Multi-objective cooperative optimization operation method and control system of AOA wastewater treatment system

Abstract

The application provides a multi-target collaborative optimization operation method and control system of an AOA process sewage treatment system, relates to the technical field of data processing, and comprises the following steps: step 1, a current working condition characteristic vector is constructed by combining current water quality indexes, operation parameters and the pollutant concentration change rates of three characteristic positions; the distance and correlation measure of the current working condition characteristic vector and each historical working condition characteristic vector are calculated, and a historical data subset is determined from a historical database to establish an online prediction model of effluent water quality, operation energy consumption and sludge production under the current working condition. The application solves the multi-target optimization problem by combining historical and current working conditions to construct an online prediction model, defining an optimization boundary and based on carbon source dynamic balance, and finally issues optimized setting values, so that the collaborative optimization of effluent water quality, operation energy consumption and sludge production of the AOA process sewage treatment system is realized, and the accuracy, stability and economy of system operation are improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a multi-objective collaborative optimization operation method and control system for an AOA process wastewater treatment system. Background Technology

[0002] In the field of wastewater treatment, the AOA (anaerobic-aerobic-anoxic) process is widely used in numerous municipal wastewater treatment plants due to its superior nitrogen and phosphorus removal performance. However, a common challenge in its actual operation and management is how to synergistically optimize effluent quality, operating energy consumption, and sludge production, especially under conditions of fluctuating influent quality. Currently, many wastewater treatment plants still rely primarily on periodic monitoring of key effluent parameters and the experience of operators for operational control. Parameter settings, such as dissolved oxygen concentration and sludge return ratio, are typically adjusted based on relatively fixed empirical values ​​or simple threshold rules. This control method may sometimes lack the timeliness and overall coordination required when facing dynamic influent, resulting in the system's carbon source utilization efficiency not consistently maintaining an optimal level.

[0003] For example, during the spring operation of a wastewater treatment plant, the influent chemical oxygen demand (COD) and total nitrogen concentration often fluctuate within and between days due to the influence of rainfall and residents' daily routines. Operators usually set the dissolved oxygen in the aerobic section based on the effluent ammonia nitrogen data from the previous shift and empirical formulas. The limitation of this approach is that it is difficult to perceive and respond in real time to the dynamic distribution and consumption status of carbon sources within the system in different biochemical processes such as anaerobic phosphorus release, aerobic phosphorus uptake, and denitrification. Sometimes, in order to ensure that the effluent total nitrogen meets the standards, a higher dissolved oxygen setpoint or sludge return ratio may be adopted. This may indirectly lead to an increase in aeration energy consumption or an increase in excess sludge production, resulting in a failure to achieve a good balance among multiple operational objectives. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a multi-objective collaborative optimization operation method and control system for AOA process wastewater treatment system, so as to reduce the intensity of manual operation and improve the intelligence level of process operation.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a multi-objective collaborative optimization operation method for an AOA process wastewater treatment system, the method comprising: Step 1: Construct a feature vector for the current operating condition by combining the current water quality indicators, operating parameters, and the rate of change of pollutant concentration at three characteristic locations; calculate the distance and correlation measure between the feature vector for the current operating condition and the feature vectors for each historical operating condition, and determine a subset of historical data from the historical database to establish an online prediction model for effluent water quality, operating energy consumption, and sludge production under the current operating condition. Step 2: Construct a multi-objective optimization problem, and based on the fluctuation characteristics of the influent water quality index data, determine the numerical boundaries of the operating state parameters in the multi-objective optimization problem, and deduce the feasible range of values ​​for the dissolved oxygen setpoint and sludge return ratio. Step 3: Within the feasible range of values, by analyzing the rate of change of pollutant concentration at three characteristic locations, a virtual equilibrium domain for the dynamic balance of carbon source in the system is generated, and the current system point is defined. The positional relationship between the current system point and the boundary of the virtual equilibrium domain is calculated to obtain a dynamic equilibrium deviation value. Step 4: Based on the pollutant concentration change rate and combined with the dynamic equilibrium deviation value, calculate the migration and transformation equilibrium of carbon source from storage to consumption in the system, and solve the multi-objective optimization problem to obtain a non-dominated solution set. Step 5: Determine the final optimization scheme from the non-dominated solution set, and issue the dissolved oxygen setpoint and sludge return ratio corresponding to the final optimization scheme as optimization setpoints for execution, thereby completing the collaborative optimization of the system operation status.

[0006] Secondly, the multi-objective collaborative optimization control system of the AOA process wastewater treatment system includes: The model building module is used to construct the current operating condition feature vector by combining the current water quality indicators, operating parameters and the pollutant concentration change rate at three characteristic locations; calculate the distance and correlation measure between the current operating condition feature vector and the feature vectors of each historical operating condition; and determine a subset of historical data from the historical database to establish an online prediction model for effluent water quality, operating energy consumption and sludge production under the current operating conditions. The deduction module is used to construct a multi-objective optimization problem and, based on the fluctuation characteristics of the influent water quality index data, determine the numerical boundaries of the operating state parameters in the multi-objective optimization problem, and deduce the feasible range of values ​​for the dissolved oxygen setpoint and the sludge return ratio. The calculation module is used to generate a virtual equilibrium domain for the dynamic balance of carbon sources in the system by analyzing the rate of change of pollutant concentration at three characteristic locations within the feasible range of values, and to define the current system point, calculate the positional relationship between the current system point and the boundary of the virtual equilibrium domain, and obtain a dynamic equilibrium deviation value. The solution module is used to calculate the migration and transformation equilibrium of carbon sources from storage to consumption within the system based on the pollutant concentration change rate and dynamic equilibrium deviation value, and to solve the multi-objective optimization problem to obtain a non-dominated solution set. The optimization module is used to determine the final optimization scheme from the non-dominated solution set, and issue the dissolved oxygen setpoint and sludge return ratio corresponding to the final optimization scheme as optimization setpoints for execution, thereby completing the collaborative optimization of the system operation status.

[0007] Thirdly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0008] The above-described solution of the present invention has at least the following beneficial effects: This system effectively improves its adaptability to fluctuations in influent water quality, addressing the shortcomings of empirical control in terms of timeliness and overall coordination when dealing with dynamic influent, ensuring stable effluent quality and avoiding imbalances in treatment effects caused by water quality fluctuations. It achieves a synergistic balance among multiple objectives, including effluent quality, operating energy consumption, and sludge production, overcoming the limitations of optimizing a single objective in control that could harm other objectives, balancing treatment effectiveness with operational economy, and reducing overall system operating costs. Furthermore, by quantifying the dynamic balance of carbon sources and calculating the carbon source migration and transformation equilibrium, it optimizes the carbon source in anaerobic, aerobic, and anoxic environments. The allocation and utilization of carbon sources in the biochemical process improves carbon source utilization efficiency, reduces carbon source waste, and enhances the core efficiency of nitrogen and phosphorus removal. It replaces the operation parameter setting mode that relies on operator experience, and achieves precise dynamic control of dissolved oxygen setpoints and sludge return ratio through data-driven prediction models and multi-objective optimization solutions, reducing the intensity of manual operation and improving the level of intelligent operation of the process. It optimizes the overall operating status of the system, reduces the ineffective equipment loss caused by unreasonable parameters, improves the stability and reliability of process operation, extends the service life of related equipment, and reduces the difficulty and cost of operation and maintenance management. Attached Figure Description

[0009] Figure 1 This is a schematic flowchart of the multi-objective collaborative optimization operation method of the AOA process wastewater treatment system provided in the embodiments of the present invention.

[0010] Figure 2 This is a schematic diagram of a multi-objective cooperative optimization control system provided in an embodiment of the present invention. Detailed Implementation

[0011] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0012] like Figure 1 As shown, embodiments of the present invention propose a multi-objective collaborative optimization operation method for an AOA process wastewater treatment system, the method comprising the following steps: Step 1: Construct a feature vector for the current operating condition by combining the current water quality indicators, operating parameters, and the rate of change of pollutant concentration at three characteristic locations; calculate the distance and correlation measure between the feature vector for the current operating condition and the feature vectors for each historical operating condition, and determine a subset of historical data from the historical database to establish an online prediction model for effluent water quality, operating energy consumption, and sludge production under the current operating condition. Step 2: Construct a multi-objective optimization problem, and based on the fluctuation characteristics of the influent water quality index data, determine the numerical boundaries of the operating state parameters in the multi-objective optimization problem, and deduce the feasible range of values ​​for the dissolved oxygen setpoint and sludge return ratio. Step 3: Within the feasible range of values, by analyzing the rate of change of pollutant concentration at three characteristic locations, a virtual equilibrium domain for the dynamic balance of carbon source in the system is generated, and the current system point is defined. The positional relationship between the current system point and the boundary of the virtual equilibrium domain is calculated to obtain a dynamic equilibrium deviation value. Step 4: Based on the pollutant concentration change rate and combined with the dynamic equilibrium deviation value, calculate the migration and transformation equilibrium of carbon source from storage to consumption in the system, and solve the multi-objective optimization problem to obtain a non-dominated solution set. Step 5: Determine the final optimization scheme from the non-dominated solution set, and issue the dissolved oxygen setpoint and sludge return ratio corresponding to the final optimization scheme as optimization setpoints for execution, thereby completing the collaborative optimization of the system operation status.

[0013] In this embodiment of the invention, the system's adaptability to fluctuations in influent water quality is effectively improved, solving the problems of insufficient timeliness and overall coordination in empirical control when dealing with dynamic influent, ensuring stable effluent water quality and avoiding imbalances in treatment effects caused by water quality fluctuations; achieving a synergistic balance among multiple objectives such as effluent water quality, operating energy consumption, and sludge production, breaking the limitation of single-objective optimization in control leading to damage to other objectives, balancing treatment effect and operational economy, and reducing the overall operating cost of the system; by quantifying the dynamic balance state of the system's carbon source and calculating the carbon source migration and transformation equilibrium, the carbon source is optimized in anaerobic and aerobic environments. This technology optimizes the distribution and utilization of carbon sources in various biochemical processes under anoxic conditions, improves carbon source utilization efficiency, reduces carbon source waste, and enhances the core efficiency of nitrogen and phosphorus removal. It replaces the operation parameter setting mode that relies on operator experience, achieving precise dynamic control of dissolved oxygen setpoints and sludge return ratios through data-driven predictive models and multi-objective optimization solutions. This reduces the intensity of manual operation and improves the intelligence level of process operation. Furthermore, it optimizes the overall system operation status, reduces ineffective equipment losses caused by unreasonable parameters, improves the stability and reliability of process operation, extends the service life of related equipment, and reduces the difficulty and cost of operation and maintenance management.

[0014] In another preferred embodiment of the present invention, prior to step 1, water quality index data of the influent and effluent of the AOA wastewater treatment system, as well as operating parameters in the anaerobic, aerobic, and anoxic zones, are collected in real time. Simultaneously, the actual pollutant concentration change rate at three characteristic locations—the inlet of the anaerobic zone, the end of the aerobic zone, and the outlet of the anoxic zone—is monitored. Specifically, this includes: firstly, multi-parameter online water quality monitoring sensors are deployed at the influent channel and the main effluent outlet of the AOA wastewater treatment system. The sensor probes are in direct contact with the water body, capturing real-time water quality index data of the influent and effluent, covering core water quality indicators such as chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, pH value, and water temperature. The sensors continuously collect data at a preset collection frequency of 5 minutes per data point. The collected raw water quality data is converted from electrical to digital signals by a signal conversion module and then transmitted to the system data acquisition terminal for preliminary storage and preprocessing. Secondly, distributed operating parameter monitoring devices are deployed in the reaction tanks of the anaerobic, aerobic, and anoxic zones of the AOA process, based on the tank structure and flow characteristics. Monitoring points were set up as follows: one at the inlet, one in the middle mixing zone, and one at the outlet in the anaerobic zone; one at the front end, one in the middle section, and one at the front end of the sedimentation zone in the aerobic zone; and one at the inlet, one in the middle of the denitrification reaction zone, and one at the effluent return end in the anoxic zone. Dissolved oxygen, oxidation-reduction potential, sludge concentration, hydraulic retention time, aeration intensity, and sludge return flow were installed at each monitoring point. All devices simultaneously collected data on the anaerobic water quality. The system collects operating parameters for each functional zone. In the anaerobic zone, parameters such as dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, and hydraulic retention time are collected. In the aerobic zone, parameters such as dissolved oxygen concentration, aeration intensity, mixed liquor sludge concentration, and hydraulic retention time are collected. In the anoxic zone, parameters such as dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, hydraulic retention time, and sludge return flow rate are collected. All operating parameters are collected by the on-site acquisition module and transmitted to the data acquisition terminal in the form of digital signals, where they are classified and stored together with water quality index data.

