Wastewater treatment artificial intelligence decision optimization method and system

Abstract

The application discloses a wastewater treatment artificial intelligence decision optimization method and system, and relates to the technical field of wastewater treatment decision, and solves the technical problem that different abnormal reasons are executed in a differentiated, safe and quantifiable hierarchical response strategy instead of one-size-fits-all control.The application diagnoses based on clear process mechanism indexes such as carbon-nitrogen ratio, sludge load and dissolved oxygen mismatch degree, each step of judgment and subsequent control action has clear physical and chemical meaning and process logic support, the credibility and acceptability of the system are improved, the priority diagnosis path of microbial inhibition state, high load impact and carbon source deficiency state is proposed, and specific, quantitative and executable control parameters and formulas are provided for each state and its subclass, realizing the leap from qualitative judgment to quantitative execution, and the system can automatically and gradually recover to normal working condition after impact relief, improving the self-recovery ability and long-term operation stability of the system.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment decision-making technology, specifically to an artificial intelligence decision-making optimization method and system for wastewater treatment. Background Technology

[0002] With increasingly stringent discharge standards for urban wastewater treatment plants, higher demands are being placed on the stability and intelligence of biological nitrogen and phosphorus removal systems. Traditional wastewater treatment control systems often rely on manual experience or simple PID feedback regulation, making it difficult to cope with complex operating conditions such as fluctuations in influent water quality and quantity, environmental changes, and sudden shocks.

[0003] In existing technologies, some systems have attempted to introduce online instruments and automatic control strategies, such as adjusting the aeration rate by monitoring parameters like DO and MLSS, or adding carbon sources based on a fixed C / N threshold. However, these methods generally suffer from the following drawbacks: Most systems only trigger intervention after the effluent exceeds the standard, failing to provide early warnings. Furthermore, they do not distinguish between different abnormal causes such as microbial inhibition, high load shocks, and insufficient carbon sources, leading to mismatched control strategies and a lack of comprehensive criteria driven by mechanisms. For example, high DO is simply attributed to excessive aeration, ignoring the fact that it may be a manifestation of inhibited microbial activity. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an artificial intelligence decision-making optimization method and system for wastewater treatment, which solves the problem of implementing differentiated, safe, and quantifiable hierarchical response strategies for different causes of anomalies, rather than a one-size-fits-all approach.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an artificial intelligence decision-making optimization method for wastewater treatment, which specifically includes the following steps: Step 1: Collect multi-source heterogeneous data in real time, including influent water quality parameters, process control parameters and environmental auxiliary data, by sensors deployed in each process unit. Perform timestamp alignment, outlier removal and missing value filling based on the dynamic 3σ principle on the multi-source heterogeneous data to obtain preprocessed data. Extract the mean, rate of change and extreme values ​​based on the preprocessed data to construct feature vectors. Step 2: Based on the diagnostic indicators driven by the feature vector computer mechanism, including carbon-nitrogen ratio, sludge load, and mismatch between dissolved oxygen and total nitrogen removal efficiency, the status is judged in priority order to obtain the microbial inhibition state, high load shock state, and carbon source deficiency state. Step 3: Implement a graded response strategy based on the judgment results. If it is a microbial inhibition state, further identify the cause of inhibition and implement corresponding measures. If it is a high-load shock state, classify the shock level according to the sludge load and influent change rate and implement graded regulation. If it is a carbon source deficiency state, dynamically calculate the carbon source addition acceleration rate and add it to the anoxic zone. At the same time, verify the effect and make dynamic adjustments based on the denitrification efficiency index.

[0006] As a further aspect of the present invention, the diagnostic indicators include: The carbon-to-nitrogen ratio represents the proportion of organic carbon to total nitrogen in the influent, and C / N = Sludge loading represents how much organic matter a unit mass of microorganisms needs to process per day, and sludge loading F / M = Where F represents organic load, M represents microbial biomass, and Q represents influent flow rate and COD. in The value represents the influent COD concentration, MLSS represents the mixed liquor suspended solids concentration, and V represents the effective volume of the biological treatment tank.

[0007] As a further aspect of the present invention, the method of performing the status judgment in priority order is as follows: Prioritize the mismatch between dissolved oxygen (DO) and removal efficiency (r) TN To determine if dissolved oxygen (DO) > 3.0 mg / L and r TN If the sludge load is >-0.5 mg / L / min, it indicates a state of microbial inhibition; otherwise, it indicates a steady state. In this case, sludge load assessment is required. If the sludge load F / M > 0.4 kg COD / kg MLSS·d, it indicates a high sludge load shock; otherwise, it indicates a steady state. Finally, the carbon-nitrogen ratio is assessed. If C / N < 3.5, it indicates a state of insufficient carbon source.

