A primary aerobic zone short-cut nitrification precision control system and method based on digital twin driving

Abstract

The application provides a kind of based on digital twin driving's first aerobic zone short-range nitrification precision control system and method. The system includes perception acquisition module, cloud digital twin module, edge control module and execution driving module. The perception acquisition module collects dissolved oxygen data, pH data and oxidation-reduction potential data in the first aerobic tank in real time. The cloud digital twin module denoises the dissolved oxygen data to eliminate bubble interference, extracts the pH change slope and oxidation-reduction potential trend as key biochemical characteristics, and constructs a virtual reaction tank model to predict nitrite nitrogen accumulation rate and biochemical state. The edge control module generates feedforward control instructions based on the prediction results, and adjusts the frequency of the aeration fan and the reflux ratio in advance when an anomaly occurs to maintain the target low-oxygen operating range. The application can protect the activity of nitrosobacteria and inhibit the activity of nitrifying bacteria in a physically disturbed environment, achieving stable operation of the system.

Description

Technical Field

[0001] This invention relates to the field of wastewater biological treatment control technology, and in particular to a precise control system and method for short-range nitrification in the primary aerobic zone based on digital twin-driven technology. Background Technology

[0002] Short-cut nitrification processes offer advantages in biological nitrogen removal from wastewater by saving carbon sources and reducing aeration energy consumption. The core of this process lies in controlling the nitrification process at the nitrite stage. In actual engineering operations, the water quality and biochemical conditions in the primary aerobic zone are highly complex. Maintaining the dominance of nitrite-oxidizing bacteria while inhibiting the growth of nitrifying bacteria requires precise control of the low-oxygen environment and operating conditions.

[0003] Traditional control methods typically rely on single sensor threshold feedback or manual adjustment based on experience. In the bubble environment generated by microporous or surface aeration, the permeable membrane of the dissolved oxygen probe deployed on-site is susceptible to interference from gas-liquid physical contact, leading to transient shifts and fluctuations in local oxygen partial pressure parameters. Using raw, unfiltered, abnormal data directly for passive feedback control easily causes frequent start-ups and shutdowns of aeration equipment or lag in frequency converter regulation, making it difficult to maintain a stable low-oxygen operating range within the primary aerobic tank.

[0004] Meanwhile, the succession and metabolic state of the microbial community within the biochemical reactor exhibit highly nonlinear and time-varying characteristics. Conventional methods often rely on periodic manual offline testing to obtain effluent ammonia nitrogen, nitrite nitrogen, and nitrate nitrogen levels. This testing method suffers from physical time delays and cannot provide early warnings before abnormal influent load surges, low-temperature conditions, or nitrite accumulation occur. When test indicators show substantial exceedances, the microbial community structure of the biochemical system has usually already shifted, and at this point, the adjustment period required to restore system stability through intervention or control logic adjustments is relatively long. Summary of the Invention

[0005] The purpose of this invention is to provide a precise control system and method for short-range nitrification in the primary aerobic zone based on digital twin drive, in order to solve the technical problems in the existing short-range nitrification control process, such as dissolved oxygen data being easily interfered with by bubbles, feedback being delayed, and difficulty in predicting and responding to abnormal biochemical conditions in advance.

[0006] In a first aspect, the present invention provides a precision control system for short-range nitrification in the first-stage aerobic zone based on digital twin-driven technology, comprising: The sensing and acquisition module is used to collect dissolved oxygen data, pH data, and redox potential data in the primary aerobic tank in real time. The cloud-based digital twin module, which is communicatively connected to the sensing and acquisition module, is used to denoise the dissolved oxygen data to eliminate abnormal fluctuations caused by bubble interference, and to extract the slope of the pH data and the trend of the redox potential data as key biochemical features. The cloud-based digital twin module is also used to construct a virtual reaction pool model based on the denoised dissolved oxygen data and the key biochemical features, and to predict the current nitrite accumulation rate and the biochemical situation within a set future time period through multi-parameter fusion. The edge control module is communicatively connected to the cloud-based digital twin module. It is used to generate feedforward control commands based on the prediction results of the cloud-based digital twin module. When an abnormal accumulation of nitrite is predicted or the system shows a trend of increasing nitrate nitrogen, it issues control signals in advance to reduce the frequency of the aeration blower and adjust the reflux ratio to maintain the target low-oxygen operating range in the primary aerobic tank. This allows for the targeted protection of nitrifying bacteria activity and the inhibition of nitrifying bacteria activity in an environment where dissolved oxygen is subject to physical detection interference. The execution drive module is used to receive the control signal and adjust the corresponding aeration blower and inlet fluid power equipment accordingly.

[0007] Optionally, the cloud-based digital twin module uses a combination of sliding window filtering and anomaly threshold removal to denoise the dissolved oxygen data; When the change in dissolved oxygen data between consecutive sampling points exceeds a preset threshold, the corresponding data is determined to be an abnormal fluctuation caused by bubble interference, and the average value of adjacent time windows is automatically used to replace the abnormal fluctuation data.