[0015] Subsequently, dedicated monitoring devices were deployed at three pre-set characteristic locations: high-precision pollutant concentration monitoring probes were installed at the inlet of the anaerobic zone, the outlet of the aerobic zone, and the return flow outlet of the anoxic zone. All three probes were of the same model and precision to ensure data consistency and comparability. Each probe continuously monitored the real-time pollutant concentration at its corresponding location. Pollutant concentrations covered core indicators related to carbon source conversion, nitrogen and phosphorus removal, such as chemical oxygen demand (COD), ammonia nitrogen, total nitrogen, and total phosphorus. Monitoring data was uploaded to the data acquisition terminal at identical time intervals, with the intervals precisely controlled to ensure synchronous data acquisition at the three characteristic locations. Upon receiving the synchronous pollutant concentration data from the three characteristic locations, the data acquisition terminal initiated a pollutant concentration change rate calculation program to perform real-time calculations of the pollutant concentration change rate at each characteristic location. The concentration of a single pollutant at each characteristic location was then calculated. The calculation process for the rate of change is as follows: Select pollutant concentration data from two consecutive time points, subtract the pollutant concentration value from the previous time point from the pollutant concentration value of the later time point to obtain the pollutant concentration change within that time interval, and then divide the pollutant concentration change by the time interval between the two time points to obtain the actual pollutant concentration change rate at that characteristic location within that time interval. The calculation formula is: Pollutant concentration change rate = (Pollutant concentration at the later time point - Pollutant concentration at the previous time point) ÷ Time interval. Following this calculation process, the concentration change rates of all core pollutants at the three characteristic locations of the anaerobic zone inlet, aerobic zone end, and anoxic zone outlet are calculated respectively. If the pollutant concentration decreases at a certain characteristic location, the calculated concentration change rate is negative, indicating that the pollutant has been degraded and consumed at that location. If the concentration increases, the rate is positive, indicating that the pollutant has accumulated or transformed at that location.

[0016] Finally, a data synchronization and verification mechanism was established. The collected influent and effluent water quality data, operating parameters of the anaerobic, aerobic, and anoxic zones, and the calculated rates of change of actual pollutant concentrations at three characteristic locations were precisely matched according to the timestamps of collection and calculation. This ensured that the water quality indicators, operating parameters, and pollutant concentration change rates at the same time dimension were synchronized data sets. Simultaneously, outlier verification was performed on all data. If any data exceeded the preset normal fluctuation range for the process, the following would be verified: Normal fluctuation ranges for water quality indicators: Chemical Oxygen Demand ±20%, Ammonia Nitrogen ±15%, Total Nitrogen ±15%, Total Phosphorus ±10%, Suspended Solids ±20%, pH 6.5 to 8.5, Water Temperature ±3℃; Normal fluctuation ranges for operating parameters: Dissolved Oxygen Concentration ±0.5 mg / L, Oxidation-Reduction Potential ±50 mV, Mixed Liquor Sludge Concentration ±200 mg / L, Aeration Intensity ±10%, Sludge Return... When the flow rate is ±10% and the hydraulic retention time is ±5%, the system automatically issues a data anomaly warning and uses linear interpolation to correct missing or abnormal data. It selects two adjacent valid monitoring data points before and after the missing or abnormal data. Let the time node of the previous valid data be t1 and the corresponding value be X1, the time node of the next valid data be t2 and the corresponding value be X2, and the time node of the missing or abnormal data be t and the value to be corrected be X. First, the time proportion coefficient k = (t - t1) / (t2 - t1) is calculated. Then, linear calculation is performed using the formula X = X1 + k × (X2 - X1) to obtain the correction value for the corresponding missing or abnormal time node. The correction value replaces the abnormal data or fills in the missing data. The complete data set after verification and correction is finally transmitted to the historical database of the AOA wastewater treatment system for structured storage and simultaneously pushed to the system's operating condition analysis module in real time.

[0017] This embodiment establishes a comprehensive and synchronized data acquisition and monitoring system, achieving full-process data capture of changes in water quality, operating parameters, and pollutant concentrations in the AOA wastewater treatment system. This overcomes the limitations of asynchronous data acquisition and incomplete data dimensions in traditional monitoring. Specific pollutant concentration monitoring is conducted at three core characteristic locations, and the actual concentration change rate is obtained through a standardized calculation process. This quantifies the pollutant transformation process within the system into specific values, addressing the pain point of not being able to perceive the pollutant degradation and transformation status in the anaerobic, aerobic, and anoxic stages in real time during operation. This achieves real-time quantitative monitoring of the system's core biochemical reaction processes. The system uses the same model and precision monitoring equipment for the three... Pollutant concentration monitoring is conducted at specific locations with data collected at equal time intervals, ensuring the consistency, comparability, and synchronization of the monitoring data. This avoids data analysis biases caused by differences in equipment and collection intervals. A data synchronization matching and outlier verification mechanism is established, using timestamps to achieve accurate correspondence of multi-dimensional data. At the same time, abnormal and missing data are corrected, ensuring the integrity and effectiveness of stored and used data and improving the reliability of data utilization. All collected and calculated data are synchronously transmitted to the historical database and the operating condition analysis module, realizing structured storage of operating data and improving the system's timeliness in responding to changes in operating status.

[0018] In a preferred embodiment of the present invention, step 1 includes: Step 100: Store the real-time collected and monitored water quality index data, operating parameters, and pollutant concentration change rates at three characteristic locations into the historical database. Based on the currently collected and monitored water quality index data, operating parameters, and pollutant concentration change rates at the three characteristic locations, construct a current operating condition feature vector. Specifically, this includes: First, completing the construction and data storage of the historical database. The historical database is constructed by standardizing, verifying, and structuring the operational data of the AOA wastewater treatment system throughout its entire lifecycle after commissioning. All data stored in the database are valid data after outlier verification and missing value correction. The data dimensions include influent and effluent water quality index data for each time period of the system. The system collects and monitors real-time influent and effluent water quality indicators, operating parameters of the anaerobic, aerobic, and anoxic zones, and pollutant concentration change rates at three characteristic locations. All data is bound to a unique timestamp and stored in a structured manner according to time series. The system also provides basic functions for data expansion, rapid retrieval, and categorized access to ensure efficient data reading, writing, and retrieval. After data synchronization and verification, the system binds the current collection timestamp to the historical database and performs batch data entry, enabling dynamic supplementation of real-time operating data to the historical database.

[0019] Based on the real-time data collected and verified above, the current operating condition feature vector is constructed. First, dimensions are extracted from all real-time data. These extracted dimensions include: influent chemical oxygen demand (COD), influent ammonia nitrogen, influent total nitrogen, influent total phosphorus, influent suspended solids, influent pH, and influent temperature; effluent COD, effluent ammonia nitrogen, effluent total nitrogen, effluent total phosphorus, effluent suspended solids, effluent pH, and effluent temperature; dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, and hydraulic retention time in the anaerobic zone; dissolved oxygen concentration, aeration intensity, mixed liquor sludge concentration, and hydraulic retention time in the aerobic zone; and dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, hydraulic retention time, and sludge return flow rate in the anoxic zone. The pollutant concentration change rates at the anaerobic zone inlet, aerobic zone end, and anoxic zone outlet are all single-value data. The values ​​for each dimension are then dimensionless to eliminate the influence of different indicator dimensions on the feature vector. This is achieved by dividing each dimension's value by the historical maximum value of that indicator in the historical database. Finally, all dimensionless values ​​are arranged in a pre-defined dimensional order to form a one-dimensional numerical sequence. This sequence is the current operating condition feature vector, where each value corresponds one-to-one with a specific water quality indicator, operating parameter, and pollutant concentration change rate dimension.

[0020] Step 101: Calculate the Euclidean distance and cosine correlation measure between the current operating condition feature vector and each historical operating condition feature vector stored in the historical database. Select multiple historical data samples whose cosine correlation measure with the current operating condition feature vector is greater than a preset threshold to form a subset of historical data. Specifically, this includes: First, determining the preset threshold for the cosine correlation measure, with a threshold value of 0.85. This is because, considering the process operation characteristics of the AOA wastewater treatment system, a cosine correlation measure value of 0.85 or higher indicates that the similarity of the system operation states corresponding to the two operating condition feature vectors is extremely high, and the changing trends and numerical characteristics of each dimension index are highly consistent. Selecting this threshold can effectively filter out historical operating condition data that has practical reference value with the current operating condition, while avoiding insufficient historical data samples due to an excessively high threshold, or introducing historical data with low similarity and limited reference value due to an excessively low threshold, thus ensuring the accuracy and effectiveness of subsequent model construction. Based on the current operating condition feature vector constructed in step 100, Euclidean distance and cosine similarity measurement are calculated sequentially between it and the historical operating condition feature vectors stored in the historical database. The dimensions of the two feature vectors are kept completely consistent. Let the current operating condition feature vector be M, containing values ​​of n dimensions, namely M1, M2, M3...Mn, and a certain historical operating condition feature vector be N, containing values ​​of n dimensions, namely N1, N2, N3...Nn. Then, the Euclidean distance between the current operating condition feature vector and the historical operating condition feature vector is calculated. The smaller the Euclidean distance value, the higher the similarity between the two feature vectors. The calculation process of the cosine similarity measurement is as follows: first, the points of the two feature vectors are calculated... The product is calculated by multiplying M1 by N1, M2 by N2, ..., Mn by Nn, and then summing the results of the products of all dimensions to obtain the dot product sum. Next, the magnitudes of the two feature vectors are calculated separately. The magnitude of the feature vector M under the current operating condition is calculated as the square root of the sum of the squares of M1, M2, ..., and Mn. The magnitude of the feature vector N under the historical operating condition is calculated as the square root of the sum of the squares of N1, N2, ..., and Nn. Finally, the dot product sum of the two feature vectors is divided by the product of the magnitudes of the feature vectors under the current operating condition and the historical operating condition. The result is the cosine similarity measure, which ranges from -1 to 1. The closer the value is to 1, the higher the similarity between the two feature vectors. Following the above calculation process, after calculating the Euclidean distance and cosine correlation measure between the current operating condition feature vector and all historical operating condition feature vectors in the historical database, the cosine correlation measure results of all historical operating condition feature vectors are filtered, and all historical operating condition feature vectors with a cosine correlation measure greater than 0.85 are selected. At the same time, the complete historical data samples corresponding to these historical operating condition feature vectors are retrieved, including all data such as the corresponding operating parameters, effluent water quality, operating energy consumption, and sludge production. These filtered historical data samples are integrated to form a historical data subset that highly matches the current operating condition.