[0008] As a further aspect of the present invention, if the state is one of microbial inhibition, the method for further identifying the cause of inhibition and implementing corresponding measures is as follows: If the low temperature inhibition is T<15℃, lock the current blower frequency and maintain the current aeration. At the same time, start the biological tank insulation measures, adjust the residual sludge discharge to 70%-80% of the current value, calculate the nitrification rate every 2 hours, and if the nitrification rate is >1.0mg / L / h for 2 consecutive times, it is considered as activity recovery. If the pH is abnormal (<6.5), add NaHCO3 to adjust it to 7.0-7.5, and follow the formula... Calculate the acceleration rate Q NaHCO3， Where k is taken as 15g / m 3 / pH, if pH>8.5, check if alkaline industrial wastewater has been mixed in, and switch the inlet valve if necessary; If a suspected toxic shock occurs, specifically manifested as a sudden increase in influent COD and simultaneous increases in effluent NH4, immediately reduce the influent volume by 50%, activate the accident regulation tank, suspend the discharge of excess sludge, retain high-concentration activated sludge to dilute the poison, and send an alarm to the central control room; If it is sludge aging, specifically manifested as the mixed liquor suspended solid concentration MLSS > 5000 mg / L and the sludge volume index SVI < 60 mL / g, start the excess sludge pump, and at the same time, according to the formula Calculate the sludge discharge rate Q of the sludge pump waste , X return is the return sludge concentration, moderately increase the sludge discharge, control the mixed liquor suspended solid concentration MLSS at 3000 - 4000 mg / L, and supplement fresh carbon source to activate endogenous respiration.

[0009] As a further solution of the present invention, the method for dividing the shock level according to the sludge load and the influent change rate is as follows: According to the formula Calculate the influent COD change rate r COD , based on the sludge load F / M and the influent COD change rate r COD Comprehensively judge the shock level. If 0.3 < F / M ≤ 0.45, and r COD ≤ 100 mg / L / h, then classify it as a mild shock. If 0.45 < F / M ≤ 0.65, and r COD > 100 mg / L / h, then classify it as a moderate shock. If F / M > 0.65, and r COD > 150 mg / L / h, then classify it as a severe shock.

[0010] As a further solution of the present invention, for the classified mild shock, increase it by 10% - 15% based on the current fan frequency, and control the dissolved oxygen DO at 2.5 - 3.5 mg / L. If the dissolved oxygen DO > 3.5 mg / L, stop the increase. At the same time, maintain the internal reflux ratio at the current set value, usually 200% - 300%, and adjust the external reflux ratio to 1.0 - 1.2 times the designed reflux ratio. The designed value is usually 100%, that is, actually set to 100% - 120%. If the calculated carbon-nitrogen ratio is less than 4.0 in real-time, start the sodium acetate dosing pump, and according to the formula Calculate the dosing rate, and set the upper limit to 20 mg / L / h; For the classified moderate shock, the PLC automatically reduces the opening of the total influent electric valve to 70% - 80%, increases the fan frequency by 20% - 30% based on the current value, and ensures that the dissolved oxygen DO is at 3.0 - 4.0 - mg / L. At the same time, stop the operation of the excess sludge discharge pump until the shock is relieved; For severe impacts, the main inlet valve opening is set to 50%, the fan frequency is set to 100%, and the powdered activated carbon (PAC) dosing system is started at a dosing rate of 30 mg / L / h.

[0011] As a further aspect of the present invention, an impact recovery verification and automatic callback mechanism is initiated based on the impact adjustment strategy, and the load recovery index LRI is calculated every 30 minutes. ,in If LRI≤1.2 and dissolved oxygen DO∈[2.0,3.0]mg / L for two consecutive cycles, the shock is considered to be relieved. At the same time, the automatic callback program is started, the inlet valve increases by 10% every 15 minutes until it reaches 100%, the blower frequency decreases by 5% every 10 minutes, returns to normal settings, resumes sludge discharge, and stops the addition of powdered activated carbon (PAC).