[0008] Optionally, the cloud-based digital twin module is also used to automatically identify the system under any of the following conditions based on the prediction results of the nitrite accumulation rate and the biochemical state: high load impact condition, low temperature suppression condition, abnormal nitrite accumulation condition, or steady-state energy-saving condition. Specifically, when the redox potential data remains in the negative range and the denoised dissolved oxygen data is lower than a set threshold, the system is determined to have entered the short-range nitrification steady-state operation state under the steady-state energy-saving condition.

[0009] Optionally, the cloud-based digital twin module also continuously receives test results and real-time sensor data transmitted back by the sensing and acquisition module, and periodically and dynamically corrects the parameters of the virtual reaction tank model based on the error between the measured value of the actual effluent data and the predicted value of the virtual reaction tank model, combined with a preset correction coefficient.

[0010] Optionally, the specific logic for the cloud-based digital twin module to dynamically adjust the frequency of the aeration blower based on the prediction results is as follows: The deviation between the target nitrite nitrogen accumulation rate and the predicted current nitrite nitrogen accumulation rate is calculated, and the initial operating frequency of the aeration blower is adaptively lowered in combination with a preset adjustment coefficient, and a feedforward frequency conversion control command is output.

[0011] Optionally, the edge control module acquires the actual influent flow rate and return flow rate to calculate the current return ratio, and when the system detects a high load shock condition, it coordinates with the frequency reduction control signal of the aeration blower to dynamically increase the return ratio to ensure stable low-oxygen operation of the system.

[0012] Optionally, when the influent ammonia nitrogen concentration is greater than a set high concentration threshold and the influent flow rate is greater than a set multiple of the average flow rate, the cloud-based digital twin module determines that the system has entered the high-load impact condition. When the predicted current nitrite accumulation rate is lower than the safety warning ratio, the system is determined to have a risk of nitrification instability and the feedforward control command is triggered.

[0013] Optionally, the sensing and acquisition module includes a multi-dimensional sensing array arranged in the primary aerobic tank, the multi-dimensional sensing array consisting of an online pH probe, an online oxidation-reduction potential probe, and an online dissolved oxygen probe; the edge control module includes a programmable logic controller. The execution drive module includes the aeration blower, inlet pump, and return pump, which are controlled by the programmable logic controller via a frequency converter.

[0014] Optionally, the dissolved oxygen concentration in the target low-oxygen operating range is limited to 0.08 to 0.15 mg / L; the feedforward control command is used to implement proactive control intervention before the effluent test indicators substantially exceed the standards.

[0015] Secondly, the present invention provides a method for precise control of short-range nitrification in the first-stage aerobic zone based on digital twin-driven technology, applicable to a precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin-driven technology as described in any one of the first aspects, the method comprising the following steps: S1. Real-time data on dissolved oxygen, pH, and redox potential in the primary aerobic tank are collected through the sensing and acquisition module. S2. The dissolved oxygen data is denoised using the cloud-based digital twin module to remove abnormal fluctuations, and the slope of the pH data and the trend of the redox potential data are extracted as key biochemical features. S3. Based on the denoised dissolved oxygen data and the key biochemical characteristics, construct a virtual reaction tank model to predict the current nitrite accumulation rate and the biochemical situation within a set future time period. S4. The edge control module generates feedforward control commands based on the prediction results, and adjusts the frequency and reflux ratio of the aeration blower in advance when abnormal nitrite accumulation or nitrate nitrogen increase trend is predicted, so as to maintain the target low oxygen operating range in the primary aerobic tank, thereby protecting the activity of nitrifying bacteria and inhibiting the activity of nitrifying bacteria.

[0016] The present invention has achieved the following beneficial effects: This invention utilizes a cloud-based digital twin module to denoise and extract features from multi-source water quality characteristic data, eliminating the transient fluctuations in dissolved oxygen parameters caused by physical detection interference such as bubbles, thus ensuring the accuracy of the underlying data. A virtual reaction tank model is constructed using key biochemical features such as the extracted pH change slope and redox potential trend. The system can predict the current nitrite accumulation rate and future biochemical status. Upon identifying high load shocks or abnormal nitrite accumulation conditions, the edge control module issues feedforward control commands in advance to coordinate the adjustment of aeration fan frequency and reflux ratio. This control logic overcomes the lag of traditional offline testing, intervening before effluent indicators exceed standards, ensuring the primary aerobic tank remains within the target low-oxygen operating range. This protects nitrite-oxidizing bacteria activity while inhibiting nitrifying bacteria activity, achieving precise and stable operation of the short-range nitrification process. Combined with a closed-loop dynamic correction mechanism for model parameters based on measured test results, this system maintains long-term consistency between the virtual model and the physical community evolution state, improving the reliability of the control strategy.