[0021] Step 102: Based on historical data subsets, construct online prediction models for effluent quality (with operating parameters as input and effluent quality as output), online prediction models for operating energy consumption (with operating parameters as input and operating energy consumption as output), and online prediction models for sludge production (with operating parameters as input and sludge production as output) under the current operating conditions. Specifically, this includes: using the operating parameters of the anaerobic, aerobic, and anoxic zones from the historical data subset as input variables for the models, including dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, and hydraulic retention time in the anaerobic zone; dissolved oxygen concentration, aeration intensity, mixed liquor sludge concentration, and hydraulic retention time in the aerobic zone; and dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, hydraulic retention time, and sludge return flow rate in the anoxic zone; and using core water quality indicators such as effluent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, pH, and water temperature from the historical data subset as output variables for the models; and using a multiple linear regression algorithm for model training, first establishing multiple linear regression models for each effluent water quality indicator. The linear regression equation uses a single effluent water quality index as the dependent variable, denoted as Y, and all operating parameters as independent variables, denoted as X1, X2, X3...Xm, where m is the total number of operating parameters. A constant term, a0, is set, and the regression coefficients corresponding to each independent variable are denoted as a1, a2, a3...am. The constructed multiple linear regression equation is Y = a0 + a1X1 + a2X2 + a3X3 + ... + amXm. This establishes a linear quantitative relationship between the dependent and independent variables. Based on a subset of historical data, the least squares method is used to solve for the regression coefficients of each regression equation. The calculation process involves minimizing the sum of squared residuals between the actual observed values ​​of the dependent variable and the predicted values ​​of the equation to obtain the regression coefficient values ​​that minimize the sum of squared residuals. This determines the specific quantitative correspondence between all operating parameters and each effluent water quality index. After constructing the regression equations and solving for the regression coefficients for all effluent water quality indicators, the equations are integrated to form an online effluent water quality prediction model with operating parameters as input and effluent water quality as output.

[0022] The operating parameters of the anaerobic, aerobic, and anoxic zones from the historical data subset are used as input variables for the model, and these input variables are consistent with those of the online effluent quality prediction model. The system operating energy consumption corresponding to the historical data subset, including aeration energy consumption, mixing energy consumption, sludge return energy consumption, and other total energy consumption data, is used as the model's output variable. The actual observed operating energy consumption is based on the comprehensive energy consumption per unit volume of wastewater treated by the AOA wastewater treatment system, assumed to be 0.2 kWh / m³. 3 0.3kWh / m 3 0.4kWh / m 3 0.5kWh / m 3 0.6kWh / m3 0.7kWh / m 3 0.8kWh / m 3 The same multiple linear regression algorithm as the online effluent water quality prediction model is used for model training. The constructed multiple linear regression equation is then used to solve the regression coefficients using the least squares method based on sample data from a subset of historical data. This minimizes the sum of squared residuals between the actual observed values ​​of operating energy consumption and the predicted values ​​of the equation. The corresponding regression coefficient values ​​are obtained by solving this problem, and the specific quantitative correspondence between operating parameters and system operating energy consumption is determined. After completion, an online operating energy consumption prediction model is obtained, which takes operating parameters as input and operating energy consumption as output.

[0023] The operating parameters of the anaerobic, aerobic, and anoxic zones in the historical data subset are used as input variables for the model, consistent with the two models mentioned above. The system sludge production data, including residual sludge discharge and sludge yield, corresponding to the historical data subset are used as output variables for the model. The model is trained using a multiple linear regression algorithm, and the constructed multiple linear regression equation is then used to solve for the regression coefficients using the least squares method based on the sample data of the historical data subset. This minimizes the sum of squared residuals between the actual observed sludge production and the predicted values ​​of the equation, thus determining the specific quantitative correspondence between the operating parameters and the system sludge production. This results in an online sludge production prediction model with operating parameters as input and sludge production as output. After the three online prediction models are built, they are all linked with the real-time data acquisition module of the AOA wastewater treatment system to achieve real-time access to input variables and online calculation of output results, ensuring the online prediction function of the model.

[0024] In this embodiment, the historical database, through the structured construction of effective operational data throughout the entire cycle and the dynamic ingestion of real-time data, achieves systematic and standardized storage of operational data for the AOA wastewater treatment system. This provides sufficient and effective data support for operational condition similarity comparison and predictive model construction. During the construction of operational condition feature vectors, full dimensional extraction and dimensionless processing eliminate dimensional differences and numerical deviations among different indicators, ensuring that the feature vectors accurately and comprehensively reflect the current actual operational conditions of the system and avoiding similarity calculation errors caused by indicator differences. The determined threshold balances the similarity of historical data with sample size, effectively filtering out a subset of historical data that highly matches the current operational conditions, eliminating historical data with limited reference value, improving the efficiency and accuracy of subsequent predictive model construction, and allowing the model to better fit the current actual operating state of the system. Simultaneously, Euclidean distance and cosine correlation coefficients are calculated to determine the similarity of operational condition feature vectors from two dimensions. Euclidean distance reflects numerical differences, while cosine correlation coefficients reflect trend similarity. This dual determination makes the selection of historical data more scientific and comprehensive, further ensuring the effectiveness of the historical data subset. Based on high matching... Three online prediction models were constructed using a subset of historical data. The input and output variables of these models correspond to the core operating indicators of the AOA process. A multiple linear regression algorithm was used to achieve a quantitative fit between the input and output variables, allowing the models to accurately reflect the intrinsic relationship between operating parameters and effluent quality, operating energy consumption, and sludge production, ensuring the accuracy and practicality of the prediction results. All three online prediction models use operating parameters as input and work in conjunction with the system's real-time data acquisition module to achieve online prediction. They can quickly output predicted results for effluent quality, operating energy consumption, and sludge production based on the current actual operating parameters, providing a basis for AOA... The real-time operation control, energy consumption management, and sludge treatment of the OA wastewater treatment system provide timely decision-making basis, effectively improving the precision and intelligence of system operation. The construction of three prediction models for effluent quality, operating energy consumption, and sludge production forms a comprehensive operation prediction system, covering the core objectives and key cost indicators of AOA process operation. It realizes all-round prediction from water quality compliance to energy consumption control and sludge management, making system operation control more targeted. It helps to achieve optimized control of energy consumption and sludge production while ensuring effluent quality, thereby improving the economy and stability of process operation.

[0025] In a preferred embodiment of the present invention, step 2 includes: Step 200: The currently collected operating parameters are used as input data to the online prediction models for effluent water quality, operating energy consumption, and sludge production. The online prediction model for effluent water quality performs linear or nonlinear transformations on the input data based on internal regression parameters to obtain the predicted value of effluent water quality. The online prediction model for operating energy consumption performs linear or nonlinear transformations on the input data based on internal regression parameters to obtain the predicted value of operating energy consumption. The online prediction model for sludge production performs linear or nonlinear transformations on the input data based on internal regression parameters to obtain the predicted value of sludge production. Specifically, this includes using the complete set of operating parameters for the anaerobic, aerobic, and anoxic zones collected in real time by the system as unified input data. The operating parameters for the anaerobic zone include dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, and hydraulic retention time. The operating parameters for the aerobic zone include dissolved oxygen concentration, aeration intensity, mixed liquor sludge concentration, and hydraulic retention time. The operating parameters for the anoxic zone include dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, hydraulic retention time, and sludge return flow rate. This set of input data is synchronously input into the online prediction models for effluent water quality, operating energy consumption, and sludge production. The data is fed into the established online prediction models for effluent quality, operational energy consumption, and sludge production. For each effluent quality indicator, the online prediction model performs a linear transformation calculation on the input data based on internally preset regression coefficients and constant terms. The regression coefficients range from -0.8 to 1.5, and the constant terms range from 0.05 to 5.0. This range is the optimal coefficient range obtained by fitting historical AOA process operational data using the least squares method, accurately matching the linear relationship between operating parameters and effluent quality indicators. The system incorporates features to adapt to the process operation patterns under different water quality loads. The calculation process involves multiplying each input operating parameter by the corresponding regression coefficient within the model to obtain the product of each operating parameter. All product results are then summed, and a constant term from within the model is added to the summation result to obtain the predicted value of a single effluent water quality indicator. This calculation process is repeated sequentially to calculate the predicted values ​​of all core water quality indicators, including effluent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, pH value, and water temperature, forming a complete set of predicted effluent water quality values. The online energy consumption prediction model performs linear transformation calculations on the input data based on internally preset regression coefficients and constant terms. The regression coefficients range from 0.1 to 2.0, and the constant terms range from 0.02 to 0.3. This range was determined through correlation analysis between AOA process energy consumption and operating parameters, accurately reflecting the influence of core parameters such as aeration intensity and sludge return flow rate on energy consumption, and closely aligning with the actual variation pattern of process energy consumption. The calculation process involves multiplying each input operating parameter by its corresponding regression coefficient within the model to obtain the product of each operating parameter. All product results are then summed, and the constant term from the model is added to the summation result to obtain the predicted value of system operating energy consumption. This predicted value uses the energy consumption per unit volume of wastewater treated as the measurement dimension, maintaining consistency with the measurement standard used during model training. The online sludge production prediction model performs linear transformation calculations on the input data based on internally preset regression coefficients and constant terms. The regression coefficients range from 0.05 to 1.2, and the constant terms range from 0.01 to 0.5. This range is obtained from the fitting analysis of sludge yield and operating parameters of the AOA process. It can accurately reflect the regulation law of parameters such as mixed liquor sludge concentration and hydraulic retention time on sludge production and match the actual characteristics of process sludge generation. The calculation process is consistent with the two models mentioned above. Each input operating parameter is multiplied by the corresponding regression coefficient in the model to obtain the product of each operating parameter. All product results are summed, and then the constant term in the model is added to the summation result to obtain the predicted value of system sludge production. This predicted value covers the quantitative results corresponding to the remaining sludge discharge and sludge production rate.

[0026] Step 201: Using the predicted values ​​of effluent water quality, operating energy consumption, and sludge production as multiple objectives to be optimized simultaneously, and the dissolved oxygen setpoint in the aerobic zone and the sludge return ratio of the system as decision variables, a multi-objective optimization problem with multiple objective functions is constructed. Based on the historical sequence and real-time collected influent water quality index data, the statistical fluctuation range of the influent water quality index data is calculated, and the upper and lower limits of the operating state parameters within the fluctuation range in the constraints of the multi-objective optimization problem are determined. Specifically, this includes using the predicted values ​​of effluent water quality, operating energy consumption, and sludge production obtained in step 200 as the three core parameters to be optimized simultaneously. The objectives are as follows: For effluent quality optimization, all predicted core water quality indicators must meet wastewater discharge standards and be within their optimal range. The optimal range is defined as follows: effluent chemical oxygen demand ≤30 mg / L, effluent ammonia nitrogen ≤1.5 mg / L, effluent total nitrogen ≤15 mg / L, effluent total phosphorus ≤0.3 mg / L, effluent suspended solids ≤20 mg / L, effluent pH 6.5 to 8.5, and effluent temperature 15 to 35℃. These values ​​adhere to the Class A standard of the Urban Wastewater Treatment Plant Pollutant Discharge Standard, and are determined in conjunction with the water quality suitability range for stable operation of the AOA process. For operational energy consumption optimization, the predicted system operating energy consumption must reach the lowest possible level, defined as ≤0.3 kWh / m³. 3 This value is based on the industry benchmark for energy-saving operation of the AOA process, and is determined in conjunction with the controllable energy consumption range of the actual process operation; the sludge production optimization target is to achieve the minimum predicted sludge production level of the system, with the minimum level set at ≤0.3 kg MLSS / m³. 3 The wastewater parameters are determined based on the operational requirements for sludge reduction in the AOA process, while also matching the reasonable range of sludge generation for pollutant removal in the wastewater. Simultaneously, the dissolved oxygen setpoint in the aerobic zone and the sludge return ratio of the system are defined as control decision variables. The initial dissolved oxygen setpoint in the aerobic zone is set at 1.5 to 2.0 mg / L, representing the optimal dissolved oxygen concentration for nitrification in the aerobic zone of the AOA process. This ensures efficient removal of ammonia nitrogen and total nitrogen while avoiding increased energy consumption due to excessively high dissolved oxygen levels. The initial sludge return ratio of the system is set at 100% to 150%, ensuring the carbon source and sludge quantity requirements for denitrification in the anoxic zone, balancing nitrogen removal efficiency with sludge return energy consumption. Based on these three optimization objectives and two decision variables, a multi-objective optimization problem is constructed, comprising three objective functions. Each objective function is matched with its corresponding optimization objective, and the decision variables serve as the core control factors for the objective functions.

