A sewage biological denitrification and dephosphorization system

Abstract

The present application provides a kind of sewage biological denitrification and phosphorus removal system, it relates to wastewater treatment technical field.The system includes on-line monitoring module, carbon source demand calculation module, carbon source distribution control module, biological reaction adjustment module and self-learning optimization module connected in sequence.On-line monitoring module collects the operating parameters of each process section in real time;Carbon source demand calculation module dynamically calculates the carbon source demand ratio by analyzing the rate change of denitrification and phosphorus accumulation process;Carbon source distribution control module automatically adjusts the carbon source dosage according to the demand ratio;Biological reaction adjustment module adjusts the aeration and hydraulic parameters in linkage, promotes the metabolic complementation of denitrification and phosphorus removal;Self-learning optimization module continuously optimizes the system operation strategy through deep learning.The present application realizes the intelligent distribution of carbon source between denitrifying bacteria and phosphorus accumulating bacteria, solves the problem of low efficiency of denitrification and phosphorus removal caused by carbon source competition, and significantly improves the wastewater treatment efficiency and operation stability.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, specifically to a biological nitrogen and phosphorus removal system for wastewater. Background Technology

[0002] In the field of biological wastewater treatment, achieving efficient simultaneous nitrogen and phosphorus removal is a complex and challenging task. Currently, the addition of carbon sources in this field is often roughly adjusted based on the influent flow rate or a single concentration of a specific pollutant; key operating parameters such as aeration and hydraulic retention time are also mostly set based on experience or adjusted slightly within a certain range. This control method treats the denitrification and polyphosphate accumulation processes as two relatively independent or even competitive stages, lacking the ability to finely perceive and coordinate the dynamic competitive relationship between the two in carbon source utilization. However, because it is impossible to identify the instantaneous proportional changes in the carbon source demand of denitrifying bacteria and polyphosphate-accumulating bacteria in real time, the system struggles to make optimal allocation decisions when carbon sources are limited. This often leads to a decline in metabolic efficiency of one group of bacteria due to insufficient carbon sources, while the other group may over-consume carbon sources, resulting in resource waste and fluctuations in effluent quality. Summary of the Invention

[0003] To achieve the above objectives, the present invention provides the following technical solution: a wastewater biological nitrogen and phosphorus removal system, comprising:

[0004] The online monitoring module is used to acquire real-time operational data from the influent, anaerobic zone, anoxic zone, and effluent during the wastewater biological reaction process. The operational data includes dissolved oxygen concentration, oxidation-reduction potential, nitrate nitrogen, nitrite nitrogen, total phosphorus, and volatile fatty acid concentration. The collected data are input into the carbon source demand calculation module in chronological order.

[0005] The carbon source demand calculation module constructs a dynamic discrimination logic for carbon source competition based on multidimensional data. By analyzing the changing trend of nitrate reduction rate in the denitrification process and the relative changes of phosphorus release rate and phosphorus uptake rate in the polyphosphate accumulation process, it calculates the instantaneous carbon source demand ratio of denitrifying bacteria and polyphosphate accumulation bacteria. Based on this, it prioritizes the allocation of carbon sources to the anoxic or anaerobic zone, or maintains the original ratio at equilibrium.

[0006] The carbon source distribution control module is connected to the carbon source injection device and the mixing aeration unit. It is used to calculate the distribution signal output by the module according to the carbon source demand, and automatically adjust the carbon source injection point and injection rate to realize dynamic carbon source supply in different reaction zones. Under the condition of insufficient carbon source, priority is given to ensuring the organic carbon absorption of polyphosphate bacteria in the anaerobic stage to maintain the energy storage and metabolic activity of polyphosphate bacteria. In the denitrification stage, the reduction efficiency of nitrate is maintained by compensating with trace carbon sources.

[0007] The biological reaction regulation module is linked with the aeration system and hydraulic regulation device. Based on the real-time carbon source distribution status, it adjusts the dissolved oxygen setpoint, aeration time ratio and hydraulic retention time, so that the denitrification process and the phosphorus removal process form a complementary relationship in metabolic rhythm.