[0017] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 A hierarchical architecture and modular diagram of a digital twin-driven short-range nitrification precision control system for the primary aerobic zone provided in this embodiment of the invention; Figure 2 This is a comparison of the effects of denoising the dissolved oxygen (DO) data before and after the denoising process in this embodiment of the invention, as well as the trend graphs of pH and oxidation-reduction potential (ORP) over time. Figure 3A flowchart illustrating an overall process for precise control of short-range nitrification in the primary aerobic zone based on digital twin-driven methods, provided in this embodiment of the invention. Figure 4 This is a comparison curve showing the effect of the method of the present invention and the traditional method on the control of nitrite accumulation rate (NAR). Detailed Implementation

[0020] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0021] like Figure 1 As shown, this embodiment provides a precise control system and method for short-range nitrification in the primary aerobic zone based on digital twin-driven architecture. The hardware and software architecture of this control system establishes data interaction and closed-loop feedback links between the physical entity layer, edge control layer, and cloud layer. The system's hardware components and software modules together constitute a control closed loop for multi-source data acquisition, processing, model prediction, and command execution. The system mainly includes a sensing and acquisition module, a cloud-based digital twin module, an edge control module, and an execution driving module. These modules are connected through an industrial communication network and perform data transmission and command delivery tasks according to a preset time period and communication protocol.

[0022] In this embodiment, the sensing and acquisition module is located in the physical entity layer. The physical entity layer includes the primary aerobic tank in the wastewater treatment system and the pipeline fluid system connecting the reaction tank. The sensing and acquisition module is configured to acquire multi-source water quality characteristic data in the primary aerobic tank in real time. This module includes a multi-dimensional sensing and acquisition array, which consists of an online pH probe, an online oxidation-reduction potential probe, and an online dissolved oxygen probe. The online pH probe is used to measure the pH data of the mixed solution; the online oxidation-reduction potential probe is used to measure the oxidation-reduction potential data of the mixed solution; and the online dissolved oxygen probe is used to measure the dissolved oxygen data of the mixed solution. Each probe is fixed at a preset depth in the primary aerobic tank by a physical support, and the specific installation position is configured on the opposite side of the aeration device or in the water flow swirl zone to obtain representative fluid characteristic data. The online pH probe, the online oxidation-reduction potential probe, and the online dissolved oxygen probe all have analog signal or digital signal output interfaces. When using analog output, the sensor outputs a 4-20mA standard current signal or a 0-10V standard voltage signal to the analog input terminal block distributed on-site. When using digital output, the sensor sends the digitized measurement value to the data concentrator via an RS485 communication bus based on the Modbus RTU protocol. The sensing and acquisition module continuously collects multi-source water quality characteristic data according to a set sampling frequency (e.g., sampling once every 5 seconds), and synchronously uploads the multi-source water quality characteristic data to the edge control module in the edge control layer and the cloud digital twin module in the cloud layer via an industrial Ethernet network.

[0023] Specifically, regarding the flow of the aforementioned multi-source water quality characteristic data, after the sensing and acquisition module completes analog-to-digital conversion and preliminary signal conditioning at the physical entity layer, it assigns a timestamp tag and sensor physical address code to each sampled data. Data packets containing timestamps, address codes, and measured values ​​are encapsulated into a standard data frame structure. This data frame is routed and forwarded via an industrial Ethernet switch deployed on-site, using the TCP / IP protocol suite. One data stream is sent directly to the edge control module deployed on-site via intranet penetration or a dedicated industrial gateway, while another data stream is uploaded to the cloud server cluster via a communication channel configured with a secure encryption protocol (such as MQTT over TLS) for use by the cloud-based digital twin module.

[0024] In this embodiment, the cloud-based digital twin module is deployed in the cloud layer. The cloud layer includes a physical server array or a virtualized cloud host system with computing, storage, and network distribution capabilities. The cloud-based digital twin module is communicatively connected to the sensing and acquisition module and is configured to receive multi-source water quality characteristic data sequences with timestamps. A time-series database is pre-configured within the cloud-based digital twin module to continuously store the received dissolved oxygen data, pH data, and redox potential data in chronological order, forming a time-series data matrix. The cloud-based digital twin module includes a signal denoising and feature decoupling unit. Since the primary aerobic tank typically has bottom microporous aeration discs or surface aerators, a large number of rising bubbles are generated during gas release. When these bubbles physically contact the permeable membrane of the online dissolved oxygen probe, it causes a transient shift in the local oxygen partial pressure parameter measured by the probe, manifesting as sudden numerical fluctuations in the time-series data. The signal denoising and feature decoupling unit is configured to use a combination of sliding window filtering and anomaly threshold removal to denoise the dissolved oxygen data.

[0025] Furthermore, regarding the denoising process for dissolved oxygen data, the signal denoising and feature decoupling unit establishes a memory structure with a length of [length missing]. A first-in, first-out (FIFO) data buffer queue. This queue forms a processing window that slides along the time axis. When new dissolved oxygen data measurements are received... Upon arrival and storage in the buffer queue, the computational logic of the signal denoising and feature decoupling unit is triggered, performing verification of the variation amplitude between consecutive sampling points: ; in, Indicates the current number Dissolved oxygen data measurements at each sampling time. Indicates the first Dissolved oxygen data measurements at each sampling time. This indicates the preset threshold for judging abnormal fluctuations.