[0027] Historical data and real-time collected data on influent water quality indicators are integrated into an influent water quality statistical dataset. This dataset contains complete data on all core indicators, including chemical oxygen demand (COD), ammonia nitrogen, total nitrogen (TNO), total phosphorus (TP), suspended solids (SSI), pH, and water temperature. For each influent water quality indicator in the dataset, the statistical fluctuation range is calculated. The calculation process is as follows: First, all data for the indicator are summed, and the sum is divided by the total number of data points to obtain the arithmetic mean. Next, the difference between each data point and the arithmetic mean is calculated. All differences are squared and summed to obtain the sum of squared differences. This sum of squared differences is then divided by the total number of data points and the square root is taken to obtain the standard deviation. Finally, the statistical fluctuation range of the indicator is calculated from the arithmetic mean minus the standard deviation to the arithmetic mean plus the standard deviation. Based on the statistical fluctuation range of various influent water quality indicators and combined with the operational characteristics of the AOA wastewater treatment process, the upper and lower limits of the operational state parameters in the constraints of the multi-objective optimization problem are determined. These operational state parameters include dissolved oxygen concentration, oxidation-reduction potential, mixed liquor sludge concentration, hydraulic retention time, aeration intensity, and sludge return flow rate in the anaerobic, aerobic, and anoxic zones. The lower limit of the process operating value for each operational state parameter is the process operating value adapted to the lower limit of the statistical fluctuation range of the corresponding influent water quality indicator, while the upper limit is the process operating value adapted to the upper limit of the statistical fluctuation range of the corresponding influent water quality indicator. Specifically, the process operating value for dissolved oxygen concentration in the anaerobic zone is set to 0 to 0.2 mg / L to ensure an anaerobic environment for phosphorus release reaction and avoid excessive dissolved oxygen inhibiting polyphosphate accumulation (PAC) activity. The process operating value for dissolved oxygen concentration in the aerobic zone is set to 0.5 to 3.0 mg / L to meet the requirements of nitrification in the aerobic zone. The dissolved oxygen requirements for the denitrification and phosphorus uptake reactions cover the concentration range for stable process operation. The dissolved oxygen concentration in the anoxic zone is set at 0 to 0.5 mg / L to ensure a low-oxygen environment for denitrification while maintaining nitrogen removal efficiency. The oxidation-reduction potential (ORP) values ​​for the anaerobic and anoxic zones are set at -300 to -100 mV and -200 to 0 mV, respectively, to match the biological reaction environment of the corresponding zones and reflect the redox state of the water. The mixed liquor sludge concentration is set at 2000 to 4000 mg / L to ensure sludge loading and treatment efficiency, avoiding excessively low concentrations leading to decreased treatment effectiveness or excessively high concentrations leading to sludge bulking. The hydraulic retention time (HRT) values ​​are 1 to 2 hours for the anaerobic zone, 6 to 8 hours for the aerobic zone, and 2 to 4 hours for the anoxic zone, matching the nitrogen and phosphorus removal reaction cycle of the AOA process to ensure sufficient removal of all pollutants. The aeration intensity values ​​are set at 2 to 5 m... 3 / (m 2•h) to meet the dissolved oxygen supply needs of the aerobic zone, while taking into account both aeration effect and energy consumption control; the sludge return flow rate is set to a value of Q to 2Q (Q is the influent flow rate) to ensure the adjustment range of the sludge return ratio and match the sludge quantity requirements for nitrogen and phosphorus removal; thereby achieving precise definition of the upper and lower limits of the operating parameters in the constraints.

[0028] Step 202: Based on the upper and lower limits of the values, and combined with process safety operation restrictions, deduce the allowable adjustment range of the dissolved oxygen setpoint as the feasible range of values ​​for the dissolved oxygen setpoint, and deduce the allowable adjustment range of the sludge return ratio as the feasible range of values ​​for the sludge return ratio. Specifically, this includes: based on the upper and lower limits of the operating status parameters determined in step 201, deriving the preliminary adjustment ranges of the dissolved oxygen setpoint in the aerobic zone and the system sludge return ratio, respectively. The preliminary lower limit of the dissolved oxygen setpoint in the aerobic zone is the process operating parameter of the dissolved oxygen concentration in the aerobic zone. The lower limit of the setpoint corresponds to the initial adjustment upper limit, which is the setpoint corresponding to the upper limit of the dissolved oxygen concentration in the aerobic zone of the operating parameters. The setpoint for dissolved oxygen in the aerobic zone is 0.5 to 3.0 mg / L. The initial adjustment lower limit of the sludge return ratio is the return ratio corresponding to the lower limit of the sludge return flow rate in the operating parameters. The initial adjustment upper limit is the return ratio corresponding to the upper limit of the sludge return flow rate in the operating parameters. The setpoint for the sludge return ratio is 50% to 200%, matching the range of the sludge return flow rate in the operating parameters, and covering the reasonable range of process control.

[0029] Based on the safe operating limits of the AOA wastewater treatment process, specifically the process safe operating limits for dissolved oxygen setpoints in the aerobic zone (not less than 0.5 mg / L and not more than 3.0 mg / L) and sludge return ratio (not less than 50% and not more than 200%), a revised range of initial adjustment values ​​is derived. This involves comparing the initial lower limit of the dissolved oxygen setpoint in the aerobic zone with the lower limit of the process safe operating limit, taking the larger value as the lower limit of the feasible range for the dissolved oxygen setpoint. The initial upper limit of the dissolved oxygen setpoint in the aerobic zone is then compared with the upper limit of the process safe operating limit. The smaller of the two values ​​is taken as the upper limit of the feasible range of dissolved oxygen setpoint. The interval between the two values ​​is the feasible range of adjustable dissolved oxygen setpoint in the aerobic zone. Using the same deduction method, the initial adjustment lower limit of sludge return ratio is compared with the lower limit of process safety operation. The larger of the two values ​​is taken as the lower limit of the feasible range of sludge return ratio. The initial adjustment upper limit of sludge return ratio is compared with the upper limit of process safety operation. The smaller of the two values ​​is taken as the upper limit of the feasible range of sludge return ratio. The interval between the two values ​​is the feasible range of adjustable sludge return ratio in the system.

[0030] This embodiment transforms the linear transformation of the model into a concrete calculation process involving multiplication, summation, and addition of constant terms. This deeply aligns with the core algorithm of the multiple linear regression model, making the prediction calculation process more concrete and avoiding calculation biases caused by abstract transformations. It also maintains the consistency of input data, ensuring that the calculation benchmarks of the three models' predictions are consistent, thus improving the accuracy and comparability of the prediction results. The constructed multi-objective optimization problem simultaneously considers three core process objectives: effluent quality, operating energy consumption, and sludge production. The decision variables focus on the two core control parameters of the AOA process: dissolved oxygen setpoint in the aerobic zone and sludge return ratio. This ensures that the optimization direction is more aligned with the actual process control needs, avoiding the problem of optimization objectives being disconnected from actual operation. The calculation of the statistical fluctuation range of influent water quality combines historical sequence data and real-time collected data, taking into account both the historical patterns of process operation and the current actual situation of influent water quality. Furthermore, the fluctuation range is defined through quantitative calculations of the arithmetic mean and standard deviation, making the results more scientific and accurate. By determining the upper and lower limits of operating parameters based on the fluctuation range of influent water quality, the constraints of the multi-objective optimization problem are adapted to the actual changes in influent water quality. This avoids the failure of optimization results caused by the mismatch between fixed constraints and influent water quality fluctuations, thus improving the practicality of the optimization problem. First, the initial adjustment range of decision variables is derived based on influent water quality fluctuations, and then corrected by combining process safety operation restrictions. This achieves a dual consideration of water quality adaptability and process safety, ensuring that the adjustment of decision variables conforms to the current fluctuation characteristics of influent water quality while avoiding process operation failures and effluent water quality exceeding standards caused by control parameters exceeding safety thresholds, thus guaranteeing the stability of process operation. The derivation of the feasible value range of decision variables adopts a quantitative method of taking the intersection, making the derivation process clearer and avoiding invalid parameter values ​​during the optimization solution process, thereby improving the efficiency and effectiveness of the optimization solution.

[0031] In a preferred embodiment of the present invention, step 3 includes: Step 300: Within the feasible range of dissolved oxygen setpoint and sludge return ratio, analyze the pollutant concentration change rates at three characteristic locations—the anaerobic zone inlet, the aerobic zone end, and the anoxic zone outlet—obtained through monitoring. Calculate the high-frequency distribution intervals of the pairwise ratios of the pollutant concentration change rates at these three characteristic locations from historical operating data, and take the median of these high-frequency distribution intervals to form a set of coordinate points as boundary points characterizing the ideal carbon source dynamics of the system. Connect these boundary points sequentially to form a closed polygonal region as a virtual equilibrium domain characterizing the internal carbon source dynamic balance of the system, specifically including... Within the feasible range of the dissolved oxygen setpoint in the aerobic zone and the sludge return ratio determined in step 202, real-time pollutant concentration data at three characteristic locations—the inlet of the anaerobic zone, the end of the aerobic zone, and the outlet of the anoxic zone—of the AOA wastewater treatment system are continuously acquired using online monitoring equipment. The pollutants include chemical oxygen demand (COD), ammonia nitrogen, total nitrogen (TNO), and total phosphorus (TP). For each pollutant at each characteristic location, the rate of change of concentration is calculated. The calculation process involves selecting pollutant concentration data from two consecutive monitoring time points, subtracting the pollutant concentration from the previous time point from the concentration at the later time point, and obtaining the concentration difference. Dividing the concentration difference by the time interval between the two monitoring points yields the concentration change rate of the pollutant at that characteristic location. This method is used to calculate the concentration change rates of all pollutants at the three characteristic locations. The complete data on the pollutant concentration change rates at the three characteristic locations is retrieved from the system's historical operating data. For each pollutant—chemical oxygen demand (COD), ammonia nitrogen, total nitrogen, and total phosphorus—pairwise ratios of the concentration change rates at the three characteristic locations are calculated: the ratio of the concentration change rate at the anaerobic zone inlet to the aerobic zone outlet, the ratio of the concentration change rate at the anaerobic zone inlet to the anoxic zone outlet, and the ratio of the concentration change rate at the aerobic zone outlet. The ratio of the concentration change rate at the end of the oxygen zone to that at the outlet of the anoxic zone is calculated by dividing the concentration change rate of the former by the concentration change rate of the latter. For each pollutant, the frequency of each of the three pairs of ratios is statistically analyzed. The intervals with a frequency of occurrence of each ratio value of at least 80% are selected as high-frequency distribution intervals. The median of each high-frequency distribution interval is then calculated by adding the lower limit and upper limit of the interval value, dividing the sum by 2, and obtaining the median of that interval. The median of the high-frequency distribution intervals for each pollutant's three pairs of ratios is used as a set of coordinate points. The numerical dimension of the coordinate points is consistent with the number of pairs of ratios. The coordinate points corresponding to all pollutants together constitute a set of boundary points characterizing the ideal carbon source dynamics of the system. All boundary points in this set are connected sequentially according to pollutant type and concentration change rate ratio type. The closed polygonal region formed by these connections is used as a virtual equilibrium domain. The boundary of this virtual equilibrium domain is defined by the values ​​of all boundary points and is used to characterize the ideal state interval of the carbon source dynamic balance within the AOA wastewater treatment system.