[0008] The self-learning optimization module collects and analyzes data throughout the entire operation cycle, establishes a parameter-performance correlation system through a deep learning model, identifies the carbon source allocation characteristics under different influent water quality conditions, and corrects the decision threshold in the carbon source demand calculation module, so that it gradually forms the ability to regulate different influent conditions.

[0009] The online monitoring module is used to acquire real-time operational data from the influent, anaerobic zone, anoxic zone, and effluent during the wastewater biological reaction process. This operational data includes, but is not limited to, dissolved oxygen concentration, oxidation-reduction potential, nitrate nitrogen, nitrite nitrogen, total phosphorus, and volatile fatty acid concentration. The online monitoring module performs high-frequency data acquisition through a multi-parameter water quality sensor array deployed in the corresponding reaction zones. The sampling frequency can be dynamically adjusted according to the influent load, ranging from 1 to 5 minutes per sampling. The acquired data are timestamped in chronological order and input into the carbon source demand calculation module to ensure the temporal consistency of the data stream, providing a reliable data foundation for subsequent dynamic analysis.

[0010] The carbon source demand calculation module constructs a dynamic discrimination logic for carbon source competition based on multidimensional data. By analyzing the changing trend of nitrate reduction rate during denitrification and the relative changes of phosphorus release and uptake rates during polyphosphate accumulation, a carbon source demand assessment model based on microbial metabolic kinetics is established. Specifically, the nitrate reduction rate is obtained by calculating the decrease in nitrate concentration per unit time in real time, while the phosphorus release and uptake rates are obtained by monitoring the increase rate of phosphorus concentration in the anaerobic zone and the decrease rate of phosphorus concentration in the aerobic zone, respectively. By comparing the instantaneous gradient of these two rates, the instantaneous carbon source demand ratio between denitrifying bacteria and polyphosphate-accumulating bacteria is calculated. This ratio calculation not only considers the rate value at the current moment but also incorporates the changing trend over a period of time for weighted correction to avoid misjudgment caused by instantaneous fluctuations. Based on this, a carbon source allocation instruction is generated, prioritizing the allocation of carbon sources to the anoxic or anaerobic zone, or maintaining the original allocation ratio when the demand of the two zones is balanced.

[0011] The carbon source distribution control module is connected to the carbon source injection device and the mixing aeration unit via an industrial fieldbus. Based on the distribution signal output by the carbon source demand calculation module, it automatically adjusts the selection of carbon source injection points and the injection rate, enabling simultaneous injection at the beginning of the anaerobic zone, the beginning of the anoxic zone, or multiple points to achieve dynamic carbon source supply to different reaction zones. The module incorporates a priority control strategy. When the system determines that the total carbon source is insufficient, it prioritizes ensuring the organic carbon absorption of polyphosphate-accumulating organisms (PAOs) during the anaerobic stage to maintain their energy storage and metabolic activity, preventing irreversible decline in phosphorus removal efficiency due to activity loss. Simultaneously, during the denitrification stage, a trace carbon source compensation program is activated to inject the minimum necessary amount of carbon source into the anoxic zone to maintain basic nitrate reduction efficiency and avoid nitrate nitrogen accumulation.

[0012] The biological reaction regulation module is linked with the aeration system and hydraulic regulation device through a control loop. Based on the real-time carbon source distribution status, it adjusts the dissolved oxygen setpoint, aeration time ratio, and hydraulic retention time. The core control objective of this module is to create a complementary metabolic rhythm between the denitrification and phosphorus removal processes through parameter adjustments. For example, when the carbon source is preferentially allocated to the anaerobic zone, the module correspondingly lowers the dissolved oxygen setpoint in the subsequent aerobic zone and appropriately extends the hydraulic retention time in the anaerobic stage, creating more favorable conditions for the full release of phosphorus and carbon absorption by polyphosphate-accumulating bacteria. Conversely, when the carbon source is tilted towards the anoxic zone, the lower limit of the dissolved oxygen setpoint is increased and the aeration sequence is optimized to ensure the full progress of the nitrification process and provide sufficient substrate for denitrification.