[0012] As a further aspect of the present invention, the method of dynamically calculating the carbon source addition acceleration rate and adding it to the anoxic zone is as follows: The target carbon-to-nitrogen ratio (C / N) is dynamically adjusted according to the effluent standard, and based on the formula... Calculate the theoretical carbon source demand ,in Indicates the target carbon-to-nitrogen ratio. Indicates the total nitrogen concentration in the influent. The value represents the influent chemical oxygen demand (COD) concentration, and 0.78 represents the COD equivalent coefficient of sodium acetate, based on the obtained theoretical carbon source demand. Introducing a denitrification efficiency factor Make corrections according to the formula. Calculate the final acceleration rate ; Continuously add the solution to the front end of the anoxic zone according to the final acceleration rate, with the preferred addition location being downstream of the internal recirculation mixing point, based on the formula. Calculate the denitrification efficiency index DENI, where Indicates the internal reflux ratio. Indicates the nitrification coefficient. Indicates the ammonia nitrogen concentration in the effluent. This indicates the total nitrogen in the influent.

[0013] As a further aspect of the present invention, the method for verifying the effect and dynamically adjusting based on the denitrification efficiency index is as follows: When 0.5 ≤ DENI < 0.7, each adjustment is 8% of the current injection rate, and the trend of the denitrification efficiency index (DENI) is continuously monitored. If DENI ≥ 0.5, the initial adjustment remains unchanged; otherwise, a second adjustment is made, each time by 6% of the current injection rate. If DENI still does not reach the threshold of ≥ 0.7 after two consecutive adjustments, efficiency factor recalibration is initiated, according to the formula... The new denitrification efficiency factor was calculated based on the formula. ,in This represents the denitrification efficiency index calculated at the current moment. This indicates the currently used denitrification efficiency factor, i.e., the initial value or the last calibration value. It is also used to revise the denitrification efficiency factor by combining the total nitrogen load, water temperature and dissolved oxygen data of the past 24 hours.

[0014] The wastewater treatment artificial intelligence decision-making optimization system includes: The multi-source data fusion and feature engineering module is used to acquire influent water quality, process parameters, and environmental data in real time. It performs timestamp alignment, outlier removal, and missing value filling on the collected multi-source data, and finally constructs feature vectors. At the same time, the feature vectors are transmitted to the wastewater treatment status intelligent diagnosis module. The intelligent diagnostic module for wastewater treatment status is used to calculate diagnostic indicators driven by computation based on the obtained feature vectors, including carbon-nitrogen ratio, sludge load, and mismatch between dissolved oxygen (DO) and removal efficiency. Based on the diagnostic indicators, the module judges the wastewater treatment status, including microbial inhibition state, high load shock state, and carbon source deficiency state. Based on the status judgment, different judgment information is generated and transmitted to the corresponding analysis unit. The microbial inhibition state classification response module is used to identify and deal with existing microbial inhibition states. It implements differentiated and safe emergency strategies for different inhibition causes. For low temperature inhibition, it locks the fan frequency and sets the sludge discharge rate. For pH abnormality, it adds materials with opposite properties according to different pH values. For toxic shock, it reduces the influent flow and suspends the discharge of excess sludge. For sludge aging, it starts the sludge pump and calculates the sludge pump discharge rate, generates corresponding classification response information, and transmits it to the intelligent decision optimization information output module. The high-load shock classification control and recovery module is used to adjust the sludge under high-load shock conditions. It calculates the influent change rate and sludge load, and judges the degree of shock by combining the two. For mild shocks, it controls the fan frequency and external return ratio. For moderate shocks, it reduces the influent valve opening and increases the fan frequency. For severe shocks, it sets the influent valve opening and fan frequency, and simultaneously starts the addition of powdered activated carbon. It generates classification control information and transmits it to the intelligent decision optimization information output module. The intelligent carbon source addition and denitrification optimization module is used to deal with the situation of insufficient carbon source. It dynamically adjusts the target carbon-nitrogen ratio according to the effluent standard, and introduces a correction factor to calculate the final addition rate. At the same time, it calculates and judges the denitrification efficiency index intermittently. If the denitrification efficiency index does not meet the requirements, it will be corrected. It also combines dissolved oxygen for linkage control to generate carbon source addition information and transmits it to the intelligent decision optimization information output module. The intelligent decision optimization information output module is used to display the acquired graded response information, graded control information, and carbon source addition information to the corresponding management personnel.

[0015] This invention provides an artificial intelligence-based decision-making optimization method and system for wastewater treatment. Compared with existing technologies, it has the following advantages: This invention diagnoses sludge by using clearly defined process mechanism indicators such as carbon-nitrogen ratio, sludge load, and dissolved oxygen mismatch. Each judgment and subsequent control action has a clear physicochemical meaning and process logic support, which improves the reliability and acceptability of the system. It proposes priority diagnostic paths for microbial inhibition state, high load shock, and carbon source deficiency state, and provides specific, quantitative, and executable control parameters and formulas for each state and its subclasses, realizing the leap from qualitative judgment to quantitative execution.