[0032] Step 301: Calculate the current pollutant conversion efficiency based on real-time collected operating parameters, and calculate the current carbon source consumption rate based on the pollutant concentration change rates at three characteristic locations obtained through synchronous monitoring; using the current pollutant conversion efficiency and the current carbon source consumption rate as coordinate values, determine a unique coordinate point in the state space, defined as the current system point; calculate the shortest Euclidean distance between the current system point and all points on the boundary of the virtual equilibrium domain, and use the value of the shortest Euclidean distance as the dynamic equilibrium deviation value, specifically including: based on the complete set of operating parameters of the anaerobic, aerobic, and anoxic zones collected in real-time by the system, combined with the online effluent water quality... The prediction model's results are used to calculate the conversion efficiency of each pollutant. The calculation process involves subtracting the pollutant concentration at the effluent from the pollutant concentration at the influent to obtain the pollutant removal concentration. This removal concentration is then divided by the influent pollutant concentration, and multiplied by 100% to obtain the conversion efficiency. This method is used to calculate the conversion efficiencies of chemical oxygen demand (COD), ammonia nitrogen, total nitrogen (TNO), and total phosphorus (TP). The arithmetic mean of all pollutant conversion efficiencies is taken as the current pollutant conversion efficiency. This is calculated by summing the conversion efficiencies of all pollutants, dividing the sum by the number of pollutant types, and obtaining the current pollutant conversion efficiency. Based on the pollutant concentration change rates at three characteristic locations—the anaerobic zone inlet, the aerobic zone end, and the anoxic zone outlet—obtained through synchronous monitoring, the current carbon source consumption rate is calculated. COD is used as the core indicator for carbon source consumption. The calculation process involves summing the COD concentration change rates at the three characteristic locations, obtaining the concentration change rate sum. The absolute value of this sum is the current carbon source consumption rate, reflecting the real-time carbon source consumption intensity of the system.A two-dimensional state space is constructed with pollutant conversion efficiency as the x-axis and carbon source consumption rate as the y-axis. The calculated current pollutant conversion efficiency is used as the x-axis value and the current carbon source consumption rate as the y-axis value. A unique coordinate point is determined in this two-dimensional state space, which is defined as the current system point. This coordinate point intuitively represents the real-time carbon source utilization and pollutant conversion state of the system. The Euclidean distance between the current system point and all points on the boundary of the virtual equilibrium domain constructed in step 300 is calculated. For each point on the boundary, its x-axis value is compared with the x-axis value of the current system point, and its y-axis value is compared with the y-axis value of the current system point. The differences between the x-axis and y-axis values ​​are calculated. The calculation process is as follows. To calculate the Euclidean distance between the current system point and the boundary point, the coordinates of the boundary point are subtracted from the corresponding coordinates of the current system point. Then, the differences in the horizontal and vertical coordinates are squared respectively. The squares of the two squares are added together to obtain the sum of squares. The square root of the sum of squares is then taken to obtain the Euclidean distance between the current system point and the boundary point. After traversing all points on the boundary of the virtual equilibrium domain and calculating all Euclidean distances, the Euclidean distance with the smallest value is selected as the shortest Euclidean distance. The value of the shortest Euclidean distance is the deviation value of the dynamic equilibrium of the carbon source in the system. The larger the deviation value, the greater the deviation of the current state of the system from the ideal state of the dynamic equilibrium of the carbon source, and vice versa.

[0033] This embodiment focuses on three core characteristic locations of the AOA process: the inlet of the anaerobic zone, the end of the aerobic zone, and the outlet of the anoxic zone. These locations reflect the pollutant conversion and carbon source utilization status at each reaction stage of the process. Based on this, the concentration change rate is calculated, making data acquisition and calculation more closely reflect the actual dynamic changes in carbon sources. The high-frequency distribution range of pairwise ratios of concentration change rates is calculated based on historical operating data. This approach considers both the historical patterns of process operation and identifies typical ratio ranges under dynamic carbon source equilibrium. Using the median as the boundary point makes the construction of the virtual equilibrium domain more representative and accurately characterizes the ideal dynamic carbon source equilibrium state of the system. The virtual equilibrium domain is presented as a closed polygonal region, transforming the abstract dynamic carbon source equilibrium into a concrete spatial region, making the definition of the system's carbon source equilibrium state more intuitive. The calculation of pollutant conversion efficiency combines real-time operating parameters and effluent water... The quality prediction results, taking the average of the conversion efficiencies of all pollutants, comprehensively reflect the overall pollutant treatment capacity of the system, avoiding the one-sidedness of a single pollutant indicator. The carbon source consumption rate, based on the core carbon source indicator chemical oxygen demand (COD) and calculated in conjunction with the concentration change rates at three characteristic locations, accurately reflects the real-time carbon source consumption status of the system, making the system state more comprehensive. By constructing a state space using pollutant conversion efficiency and carbon source consumption rate as coordinate values, the real-time operating state of the system is transformed into a unique current system point, realizing the quantitative and spatial representation of the process operating state, making the comparative analysis of the system state more operable. The deviation between the current system point and the boundary of the virtual equilibrium domain is calculated using Euclidean distance. The calculation process is quantitative and accurate, objectively reflecting the magnitude of the deviation between the current system state and the ideal carbon source equilibrium state, avoiding errors from subjective experience judgment.

[0034] In a preferred embodiment of the present invention, step 4 includes: Step 400: Based on the pollutant concentration change rates at the anaerobic zone inlet and the anoxic zone outlet, calculate the ratio of the pollutant concentration change rate at the anaerobic zone inlet to that at the anoxic zone outlet, as the carbon source migration and conversion rate ratio. Then, weighted and fused with the dynamic equilibrium deviation value and the carbon source migration and conversion rate ratio, obtain a comprehensive characterization of the coordination of carbon source migration and conversion equilibrium from storage to consumption within the system. Specifically, this includes: calculating the carbon source migration and conversion rate ratio based on pollutant concentration change rate data obtained from online monitoring at the anaerobic zone inlet and the anoxic zone outlet, where chemical oxygen demand (COD) is the core indicator of carbon source. The calculation process involves dividing the rate of change of chemical oxygen demand (COD) concentration at the inlet of the anaerobic zone by the rate of change of COD concentration at the outlet of the anoxic zone. The result is the carbon source migration and conversion rate ratio. This ratio directly reflects the rate matching characteristics of carbon source migration and conversion from the anaerobic zone to the anoxic zone within the system. When the ratio approaches 1, it indicates that the carbon source release rate in the anaerobic zone and the carbon source consumption rate in the anoxic zone are relatively well matched, and the coordination of carbon source migration and conversion is good. The greater the deviation of the ratio from 1, the worse the rate matching between carbon source release in the anaerobic zone and carbon source consumption in the anoxic zone, and the more likely the carbon source migration and conversion process will result in carbon source excess or carbon source deficiency. The dynamic equilibrium deviation value calculated in step 301 is weighted and fused with the carbon source migration and conversion rate ratio calculated in this step to obtain the migration and conversion balance. This index comprehensively characterizes the degree of coordination of the carbon source from storage to consumption within the system. The weighted fusion calculation process is as follows: First, the weight coefficient of the dynamic equilibrium deviation value is set to 0.6, and the weight coefficient of the carbon source migration and conversion rate ratio is set to 0.4. The dynamic equilibrium deviation value directly reflects the degree of deviation between the overall dynamic and ideal equilibrium state of the carbon source and is the core evaluation index of the carbon source coordination state, so it is given a higher weight. The carbon source migration and conversion rate ratio focuses on the rate matching of carbon source cross-regional migration and is an auxiliary evaluation index, so it is given a relatively lower weight. The two weight coefficients are between 0 and 1 and their sum is 1. Then, the dynamic equilibrium deviation value is multiplied by its corresponding weight coefficient of 0.6 to obtain the weighted result of the dynamic equilibrium deviation value. The carbon source migration and conversion rate ratio is multiplied by its corresponding weight coefficient of 0.4 to obtain the weighted result of the carbon source migration and conversion rate ratio. Finally, the two weighted results are added together, and the sum is the migration and conversion balance.

[0035] Step 401: Based on the magnitude and direction of the dynamic equilibrium deviation, the virtual equilibrium domain is geometrically divided to form at least one sub-region centered on the current system point. This sub-region is then combined with the undivided portion of the virtual equilibrium domain to define a correction region for guiding the optimization search. Specifically, this includes: determining the magnitude and direction of the dynamic equilibrium deviation calculated in Step 301. The deviation direction refers to the spatial orientation of the current system point relative to the virtual equilibrium domain, specifically categorized into five spatial orientations: positive x-axis, negative x-axis, positive y-axis, negative y-axis, and oblique y-axis directions. The deviation magnitude is further divided into low deviation (≤0.2) and medium deviation (<0.2). The virtual equilibrium domain constructed in step 300 is geometrically segmented into three levels: value ≤ 0.5, high deviation (value > 0.5). The segmentation process takes the current system point as the core reference. First, the segmentation direction is determined according to the deviation direction. The area in the virtual equilibrium domain that is consistent with the deviation direction of the current system point is defined as the basic segmentation area. Then, the coverage range of the sub-region is determined according to the magnitude of the deviation value. The coverage range of the sub-region is 30% of the basic segmentation area when the deviation level is low, 50% when the deviation level is medium, and 80% when the deviation level is high. Finally, at least one sub-region is divided with the current system point as the geometric center and the coverage range and deviation degree are adapted. The segmented sub-regions are spatially combined with the remaining unsegmented regions in the virtual equilibrium domain, preserving the boundary integrity of each region and performing closure processing to form a new closed spatial region. This region is defined as the correction region, which serves as the core guiding region for the optimization search of decision variables in subsequent multi-objective optimization problems. The sub-regions are the core focusing regions for optimization search, and the remaining regions in the virtual equilibrium domain are auxiliary exploration regions. The combination of the two is used to narrow the optimization search range, improve search targeting, and avoid search blind spots caused by over-focusing on local regions.

[0036] Step 402a involves weighted summation of the dynamic equilibrium deviation and the migration-conversion equilibrium to calculate an adaptive coefficient. A rectangular region is then defined using the geometric center of the correction region as a reference and its boundary as the initial search space for the two decision variables in the multi-objective optimization problem: dissolved oxygen setpoint and sludge return ratio. Specifically, this includes: weighted summation of the dynamic equilibrium deviation obtained in step 301 and the migration-conversion equilibrium obtained in step 400 to calculate the adaptive coefficient. This coefficient is used to dynamically adjust the step size of the optimization search. The calculation process involves first setting the weighting coefficient of the dynamic equilibrium deviation to 0.7, and the weighting coefficient of the migration-conversion equilibrium to... The adaptive coefficient, with a weight of 0.3, plays a crucial role in matching the degree of deviation of the carbon source dynamics to adjust the search step size. The dynamic equilibrium deviation value is a direct quantitative indicator of the degree of deviation and has a more significant impact on step size adjustment, hence it is assigned a higher weighting coefficient. The migration and transformation equilibrium comprehensively reflects the coordination state of the carbon source and provides an auxiliary reference for step size adjustment, hence it is assigned a lower weighting coefficient. Both weighting coefficients are constants greater than 0. The dynamic equilibrium deviation value is then multiplied by its corresponding weighting coefficient of 0.7 to obtain the weighted term of the dynamic equilibrium deviation value. The migration and transformation equilibrium is then multiplied by its corresponding weighting coefficient of 0.3 to obtain the weighted term of the migration and transformation equilibrium. Finally, the two weighted terms are added together, and the sum is the adaptive coefficient. The geometric center of the modified region constructed in step 401 is determined by calculating the arithmetic mean of the abscissas of all boundary points of the modified region to obtain the abscissa of the geometric center, and the arithmetic mean of the ordinates of all boundary points of the modified region to obtain the ordinate of the geometric center. The combination of the abscissas and ordinates is the geometric center of the modified region. This geometric center is used as the center reference point of the rectangular region. The maximum lateral boundary difference of the modified region is used as the lateral side length of the rectangular region. The maximum lateral boundary difference is the maximum abscissa value minus the minimum abscissa value of the modified region. The maximum longitudinal boundary difference of the modified region is used as the longitudinal side length of the rectangular region. The maximum longitudinal boundary difference is the maximum ordinate value minus the minimum ordinate value of the modified region. A rectangular region adapted to the space of the modified region is constructed. This rectangular region is directly used as the initial search space for the two decision variables, dissolved oxygen setpoint in the aerobic zone and sludge return ratio in the system, in the multi-objective optimization problem. The initial search space defines the initial value range of the two decision variables for optimization search.