[0026] This judgment logic compares the absolute value of the difference between two adjacent discrete time points with a set anomaly threshold. If the absolute value of the difference is greater than... Then the signal denoising and feature decoupling unit will decouple the current data points Data points marked as abnormal fluctuations caused by physical disturbances from bubbles.

[0027] In this embodiment, parameters The value is set as follows: ; in, This indicates the preset threshold for judging abnormal fluctuations, with a value range of 0.3 mg / L to 0.5 mg / L.

[0028] Specifically, when the above conditions are met—that is, when the change in dissolved oxygen data between consecutive sampling points exceeds a preset anomaly threshold—the signal denoising and feature decoupling unit will not directly block the data flow, but will instead initiate a substitution value calculation subroutine. This subroutine calculates a smoothed substitution value by extracting historical sampling data that has not been marked as an anomaly within a sliding time window and performing an arithmetic average operation. The specific substitution algorithm model is as follows: ; in, This represents the calculated dissolved oxygen value used to replace outlier data after filtering. This indicates the number of sampling data points included within the set sliding time window. This represents the index variable of the sampling points within the time window. Indicates the first Effective dissolved oxygen data measurements at each historical sampling time.

[0029] The signal denoising and feature decoupling unit utilizes the calculated... Replace the original abnormal measurement value The smoothed continuous dissolved oxygen data sequence is then written into the feature buffer. Through this logic, low-amplitude abnormal fluctuations caused by bubble interference are isolated from subsequent model input parameters. Combined with... Figure 2 The DO / pH / ORP trend graph shown shows that the original DO data exhibited drastic transient changes due to bubble interference, while the DO curve after digital twin filtering effectively eliminated these spike noises, presenting a smooth and realistic fluid characteristic. In addition, the graph also intuitively reflects the key characteristic change trends of pH and ORP on the same time axis.

[0030] After denoising the dissolved oxygen data, the metabolic window locking model configured in the cloud-based digital twin module performs feature extraction. Since the biochemical reactions in the primary aerobic tank (e.g., the conversion of ammonia nitrogen to nitrite by ammonia-oxidizing bacteria) are accompanied by the release of hydrogen ions and changes in the redox potential of the mixed liquor, the rate of change in the time series data can directly characterize the reaction kinetics. The metabolic window locking model is equipped with a numerical differential calculation component, which uses a first-order or higher-order finite difference algorithm to calculate the derivative of the pH data within a set time step, thereby extracting the slope of the pH data change. Similarly, this component performs a difference operation on the redox potential time series to extract the trend (i.e., the gradient) of the redox potential data change. After the above processing, the denoised dissolved oxygen data, the slope of the pH data change, and the gradient of the redox potential data together constitute the key biochemical feature matrix.

[0031] The metabolic window locking model is used to dynamically construct a short-range nitrification state prediction function based on the extracted key biochemical characteristics, current influent flow rate, reflux ratio, and system temperature. The metabolic window locking model includes a multivariate nonlinear mapping algorithm structure (e.g., a long short-term memory network or a partial least squares regression model) trained on historical data. This prediction function is invoked in each calculation cycle to achieve advanced prediction of the current nitrite accumulation rate and the biochemical status within a set future time period. The specific prediction function relationship is expressed as follows: ; in, This represents the predicted nitrite accumulation rate at the current moment. This represents the slope of the pH data over time. This represents the gradient of redox potential data over time. This represents the denoised dissolved oxygen data. Indicates the inflow rate. Indicates the reflux ratio. Indicates the system temperature. This represents a short-range nitration state multivariate nonlinear prediction function that has been pre-trained and configured within a cloud-based digital twin module.

[0032] In this execution logic, the inflow rate and system temperature Data is collected by electromagnetic flow meters and temperature sensors installed in the inlet pipe and uploaded to the cloud. Return ratio. The calculations are based on the pump unit operating parameters fed back in real time by the edge control module. The model not only outputs the time... of The value also outputs future values ​​based on time series prediction algorithms. to ( The preset prediction step size corresponds to the prediction vector set within a time range of 1 to 2 hours in the future.

[0033] Based on the prediction results output by the metabolic window locking model and the current key biochemical characteristics, the cloud-based digital twin module is configured with a state machine logic for operating condition identification. This state machine is used to automatically identify any of the following operating conditions: high-load shock condition, low-temperature suppression condition, abnormal nitrite accumulation condition, or steady-state energy-saving condition.

[0034] To identify high-load shock conditions, the system analyzes influent ammonia nitrogen concentration data and influent flow rate data uploaded by external online influent monitoring equipment. The state machine is configured with the following trigger conditions: ; in, This indicates the influent ammonia nitrogen concentration obtained through online sensors or predictive models, 700 This indicates the set high-concentration ammonia nitrogen shock threshold.

[0035] Meanwhile, the criteria for detecting a surge in traffic are configured as follows: ; in, This indicates the inflow rate measured by the system in real time at the current moment. 1.2 represents the average inflow rate of the system within a preset historical time period, and 1.2 represents the set flow surge judgment coefficient.

[0036] Specifically, when the real-time data received by the cloud-based digital twin module simultaneously meets the condition that the influent ammonia nitrogen concentration is greater than 700... Furthermore, when the real-time inflow rate is greater than 1.2 times the average flow rate, the logic operation unit in the state machine outputs a truth signal, determining that the system has entered a high-load impact condition.