[0016] This invention innovatively introduces verification indicators such as load recovery index and denitrification efficiency index, and designs an automatic callback program. After the shock is relieved, the system can automatically and gradually recover to normal operating conditions, forming a complete closed loop of perception, diagnosis, control, verification and recovery, which improves the system's self-recovery capability and long-term operational stability. Attached Figure Description

[0017] Figure 1 This is a flowchart of the wastewater treatment artificial intelligence decision optimization method of the present invention; Figure 2 This is a block diagram of the wastewater treatment artificial intelligence decision optimization system of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0019] First Embodiment Please see Figure 1 This application provides an artificial intelligence decision-making optimization method for wastewater treatment, which specifically includes the following steps: Step 1: Real-time collection of multi-source heterogeneous data is achieved through sensors, online instruments, and SCADA systems deployed in various process units of the wastewater treatment plant. This multi-source heterogeneous data includes influent water quality parameters, process control parameters, and environmental auxiliary data. Specific influent water quality parameters include COD, BOD5, TN, TP, pH, and flow rate. Process control parameters include dissolved oxygen (DO), MLSS, ORP, sludge return ratio, aeration rate, and chemical dosage. Environmental auxiliary data includes air temperature and rainfall. The obtained multi-source heterogeneous data is then preprocessed by aligning timestamps, removing outliers, and imputing missing values. Outlier removal follows the 3σ principle. Based on the obtained preprocessed data, a sliding window statistical analysis is performed with a sliding window value of 2 hours. The corresponding mean, rate of change, and extreme values ​​are extracted, and a feature vector is constructed based on the obtained features.

[0020] Step 2: Based on the obtained feature vectors, calculate the diagnostic indicators driven by computation, specifically including the carbon-to-nitrogen ratio, sludge load, and the mismatch between dissolved oxygen (DO) and removal efficiency (r). TN The carbon-to-nitrogen ratio represents the proportion of organic carbon to total nitrogen in the influent, and C / N = Sludge loading represents how much organic matter a unit mass of microorganisms needs to process per day, and sludge loading F / M = Where F represents organic load, M represents microbial biomass, and Q represents influent flow rate and COD. in The influent COD concentration is represented by , MLSS by mixed liquor suspended solids concentration, and V by effective volume of the biological treatment tank. The wastewater treatment status is judged by comprehensively considering the carbon-nitrogen ratio, sludge load, and the mismatch between dissolved oxygen (DO) and removal efficiency. Prioritize the mismatch between dissolved oxygen (DO) and removal efficiency (r) TN To determine if dissolved oxygen (DO) > 3.0 mg / L and r TN If the concentration is >-0.5 mg / L / min, it indicates a state of microbial inhibition. This state is the most insidious and harmful, and requires the highest priority treatment. Conversely, if it is not inhibited, it indicates a steady state, and sludge loading needs to be assessed. If the sludge loading F / M > 0.4 kg COD / kg MLSS·d, it indicates a high sludge load shock, and priority treatment should be given. Otherwise, it indicates a steady state. Finally, the carbon-nitrogen ratio is assessed. If C / N < 3.5, it indicates a carbon source deficiency state, and organic carbon needs to be supplemented by adding an external carbon source to adjust the carbon-nitrogen ratio to a suitable range. Based on the diagnostic indicators, the corresponding causes of the abnormalities are determined, and a graded response strategy is implemented according to the differences in causes.

[0021] Step 3: Regarding the microbial inhibition state, if it is low-temperature inhibition, specifically manifested as water temperature T < 15℃, then maintain the current aeration, prohibit increasing the blower frequency, simultaneously activate the biological tank insulation measures, and appropriately reduce the sludge discharge. The specific treatment method is as follows: Lock the current fan frequency, prohibit any upward adjustment commands, adjust the excess sludge discharge volume to 70%-80% of the current value, calculate the nitrification rate every 2 hours. If the nitrification rate > 1.0 mg / L / h for two consecutive times, it is regarded as the restoration of activity; If it is abnormal pH, if pH < 6.5, add NaHCO3 to adjust it to 7.0 - 7.5, and according to the formula Calculate the dosing rate Q NaHCO3， where k takes the value of 15g / m 3 / pH. If pH > 8.5, check whether there is alkaline industrial wastewater mixed in, and switch the inlet valve if necessary; If it is suspected of toxic shock, specifically manifested as a sudden increase in influent COD and a simultaneous increase in effluent NH4, immediately reduce the influent volume by 50%, activate the accident adjustment tank, suspend the excess sludge discharge, retain the high-concentration activated sludge to dilute the poison, and push an alarm to the central control room; If it is sludge aging, specifically manifested as the mixed liquor suspended solid concentration MLSS > 5000mg / L and the sludge volume index SVI < 60mL / g, start the excess sludge pump, and at the same time according to the formula Calculate the sludge pump sludge discharge rate Q waste , X return is the return sludge concentration, moderately increase the sludge discharge, control MLSS at 3000 - 4000mg / L, and supplement fresh carbon source to activate endogenous respiration.