[0037] Step 402b involves calculating the distance and correlation metrics between the current system point and the historical operating condition feature vectors stored in the historical database. Based on the calculation results, a set of historical optimization solutions is selected from the historical optimization information, and a portion of the solutions from the historical optimization solution set is introduced into the initial population. Specifically, this includes: extracting the operating condition feature vector corresponding to the current system point, which contains core operating condition indicators such as the current pollutant conversion efficiency, carbon source consumption rate, dynamic equilibrium deviation, carbon source migration-conversion rate ratio, and migration-conversion equilibrium; simultaneously retrieving the feature vectors of all historical operating conditions stored in the historical database. The indicator dimensions of each historical operating condition feature vector are consistent with those of the current operating condition feature vector, and all have undergone standardization processing to eliminate index... Scalar differences; calculate the spatial distance between the current operating condition feature vector and each historical operating condition feature vector, and then obtain the correlation measure between the two by calculating the correlation coefficient between the vectors. The correlation coefficient is calculated by dividing the covariance of each index by the product of the standard deviation of the current operating condition feature vector and the standard deviation of the corresponding historical operating condition feature vector. The result is the correlation measure value. Based on the calculation results of spatial distance and correlation measure, select the optimization solutions corresponding to several historical operating conditions with the smallest spatial distance and a correlation measure value ≥ 0.8 to form a historical optimization solution set. Randomly select 30% of the effective solutions from this historical optimization solution set and directly introduce them into the initial population of multi-objective optimization as the core solution unit of the initial population.

[0038] Step 402c involves employing an optimization search framework based on population iteration and survival of the fittest. An adaptive coefficient is used to dynamically adjust the search step size for generating new solutions in each iteration. The boundary range of the correction region is used as a hard constraint on the decision variable values ​​when generating new solutions. The spatial distribution pattern of solutions in the historical optimization solution set guides the direction of new solution generation. Under the premise of satisfying all constraints in the constructed multi-objective optimization problem, the process of population iteration and survival of the fittest is repeated until the preset number of iterations is reached. Finally, a set of non-dominated solutions is obtained, consisting of multiple solutions that are non-dominant to each other on the three objectives of predicted effluent quality, predicted operating energy consumption, and predicted sludge production. Specifically, this includes: conducting a multi-objective optimization search using an optimization search framework based on population iteration and survival of the fittest, with a preset number of iterations of 200; firstly, the adaptive coefficient obtained in step 402a is used as the search step size... Based on this, the search step size for generating new solutions in each population iteration is dynamically adjusted. The larger the adaptive coefficient, the smaller the search step size, and vice versa. Specifically, the actual search step size is obtained by multiplying the base step size by the reciprocal of the adaptive coefficient, thus achieving dynamic adaptation of the search step size. This allows for fine-grained searching with small steps when the system deviates significantly, and rapid exploration with large steps when the deviation is small. The boundary range of the correction region constructed in step 401 is used as a hard constraint condition when generating new solutions. When generating new solutions for the dissolved oxygen setpoint and sludge return ratio in the aerobic zone, it is first determined whether the value of the solution is within the value range of the decision variables corresponding to the correction region. If it exceeds the range, the solution is directly discarded. This ensures that the values ​​of the two decision variables in the generated new solution are within the boundary range of the correction region, while also adhering to all operational constraints constructed in the multi-objective optimization problem, and solutions that exceed the constraint range are not allowed.

[0039] Based on the historical optimized solution set selected in step 402b, the specific values ​​of the two decision variables, namely the dissolved oxygen setpoint in the aerobic zone and the sludge return ratio in the system, are extracted for all solutions in the solution set, forming a decision variable value dataset. A two-dimensional statistical analysis is carried out on this dataset. First, the distribution of the values ​​of the two decision variables is fitted with density using the kernel density estimation method. The preset threshold for kernel density fitting is 0.7. This threshold has been statistically verified by the historical optimized solution set. It can accurately screen out the core areas with high solution aggregation density, avoiding the narrow range of high aggregation areas and the omission of high-quality solutions due to an excessively high threshold. It can also prevent the high aggregation area from being too wide due to an excessively low threshold, thus losing the aggregation guidance significance. The areas with fitted kernel density values ​​higher than the preset threshold of 0.7 are clearly defined as high aggregation areas where the solution is within the range of decision variable values. Further analysis was conducted using linear fitting and trend testing to explore the correlation between the values ​​of the two decision variables and the changes in process operating status. Specifically, the analysis focused on two core indicators in the process operating status: carbon source consumption rate and pollutant conversion efficiency. The baseline values ​​for carbon source consumption rate and pollutant conversion efficiency were set at 0.8 mg / (L·h) and 85%, respectively. Both baseline values ​​were determined by statistically averaging the historical operating conditions of the AOA wastewater treatment system under long-term stable operation, closely reflecting the actual carbon source utilization and pollutant treatment capacity of the process, and serving as core reference values ​​for determining the process operating status. The analysis also examined the linkage between changes in the dissolved oxygen setpoint in the aerobic zone, the system sludge return ratio, and these two process status indicators. The analysis revealed that when the carbon source consumption rate was higher than the process operating baseline value and the pollutant conversion efficiency was lower than the process operating baseline value, the correlation between these two indicators was broken. When the target value is met, the dissolved oxygen setpoint in the aerobic zone needs to be increased by 0.2 to 0.3 mg / L, and the sludge return ratio needs to be reduced by 10% to 15% to match the carbon source consumption rate. When the carbon source consumption rate is lower than the process operating baseline value and the pollutant conversion efficiency is higher than the process operating baseline value, the dissolved oxygen setpoint in the aerobic zone can be reduced by 0.2 to 0.3 mg / L, and the sludge return ratio needs to be increased by 10% to 15% to balance treatment effect and energy consumption control. Through the trend test of the linear fitting of this linkage law (the significance level is set at 0.05 to ensure the reliability of the trend), the optimal distribution trend of the solution in the two-dimensional space of decision variables is finally obtained. This trend clarifies the optimal value combination direction of the two decision variables under different process operating conditions, and intuitively presents the optimal linkage change trajectory of dissolved oxygen setpoint and sludge return ratio.Using this highly clustered region and the optimal distribution trend as a guide for generating new solutions, in the new solution generation stage, the probability of generating new solutions in the highly clustered region is increased to 1.5 to 2 times the preset base probability by adjusting the random generation probability, so that the proportion of new solutions generated in this region is significantly increased. At the same time, with the optimal distribution trend as the extension guide, the range of values ​​for generating new solutions is expanded along the positive and negative directions of the trend line, and the expansion range is strictly controlled within the boundary of the correction region, ensuring that the overall generation of new solutions is tilted towards the clustered region of historical optimized solutions. This approach not only relies on historical optimization experience to lock in the potential generation range of high-quality solutions, but also explores appropriately along the optimal trend, effectively improving the generation effectiveness and target performance quality of new solutions.

[0040] Under the premise of satisfying all constraints, a population iteration and survival-of-the-fittest process is executed. In each iteration, based on the dynamically adjusted actual search step size and the historical solution distribution pattern, new solutions are generated through crossover and mutation. The number of new solutions is 50% of the current population size. Then, the new solutions are merged with the solutions in the current population to obtain a merged solution set. Subsequently, the optimization requirements of three objectives are met: predicted effluent quality, predicted operating energy consumption, and predicted sludge production. Specifically, the predicted effluent quality must meet the preset standard of Class A discharge, namely, chemical oxygen demand ≤ 50 mg / L, ammonia nitrogen ≤ 5 mg / L, total nitrogen ≤ 15 mg / L, and total phosphorus ≤ 0.5 mg / L; the predicted operating energy consumption must approach the theoretical minimum energy consumption value, which is taken as 0.6 kWh / m³. 3 The energy consumption should align with the theoretical lower limit of long-term operational statistics for small and medium-sized AOA wastewater treatment systems, balancing treatment effectiveness with reasonable energy consumption. The predicted sludge production value should approach the theoretical minimum production value, which is set at 0.3 kgDS / m³. 3 Based on the biodegradation characteristics of the AOA process and the reasonable lower limit calculated using sludge reduction theory (considering all three factors synergistically without favoring a single objective), the merits of all solutions in the merged solution set are evaluated. Invalid solutions that do not meet the operational constraints are first eliminated. Then, the valid solutions are ranked according to their target performance. Solutions with the same initial population size are retained in descending order of performance to form a new population. This population iteration and selection process is repeated until the population iteration reaches a preset 200 times, at which point iteration stops. Finally, the final population is selected from the remaining solutions. The non-dominated sorting method is used to screen out all solutions that are not mutually dominant on the three optimization objectives. That is, there is no solution whose performance is better than another solution on all objectives. The set of these non-dominated solutions is called the non-dominated solution set. This solution set contains multiple sets of decision variable values ​​that are suitable for the operation of the AOA wastewater treatment system (i.e., dissolved oxygen setpoint in the aerobic zone and sludge return ratio in the system). Each set of values ​​can achieve a synergistic balance among the three optimization objectives of effluent quality, operating energy consumption and sludge production, and can serve as a multi-objective optimal decision reference for process control.

[0041] This embodiment calculates the carbon source migration-conversion rate ratio using chemical oxygen demand (COD) as the core indicator of carbon source. It then combines this with dynamic equilibrium deviation values ​​for weighted fusion to obtain the migration-conversion equilibrium degree. This achieves a comprehensive quantification of carbon source migration-conversion rate and dynamic equilibrium state, characterizing the overall coordination of carbon source storage and consumption within the system. Based on the magnitude and direction of the dynamic equilibrium deviation value, the virtual equilibrium domain is geometrically segmented, and a correction region is constructed. This ensures the optimization search area matches the current dynamic deviation state of the carbon source in the system, effectively narrowing the scope of subsequent optimization searches, avoiding meaningless global searches, and improving the targeting and efficiency of optimization searches. The dynamic equilibrium deviation value and the migration-conversion equilibrium degree are weighted and summed to obtain an adaptive coefficient, achieving dynamic coefficient generation based on the real-time state of the system. Simultaneously, the correction region... The initial search space for decision variables is determined based on the system's actual operating state. Historical optimization solutions are selected by calculating the distance and correlation metrics of the operating condition feature vectors, and some solutions are introduced into the initial population. This fully utilizes the optimization experience from historical process operation, avoiding the inefficiency of starting with random solutions and improving the initial quality and convergence speed of the population. An optimization framework of population iteration and survival of the fittest is adopted, with adaptive coefficients dynamically adjusting the search step size. This achieves dynamic matching between the search step size and the real-time system state; when the deviation is large, the step size is reduced for refined searching, and when the deviation is small, the step size is increased for rapid exploration. Simultaneously, the correction region is used as a hard constraint, and the distribution pattern of historical solutions guides the direction of new solution generation, effectively improving the efficiency and accuracy of the optimization search.