[0037] For identifying abnormal nitrite accumulation conditions, the state machine relies on the predicted nitrite nitrogen accumulation rate data stream output by the metabolic window locking model. Its triggering conditions are configured as follows: ; in, This represents the current nitrite accumulation rate predicted in real time by the cloud-based digital twin module, while 85% represents the set lower threshold for nitrification instability risk warning.

[0038] Specifically, when the model predicts that the current nitrite nitrogen accumulation rate is less than 85%, or when the predicted value curve of the effluent nitrite nitrogen concentration in the future set time period shows a continuous upward trend and the slope exceeds the set threshold, the cloud-based digital twin module determines that the system has a risk of nitrification instability, and the state machine jumps and locks to the abnormal nitrite accumulation condition.

[0039] For low-temperature suppression conditions, the state machine continuously monitors the uploaded system water temperature. .when When the temperature is below the preset low temperature threshold (e.g., 15 degrees Celsius) and continues for more than the set time window length, the state machine recognizes that the current condition is a low temperature suppression condition.

[0040] For steady-state energy-saving operation, the state machine is configured with combined conditions. When the redox potential data is continuously maintained in the negative range, and the denoised dissolved oxygen data is lower than the set upper limit threshold of steady-state dissolved oxygen (e.g., 0.15 mg / L), and the triggering conditions of the other three abnormal operating conditions mentioned above are not met, the state machine determines that the system enters the short-range nitrification steady-state operation state under steady-state energy-saving operation.

[0041] Based on the current output state of the state machine, the cloud-based digital twin module generates corresponding feedforward control commands and sends them to the edge control layer. These feedforward control commands contain specific equipment adjustment parameter settings. When the state machine detects an abnormal nitrite accumulation condition or a rising trend in nitrite nitrogen, the feedforward adaptive frequency adjustment module within the cloud-based digital twin module is activated. This module calculates control parameters to implement control strategies that reduce the aeration fan frequency and adjust the recirculation ratio.

[0042] Specifically, the feedforward adaptive frequency adjustment module calculates the deviation parameter between the target nitrite accumulation rate and the predicted current nitrite accumulation rate, and executes the following control law equation: ; in, This indicates the target adjustment frequency of the aeration blower calculated and output by the feedforward adaptive frequency adjustment module. This indicates the current initial operating frequency of the aeration blower. This represents the preset frequency adjustment weighting coefficient. This indicates the target nitrite accumulation rate set by the system. This indicates the predicted current nitrite accumulation rate.

[0043] This equation uses the difference between the system's expected target nitrite accumulation rate and the currently predicted actual nitrite accumulation rate as a negative feedback term, weighted by coefficients. After scaling, the current operating frequency of the aeration blower is... Dynamic reduction is performed. The cloud-based digital twin module will calculate the optimal stepping frequency command. It is encapsulated in the feedforward control command data packet.

[0044] Simultaneously, the cloud-based digital twin module also generates control commands for the recirculation ratio parameter. The physical correlation calculation equation for the recirculation ratio is configured as follows: ; in, This represents the calculated reflux ratio. This indicates the real-time collected return flow rate delivered by the return pump. This indicates the inflow rate collected by the system at the same time.

[0045] When the system is under high load impact conditions, the cloud-based digital twin module calculates the required increase in return flow rate while calculating the frequency reduction command, so that the return flow ratio... Increase the frequency to meet the hydraulic load scheduling requirements of the system. The aforementioned target adjustment frequency and return ratio setpoint together constitute the feedforward control command, which is sent to the edge control layer via a wide area network or dedicated line network.

[0046] The edge control module is deployed in a field control cabinet or local computer room close to the physical entity layer. This module mainly consists of a programmable logic controller (PLC) or an industrial edge computing gateway. The edge control module is equipped with a communication interface and a lower-level control program, communicating with the cloud-based digital twin module, the sensing and acquisition module, and the execution and driving module, respectively. The edge control module receives feedforward control commands from the cloud-based digital twin module and executes the command parsing locally. Due to the mechanical inertia and electrical limitations of industrial field equipment, the edge control module adjusts the received target frequency (such as the output of the feedforward adaptive frequency adjustment module) accordingly. The signal is compared with the local set maximum / minimum frequency limit of the inverter, and a device drive signal (such as a 4-20mA analog input signal or Modbus / Profibus digital communication write command) is generated within the limit range.

[0047] The execution drive module is located at the physical entity layer and connected to the output terminals of the edge control module. The execution drive module includes an aeration blower, an influent pump, and a return pump, as well as a matching motor frequency converter assembly. When the edge control module sends a control signal, the frequency converter configured at the front end of the aeration blower receives the signal and changes the stator winding power supply frequency to smoothly reduce the aeration blower speed. Similarly, the frequency converters or contactors controlling the influent pump and return pump receive commands to change the pump's operating frequency or the number of pumps in operation, dynamically increasing the return ratio. By adjusting the corresponding aeration blower and influent fluid dynamic equipment, the system controls the gas and liquid phase mass transfer input in the primary aerobic tank, thereby maintaining the target low-oxygen operating range. The dissolved oxygen concentration in this target low-oxygen operating range is limited to 0.08 to 0.15 mg / L. In environments with physical detection interference, this specific oxygen partial pressure control strategy can directionally protect the biological activity of nitrifying bacteria and inhibit the biological activity of nitrifying bacteria. The feedforward control command takes effect before the physical time point when the test indicators substantially exceed the standard, thus implementing control intervention.