[0022] Step four: For the situation of high-load shock of sludge, according to the formula Calculate the influent COD change rate r COD , based on the sludge load F / M and the influent COD change rate r COD Comprehensively judge the shock level. If 0.3 < F / M ≤ 0.45, and r COD ≤ 100mg / L / h, then classify it as a mild shock. If 0.45 < F / M ≤ 0.65, and r COD > 100mg / L / h, then classify it as a moderate shock. If F / M > 0.65, and r COD > 150mg / L / h, then classify it as a severe shock; For the classified mild shock, increase it by 10% - 15% based on the current fan frequency, and control the dissolved oxygen DO at 2.5 - 3.5mg / L. If the dissolved oxygen DO > 3.5mg / L, stop the increase. At the same time, maintain the current set value of the internal reflux ratio, usually 200% - 300%, and adjust the external reflux ratio to 1.0 - 1.2 times the designed reflux ratio. The designed value is usually 100%, that is, actually set it to 100% - 120%. If the calculated carbon-nitrogen ratio obtained in real time is less than 4.0, then start the sodium acetate dosing pump, and according to the formula Calculate the dosing rate, and set the upper limit to 20 mg / L / h; For moderate impacts, the PLC automatically reduces the opening of the main inlet electric valve to 70%-80%, increases the fan frequency by 20%-30% from the current value, ensures dissolved oxygen (DO) is at 3.0-4.0 mg / L, and stops the operation of the residual sludge discharge pump until the impact subsides. For severe impacts, the main inlet valve opening is set to 50%, the fan frequency is set to 100%, and the powdered activated carbon (PAC) dosing system is started at a dosing rate of 30 mg / L / h for 2 hours. Based on the above-mentioned shock adjustment strategy, a shock recovery verification and automatic callback mechanism is initiated, and the load recovery index (LRI) is calculated every 30 minutes. ,in If LRI≤1.2 and dissolved oxygen DO∈[2.0,3.0]mg / L for two consecutive cycles, the shock is considered to be relieved. At the same time, the automatic callback program is started, the inlet valve increases by 10% every 15 minutes until it reaches 100%, the blower frequency decreases by 5% every 10 minutes, returns to normal settings, resumes sludge discharge, and stops the addition of powdered activated carbon (PAC).

[0023] Step 5: For cases with insufficient carbon source, dynamically adjust the target carbon-to-nitrogen ratio (C / N) according to the effluent standard, and apply the formula... Calculate the theoretical carbon source demand ,in Indicates the target carbon-to-nitrogen ratio. Indicates the total nitrogen concentration in the influent. The value represents the influent chemical oxygen demand (COD) concentration, and 0.78 represents the COD equivalent coefficient of sodium acetate, based on the obtained theoretical carbon source demand. Introducing a denitrification efficiency factor Make corrections according to the formula. Calculate the final acceleration rate ; Based on the calculated final acceleration rate, the carbon source dosing pump is started, and the carbon source is continuously added to the front end of the anoxic zone according to the final acceleration rate. The dosing location is preferably downstream of the internal reflux mixing point. The denitrification efficiency index is calculated every 15 minutes, according to the formula. Calculate the denitrification efficiency index DENI, where Indicates the internal reflux ratio. Indicates the nitrification coefficient. Indicates the ammonia nitrogen concentration in the effluent. This indicates the total nitrogen in the influent, and the effectiveness is judged based on the denitrification efficiency index (DENI). If DENI ≥ 0.7 and remains so for 30 minutes, the carbon source addition is considered effective. If DENI < 0.5 and the carbon source has been added at a rate ≥ 20 mg / L / h, a suspected carbon source ineffectiveness alarm is triggered. When 0.5 ≤ DENI < 0.7, the system will automatically enter dynamic adjustment mode, gradually fine-tuning the carbon source feed rate by 8% of the current feed rate each time, while continuously monitoring the DENI trend. If the denitrification efficiency index (DENI) ≥ 0.5, the initial adjustment remains unchanged; otherwise, a second adjustment is performed, with each adjustment being 6% of the current feed rate. If DENI still does not reach the threshold of ≥ 0.7 after two consecutive adjustments, efficiency factor recalibration is initiated, according to the formula... The new denitrification efficiency factor was calculated based on the formula. ,in This represents the denitrification efficiency index calculated at the current moment. This indicates the currently used denitrification efficiency factor, i.e., the initial value or the last calibration value. It is then revised based on total nitrogen load, water temperature, and dissolved oxygen data from the past 24 hours. The specific revision method is as follows: If the TN load fluctuation is >30%, then the corrected denitrification efficiency factor is: If the water temperature is <12℃, then the corrected denitrification efficiency factor is: If the average dissolved oxygen (DO) in the hypoxic area avg If the concentration is >0.4 mg / L, then the corrected denitrification efficiency factor is: During the carbon source addition process, the dissolved oxygen concentration change in the anoxic zone is recorded simultaneously. When the dissolved oxygen is >0.5mg / L, the submersible mixer power is automatically increased to 110%–120% of the rated power and maintained for 10 minutes. If the dissolved oxygen DO is still >0.5mg / L, check whether the dissolved oxygen DO at the end of the aerobic zone is too high. If so, reduce the aeration intensity of the aerobic zone.