[0013] The self-learning optimization module collects and stores data throughout the entire operation cycle. It establishes a nonlinear correlation system between system operating parameters and nitrogen and phosphorus removal performance indicators by constructing a deep learning model. This deep learning model adopts a long short-term memory network structure to identify carbon source allocation characteristics under different influent water quality conditions, such as low carbon-nitrogen ratio and high phosphorus load. Through periodic model training and verification, it outputs correction parameters for the decision threshold in the carbon source demand calculation module, enabling it to gradually form an adaptive control capability for influent water quality fluctuations and realize the evolution from rule-based control to data-driven intelligent control.

[0014] As a further technical solution, the carbon source demand calculation module is equipped with a dynamic threshold update unit. When the denitrification and polyphosphate accumulation stages alternate, this unit uses time series data on dissolved oxygen, redox potential, and nitrate reduction rate from the historical operating database to analyze the time offset patterns through correlation analysis. Based on these patterns, the trigger threshold for priority carbon source allocation is slightly adjusted, with the adjustment range not exceeding ±15% of the original threshold. This dynamic threshold update process compares the carbon source utilization rate change trend of the current operating cycle with that of the previous cycle, and calculates the difference range between the carbon source utilization saturation point of denitrifying bacteria and the carbon storage stability point of polyphosphate accumulation bacteria using a sliding window algorithm. The stable interval of this difference over multiple consecutive cycles is used as the new decision interval, thereby freeing the carbon source allocation strategy from dependence on fixed proportional parameters and achieving flexible control based on the actual operating state of the system.

[0015] As a further technical solution, the online monitoring module includes a set of time-synchronized sampling logic. This logic coordinates the sampling devices at the inlet and outlet through a unified clock signal to achieve synchronous collection of key water quality indicators within the same hydraulic retention cycle. By normalizing the dissolved oxygen, total phosphorus, and volatile fatty acid concentration data within the same cycle, analytical biases caused by differences in dimensions and orders of magnitude are eliminated, generating a carbon source utilization efficiency sequence reflecting the conversion efficiency of carbon sources in the system. The carbon source demand calculation module identifies the temporal characteristics of carbon source consumption in different reaction zones based on this sequence. When the peak carbon source consumption peak is identified by the peak detection algorithm as being earlier or later than the expected time point by more than one sampling cycle, the data sampling time step is automatically adjusted. By shortening or extending the sampling interval, the response rate of the subsequent carbon source prediction model is corrected, ensuring that the control system can track changes in the metabolic state of microorganisms in a timely manner.

[0016] As a further technical solution, the carbon source allocation control module is equipped with feedback regulation logic. This logic monitors the fluctuation range of oxidation-reduction potential in the anoxic and anaerobic zones in real time after the carbon source injection rate changes. A dynamic stability range of oxidation-reduction potential based on historical stable operating data is set. When the fluctuation range of oxidation-reduction potential in any region exceeds the set range for three consecutive sampling cycles, a micro-compensation program is triggered. Before the carbon source supply reaches stability again, this program temporarily limits the output intensity of the aeration system and extends the design hydraulic retention time of the anoxic zone by 10%-20%, allowing denitrifying bacteria and polyphosphate-accumulating bacteria communities sufficient time to re-establish metabolic balance under the new carbon source allocation ratio. The micro-changes in oxidation-reduction potential indirectly reflect the real-time fluctuations of microbial metabolic state, achieving synchronous convergence of carbon source allocation operation and biological and biochemical reactions.