[0048] To ensure the long-term stability of the cloud-based model's prediction accuracy, this embodiment's system is configured with a parameter closed-loop correction mechanism. The cloud-based digital twin module is equipped with a data feedback receiving interface. The edge control module packages its internal real-time sensor data and actual equipment operating status feedback values, and continuously uploads them to the cloud-based digital twin module as actual operating data through this interface. In addition, a laboratory data entry terminal is configured on-site. Operators periodically upload the test results of effluent indicators (such as the actual concentrations of nitrate nitrogen and nitrite nitrogen obtained from offline manual testing) through the entry terminal. The cloud-based digital twin module marks the received periodic test results as measured baseline data.

[0049] The cloud-based digital twin module is equipped with a model parameter update algorithm. When new actual effluent test measurements are received, the algorithm engine extracts the historical predicted test index values ​​corresponding to the timestamp, calculates the error between the two, and periodically and dynamically corrects the parameters of the virtual reaction tank model using the following correction formula: ; in, This indicates the updated parameter set of the virtual reaction tank model generated after this round of corrections. This represents the currently active set of parameters for the virtual reaction tank model. This indicates the step size coefficient for correcting the model parameters. This indicates the actual effluent test measurement values ​​obtained through offline laboratory testing. This represents the predicted test index values ​​synchronously output by the virtual reaction pool model at the same sampling time.

[0050] Parameter set Includes multivariate prediction functions The algorithm incorporates internal weights, bias terms, and biochemical reaction kinetic rate constants. When a residual exists between the measured and predicted values, the update algorithm uses iterative methods such as gradient descent or Kalman filtering, employing a set correction step size coefficient. Update the parameter set. Corrected model parameters. It will be loaded into the metabolic window locking model for calculation in the next time period, thereby ensuring the consistency between the virtual reaction pool model and the long-term evolutionary state of the biological community in the physical entity layer.

[0051] Based on the above system architecture and the interaction logic of its modules, this embodiment of the invention also provides a method for precise control of short-range nitrification in the first-stage aerobic zone based on digital twin-driven principles. For example... Figure 3 The operation flow of the anaerobic ammonia oxidation system is shown below. This method is applied to the aforementioned control system and runs as an executable instruction sequence in industrial computing devices equipped with processors and memory, as well as cloud servers. The specific execution flow of the control method includes the following technical steps.

[0052] In the first execution phase (S1), the system performs multi-source data synchronization and feature decoupling operations. The sensing and acquisition module acquires discrete-period analog measurement voltage or current signals at a preset sampling frequency through a multi-dimensional sensing array (online pH probe, online redox potential probe, and online dissolved oxygen probe) deployed in the primary aerobic tank. The front-end analog-to-digital conversion module digitizes the analog signals, generating preliminary dissolved oxygen, pH, and redox potential data. At this stage, the data typically contains broadband noise caused by physical environmental factors (such as the impact of aeration airflow on the permeable membrane). The sensing and acquisition module adds a time-series stamp and device physical address identification code to each data frame, forming a communication data packet. This communication data packet is sent to the data buffer register of the edge control module based on the industrial Ethernet network communication protocol, and simultaneously transmitted through a secure tunnel to the message queue receiver of the cloud digital twin module in the cloud layer.

[0053] In the second execution phase (S2), the cloud-based digital twin module performs data cleaning and key feature extraction logic. The cloud receiving process pulls raw time-series data blocks from the message queue and loads them into the internally configured signal denoising and feature decoupling unit. First, the processing unit applies a sliding window filtering algorithm to the dissolved oxygen data channel. This algorithm extracts the values ​​of two consecutive sampling points and performs a subtraction operation to obtain the absolute value of the difference. When the comparison logic determines that this absolute value is greater than an anomaly threshold... When the parameter range is pre-configured to be between 0.3 and 0.5 mg / L, the dissolved oxygen measurement value corresponding to the current timestamp is identified as invalid abnormal fluctuation data caused by physical interference from bubbles. The processing unit then extracts the most recent value within a fixed time window. The system calculates the arithmetic mean of historical sampling points that were not marked as abnormal, and then injects the calculated average into the original timestamp data as a smoothing substitute value, thereby eliminating abnormal fluctuations. Subsequently, the processing unit extracts the time series sequences of pH and redox potential data. By performing polynomial fitting or direct difference calculation on the data sequences within a fixed time window, the system obtains the slope of pH changes and the trend of redox potential changes within that time window, respectively. Since the biochemical reaction process in the water alters the pH and oxidized state balance in the mixed solution, these slopes and trends are encapsulated and defined as key biochemical characteristics characterizing the current biochemical activity of the system.

