption rate is calculated based on the dynamic trend of DO to support impact identification and SRT decision-making.

[0028] Furthermore, based on the control guidance mechanism model, the following operations can be performed: mapping the trend characteristic parameters to a load index characterizing the impact intensity of COD increase, wherein the load index includes at least: the integral of the oxygen consumption rate OUR within a preset time window, and the impact intensity index composed of the change in oxygen consumption rate and the drop in DO.

[0029] The method of the present invention can be further incorporated into a machine learning model to form a collaborative decision-making and adaptive update mechanism with the control-oriented mechanism model.

[0030] The method of the present invention includes establishing and calling a machine learning model, wherein the input of the machine learning model includes at least the trend feature parameters, the aeration oxygen supply input signal and the influent flow rate, the output is the load index, the adjusted target sludge age SRT, and training, correction and / or online updating using historical operating data and manual COD sampling data.

[0031] Furthermore, the generation rule of the target sludge age SRT is determined at least by the control-oriented mechanism model and can be corrected or updated by the machine learning model. In the control process of generating the sludge discharge volume Qw ​​based on the target sludge age SRT, preset operating constraints must be followed. The operating constraints include at least the DO working range constraint, the SRT constraint, and the limit constraint of the sludge discharge volume Qw.

[0032] Furthermore, the machine learning model is used to train, correct, or update the DO identifiable interval parameters, impact identification threshold, mapping relationship, and target sludge age (SRT) generation rules based on historical operating data and manual COD sampling data, thereby improving the robustness of identification and decision-making under different operating conditions. The machine learning model preferably employs at least one of random forest regression, gradient boosting tree, or long short-term memory network.

[0033] The present invention also provides a sludge age control system for implementing the aforementioned method in the HRAS process, comprising:

[0034] a. Online monitoring module, used to collect DO data in real time, and optionally collect ORP, influent flow rate and aeration oxygen supply input signals;

[0035] b. DO Recognizable Range Determination Module, used to construct or determine the DO recognizable range under the condition that the influent flow rate is stable and the aeration oxygen supply input is maintained at the set value or within the allowable fluctuation range;

[0036] c. Feature calculation module, used to filter the DO signal and perform sliding window feature extraction, outputting trend feature parameters;

[0037] d. Load identification module, used to identify influent load shock events caused by COD increase based on trend characteristic parameters and the persistence criterion of oxidation-reduction potential (ORP) signal;

[0038] e. SRT decision module, used to calculate the target sludge age SRT according to the target sludge age SRT generation rule after the influent load shock event is identified;

[0039] f. Sludge discharge execution module, used to convert the target sludge age SRT into sludge discharge volume Qw ​​and control the sludge discharge execution mechanism to output Qw;

[0040] g. Aeration auxiliary adjustment module, used to maintain DO within the preset working range to ensure identification reliability and operational stability;

[0041] h. Model and parameter generation module, used to build and call control-oriented mechanism model and machine learning model to determine DO identifiable interval parameters, identify threshold and target sludge age SRT and participate in online control.

[0042] Furthermore, the model and parameter generation module includes a mechanism model submodule, which at least includes an oxygen transfer term and an oxygen consumption term, used to calculate the oxygen consumption rate OUR or its increment ΔOUR and generate the identification threshold and load index accordingly;

[0043] Furthermore, the model and parameter generation module includes a machine learning submodule, which trains, corrects, or updates the identification threshold, mapping relationship, and target sludge age (SRT) generation based on historical operating data and manual COD sampling data.

[0044] Beneficial effects of the present invention

[0045] Compared with the prior art, the present invention has the following significant advantages:

[0046] The control method of this invention is timely and proactive: without relying on online COD measurement, it realizes online identification of COD rise shocks through DO trend characteristics, transforming SRT control from "lagging intervention" to "rapid response after identification triggering", effectively reducing control lag.

[0047] The control method of the present invention is feasible and stable: by constraining the DO identification range and continuous criteria such as ORP, the influence of aeration disturbance and noise on the DO signal is reduced, misjudgment and frequent adjustment are reduced, and the stability of closed-loop control is improved.

[0048] The control method of the present invention is interpretable and transferable: a control-guided mechanism model is introduced to determine the threshold, interval and SRT rule, so that the key parameters have clear physical meaning and are easy to adjust and reuse under different operating conditions.

[0049] The control method of the present invention is robust: by combining a machine learning model to train, correct or update the threshold and rules, the adaptability under long-term operation and changing operating conditions can be improved.