[0017] As a further technical solution, the biological reaction regulation module also includes a phased aeration adjustment unit. This unit is activated when the carbon source demand calculation module determines that polyphosphate-accumulating bacteria have a dominant carbon source demand. After activation, the dissolved oxygen setpoint in the aerobic zone is reduced by 0.5-1.0 mg / L, and the actual hydraulic retention time in the anaerobic stage is temporarily extended by 15%-30% by controlling the inlet valve or the internal return valve. Based on the carbon source consumption rate curve output by the carbon source demand calculation module, the aeration adjustment unit judges the carbon source consumption trend and slows down the aeration intensity in advance before the predicted peak carbon absorption of polyphosphate-accumulating bacteria is reached, so as to achieve a smooth metabolic transition from the anaerobic environment to the hypoxic environment. This control strategy allows polyphosphate-accumulating bacteria to complete most of the energy storage when the peak carbon source distribution is reached. Subsequently, denitrifying bacteria can make full use of the intermediate products released by polyphosphate-accumulating bacteria metabolism, such as low molecular weight fatty acids, to carry out denitrification reactions in the hypoxic stage, thereby forming a dynamic and complementary biological metabolic rhythm in time sequence.

[0018] As a further technical solution, the self-learning optimization module, when modeling the operating cycle data using deep learning, establishes an operating mode recognition logic by utilizing the difference characteristics between data from multiple cycles. This logic analyzes the changing trends of the influent total nitrogen to total phosphorus ratio, classifying it into different water quality modes, such as high-nitrogen low-carbon mode and nitrogen-phosphorus balanced mode, and selects the most suitable carbon source allocation strategy verified in historical data as the initial control parameter for each mode. When the online monitoring module detects that the rate of change of key water quality parameters, such as influent COD and ammonia nitrogen, exceeds the preset range, the system temporarily switches from the conventional optimization mode to the high-response mode. In this mode, the data update frequency is doubled, and the decision adjustment cycle of the carbon source allocation control module is shortened accordingly, thereby quickly adapting to sudden changes in influent water quality and preventing a sharp deterioration in system performance.

[0019] As a further technical solution, the deep learning model of the self-learning optimization module adopts a phased target update logic during parameter training. This logic divides the training process into three progressive stages: in the early stage of training, the carbon source utilization rate of the system is taken as the main optimization target, aiming to quickly establish the basic mapping relationship between carbon source addition and consumption; in the middle stage of training, the optimization criterion is to minimize the difference between the denitrification rate and the phosphorus absorption rate, guiding the model to learn how to balance the carbon source demand of the two competing processes; in the later stage of training, the comprehensive ratio of the system's unit energy consumption and nitrogen and phosphorus removal rate is taken as the final optimization target, driving the model to achieve the leap from local optimization to global energy efficiency optimization. This phased target update logic enables the self-learning process to proceed gradually, transitioning from basic carbon source efficiency control to complex comprehensive optimization of the entire process.

[0020] As a further technical solution, a two-way data verification process is provided between the carbon source demand calculation module and the self-learning optimization module to periodically correct prediction deviations. This process is automatically executed at the end of each complete operating cycle. The system analyzes the difference between the actual operation records of the carbon source allocation control module and the corresponding effluent nitrogen and phosphorus removal results, such as total nitrogen and total phosphorus removal rates. When the difference exceeds the dynamic threshold set based on historical performance data, the self-learning optimization module starts the correction program, regenerates a new set of prediction correction parameters and feeds them back to the decision core of the carbon source demand calculation module, replacing the poorly performing parameter set. This forms a closed-loop optimization system from decision-making to execution to verification and correction, ensuring the continuous improvement and long-term stability of system performance.

[0021] This invention provides a biological nitrogen and phosphorus removal system for wastewater, which has the following beneficial effects:

[0022] 1. This invention solves the problem of rigid carbon source allocation caused by the inability to perceive the competitive situation between denitrifying bacteria and polyphosphate-accumulating bacteria in real time by constructing a dynamic carbon source demand discrimination logic based on multidimensional real-time data and calculating the instantaneous carbon source demand ratio, thereby improving the accuracy and utilization efficiency of carbon source addition.