[0054] In the third execution phase (S3), the system performs virtual simulation and proactive situation prediction. The cloud-based digital twin module extracts the pre-processed, denoised dissolved oxygen data, influent flow rate, reflux ratio data, system temperature, and the two key biochemical characteristics mentioned above. These data are fed into the internal metabolic window locking model as input vectors. This model numerically reconstructs the biochemical and physical processes of the primary aerobic tank in virtual space. Based on the configured multivariate nonlinear prediction function, the model outputs two-dimensional prediction results within the current calculation cycle: the first dimension is the real-time estimate of the nitrite nitrogen accumulation rate at the current time point; the second dimension is the biochemical situation change curves of the nitrite nitrogen accumulation rate and effluent nitrite nitrogen concentration for a future set time period (e.g., within the next 120 minutes) obtained based on a time series extrapolation algorithm. Subsequently, the state machine within the cloud-based digital twin module combines the influent parameters and prediction results to compare with preset high load thresholds, low temperature thresholds, and safety warning ratios. When the state machine detects that the system state parameters meet any configuration condition, it automatically updates the system operating condition flag to the corresponding item among high load impact condition, low temperature suppression condition, nitrite accumulation abnormal condition, or steady-state energy-saving condition.

[0055] In the fourth execution phase (S4), the system issues abnormal early warning commands and implements virtual-real closed-loop control. When the operating condition indicator shows that the system is in an abnormal nitrite accumulation state, or when the model predicts that the effluent nitrite nitrogen trend curve will exceed the preset tolerance upper limit threshold at some future time, the feedforward control command generation program is triggered. The feedforward adaptive frequency adjustment module inside the program calculates the control deviation between the currently predicted nitrite nitrogen accumulation rate and the set target value, introduces preset weighting coefficients to calculate the frequency adjustment reduction of the aeration blower for the current moment, and generates the target adjustment frequency command based on this. At the same time, based on the real-time ammonia nitrogen load, a new return flow rate and influent flow rate ratio is calculated, and a return flow ratio adjustment command is generated. The cloud sends a data packet containing the above command sequence to the edge control layer. The edge control module receives the data packet and parses out the frequency and flow rate setpoints. The output logic port of the edge control module generates analog standard signals or bus communication control words for the control equipment according to the setpoints. After receiving the control word, the corresponding execution drive module (including aeration blowers and return pumps connected to the frequency converter, and other influent fluid power equipment) performs the action. The rotor speed of the aeration blower decreases according to a set acceleration / deceleration time curve, reducing the volumetric air flow rate supplied to the reaction tank; the return pump changes its speed to increase the mixed liquor return flow rate. Through the coordinated action of mechanical and electrical components, the gas-liquid mass transfer efficiency and hydrodynamic residence time in the primary aerobic tank are altered. This physical control result causes the dissolved oxygen concentration in the tank to decrease and stabilize within the target low-oxygen operating range of 0.08 to 0.15 mg / L. Under this dissolved oxygen gradient, oxygen permeation to the surface and interior of the biofilm in the aerobic tank is restricted, thereby inhibiting the proliferation and oxidative activity of nitrifying bacteria and maintaining a metabolic environment dominated by nitrifying bacteria. To visually demonstrate the effectiveness of the above control strategy, such as... Figure 4 As shown in the graph of the nitrite nitrogen accumulation rate over time, when faced with system fluctuations, the system using the traditional method exhibits a significant irreversible decline in the nitrite nitrogen accumulation rate (falling below 60%), while the system using the method of this invention can continuously and stably control the nitrite nitrogen accumulation rate (NAR) at an extremely high level (above 95%), successfully avoiding nitrification instability.

[0056] Meanwhile, the control method includes a long-term data correction feedback process. During continuous equipment operation, operators periodically input the laboratory's offline test results (i.e., actual effluent indicator measurements) into the cloud database system. The cloud-based digital twin module synchronously extracts the model's predicted indicator values ​​at the corresponding test sampling time and calculates the error term between the actual measured value and the predicted value. Based on the set step size correction coefficient, the parameter update algorithm incrementally modifies the underlying state transition matrix or function weight parameter set of the virtual reaction tank model. The updated parameters overwrite the original model configuration and are stored in memory for the next real-time prediction calculation, completing the overall data closed-loop flow from on-site control execution to model adaptive correction.