[0050] The control method of this invention reduces reliance on manual experience: it transforms traditional experience-based sludge discharge and post-processing into automated decision-making and execution based on online signals and model rules, thereby improving operational management efficiency. Attached Figure Description

[0051] Figure 1 System overall block diagram and closed-loop control flow diagram;

[0052] Figure 2 System impact determination flowchart;

[0053] Figure 3 Schematic diagram of the control process for target sludge age (SRT), sludge discharge execution, and aeration-assisted maintenance;

[0054] Figure 4 A schematic diagram of a continuous flow HRAS system. Detailed Implementation

[0055] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0056] In the following description, specific details, such as particular internal procedures and techniques, are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will appreciate that the invention may be practiced in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, and methods have been omitted so as not to obscure the description of the invention with unnecessary detail.

[0057] This invention provides a method and system for online identification of organic load and adaptive control of sludge age (SRT) based on dissolved oxygen (DO) trend characteristics. This method addresses operating conditions where COD is difficult to measure continuously and reliably online and where COD surges occur. Under stable influent flow conditions, it utilizes the dynamic trend characteristics of DO as a representation of organic load changes to achieve online identification of organic load surges caused by COD increases. Upon triggering the identification, it adjusts the sludge discharge rate to regulate SRT, thereby reducing identification and adjustment lag and improving the operational stability and shock resistance of the HRAS under high load and fluctuating conditions.

[0058] like Figure 1As shown, during implementation, the system first collects data through an online monitoring module, acquiring DO signals in the HRAS reactor in real time at a fixed sampling frequency. Optional signals include oxidation-reduction potential (ORP), influent flow rate (Q), and aeration oxygen input. The acquired online signals are processed by a data preprocessing module for filtering and timestamp alignment, forming a standardized sequence suitable for online feature extraction. Subsequently, the system enters the DO identifiable interval determination process: to avoid misjudging DO fluctuations caused by oxygen supply disturbances, mixing fluctuations, or sensor noise as COD surges, this invention first constrains the oxygen supply side in its control strategy, ensuring stable aeration, and then interprets the DO trend under controlled conditions. Specifically, the DO identifiable interval determination module satisfies at least the following conditions when determining a DO identifiable interval: stable influent flow rate (within the allowable fluctuation range), stable aeration oxygen input (maintaining the set value or within the allowable fluctuation range, with limited deviation or rate of change), and DO concentration within a preset working range and continuously not less than a preset time window. Through these constraints, changes in DO within the window better reflect changes on the oxygen consumption side, thereby ensuring the identifiability of trend features and the reliability of shock identification. Once the system enters the DO identifiable range, the feature calculation module performs sliding window feature extraction on the DO signal. Linear regression is used to fit the change in DO over time within the judgment period, obtaining the regression slope as the trend feature parameter dDO / dt. Optional auxiliary features such as the DO drop amplitude ΔDO and the fluctuation intensity within the time period can also be calculated. Subsequently, the load identification module performs organic load impact identification (i.e., impact determination) based on the trend feature parameter and the persistence criterion of the oxidation-reduction potential (ORP) signal: when dDO / dt meets the impact triggering condition, and DO shows a significant trend within the time period (e.g., a continuous decrease exceeding a preset threshold), it enters the candidate impact state. Simultaneously, it detects whether the ORP meets the preset same-direction change condition and remains at least at a preset holding time. Only when the persistence criterion of the ORP signal is met is an organic load impact event caused by a rise in COD determined, and the impact identification result is output, such as the impact identification signal and the impact intensity characterization quantity. This combined strategy of trend change triggering and persistence verification prevents misjudgments caused by DO fluctuations not due to impacts, reducing the risk of false triggering and frequent adjustments caused by short-term disturbances.

[0059] After identifying an organic load shock event, the system enters the SRT adaptive decision-making and sludge discharge execution phase, such as... Figure 1As shown, the SRT decision module calls the model and parameter generation module to generate the target sludge age SRT. The sludge discharge execution module converts the target sludge age SRT into a sludge discharge volume Qw ​​and sends it to the sludge discharge execution mechanism for execution, thereby adjusting the biomass inventory in the reactor to ensure that the treatment capacity matches the impact of rising COD. For scenarios with rising COD, this invention preferably increases the target sludge age SRT within the operational constraints to enhance the system's capacity to bear and buffer organic loads. The sludge discharge execution module discharges the corresponding sludge, directly adjusting the sludge concentration and sludge age SRT in the reactor to achieve an active response to organic load impacts. At the same time, real-time data such as dissolved oxygen (DO) and sludge status updated in the reactor after sludge discharge are re-collected through the "online monitoring module". Simultaneously, the aeration auxiliary adjustment module uses aeration as an auxiliary adjustment means to ensure the reliability of identification and operational stability, enabling the system to quickly restore DO to the identifiable range and enter the next control cycle (dissolved oxygen recovery, re-entering data acquisition), creating conditions for subsequent continuous online identification and control. The above control process operates cyclically with a fixed control cycle, forming a closed-loop regulation.