[0023] 2. This invention solves the problem of the difficulty in coordinating the metabolic rhythms of the two biological processes of denitrification and phosphorus removal by dynamically adjusting the carbon source allocation according to the instantaneous carbon source demand ratio and adjusting biological reaction parameters such as dissolved oxygen and hydraulic retention time in conjunction with it. It improves the metabolic complementarity between different reaction units within the plant, so that the energy reserves of polyphosphate-accumulating bacteria and the nitrate reduction process of denitrifying bacteria are dynamically coupled. This improves the overall stability of nitrogen and phosphorus removal while reducing the risk of fluctuations caused by endogenous competition.

[0024] 3. This invention solves the core bottleneck of insufficient adaptive capability when facing complex and variable water inflow conditions by establishing a parameter and performance correlation system through a self-learning optimization module and using a deep learning model to continuously correct the decision threshold, thereby improving the system's ability to autonomously learn from dynamic operating data and optimize its core control strategy. Detailed Implementation

[0025] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] The online monitoring module acquires real-time operational data, including dissolved oxygen concentration, oxidation-reduction potential, nitrate nitrogen concentration, nitrite nitrogen concentration, total phosphorus concentration, and volatile fatty acid concentration, through sensors deployed at the inlet, anaerobic zone, anoxic zone, and outlet. The dissolved oxygen sensor has a measurement accuracy of ±0.1 mg / L, the oxidation-reduction potential measurement range covers -1000 mV to 1000 mV, and the nitrate nitrogen analyzer has a detection limit of 0.1 mg / L. All sensors are synchronized via a unified clock signal, and the data acquisition frequency is set to once every 3 minutes. After filtering, the acquired data is transmitted to the carbon source demand calculation module in chronological order.

[0027] The carbon source demand calculation module constructs a dynamic discrimination logic for carbon source competition based on received multidimensional data. This module calculates the instantaneous carbon source demand ratio between denitrifying bacteria and polyphosphate-accumulating bacteria by analyzing the changing trend of nitrate reduction rate during denitrification and the relative changes in phosphorus release and uptake rates during polyphosphate accumulation. The nitrate reduction rate is calculated using monitoring data from the most recent 30 minutes, with the slope of change obtained through a linear regression algorithm. The phosphorus release rate is calculated by monitoring the rate of increase in phosphorus concentration in the anaerobic zone. When the nitrate reduction rate... When the phosphorus release rate continues to increase while the rate of phosphorus release remains stable, it is determined that the carbon source demand of denitrifying bacteria is dominant; when the phosphorus release rate increases rapidly while the nitrate reduction rate does not change much, it is determined that the carbon source demand of polyphosphate-accumulating bacteria is dominant; the system sets the carbon source demand ratio thresholds to 1.15 and 0.85. When the ratio value calculated for three consecutive monitoring cycles is greater than 1.15, an instruction is generated to preferentially allocate the carbon source to the anoxic zone; when the ratio value is less than 0.85, an instruction is generated to preferentially allocate the carbon source to the anaerobic zone; when the ratio value is between 0.85 and 1.15, the original allocation ratio is maintained.

[0028] The carbon source distribution control module automatically adjusts the carbon source injection point and injection rate based on the distribution signal output by the carbon source demand calculation module. This module is directly connected to the carbon source injection device and independently controls the carbon source addition in the anaerobic and anoxic zones. During implementation, when it is necessary to increase the carbon source addition in the anoxic zone, the addition rate at the corresponding injection point is increased by 30% from the baseline value, while the addition rate in the anaerobic zone is decreased by 30%. Under conditions of insufficient carbon source, the system prioritizes the carbon source supply to the anaerobic zone, ensuring that its distribution ratio is not less than 60% of the total carbon source. Simultaneously, during the denitrification stage, the minimum necessary amount of carbon source is maintained, and a micro-carbon source compensation mechanism ensures the basic efficiency of the nitrate reduction reaction. The carbon source addition rate is adjusted gradually, with a system stabilization time allowed after each adjustment, and the control effect is verified by monitoring changes in the oxidation-reduction potential.