[0042] In a preferred embodiment of the present invention, step 5 includes: Step 500: For each solution in the non-dominated solution set, the online effluent quality prediction model is used to predict the detailed effluent quality indicators corresponding to the solution in the non-dominated solution set. Solutions whose predicted effluent quality indicators all meet the preset discharge standards are selected to form a candidate solution set. Specifically, this includes: First, retrieving the non-dominated solution set obtained in step 402c. This solution set contains multiple solutions that are not mutually dominant on the three objectives of predicted effluent quality, predicted operating energy consumption, and predicted sludge production. Each solution corresponds to a set of specific values ​​for dissolved oxygen setpoints in the aerobic zone and the sludge return ratio of the system. For each solution in the non-dominated solution set, it is input into the trained and validated online effluent quality prediction model. The model is used to predict the detailed effluent quality indicators corresponding to each solution. The detailed effluent quality indicators are consistent with the effluent quality indicators of the optimization objectives in step 402c, specifically including four core indicators: chemical oxygen demand, ammonia nitrogen, total nitrogen, and total phosphorus, to ensure the comprehensiveness and relevance of the predicted indicators. The prediction process strictly follows the input-output logic preset by the model to ensure the accuracy of the prediction results. A pre-defined emission standard is established, which is the Class A compliance standard preset for the AOA wastewater treatment system. The specific preset emission limits are: Chemical Oxygen Demand (COD) ≤ 50 mg / L, Ammonia Nitrogen (ANOV) ≤ 5 mg / L, Total Nitrogen (TN) ≤ 15 mg / L, and Total Phosphorus (TP) ≤ 0.5 mg / L. This standard aligns with the design treatment capacity of the AOA wastewater treatment system, balancing effluent quality compliance with process operational feasibility, and represents the core water quality baseline that the system must achieve. For each solution, the four predicted effluent water quality indicators are individually checked and screened. The screening criteria are: the predicted values ​​for COD, ANOV, TN, and TP for each solution must be less than or equal to the corresponding preset emission limits; none can be omitted. If any predicted effluent water quality indicator in a solution exceeds the corresponding limit, the solution is discarded. If all predicted effluent water quality indicators meet the preset emission standard, the solution is retained. All retained solutions are integrated and summarized to form a candidate solution set. All solutions in this solution set meet the effluent quality standards, while also taking into account the optimization potential of operating energy consumption and sludge production.

[0043] Step 501: Perform a multi-attribute comprehensive evaluation and ranking of the predicted values ​​of effluent quality, operating energy consumption, and sludge production for each solution in the candidate solution set. Select the solution ranked first in the comprehensive evaluation as the final optimization scheme. Specifically, this includes: retrieving the candidate solution set obtained in step 500, identifying the three core evaluation indicators corresponding to each solution in the candidate solution set, namely, the predicted value of effluent quality, the predicted value of operating energy consumption, and the predicted value of sludge production. The predicted value of effluent quality is based on the average of the predicted values ​​of the four core indicators (chemical oxygen demand, ammonia nitrogen, total nitrogen, and total phosphorus) as the comprehensive characterization value. The predicted value of operating energy consumption is based on the energy consumption per unit of treated water volume (kWh / The sludge production forecast is expressed as the sludge production per unit volume of treated water (kgDS / kg) as the characterization value. The three core evaluation indicators are used as the representation value. A multi-attribute comprehensive evaluation and ranking is carried out on the three core evaluation indicators. First, the weight coefficients of each evaluation indicator are set. The weight coefficients are determined in combination with the operation priority of the AOA wastewater treatment system. Specifically, the weight coefficients are: 0.5 for the comprehensive prediction value of effluent water quality, 0.3 for the prediction value of operating energy consumption, and 0.2 for the prediction value of sludge production. Among them, the effluent water quality compliance is the core bottom line of process operation and should be given the highest weight. Operating energy consumption and sludge production are the key to energy saving, consumption reduction and operation cost reduction, and are given corresponding weights respectively. The sum of the three weight coefficients is 1, and all are between 0 and 1. For each solution, the three core evaluation indicators are standardized to eliminate the impact of differences in the dimensions of different indicators. The standardization process is as follows: For the comprehensive predicted value of effluent water quality (the smaller the better), the standardized value = (maximum value of indicator - value of indicator in this solution) ÷ (maximum value of indicator - minimum value of indicator); For the predicted value of operating energy consumption (the smaller the better), the standardized value = (maximum value of indicator - value of indicator in this solution) ÷ (maximum value of indicator - minimum value of indicator); For the predicted value of sludge production (the smaller the better), the standardized value = (maximum value of indicator - value of indicator in this solution) ÷ (maximum value of indicator - minimum value of indicator). Through this calculation process, the values ​​of all indicators are standardized to between 0 and 1. The closer the standardized value is to 1, the better the performance of that indicator. The comprehensive evaluation score for each solution is calculated as follows: Comprehensive evaluation score = (standardized value of comprehensive predicted effluent water quality × 0.5) + (standardized value of predicted operating energy consumption × 0.3) + (standardized value of predicted sludge production × 0.2). The three weighted scores of each solution are added together to obtain the final comprehensive evaluation score for that solution. The comprehensive evaluation scores of all solutions in the candidate solution set are sorted from highest to lowest. The solution ranked first in the comprehensive evaluation is selected as the final optimized scheme. This scheme not only meets the requirements for effluent quality, but also achieves the optimal balance between operating energy consumption and sludge production. It is the optimal control scheme that adapts to the actual operating needs of the AOA process. The scheme includes optimized setpoints for dissolved oxygen in the aerobic zone and the sludge return ratio of the system, providing core parameters for subsequent system control.

[0044] Step 502: The values ​​of dissolved oxygen setpoint and sludge return ratio corresponding to the final optimized scheme are sent to the bottom process controller of the AOA wastewater treatment system as optimized setpoints. The bottom process controller adjusts the aeration equipment in the aerobic zone to control the dissolved oxygen concentration and adjusts the sludge return pump to control the sludge return ratio based on the received optimized setpoints, thereby completing the coordinated optimization of the system operation status. Specifically, this includes: extracting the core control parameters in the final optimized scheme determined in step 501, namely the optimized setpoints of dissolved oxygen setpoint in the aerobic zone and sludge return ratio in the system. The specific process for obtaining these optimized setpoints is as follows: First, a non-dominated solution set is generated through multi-objective optimization in step 402c, with each solution corresponding to a set of decision variables (dissolved oxygen setpoint in the aerobic zone and sludge return ratio in the system); then, candidate solutions that meet the effluent quality standards are obtained through screening in step 500, and all valid decision variable values ​​in the candidate solution set are retained; finally, the solution with the highest score is selected through multi-attribute comprehensive evaluation and sorting in step 501, and the two sets of decision variable values ​​corresponding to this solution are the optimized setpoints extracted in this step. During extraction, ensure that the values ​​of the two parameters are consistent with the range of the decision variables in step 402c, and completely match the parameters of the corresponding solutions in the candidate solution set to avoid parameter extraction errors. The extracted dissolved oxygen setpoint and sludge return ratio optimization setpoint are sent to the underlying process controller of the AOA wastewater treatment system through the system data transmission module. The data transmission process ensures real-time performance and accuracy to avoid parameter loss or transmission delay. The underlying process controller, as the core execution unit for system operation and control, receives the optimized setpoints in real time.The underlying process controller automatically adjusts the operating parameters of the aeration equipment in the aerobic zone based on the received optimized dissolved oxygen setpoint. Specifically, if the actual dissolved oxygen concentration in the aerobic zone is lower than the optimized setpoint, the aeration intensity of the aeration equipment is increased (e.g., increasing the operating power of the aeration blower, adjusting within 50% to 100% of the rated power) to accelerate oxygen supply until the actual dissolved oxygen concentration stabilizes within ±0.1 mg / L of the optimized setpoint. If the actual dissolved oxygen concentration is higher than the optimized setpoint, the aeration intensity of the aeration equipment is reduced to decrease oxygen supply, ensuring the dissolved oxygen concentration remains stable within the optimized setpoint range, thus achieving precise control of the dissolved oxygen concentration in the aerobic zone. Simultaneously, the underlying process controller automatically adjusts the operating parameters of the sludge return pump based on the received optimized sludge return ratio setpoint. Specifically, if the actual sludge return ratio is higher than the optimized setpoint, the aeration intensity of the aeration equipment is increased (e.g., increasing the operating power of the aeration blower, adjusting within 50% to 100% of the rated power) to accelerate oxygen supply until the actual dissolved oxygen concentration stabilizes within ±0.1 mg / L of the optimized setpoint. If the sludge return ratio is lower than the optimized set value, the operating frequency of the sludge return pump is increased (adjustment range is 50Hz to 60Hz) to increase the sludge return flow rate until the actual sludge return ratio stabilizes within ±1% of the optimized set value. If the current actual sludge return ratio is higher than the optimized set value, the operating frequency of the sludge return pump is reduced to decrease the sludge return flow rate, ensuring that the sludge return ratio stabilizes within the optimized set value range, thus achieving precise control of the sludge return ratio. During the adjustment process, the underlying process controller collects the actual values ​​of dissolved oxygen concentration and sludge return ratio in the aerobic zone in real time, compares them with the optimized set value, and continuously fine-tunes the operating parameters of the aeration equipment and sludge return pump to ensure that the two core control parameters stabilize within the optimized set value range. This completes the coordinated optimization of the AOA wastewater treatment system's operating status, achieving a coordinated balance between carbon source storage and consumption, effluent quality, energy consumption, and sludge production.

[0045] This example uses an online effluent quality prediction model to predict detailed effluent indicators and selects candidate solutions according to the Class A discharge standard. This ensures that all solutions in the candidate solution set meet the core baseline of process operation (effluent compliance), effectively avoiding the problem of some solutions in the non-dominated solution set failing to meet effluent quality standards and being unusable in practice. This improves the practicality and reliability of the optimized solutions, while focusing on four core water quality indicators. A multi-attribute comprehensive evaluation ranking method is used to select the final optimized solution. By setting weight coefficients (highlighting the core position of effluent quality while considering energy consumption and sludge production), combining standardized treatment to eliminate dimensional differences, and then calculating a comprehensive score through weighted summation, the candidate solution set is accurately evaluated. Precise sorting avoids the deviation caused by evaluation of a single indicator, ensuring that the final selected scheme not only meets the effluent standards but also achieves the goals of energy saving, consumption reduction, and sludge production reduction, thus adapting to the core requirements of AOA process co-optimization. The control parameters of the final optimized scheme are directly sent to the underlying process controller, which automatically adjusts the aeration equipment and sludge return pump, enabling rapid implementation of the optimized scheme without manual adjustment. This improves the automation and intelligence level of system operation. At the same time, real-time fine-tuning ensures that the dissolved oxygen concentration and sludge return ratio remain stable within the optimized set value range, achieving precise co-optimization of system operation status and ensuring the stability of process operation.

[0046] like Figure 2 As shown, embodiments of the present invention also provide a multi-objective collaborative optimization operation control system for an AOA process wastewater treatment system, including: The model building module is used to construct the current operating condition feature vector by combining the current water quality indicators, operating parameters and the pollutant concentration change rate at three characteristic locations; calculate the distance and correlation measure between the current operating condition feature vector and the feature vectors of each historical operating condition; and determine a subset of historical data from the historical database to establish an online prediction model for effluent water quality, operating energy consumption and sludge production under the current operating conditions. The deduction module is used to construct a multi-objective optimization problem and, based on the fluctuation characteristics of the influent water quality index data, determine the numerical boundaries of the operating state parameters in the multi-objective optimization problem, and deduce the feasible range of values ​​for the dissolved oxygen setpoint and the sludge return ratio. The calculation module is used to generate a virtual equilibrium domain for the dynamic balance of carbon sources in the system by analyzing the rate of change of pollutant concentration at three characteristic locations within the feasible range of values, and to define the current system point, calculate the positional relationship between the current system point and the boundary of the virtual equilibrium domain, and obtain a dynamic equilibrium deviation value. The solution module is used to calculate the migration and transformation equilibrium of carbon sources from storage to consumption within the system based on the pollutant concentration change rate and dynamic equilibrium deviation value, and to solve the multi-objective optimization problem to obtain a non-dominated solution set. The optimization module is used to determine the final optimization scheme from the non-dominated solution set, and issue the dissolved oxygen setpoint and sludge return ratio corresponding to the final optimization scheme as optimization setpoints for execution, thereby completing the collaborative optimization of the system operation status.