[0060] like Figure 1As shown, the key parameters, rules, and model relationships in the above decision-making and control process are all supported by the model and parameter generation module. This module integrates machine learning models and control-oriented mechanism models, providing data-driven and theory-driven decision-making basis for the entire process. The machine learning model, based on historical operating data and manual COD sampling data, trains and corrects the DO identifiable interval parameters, shock identification threshold, the mapping relationship between DO trend characteristics and shock intensity, and the target sludge age SRT generation rules. The control-oriented mechanism model is used to determine the mapping relationship between DO identifiable interval parameters, shock identification threshold, trend characteristic parameters, and shock intensity, as well as the target sludge age SRT generation rules. The determination of the above rules and parameter intervals is based on the initial parameters calculated by the control-oriented mechanism model under baseline operating conditions, and periodically corrected and updated based on historical operating data and manual COD sampling data. Furthermore, to ensure the interpretability and transferability of the parameters for determining the DO change trend characteristics, shock intensity, and target sludge age SRT, this invention establishes a simplified control-oriented mechanism model based on the ASM1 carbon-oxygen subsystem. This model is used to determine the mapping relationship between DO identifiable interval parameters, shock identification threshold, trend characteristic parameters, and shock intensity, as well as the generation rules for the target sludge age SRT. The determination of these rules and parameter intervals is based on initial parameters calculated by the mechanism model under baseline operating conditions, and periodically corrected and updated based on historical operating data and manual COD sampling data. The mechanism model is not limited to a fixed form, but should at least include: 1. Oxygen transfer relationship on the oxygen supply side; 2. Characterization relationship of oxygen consumption rate (OUR); 3. Conversion relationship between SRT and sludge discharge volume Qw. Under stable influent flow rate Q, and with aeration oxygen input remaining constant or subject to limited fluctuation within each identification interval (i.e., the oxygen mass transfer coefficient (kLa) remaining constant or subject to limited change), the model supports the determination of DO identifiable interval parameters, shock identification threshold, and the generation rules for the target sludge age SRT. After completing the phased SRT adjustment, the DO is adjusted to the preset working range by aeration to enter the next round of identification and control.

[0061] The mechanistic model states at least include a soluble substrate. Particulate substrates heterotrophic bacteria concentration With dissolved oxygen concentration ,Right now:

[0062]

[0063] Inlet and outlet water meet ,in This refers to the overflow water flow rate. The sludge discharge flow rate is given. Considering the entrainment of particulate matter with the overflow effluent, a solids escape coefficient is introduced. The particulate COD carried away by the effluent is characterized as follows:

[0064]

[0065] Based on this, sludge age SRT and sludge discharge volume The following conditions must be met:

[0066]

[0067] Therefore, the sludge discharge volume corresponding to the target sludge age SRT can be converted as follows:

[0068]

[0069] Where V is the effective volume of the HRAS reactor. When it approaches zero, in the above formula This corresponds to the situation where solids are mainly carried out with the sludge discharge. The above conversion formula is directly used to control the execution layer, realizing the conversion from the target sludge age (SRT) to the sludge discharge rate. Executable conversion.

[0070] To characterize the relationship between DO trends and changes in oxygen consumption, the mechanistic model includes at least an oxygen transfer term and an oxygen consumption term. The oxygen transfer term is preferably represented as:

[0071]

[0072] in This represents the saturated dissolved oxygen concentration. This is the equivalent oxygen mass transfer coefficient. Based on this, the dissolved oxygen dynamics can be expressed in the following form:

[0073]

[0074] in , This refers to the oxygen consumption rate. Preferably, It consists at least of the oxygen consumption for the aerobic growth of heterotrophic bacteria and the oxygen consumption for maintenance, that is:

[0075]

[0076]

[0077] in For heterotrophic bacteria yield, These are kinetic parameters. As can be seen from the dissolved oxygen equation above, within the identifiable range of DO, when... When the influent flow rate Q is approximately constant or subject to limited variation, the trend characteristics of DO can characterize the degree of imbalance between the oxygen mass transfer term and the oxygen consumption term. This can be used to construct shock identification thresholds and shock intensity characterization quantities, and further used to generate decision rules for the target sludge age (SRT). In other words, this invention transforms DO from a perturbed online signal into a controllable organic load proxy measurement signal by first constraining the oxygen supply input to form an identifiable interval, and then identifying the DO change trend within that interval.