[0029] The bioreactor adjustment module adjusts the system's operating parameters based on the real-time carbon source distribution status. This module works in conjunction with the aeration system and hydraulic regulating device, adjusting the dissolved oxygen setpoint, aeration time ratio, and hydraulic retention time according to the carbon source distribution ratio in different reaction zones. When the carbon source is mainly distributed to the anaerobic zone, the bioreactor adjustment module lowers the dissolved oxygen setpoint in the aerobic zone from 2.0 mg / L to 1.3 mg / L and extends the hydraulic retention time in the anaerobic zone from 1.5 hours to 1.7 hours by adjusting the inlet valve. When the carbon source is mainly distributed to the anoxic zone, the dissolved oxygen setpoint in the aerobic zone is raised to 2.3 mg / L, while simultaneously increasing the internal reflux ratio of the mixed liquor. These adjustments create a complementary metabolic rhythm between the denitrification and phosphorus removal processes, improving the overall treatment efficiency of the system.

[0030] The self-learning optimization module collects and analyzes data throughout the entire operation cycle, establishing a correlation system between parameters and performance through a deep learning model. This module uses a time-series prediction model to analyze historical operating data and identify carbon source allocation characteristics under different influent water quality conditions. During the deep learning model training process, the system uses the balance between carbon source utilization rate, denitrification rate, and phosphorus absorption rate, as well as the overall system energy efficiency, as phased optimization targets. When influent water quality changes, the self-learning optimization module analyzes the changing trend of the total nitrogen to total phosphorus ratio and automatically selects the most suitable carbon source allocation strategy, which is an existing strategy based on experience and historical cases. When the rate of change of water quality parameters exceeds the normal range, the system switches to a high-response mode, increasing the data update frequency and shortening the carbon source adjustment cycle.

[0031] The online monitoring module includes time-synchronized sampling logic to ensure that key water quality indicators are collected at the inlet and outlet within the same hydraulic retention cycle. The collected dissolved oxygen, total phosphorus, and volatile fatty acid concentrations are normalized to generate a carbon source utilization efficiency sequence. When a significant deviation is detected between the peak carbon source consumption time and the expected time, the system automatically adjusts the data sampling time step from once every 3 minutes to once every 2 minutes to correct the response rate of subsequent carbon source predictions.

[0032] The carbon source demand calculation module includes a dynamic threshold update unit. This unit makes minor adjustments to the carbon source priority allocation threshold based on historical operating data as the denitrification and polyphosphate accumulation stages alternate. The adjustment process compares the carbon source utilization rate trends of the current cycle with those of the previous cycle to calculate the difference between the carbon source utilization saturation point of denitrifying bacteria and the carbon storage stability point of polyphosphate accumulation bacteria. The stable difference range is used as the new decision interval. For example, when a continuous increase in the carbon source utilization efficiency of denitrifying bacteria is detected, the carbon source demand ratio threshold is increased accordingly, making the system more inclined to allocate carbon sources to the anoxic zone.

[0033] The carbon source distribution control module is equipped with feedback regulation logic, which monitors the fluctuation range of oxidation-reduction potential in the anoxic and anaerobic zones in real time after the carbon source injection rate changes. When the fluctuation range of oxidation-reduction potential exceeds the set stable range, a micro-compensation program is triggered to temporarily limit the aeration intensity and prolong the reaction time in the anoxic zone. The stable range of oxidation-reduction potential is determined based on the statistical characteristics of historical data at this point, taking the range of 25% to 75% of the historical data.

[0034] When the bioreaction regulation module determines that polyphosphate-accumulating bacteria have a dominant carbon source demand, the phased aeration adjustment unit automatically reduces the dissolved oxygen setpoint and briefly extends the anaerobic stage residence time. Based on the rate of change output by the carbon source demand calculation module, it judges the carbon source consumption trend and slows down aeration before the peak carbon absorption of polyphosphate-accumulating bacteria to achieve metabolic transition between the anaerobic and anoxic zones. In specific implementation, when the concentration of volatile fatty acids is detected to begin to decrease, the aeration adjustment program is started to gradually reduce the dissolved oxygen concentration in the aerobic zone.