[0024] Second Embodiment Please see Figure 2 This application provides an artificial intelligence decision-making optimization system for wastewater treatment, including: a multi-source data fusion and feature engineering module, a wastewater treatment status intelligent diagnosis module, a microbial inhibition state graded response module, a high-load shock graded control and recovery module, a carbon source intelligent dosing and denitrification optimization module, and an intelligent decision-making optimization information output module, and combined with Figure 2 It can be seen that the information between the above functional modules is transmitted in one direction only.

[0025] The multi-source data fusion and feature engineering module is used to acquire influent water quality, process parameters, and environmental data in real time. It performs timestamp alignment, outlier removal, and missing value filling on the collected multi-source data, and finally constructs feature vectors. At the same time, the feature vectors are transmitted to the wastewater treatment status intelligent diagnosis module, and the specific processing method is the same as the processing process in step one. The intelligent diagnostic module for wastewater treatment status is used to calculate diagnostic indicators driven by computation based on the obtained feature vectors, including carbon-nitrogen ratio, sludge load, and mismatch between dissolved oxygen (DO) and removal efficiency. Based on the diagnostic indicators, the module judges the wastewater treatment status, including microbial inhibition state, high load shock state, and carbon source deficiency state. Based on the status judgment, different judgment information is generated and transmitted to the corresponding analysis unit. The specific treatment method is the same as the treatment process in step two. The microbial inhibition state graded response module is used to identify and deal with existing microbial inhibition states. It implements differentiated and safe emergency strategies for different inhibition causes. For low temperature inhibition, it locks the fan frequency and sets the sludge discharge rate. For pH abnormality, it adds materials with opposite properties according to different pH values. For toxic shock, it reduces the influent flow and suspends the discharge of excess sludge. For sludge aging, it starts the sludge pump and calculates the sludge pump discharge rate, generates corresponding graded response information, and transmits it to the intelligent decision optimization information output module. The specific processing method is the same as the processing process in step three. The high-load shock classification control and recovery module is used to adjust the sludge under high-load shock conditions. It calculates the influent change rate and sludge load, and judges the degree of shock by combining the two. For mild shocks, it controls the fan frequency and external return ratio. For moderate shocks, it reduces the influent valve opening and increases the fan frequency. For severe shocks, it sets the influent valve opening and fan frequency, and simultaneously starts the addition of powdered activated carbon. It generates classification control information and transmits it to the intelligent decision optimization information output module. The specific processing method is the same as the processing process in step four. The intelligent carbon source addition and denitrification optimization module is used to deal with the situation of insufficient carbon source. It dynamically adjusts the target carbon-nitrogen ratio according to the effluent standard, and introduces a correction factor to calculate the final addition rate. At the same time, it calculates the denitrification efficiency index intermittently and judges it. If the denitrification efficiency index does not meet the requirements, it will be corrected. It also combines dissolved oxygen for linkage control to generate carbon source addition information and transmits it to the intelligent decision optimization information output module. The specific processing method is the same as the processing process in step five. The intelligent decision optimization information output module is used to display the acquired graded response information, graded control information, and carbon source addition information to the corresponding management personnel.