[0078] To further improve the robustness of identification and decision-making under different operating conditions, this invention can also introduce a machine learning model as a correction and update mechanism for the mechanistic model. The machine learning model is trained and corrected based on historical operating data and manual COD sampling data for parameters of the DO identifiable interval, shock identification threshold, the mapping relationship between DO trend characteristics and shock intensity, and the target sludge age (SRT) generation rule. The mechanistic model provides structurally interpretable boundary constraints and physical meaning, while the machine learning model provides the ability to fit and correct for long-term drift, sensor bias, and complex operating conditions. The two work together to improve long-term operational adaptability.

[0079] Furthermore, to avoid frequent sludge discharge adjustments caused by short-term disturbances, this invention can set anti-oscillation strategies, including but not limited to: trigger dead zone, i.e., triggering only when dDO / dt exceeds a certain threshold, minimum adjustment interval, and sludge discharge volume limit. SRT limiting ( (etc.) After sludge discharge, the DO is restored and stabilized within the preset working range by aeration assistance, and the system only enters the next identification window after a limited adjustment time interval, thereby suppressing false triggering and frequent adjustments and maintaining closed-loop stable operation.

[0080] Figure 2The flowchart describes the judgment logic of system impact identification. The process starts from the "Judgment Start" node and enters the subsequent data acquisition and logical judgment stages. First, the "Data Acquisition and Preprocessing" operation is performed to collect DO signals in the HRAS reactor. Optional signals include oxidation-reduction potential (ORP), influent flow rate (Q), and aeration oxygen input. Filtering and timestamp alignment are used to form a standardized sequence that can be used for online feature extraction, providing reliable data for subsequent analysis. The preprocessed data needs to be judged to "whether it has entered the identifiable range". If not, it directly enters "End and enter the next cycle", waiting for the next cycle of data acquisition and analysis. If yes, it proceeds to the next step, "Calculate the trend characteristic parameter dDO / dt". When the DO is judged to be in the identifiable range, at least the following conditions must be met: influent flow rate (Q) is stable (within the allowable fluctuation range), aeration oxygen input is stable (maintaining the set value or within the allowable fluctuation range, and its deviation or rate of change is limited), and DO concentration is within the preset working range and remains at least within the preset time window. When data enters the identifiable range, the trend characteristic parameter dDO / dt is calculated. Based on the calculated dDO / dt, it is determined whether the impact triggering condition is met (DO shows a significant trend within a time period, such as a continuous decrease exceeding a preset threshold, indicating entry into the candidate impact state). If not, the process returns to "End and Enter the Next Cycle," and data is re-acquired. If yes, the process enters the "Enter Candidate Impact State." After entering the candidate impact state, it is necessary to check whether the ORP meets the preset unidirectional change condition and remains for no less than the preset holding time. If not, the process returns to "End and Enter the Next Cycle," and data is re-acquired. If yes, the impact identification result is output (e.g., outputting the impact identification signal and impact intensity characterization quantity), completing the continuity criterion for the redox potential ORP signal, and preventing DO misjudgment caused by non-impact factors (such as environmental fluctuations and sensor noise). Regardless of whether the impact result is output, the process will eventually enter "End and Enter the Next Cycle," achieving periodic monitoring.

[0081] Figure 3This document describes the sludge age (SRT) adjustment and sludge discharge control process in response to organic load shock events within the system. The process begins with the question, "Is this an organic load shock event?" If the result is "no," the current SRT and sludge discharge strategy are maintained, and the current adjustment cycle ends immediately after execution. If the result is "yes," the subsequent adjustment process for the shock event begins. When an organic load shock event is identified, a control-oriented mechanism model / machine learning model is invoked, inputting "threshold, SRT rules, and the relationship between shock intensity and SRT," etc., and outputting a target SRT based on the identified shock intensity. After the target SRT is output, the system enters the operational constraint stage (e.g., adjusting the SRT range, minimum adjustment interval, preventing adjustment oscillations and adjustment limits) to verify the reasonableness of the target SRT (e.g., avoiding excessive adjustment amplitude, exceeding the feasible SRT range, excessive adjustment frequency, etc.) to ensure the stability and feasibility of the adjustment strategy. After the operational constraint verification passes, the system proceeds to convert the target SRT into a specific sludge discharge volume; subsequently, the sludge discharge actuator executes this discharge volume, completing the sludge discharge adjustment. After sludge removal, aeration is used to assist in adjusting dissolved oxygen (DO) levels, promoting DO recovery to prepare for the next round of identification. The DO level is adjusted through aeration to restore the system to a stable state where shocks can be identified again. Finally, the adjustment round ends after the aeration adjustment is completed.