[0035] The deep learning model in the self-learning optimization module adopts a phased target update logic during parameter training. In the early stage of training, the carbon source utilization rate is the main optimization target. In the middle stage, the minimum difference between the denitrification rate and the phosphorus absorption rate is the optimization criterion. In the later stage, the comprehensive ratio of system energy consumption and nitrogen and phosphorus removal rate is the final target. The duration of each training stage is dynamically adjusted according to the model convergence. Typically, the early training lasts for 100 cycles, the middle training lasts for 50 cycles, and the later training lasts for 20 cycles.

[0036] A two-way data verification process is set between the carbon source demand calculation module and the self-learning optimization module. At the end of each operating cycle, the system performs a difference analysis between the carbon source allocation record and the corresponding nitrogen and phosphorus removal results. When the difference exceeds the set threshold, the self-learning optimization module regenerates new prediction correction parameters and feeds them back to the carbon source demand calculation module. The difference threshold is usually set at 15%. When the predicted value of total nitrogen or total phosphorus in the effluent deviates from the measured value for three consecutive cycles, the parameter correction procedure is triggered.

[0037] The coordinated operation of the various modules in this invention ensures that the system maintains excellent treatment performance under different influent conditions. At the same time, the self-learning function continuously optimizes the operating strategy to achieve efficient and stable biological nitrogen and phosphorus removal.

[0038] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A sewage biological denitrification and dephosphorization system, characterized in that, include: The online monitoring module is used to acquire real-time operational data from the influent, anaerobic zone, anoxic zone, and effluent during the wastewater biological reaction process. The operational data includes dissolved oxygen concentration, oxidation-reduction potential, nitrate nitrogen, nitrite nitrogen, total phosphorus, and volatile fatty acid concentration. The collected data are input into the carbon source demand calculation module in chronological order. The carbon source demand calculation module constructs a dynamic discrimination logic for carbon source competition based on multidimensional data. By analyzing the changing trend of nitrate reduction rate in the denitrification process and the relative changes of phosphorus release rate and phosphorus uptake rate in the polyphosphate accumulation process, it calculates the instantaneous carbon source demand ratio of denitrifying bacteria and polyphosphate accumulation bacteria. Based on this, carbon sources are preferentially allocated to anoxic or anaerobic areas, or the original ratio is maintained at equilibrium. The carbon source distribution control module is connected to the carbon source injection device and the mixing aeration unit. It is used to calculate the distribution signal output by the module according to the carbon source demand, and automatically adjust the carbon source injection point and injection rate to realize dynamic carbon source supply in different reaction zones. Under the condition of insufficient carbon source, priority is given to ensuring the organic carbon absorption of polyphosphate bacteria in the anaerobic stage to maintain the energy storage and metabolic activity of polyphosphate bacteria. In the denitrification stage, the reduction efficiency of nitrate is maintained by compensating with trace carbon sources. The biological reaction regulation module is linked with the aeration system and hydraulic regulation device. Based on the real-time carbon source distribution status, it adjusts the dissolved oxygen setpoint, aeration time ratio and hydraulic retention time, so that the denitrification process and the phosphorus removal process form a complementary relationship in metabolic rhythm. The self-learning optimization module collects and analyzes data throughout the entire operation cycle, establishes a parameter-performance correlation system through a deep learning model, identifies the carbon source allocation characteristics under different influent water quality conditions, and corrects the decision threshold in the carbon source demand calculation module, so that it gradually forms the ability to regulate different influent conditions.

2. The system for biological nitrogen and phosphorus removal from wastewater according to claim 1, characterized in that: The carbon source demand calculation module includes a dynamic threshold update unit. This unit adjusts the carbon source priority allocation threshold slightly based on the time offset between dissolved oxygen, redox potential, and nitrate reduction rate in historical operating data as the denitrification and polyphosphate accumulation stages alternate. This allows the system to automatically correct the carbon source supply strategy under conditions of significant water quality fluctuations. The dynamic threshold update process calculates the difference between the carbon source utilization saturation point of denitrifying bacteria and the carbon storage stability point of polyphosphate accumulation bacteria by comparing the current cycle with the previous cycle. The stable difference zone is used as the new decision interval, so that carbon source allocation no longer depends on a fixed ratio.