[0026] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0027] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An artificial intelligence decision-making optimization method for wastewater treatment, characterized in that, The method specifically includes the following steps: Step 1: Collect multi-source heterogeneous data in real time, including influent water quality parameters, process control parameters and environmental auxiliary data, by sensors deployed in each process unit. Perform timestamp alignment, outlier removal and missing value filling based on the dynamic 3σ principle on the multi-source heterogeneous data to obtain preprocessed data. Extract the mean, rate of change and extreme values ​​based on the preprocessed data to construct feature vectors. Step 2: Based on the diagnostic indicators driven by the feature vector computer mechanism, including carbon-nitrogen ratio, sludge load, and mismatch between dissolved oxygen and total nitrogen removal efficiency, the status is judged in priority order to obtain the microbial inhibition state, high load shock state, and carbon source deficiency state. Step 3: Implement a graded response strategy based on the judgment results. If it is a microbial inhibition state, further identify the cause of inhibition and implement corresponding measures. If it is a high-load shock state, classify the shock level according to the sludge load and influent change rate and implement graded regulation. If it is a carbon source deficiency state, dynamically calculate the carbon source addition acceleration rate and add it to the anoxic zone. At the same time, verify the effect and make dynamic adjustments based on the denitrification efficiency index.

2. The wastewater treatment artificial intelligence decision-making optimization method according to claim 1, characterized in that, The diagnostic indicators include: The carbon-to-nitrogen ratio represents the proportion of organic carbon to total nitrogen in the influent, and C / N = Sludge loading represents how much organic matter a unit mass of microorganisms needs to process per day, and sludge loading F / M = Where F represents organic load, M represents microbial biomass, and Q represents influent flow rate and COD. in The value represents the influent COD concentration, MLSS represents the mixed liquor suspended solids concentration, and V represents the effective volume of the biological treatment tank.

3. The wastewater treatment artificial intelligence decision-making optimization method according to claim 1, characterized in that, The method of performing status judgments in priority order is as follows: Prioritize the mismatch between dissolved oxygen (DO) and removal efficiency (r) TN To determine if dissolved oxygen (DO) > 3.0 mg / L and r TN If the sludge load is >-0.5 mg / L / min, it indicates a state of microbial inhibition; otherwise, it indicates a steady state. In this case, sludge load assessment is required. If the sludge load F / M > 0.4 kg COD / kg MLSS·d, it indicates a high sludge load shock; otherwise, it indicates a steady state. Finally, the carbon-nitrogen ratio is assessed. If C / N < 3.5, it indicates a state of insufficient carbon source.

4. The wastewater treatment artificial intelligence decision-making optimization method according to claim 1, characterized in that, If the condition is one of microbial inhibition, the method for further identifying the cause of inhibition and implementing corresponding measures is as follows: If the low temperature inhibition is T<15℃, lock the current blower frequency and maintain the current aeration. At the same time, start the biological tank insulation measures, adjust the residual sludge discharge to 70%-80% of the current value, calculate the nitrification rate every 2 hours, and if the nitrification rate is >1.0mg / L / h for 2 consecutive times, it is considered as activity recovery. If the pH is abnormal (<6.5), add NaHCO3 to adjust it to 7.0-7.5, and follow the formula... Calculate the acceleration rate Q NaHCO3， Where k is taken as 15g / m 3 / pH, if pH>8.5, check if alkaline industrial wastewater has been mixed in, and switch the inlet valve if necessary; If a toxic shock is suspected, specifically manifested as a sudden increase in influent COD and a simultaneous increase in effluent NH4, immediately reduce the influent flow by 50%, activate the emergency equalization tank, suspend the discharge of excess sludge, retain high-concentration activated sludge to dilute the toxins, and send an alarm to the central control room. If the issue is sludge aging, specifically characterized by a mixed liquor suspended solids concentration (MLSS) > 5000 mg / L and a sludge volume index (SVI) < 60 mL / g, start the excess sludge pump and simultaneously apply the formula... Calculate the sludge discharge rate Q of the sludge pump waste X return To improve the concentration of reflux sludge, the sludge discharge should be increased appropriately, and the mixed liquor suspended solids concentration (MLSS) should be controlled at 3000-4000 mg / L. Fresh carbon sources should be added to activate endogenous respiration.

5. The wastewater treatment artificial intelligence decision-making optimization method according to claim 1, characterized in that, The method for classifying shock levels based on sludge load and influent change rate is as follows: According to the formula calculate the change rate r of the influent COD COD , based on the sludge loading F / M and the change rate r of the influent COD COD comprehensively judge the shock level. If 0.3 < F / M ≤ 0.45 and r COD ≤ 100 mg / L / h, then classify it as a mild shock. If 0.45 < F / M ≤ 0.65 and r COD > 100 mg / L / h, then classify it as a moderate shock. If F / M > 0.65 and r COD > 150 mg / L / h, then classify it as a severe shock.