[0082] Example 1

[0083] This embodiment provides a laboratory implementation of a method for online identification of organic load and adaptive control of sludge age (SRT) based on dissolved oxygen (DO) trend characteristics for high-load activated sludge processes (HRAS). Specifically, a continuous-flow HRAS system is built in the laboratory. Figure 4As shown, the experimental setup mainly includes an inlet unit, a continuous flow completely mixed aeration reactor 3 (HRAS reactor), an outlet unit, a sludge removal unit, an aeration and oxygen supply unit, and an online monitoring and data acquisition control unit. The inlet unit includes an inlet tank 1 and a first peristaltic pump 2 for pumping raw water from the inlet tank to the continuous flow completely mixed aeration reactor 3. The outlet unit is an overflow weir 4 (overflow outlet) installed on the side wall of the continuous flow completely mixed aeration reactor 3. The aeration and oxygen supply unit includes a microporous aerator 5 installed at the bottom of the continuous flow completely mixed aeration reactor 3, a variable frequency blower 6 supplying air to the microporous aerator 5, and a gas flow meter 7 for monitoring and adjusting the aeration rate. The sludge removal unit is a second peristaltic pump 8. The online monitoring and data acquisition control unit includes: an online monitoring module 9, which includes at least a DO sensor installed within the continuous flow completely mixed aeration reactor 3; and a control unit 10 (such as a programmable logic controller PLC), electrically connected to the online monitoring module 9. The PLC integrates a DO (Dissolved Oxygen) identification interval determination module, a feature calculation module, a load identification module, and an SRT (Self-Reliance Time) decision module. The reaction tank is a rectangular tank made of plexiglass, 30cm long, 20cm wide, and 25cm high, with an effective water depth of approximately 18cm and an effective volume of 10L. Microporous aerators 5 are installed at the bottom of the tank to provide dissolved oxygen. The aeration rate is jointly adjusted and recorded by a gas flow meter 7 and a variable frequency blower 6. Water is continuously supplied via a first peristaltic pump 2, and discharged via continuous overflow. Sludge removal is performed by a separate second peristaltic pump 8 according to control commands. The online monitoring module 9 also includes sensors for collecting ORP (Oxygen Response Rate) signals, influent flow signals, and oxygen supply input signals such as aeration rate or blower frequency. Examples include the influent flow meter 11 and the sludge flow meter 12 shown in the figure. In this embodiment, the system operates with a constant influent flow rate to maintain a stable HRT (Hydrogen Temperature Response). The basic operating SRT is set to 1 day, and the allowable adjustment range of SRT is set to 1 day to 2.5 days to avoid excessive fluctuations that could cause control oscillations. The control unit 10 is specifically configured to perform the following steps: receive online data from the online monitoring module 9; determine organic load shock events based on the online data; determine the sludge discharge volume Qw ​​based on the target sludge age SRT and generate control commands; and send the control commands to the sludge discharge execution unit and / or aeration and oxygen supply unit to adjust the sludge discharge and / or aeration operations.

[0084] During system operation, the online monitoring module (sensor monitoring instrument) collects DO signals in the reaction tank in real time at a fixed sampling frequency, and simultaneously collects ORP signals, influent flow rate Q, and oxygen supply input signals such as aeration volume or blower frequency. The collected signals are then fed into the control unit for filtering, noise reduction, and timestamp alignment. To ensure that DO changes primarily reflect changes in oxygen consumption rather than oxygen supply disturbances, the control unit first determines whether the system has entered the DO recognition interval within each control cycle. The system is considered to have entered the DO recognition interval when the influent flow rate Q is within the allowable fluctuation range, the aeration oxygen supply input remains at the set value or is within the allowable fluctuation range and its changes are limited, and the DO is within the preset operating range and remains at least within the preset time window. Within this interval, the system performs DO trend identification and calculation to reduce false triggering caused by short-term disturbances and noise.

[0085] Once the system determines that the DO (Displacement Oxygen Demand) has entered the identifiable range, it performs sliding window feature extraction on the DO signal. Specifically, linear regression is used to fit the relationship between DO and time, and the regression slope is used as the trend feature parameter dDO / dt. Subsequently, the system performs shock identification. When dDO / dt meets a preset trigger condition, such as a continuous decrease exceeding a threshold, it enters the candidate shock state. Simultaneously, it checks whether the ORP (Organic Oxidation Potential) meets a preset condition for continuous change. Only when the persistence criterion of the ORP signal passes is the system determined to have caused an organic load shock event due to an increase in COD, and the shock identification result is output, such as the shock identification signal and the shock intensity characterization quantity. This combined strategy of trend change triggering and persistence verification effectively avoids misjudging transient DO drops caused by fluctuations in oxygen input, mixing fluctuations, or sensor noise as COD shocks.