[0057] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A precision control system for short-range nitrification in the primary aerobic zone based on digital twin-driven architecture, characterized in that, include: The sensing and acquisition module is used to collect dissolved oxygen data, pH data, and redox potential data in the primary aerobic tank in real time. The cloud-based digital twin module, which is communicatively connected to the sensing and acquisition module, is used to denoise the dissolved oxygen data to eliminate abnormal fluctuations caused by bubble interference, and to extract the slope of the pH data and the trend of the redox potential data as key biochemical features. The cloud-based digital twin module is also used to construct a virtual reaction pool model based on the denoised dissolved oxygen data and the key biochemical features, and to predict the current nitrite accumulation rate and the biochemical situation within a set future time period through multi-parameter fusion. The edge control module is communicatively connected to the cloud-based digital twin module. It is used to generate feedforward control commands based on the prediction results of the cloud-based digital twin module. When an abnormal accumulation of nitrite is predicted or the system shows a trend of increasing nitrate nitrogen, it issues control signals in advance to reduce the frequency of the aeration blower and adjust the reflux ratio to maintain the target low-oxygen operating range in the primary aerobic tank. This allows for the targeted protection of nitrifying bacteria activity and the inhibition of nitrifying bacteria activity in an environment where dissolved oxygen is subject to physical detection interference. The execution drive module is used to receive the control signal and adjust the corresponding aeration blower and inlet fluid power equipment accordingly.

2. The precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin drive according to claim 1, characterized in that, The cloud-based digital twin module uses a combination of sliding window filtering and anomaly threshold removal to denoise the dissolved oxygen data. When the change in dissolved oxygen data between consecutive sampling points exceeds a preset threshold, the corresponding data is determined to be an abnormal fluctuation caused by bubble interference, and the average value of adjacent time windows is automatically used to replace the abnormal fluctuation data.

3. The precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin drive according to claim 1, characterized in that, The cloud-based digital twin module is also used to automatically identify the system under any of the following conditions based on the prediction results of the nitrite accumulation rate and the biochemical state: high load impact condition, low temperature suppression condition, abnormal nitrite accumulation condition, or steady-state energy-saving condition. Specifically, when the redox potential data remains in the negative range and the denoised dissolved oxygen data is lower than a set threshold, the system is determined to have entered the short-range nitrification steady-state operation state under the steady-state energy-saving condition.

4. The precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin drive according to claim 1, characterized in that, The cloud-based digital twin module also continuously receives test results and real-time sensor data transmitted back by the sensing and acquisition module. Based on the error between the measured value of the actual effluent data and the predicted value of the virtual reaction tank model, it periodically and dynamically corrects the parameters of the virtual reaction tank model in combination with a preset correction coefficient.

5. A precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin drive, as described in claim 1, is characterized in that... The specific logic behind the cloud-based digital twin module dynamically adjusting the frequency of the aeration blower based on the prediction results is as follows: The deviation between the target nitrite nitrogen accumulation rate and the predicted current nitrite nitrogen accumulation rate is calculated, and the initial operating frequency of the aeration blower is adaptively lowered in combination with a preset adjustment coefficient, and a feedforward frequency conversion control command is output.

6. A precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin drive, as described in claim 1, is characterized in that... The edge control module acquires the actual influent flow rate and return flow rate to calculate the current return ratio. When the system detects a high load impact condition, it coordinates with the frequency reduction control signal of the aeration blower to dynamically increase the return ratio to ensure stable low-oxygen operation of the system.

7. A precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin drive, as described in claim 3, is characterized in that... When the influent ammonia nitrogen concentration exceeds the set high concentration threshold and the influent flow rate exceeds a set multiple of the average flow rate, the cloud-based digital twin module determines that the system has entered the high-load impact condition. When the predicted current nitrite accumulation rate is lower than the safety warning ratio, the system is determined to have a risk of nitrification instability and the feedforward control command is triggered.

8. A precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin drive, as described in claim 1, is characterized in that... The sensing and acquisition module includes a multi-dimensional sensing array arranged in the primary aerobic tank, which consists of an online pH probe, an online oxidation-reduction potential probe, and an online dissolved oxygen probe; the edge control module includes a programmable logic controller. The execution drive module includes the aeration blower, inlet pump, and return pump, which are controlled by the programmable logic controller via a frequency converter.

9. A precise control system for short-range nitrification in the first-stage aerobic zone based on digital twin drive, as described in claim 1, is characterized in that... The dissolved oxygen concentration in the target low-oxygen operating range is limited to 0.08 to 0.15 mg / L; the feedforward control command is used to implement proactive control intervention before the effluent test indicators substantially exceed the standards.

10. A method for precise control of short-cut nitrification in the first-stage aerobic zone based on digital twin-driven technology, characterized in that... The method applied to the digital twin-driven first-stage aerobic zone short-range nitrification precision control system according to any one of claims 1 to 9 includes the following steps: S1. Real-time data on dissolved oxygen, pH, and redox potential in the primary aerobic tank are collected through the sensing and acquisition module. S2. The dissolved oxygen data is denoised using the cloud-based digital twin module to remove abnormal fluctuations, and the slope of the pH data and the trend of the redox potential data are extracted as key biochemical features. S3. Based on the denoised dissolved oxygen data and the key biochemical characteristics, construct a virtual reaction tank model to predict the current nitrite accumulation rate and the biochemical situation within a set future time period. S4. The edge control module generates feedforward control commands based on the prediction results, and adjusts the frequency and reflux ratio of the aeration blower in advance when abnormal nitrite accumulation or nitrate nitrogen increase trend is predicted, so as to maintain the target low oxygen operating range in the primary aerobic tank, thereby protecting the activity of nitrifying bacteria and inhibiting the activity of nitrifying bacteria.

Images