3. The system according to claim 1, characterized in that: The online monitoring module includes a set of time-synchronized sampling logic, which is used to collect key water quality indicators at both the inlet and outlet ends. By normalizing the concentrations of dissolved oxygen, total phosphorus, and volatile fatty acids within the same period, a carbon source utilization efficiency sequence is generated. The carbon source demand calculation module identifies the temporal characteristics of carbon source consumption in different reaction zones based on the carbon source utilization efficiency sequence. When the peak carbon source consumption is detected to be earlier or later, the response rate of subsequent carbon source prediction is corrected by adjusting the data sampling time step.

4. The system according to claim 1, characterized in that: The feedback adjustment logic is arranged in the carbon source distribution control module, and is used for monitoring the fluctuation amplitude of the redox potential of the anoxic zone and the anaerobic zone in real time after the carbon source injection rate is changed; when it is detected that the change amplitude of the redox potential exceeds a set stable interval, a trace compensation program is triggered, the aeration intensity is temporarily limited and the anoxic zone reaction time is prolonged before the carbon source supply is stabilized, so that the flora reestablishes metabolic balance under the new carbon source distribution ratio; the real-time fluctuation of the metabolic state of the microorganism is reflected through the trace change of the redox potential, and the synchronous convergence of the carbon source distribution and the biological reaction is realized.

5. The system for biological nitrogen and phosphorus removal from wastewater according to claim 1, characterized in that: The phase aeration adjustment unit is further arranged in the biological reaction adjustment module, and is used for automatically reducing the dissolved oxygen setting value and temporarily prolonging the anaerobic stage residence time when it is judged that the carbon source demand of the phosphorus accumulating bacteria is dominant; the aeration adjustment unit judges the carbon source consumption trend according to the change rate output by the carbon source demand calculation module, slows down the aeration in advance before the carbon absorption peak value of the phosphorus accumulating bacteria, realizes the metabolic transition between the anaerobic zone and the anoxic zone, and makes the phosphorus accumulating bacteria complete energy storage before the carbon source distribution peak value is reached, and then the denitrifying bacteria can fully utilize the intermediate products released by the phosphorus accumulating bacteria to perform denitrification reaction in the anoxic stage, forming a dynamic complementary biological metabolic rhythm.

6. The system for biological nitrogen and phosphorus removal from wastewater according to claim 1, characterized in that: The self-learning optimization module uses the multi-cycle data difference to establish the operation mode recognition logic when the deep learning model is modeled on the operation cycle data, selects the most suitable carbon source distribution strategy by recognizing the change trend of the total nitrogen and total phosphorus ratio of the influent, and updates the calculation parameters of the carbon source demand calculation module in real time; when it is detected that the change rate of the water quality parameter exceeds the preset range, the system is temporarily switched to the high response mode, the data update frequency is increased and the carbon source adjustment cycle is shortened, and the influent water quality mutation is quickly adapted.

7. The system according to claim 6, wherein the system is characterized by: The deep learning model of the self-learning optimization module adopts the phased target updating logic in the parameter training process, that is, the carbon source utilization rate is taken as the main optimization target in the early stage of the deep learning model training, the difference between the denitrification rate and the phosphorus absorption rate is minimized as the optimization criterion in the middle stage, and the comprehensive ratio of the system energy consumption and the nitrogen and phosphorus removal rate is taken as the final target in the later stage; the phased target updating logic enables the system to gradually transition from carbon source efficiency control to overall process optimization.

8. The system according to claim 1, wherein the system is characterized by, The bidirectional data verification process is arranged between the carbon source demand calculation module and the self-learning optimization module, and is used for periodically correcting the prediction deviation; at the end of each operation cycle, the system performs difference analysis on the carbon source distribution record and the corresponding denitrification and phosphorus removal result, and when the difference exceeds the set threshold, the self-learning optimization module generates new prediction correction parameters and feeds them back to the carbon source demand calculation module.