[0086] Once the impact identification is triggered, the system enters the SRT adaptive decision-making and sludge discharge execution phase. The control unit calls the control guidance mechanism model to generate the target sludge age SRT, and combines the upper and lower limits of SRT, sludge discharge volume limits, and minimum adjustment intervals to make the target sludge age SRT executable. Subsequently, based on the SRT and... Calculation of sludge discharge volume using conversion relationships The commands are then sent to the sludge pump for execution, adjusting the biomass within the reactor to synchronize the treatment capacity with the COD surge. For COD surge scenarios, this embodiment preferably increases the target sludge age (SRT) within constraints to improve biomass carrying capacity and enhance shock resistance. Simultaneously, aeration serves as an auxiliary adjustment method to maintain dissolved oxygen (DO) within the preset operating range, allowing the system to quickly recover to the identifiable range and enter the next round of identification and control. The above process is executed cyclically with a fixed control cycle, forming a closed-loop control system.

[0087] In the verification operation of this embodiment, a typical upward shock was simulated by artificially setting a step increase in influent COD. The results showed that within one or a few control cycles after the shock occurred, the system was able to identify the shock using the persistence criteria of DO trend characteristics and oxidation-reduction potential (ORP) signals, and trigger the increase of the target sludge age (SRT) and sludge discharge rate. Adjustment. After adjustment, DO can be maintained and returned to the preset working range, thereby achieving continuous online identification of subsequent shocks and adaptive SRT control. This embodiment illustrates that the method of the present invention can achieve timely identification of COD rise shocks and SRT linkage control in a laboratory continuous flow rectangular fully mixed HRAS reaction system, verifying the feasibility of the method.

[0088] Example 2

[0089] This embodiment provides an application of an online organic load identification and sludge age (SRT) adaptive control method based on dissolved oxygen (DO) trend characteristics in a high-load activated sludge process (HRAS) system on an engineering scale. A continuous flow completely mixed HRAS unit is installed in a wastewater treatment process. The device includes an HRAS reactor, an aeration system, a sludge discharge system, online monitoring instruments, and a host computer control system.

[0090] The effective volume of the HRAS reactor is 60m³. 3 It operates using a continuous inlet and outlet water method, with a treatment flow rate of approximately 900 m³ / h. 3 / d, corresponding to an HRT of approximately 1.6h. The system is equipped with online DO and ORP probes, and is connected to the control signal for aeration and oxygen supply and the influent flow rate Q signal, with a sampling interval of 1min. The system executes a control strategy in a fixed control cycle of 3h, with a basic operating SRT set to 0.8d, an allowable SRT range set to 0.5d to 2d, and sludge discharge limits and minimum adjustment intervals set to suppress frequent adjustments. Preset parameters:

[0091] DO identification range conditions: influent flow rate Q fluctuation <±5%, aeration rate fluctuation <±2%, and DO within the range of 2.0-4.0 mg / L for at least 30 minutes.

[0092] Impact recognition threshold: dDO / dt ≤ -0.25 mg / (L·min), and the state lasts for 10 minutes; ORP must also show a downward trend for 10 minutes.

[0093] Target SRT decision model: A simplified control-oriented mechanism model based on the ASM1 carbon oxygen system is adopted. When an impact is identified, the target sludge age SRT is calculated based on this model.

[0094] SRT conversion: Calculate the sludge discharge volume according to the formula Qw = V / SRT.

[0095] Auxiliary module: During and after the identification and sludge discharge control, the aeration auxiliary module is activated. It uses PID control to fine-tune the blower frequency and stably maintain the DO concentration within the preset working range of 2.0-4.0 mg / L, providing a reliable environment for subsequent continuous monitoring and identification.

[0096] Within each control cycle, the system first determines whether the DO (Dissolved Oxygen) identifiable range has been entered, ensuring stable influent flow, limited aeration input, and DO within the operating range. Further, within the identifiable range, dDO / dt is calculated and combined with the persistence criterion of the oxidation-reduction potential (ORP) signal to identify shocks caused by COD increases. Upon detection of a shock trigger, the control-guided mechanism model is invoked to determine the target sludge age (SRT) and converted into a sludge discharge rate (Qw) for execution. After execution, aeration is adjusted to maintain DO within the operating range, ensuring the reliability of subsequent identifications. The mechanism model generates DO identifiable range parameters, identification thresholds, and rules relating shock intensity to the target sludge age (SRT). A machine learning model, as an optional module, is used to periodically correct and update the thresholds and rules based on historical operating data and manual COD sampling data.

[0097] During the 90-day operational validation period, the system experienced multiple influent COD surge events, with a typical surge occurring on day 45: the influent COD rose from approximately 650 mg / L to approximately 1200 mg / L within a short period and remained elevated for about 5 hours. The implementation process of this method is detailed below using this typical high-intensity surge event as an example:

[0098] Days 0-44 (before shock): The system operated stably, with influent COD approximately 650 mg / L. The controller's judgment conditions were met, and the system entered the "DO identifiable range." The calculated dDO / dt fluctuated slightly near zero, ORP remained stable, and no shock judgment was made. The system began sludge discharge at a stable sludge discharge rate of 0.8 days based on the baseline SRT.