[0047] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0048] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A multi-objective collaborative optimization operation method for an AOA process wastewater treatment system, characterized in that, include: Step 1: Construct a feature vector for the current operating condition by combining the current water quality indicators, operating parameters, and the rate of change of pollutant concentration at three characteristic locations; calculate the distance and correlation measure between the feature vector for the current operating condition and the feature vectors for each historical operating condition, and determine a subset of historical data from the historical database to establish an online prediction model for effluent water quality, operating energy consumption, and sludge production under the current operating condition. Step 2: Construct a multi-objective optimization problem, and based on the fluctuation characteristics of the influent water quality index data, determine the numerical boundaries of the operating state parameters in the multi-objective optimization problem, and deduce the feasible range of values ​​for the dissolved oxygen setpoint and sludge return ratio. Step 3: Within the feasible range of values, by analyzing the rate of change of pollutant concentration at three characteristic locations, a virtual equilibrium domain for the dynamic balance of carbon source in the system is generated, and the current system point is defined. The positional relationship between the current system point and the boundary of the virtual equilibrium domain is calculated to obtain a dynamic equilibrium deviation value. Step 4: Based on the pollutant concentration change rate and combined with the dynamic equilibrium deviation value, calculate the migration and transformation equilibrium of carbon source from storage to consumption in the system, and solve the multi-objective optimization problem to obtain a non-dominated solution set. Step 5: Determine the final optimization scheme from the non-dominated solution set, and issue the dissolved oxygen setpoint and sludge return ratio corresponding to the final optimization scheme as optimization setpoints for execution, thereby completing the collaborative optimization of the system operation status.

2. The multi-objective collaborative optimization operation method for the AOA process wastewater treatment system according to claim 1, characterized in that, Before step 1, real-time water quality data of the influent and effluent of the AOA wastewater treatment system, as well as the operating parameters in the anaerobic, aerobic, and anoxic zones, are collected. The actual pollutant concentration change rate at three characteristic locations—the inlet of the anaerobic zone, the end of the aerobic zone, and the outlet of the anoxic zone—is monitored simultaneously.

3. The multi-objective collaborative optimization operation method for the AOA process wastewater treatment system according to claim 2, characterized in that, Step 1 includes: The real-time collected and monitored water quality index data, operating parameters, and pollutant concentration change rates at three characteristic locations are stored in the historical database. Based on the currently collected and monitored water quality index data, operating parameters, and pollutant concentration change rates at three characteristic locations, a current operating condition feature vector is constructed. Calculate the Euclidean distance and cosine correlation measure between the current working condition feature vector and each historical working condition feature vector stored in the historical database. Select multiple historical data samples whose cosine correlation measure with the current working condition feature vector is greater than a preset threshold to form a subset of historical data. Based on a subset of historical data, we constructed online prediction models for effluent quality (with operating parameters as input and effluent quality as output), online prediction models for operating energy consumption (with operating parameters as input and operating energy consumption as output), and online prediction models for sludge production (with operating parameters as input and sludge production as output) under the current operating conditions.

4. The multi-objective collaborative optimization operation method for the AOA process wastewater treatment system according to claim 3, characterized in that, Step 2 includes: The currently collected operating parameters are used as input data to the online prediction models for effluent water quality, operating energy consumption, and sludge production, respectively. The online prediction model for effluent water quality performs linear or nonlinear transformations on the input data based on internal regression parameters to obtain the predicted value of effluent water quality; the online prediction model for operating energy consumption performs linear or nonlinear transformations on the input data based on internal regression parameters to obtain the predicted value of operating energy consumption; and the online prediction model for sludge production performs linear or nonlinear transformations on the input data based on internal regression parameters to obtain the predicted value of sludge production. Using the predicted values ​​of effluent water quality, operating energy consumption, and sludge production as multiple objectives to be optimized simultaneously, and the dissolved oxygen setpoint in the aerobic zone and the sludge return ratio of the system as decision variables, a multi-objective optimization problem containing multiple objective functions is constructed. Based on the historical sequence of influent water quality index data and the real-time collected influent water quality index data, the statistical fluctuation range of the influent water quality index data is calculated, and the upper and lower limits of the operating state parameters in the constraints of the multi-objective optimization problem within the fluctuation range are determined. Based on the upper and lower limits of the values, and combined with the process safety operation restrictions, the allowable range of values ​​for the dissolved oxygen setpoint is deduced as the feasible range of values ​​for the dissolved oxygen setpoint. Similarly, the allowable range of values ​​for the sludge return ratio is deduced as the feasible range of values ​​for the sludge return ratio.

5. The multi-objective collaborative optimization operation method for the AOA process wastewater treatment system according to claim 4, characterized in that, Step 3 includes: Within the feasible range of dissolved oxygen setpoint and sludge return ratio, the pollutant concentration change rates at three characteristic locations—the anaerobic zone inlet, the aerobic zone end, and the anoxic zone outlet—obtained through analysis and monitoring are used to calculate the high-frequency distribution intervals of the pairwise ratios of the pollutant concentration change rates at these three characteristic locations in historical operating data. The median of these high-frequency distribution intervals is then used to construct a set of coordinate points as boundary points characterizing the dynamics of the ideal carbon source in the system. These boundary points are then connected sequentially to form a closed polygonal region, which serves as a virtual equilibrium domain characterizing the dynamic balance of the carbon source within the system. The current pollutant conversion efficiency is calculated based on the real-time collected operating parameters, and the current carbon source consumption rate is calculated based on the pollutant concentration change rate at three characteristic locations obtained by synchronous monitoring. Using the current pollutant conversion efficiency and the current carbon source consumption rate as coordinate values, a unique coordinate point is determined in the state space and defined as the current system point. The shortest Euclidean distance between the current system point and all points on the boundary of the virtual equilibrium domain is calculated, and the value of the shortest Euclidean distance is used as the dynamic equilibrium deviation value.

6. The multi-objective collaborative optimization operation method for the AOA process wastewater treatment system according to claim 5, characterized in that, Step 4 includes: Based on the pollutant concentration change rates at the anaerobic zone inlet and the anoxic zone outlet, the ratio of the pollutant concentration change rate at the anaerobic zone inlet to that at the anoxic zone outlet is calculated as the carbon source migration and conversion rate ratio. The dynamic equilibrium deviation value and the carbon source migration and conversion rate ratio are weighted and fused to obtain a comprehensive characterization of the migration and conversion equilibrium degree of carbon source in the system from storage to consumption. Based on the magnitude and direction of the dynamic equilibrium deviation, the virtual equilibrium domain is geometrically divided to form at least one sub-region centered on the current system point. The sub-region is then combined with the undivided part of the virtual equilibrium domain to define a correction region for guiding the optimization search. By using dynamic equilibrium deviation, migration and transformation equilibrium, and geometric features of the correction region as search-guided knowledge, and combining structured knowledge extracted from historical optimization information, a multi-objective evolutionary search and solution is performed on the multi-objective optimization problem, ultimately obtaining a set containing multiple non-dominated solutions, i.e., the non-dominated solution set.

7. The multi-objective collaborative optimization operation method for the AOA process wastewater treatment system according to claim 6, characterized in that, By utilizing dynamic equilibrium deviation, migration and transformation equilibrium, and the geometric characteristics of the correction region as search-guided knowledge, combined with structured knowledge extracted from historical optimization information, a multi-objective evolutionary search and solution is performed on the multi-objective optimization problem. This ultimately yields a set containing multiple non-dominated solutions, i.e., the non-dominated solution set, which includes: The dynamic equilibrium deviation value and the migration and transformation equilibrium degree are weighted and summed to obtain an adaptive coefficient. A rectangular area is determined based on the geometric center of the correction area and the boundary range of the correction area. The rectangular area is used as the initial search space for the two decision variables, dissolved oxygen setpoint and sludge return ratio, in the multi-objective optimization problem. By calculating the distance and correlation measure between the current system point and the feature vectors of each historical working condition stored in the historical database, and based on the calculation results of the distance and correlation measure, the historical optimization solution set is selected from the historical optimization information, and a portion of the solutions of the historical optimization solution set is introduced into the initial population. An optimization search framework based on population iteration and survival of the fittest is adopted. The search step size for generating new solutions in each iteration is dynamically adjusted with adaptive coefficients. The boundary range of the correction region is used as a hard constraint on the values ​​of decision variables when generating new solutions. The spatial distribution pattern of solutions in the historical optimization solution set is used to guide the generation direction of new solutions. Under the premise of satisfying all constraints in the constructed multi-objective optimization problem, the process of population iteration and survival of the fittest is repeated until the preset number of iterations is reached. Finally, a set consisting of multiple solutions that are not mutually dominant on the three objectives of predicted effluent water quality, predicted operating energy consumption and predicted sludge production is obtained, namely, the non-dominated solution set.

8. The multi-objective collaborative optimization operation method for the AOA process wastewater treatment system according to claim 7, characterized in that, Step 5 includes: For each solution in the non-dominated solution set, the detailed effluent water quality index corresponding to the solution in the non-dominated solution set is predicted using the online effluent water quality prediction model. Solutions in which all predicted effluent water quality indexes meet the preset discharge standards are selected to form a candidate solution set. For each solution in the candidate solution set, the predicted values ​​of effluent quality, operating energy consumption and sludge production are evaluated and ranked using a multi-attribute comprehensive evaluation. The solution ranked first in the comprehensive evaluation is selected as the final optimization scheme. The dissolved oxygen setpoint and sludge return ratio corresponding to the final optimized scheme are sent to the bottom process controller of the AOA wastewater treatment system as optimized setpoints. The bottom process controller adjusts the aeration equipment in the aerobic zone to control the dissolved oxygen concentration and adjusts the sludge return pump to control the sludge return ratio according to the received optimized setpoints, thereby completing the collaborative optimization of the system operation status.

9. A multi-objective collaborative optimization operation control system for an AOA process wastewater treatment system, wherein the system implements the method as described in any one of claims 1 to 8, characterized in that, include: The model building module is used to construct the current operating condition feature vector by combining the current water quality indicators, operating parameters and the pollutant concentration change rate at three characteristic locations; calculate the distance and correlation measure between the current operating condition feature vector and the feature vectors of each historical operating condition; and determine a subset of historical data from the historical database to establish an online prediction model for effluent water quality, operating energy consumption and sludge production under the current operating conditions. The deduction module is used to construct a multi-objective optimization problem and, based on the fluctuation characteristics of the influent water quality index data, determine the numerical boundaries of the operating state parameters in the multi-objective optimization problem, and deduce the feasible range of values ​​for the dissolved oxygen setpoint and the sludge return ratio. The calculation module is used to generate a virtual equilibrium domain for the dynamic balance of carbon sources in the system by analyzing the rate of change of pollutant concentration at three characteristic locations within the feasible range of values, and to define the current system point, calculate the positional relationship between the current system point and the boundary of the virtual equilibrium domain, and obtain a dynamic equilibrium deviation value. The solution module is used to calculate the migration and transformation equilibrium of carbon sources from storage to consumption within the system based on the pollutant concentration change rate and dynamic equilibrium deviation value, and to solve the multi-objective optimization problem to obtain a non-dominated solution set. The optimization module is used to determine the final optimization scheme from the non-dominated solution set, and issue the dissolved oxygen setpoint and sludge return ratio corresponding to the final optimization scheme as optimization setpoints for execution, thereby completing the collaborative optimization of the system operation status.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 8.

Images