[0099] Day 45 (impact arrives): The COD concentration in the system influent rises rapidly, from approximately 650 mg / L to approximately 1200 mg / L and remains elevated for about 5 hours.

[0100] Fifteen minutes after the impact (impact identification phase): Due to increased microbial oxygen consumption, DO begins to decline rapidly and continuously. Within the sliding window, the calculated dDO / dt quickly becomes negative and remains below -0.25 mg / (L·min). The ORP value simultaneously shows a significant and continuous decline from +150 mV. The persistence criteria for both DO and ORP simultaneously meet the preset thresholds and durations. The impact identification module determines that a COD increase impact event has occurred. The system confirms that a "high-intensity" COD increase impact event has occurred.

[0101] Decision-making and execution phase: The SRT decision module calculates, based on a preset model, that the target sludge age (SRT) should be adjusted to 1.6 days. The sludge discharge execution module calculates, based on the new target sludge age (SRT) and the current reactor volume (V), that the sludge discharge rate (Qw) is approximately 37.5 m³. 3 The system then sends a command to the sludge pump to execute the quantitative sludge discharge. The aeration assist module operates synchronously, increasing the aeration rate by approximately 15%, successfully pulling the dissolved oxygen (DO) back from a low of nearly 1.75 mg / L and stabilizing it at around 2.6 mg / L. After approximately 1-2 hRTs, the increased biomass began to match the high load, the downward trend in DO was fundamentally halted, and the system operated in a new steady state adapted to the higher influent load. After the shock event, the system, through continuous monitoring, gradually reduced the solid reflux rate (SRT) back to near the baseline value in subsequent control cycles.

[0102] The control system identifies the impact based on DO trend characteristics and completes SRT adjustment within one control cycle (3 hours), increasing the target sludge age SRT from 0.8 days to 1.6 days, thereby increasing biomass to match the impact load while maintaining DO within the preset operating range. Compared to the control strategy (effluent COD exceeding 700 mg / L, requiring more than 18 hours to recover to steady state), which uses a fixed SRT and relies solely on changing the aeration rate to maintain DO, this embodiment significantly reduces the peak effluent COD (maximum effluent COD 450 mg / L) and fluctuation amplitude (smooth fluctuation) during the impact, and shortens the system recovery time to stable operation (approximately 6-8 hours after the impact). During the test period, the overall system operation remained stable, the impact event was promptly identified and triggered SRT regulation, significantly reducing the frequency of manual intervention. The above embodiments demonstrate that the method of the present invention can effectively improve the timeliness of HRAS identification and the effectiveness of regulation of COD rise impacts under engineering operating conditions, and has good engineering application prospects.

Claims

1. A sludge age self-adaptive control method based on dissolved oxygen trend identification, characterized in that, The method includes the following steps: S1. Data Acquisition and Preprocessing: Real-time acquisition of operating data of the HRAS reactor, including at least dissolved oxygen (DO) signals, and preprocessing of the operating data to obtain a standardized signal sequence; S2. Determination of DO Identifiable Range: Under the condition that the influent flow rate remains stable, the DO identifiable range is determined, wherein within the DO identifiable range, the aeration oxygen supply input remains stable, and the DO concentration is within the preset working range and continues to be no less than the preset time window. S3. Trend feature extraction: Within the identifiable range of DO, feature extraction is performed on the DO signal, and trend feature parameters characterizing the change of DO over time are calculated; S4. Organic load shock identification: Based on the trend characteristic parameters and combined with the persistence criterion of the redox potential (ORP) signal, determine whether an organic load shock event caused by the increase in chemical oxygen demand (COD) has occurred. S5. SRT Decision and Sludge Discharge Execution: After identifying the organic load shock event, determine the target sludge age SRT, convert the target sludge age SRT into sludge discharge volume Qw, and control the sludge discharge execution mechanism to output the sludge discharge volume Qw. Among them, the DO identifiable interval parameter, the impact identification threshold of the organic load shock event, the relationship between the trend characteristic parameter and the load shock intensity, and the target sludge age SRT are at least determined by the control-oriented mechanism model and participate in online control; the DO identifiable interval parameter includes: the allowable fluctuation range for influent flow stability determination, the set value and allowable fluctuation range of aeration oxygen supply input, the preset working range of DO concentration, and the preset time window for determining that DO is continuously within the working range; In step S4, the persistence criterion for the redox potential (ORP) signal includes: when the trend characteristic parameter meets the preset impact triggering condition, and the ORP signal meets the preset same-direction change condition and lasts for no less than the preset holding time, it is determined that an organic load impact event caused by the increase in COD has occurred. In step S5, the target sludge age SRT is determined based on a control-oriented mechanism model. This model includes at least an oxygen transfer term determined based on aeration input and an oxygen consumption term determined based on DO (dissolved oxygen) change trends. The control-oriented mechanism model is configured to perform the following functions: based on the equivalent oxygen transfer coefficient kLa and saturated dissolved oxygen concentration SO₂ determined by the aeration input. * Oxygen transfer relationships are constructed, and the oxygen consumption rate (OUR) or the change in oxygen consumption rate is calculated based on the DO change trend to determine the impact identification threshold and SRT decision. The oxygen transfer term is represented as: wherein is the saturated dissolved oxygen concentration, is the dissolved oxygen concentration, is the equivalent oxygen mass transfer coefficient; the dissolved oxygen dynamics are represented in the form of: wherein , Q is the influent flow rate, V is the effective volume of the HRAS reactor, is the oxygen uptake rate; It consists of the oxygen consumption for the aerobic growth of heterotrophic bacteria and the oxygen consumption for maintenance, namely: in For heterotrophic bacteria yield, For dynamic parameters, It is a soluble substrate. This represents the concentration of heterotrophic bacteria.

2. The method of claim 1, wherein, The trend characteristic parameters include, but are not limited to, the regression slope dDO / dt of DO changing over time, which is obtained by performing linear regression fitting on the DO data within the sliding window.

3. The method of claim 1, wherein, The determination of the DO identifiable range requires the following conditions to be met simultaneously: the deviation of the aeration oxygen supply input signal from the set value does not exceed the preset threshold; the DO is within the preset working range and the duration is not less than the preset time window; the influent flow rate is within the preset allowable fluctuation range.

4. The method according to claim 1, characterized in that, The trend characteristic parameters are mapped to load indicators that characterize the impact intensity of COD increase. The load indicators include at least: the integral of the oxygen consumption rate (OUR) within a preset time window, and the impact intensity indicator composed of the change in oxygen consumption rate and the drop in DO.

5. The method of claim 4, wherein, The method further establishes and invokes a machine learning model. The input of the machine learning model includes at least the trend feature parameters, the aeration and oxygen supply input signal, and the influent flow rate. The output is the load index, the adjusted target sludge age SRT, and the model is trained, corrected, and / or updated online using historical operating data and manual COD sampling data.

6. The method of claim 5, wherein, The generation rule of the target sludge age SRT is determined at least by the control guidance mechanism model and can be determined by correction or update by the machine learning model. In the control process of generating sludge discharge volume Qw ​​based on the target sludge age SRT, preset operating constraints must be followed. The operating constraints include at least the DO working range constraint, SRT constraint and the limit constraint of sludge discharge volume Qw.

7. A sludge age adaptive control system for implementing the method according to any one of claims 1-6, characterized in that, include: a. Online monitoring module, used to collect DO data in real time, and optionally collect ORP, influent flow rate and aeration oxygen supply input signals; b. DO Recognizable Range Determination Module, used to construct or determine the DO recognizable range under the condition that the influent flow rate is stable and the aeration oxygen supply input is maintained at the set value or within the allowable fluctuation range; c. Feature calculation module, used to filter the DO signal and perform sliding window feature extraction, outputting trend feature parameters; d. Load identification module, used to identify influent load shock events caused by COD increase based on trend characteristic parameters and the persistence criterion of oxidation-reduction potential (ORP) signal; e. SRT decision module, used to calculate the target sludge age SRT according to the target sludge age SRT generation rule after the influent load shock event is identified; f. Sludge discharge execution module, used to convert the target sludge age SRT into sludge discharge volume Qw ​​and control the sludge discharge execution mechanism to output Qw; g. Aeration auxiliary adjustment module, used to maintain DO within the preset working range to ensure identification reliability and operational stability; h. Model and parameter generation module, used to build and call control-oriented mechanism model and machine learning model to determine DO identifiable range parameters, shock identification threshold and target sludge age SRT and participate in online control.

8. The system of claim 7, wherein, The model and parameter generation module includes a mechanism model submodule, which contains at least an oxygen transfer term and an oxygen consumption term, used to calculate the oxygen consumption rate OUR or its increment ΔOUR and generate the impact identification threshold and load index accordingly.

9. The system of claim 7, wherein, The model and parameter generation module includes a machine learning submodule, which trains, corrects, or updates the shock identification threshold, mapping relationship, and target sludge age (SRT) generation based on historical operating data and manual COD sampling data.

Images