6. The wastewater treatment artificial intelligence decision-making optimization method according to claim 5, characterized in that, For mild impacts, increase the fan frequency by 10%-15% based on the current frequency, and control the dissolved oxygen (DO) at 2.5-3.5 mg / L. If the DO > 3.5 mg / L, stop increasing the frequency. Simultaneously, maintain the internal recirculation ratio at the current set value, typically 200%-300%, and adjust the external recirculation ratio to 1.0-1.2 times the design recirculation ratio. The design value is typically 100%, meaning the actual setting is 100%-120%. If the real-time calculated carbon-to-nitrogen ratio is less than 4.0, start the sodium acetate dosing pump and follow the formula... Calculate the dosing rate, and set the upper limit to 20 mg / L / h; For moderate impacts, the PLC automatically reduces the opening of the main inlet electric valve to 70%-80%, increases the fan frequency by 20%-30% from the current value, ensures dissolved oxygen (DO) is at 3.0-4.0 mg / L, and stops the operation of the residual sludge discharge pump until the impact subsides. For severe impacts, the main inlet valve opening is set to 50%, the fan frequency is set to 100%, and the powdered activated carbon (PAC) dosing system is started at a dosing rate of 30 mg / L / h.

7. The wastewater treatment artificial intelligence decision-making optimization method according to claim 6, characterized in that, Based on the shock adjustment strategy, a shock recovery verification and automatic callback mechanism are initiated, and the load recovery index (LRI) is calculated every 30 minutes. ,in If LRI≤1.2 and dissolved oxygen DO∈[2.0,3.0]mg / L for two consecutive cycles, the shock is considered to be relieved. At the same time, the automatic callback program is started, the inlet valve increases by 10% every 15 minutes until it reaches 100%, the blower frequency decreases by 5% every 10 minutes, returns to normal settings, resumes sludge discharge, and stops the addition of powdered activated carbon (PAC).

8. The wastewater treatment artificial intelligence decision-making optimization method according to claim 1, characterized in that, If the carbon source is insufficient, the method for dynamically calculating the carbon source addition acceleration rate and adding it to the anoxic zone is as follows: The target carbon-to-nitrogen ratio (C / N) is dynamically adjusted according to the effluent standard, and based on the formula... Calculate the theoretical carbon source demand ,in Indicates the target carbon-to-nitrogen ratio. Indicates the total nitrogen concentration in the influent. The value represents the influent chemical oxygen demand (COD) concentration, and 0.78 represents the COD equivalent coefficient of sodium acetate, based on the obtained theoretical carbon source demand. Introducing a denitrification efficiency factor Make corrections according to the formula. Calculate the final acceleration rate ; Continuously add the solution to the front end of the anoxic zone according to the final acceleration rate, with the preferred addition location being downstream of the internal recirculation mixing point, based on the formula. Calculate the denitrification efficiency index DENI, where Indicates the internal reflux ratio. Indicates the nitrification coefficient. Indicates the ammonia nitrogen concentration in the effluent. This indicates the total nitrogen in the influent.

9. The wastewater treatment artificial intelligence decision-making optimization method according to claim 1, characterized in that, The method for verifying the effect and dynamically adjusting based on the denitrification efficiency index is as follows: When 0.5 ≤ DENI < 0.7, each adjustment is 8% of the current injection rate, and the trend of the denitrification efficiency index (DENI) is continuously monitored. If DENI ≥ 0.5, the initial adjustment remains unchanged; otherwise, a second adjustment is made, each time by 6% of the current injection rate. If DENI still does not reach the threshold of ≥ 0.7 after two consecutive adjustments, efficiency factor recalibration is initiated, according to the formula... The new denitrification efficiency factor was calculated based on the formula. ,in This represents the denitrification efficiency index calculated at the current moment. This indicates the currently used denitrification efficiency factor, i.e., the initial value or the last calibration value. It is also used to revise the denitrification efficiency factor by combining the total nitrogen load, water temperature and dissolved oxygen data of the past 24 hours.

10. A wastewater treatment artificial intelligence decision-making optimization system, characterized in that, The wastewater treatment artificial intelligence decision optimization method according to any one of claims 1-9 includes: The multi-source data fusion and feature engineering module is used to acquire influent water quality, process parameters, and environmental data in real time. It performs timestamp alignment, outlier removal, and missing value filling on the collected multi-source data, and finally constructs feature vectors. At the same time, the feature vectors are transmitted to the wastewater treatment status intelligent diagnosis module. The intelligent diagnostic module for wastewater treatment status is used to calculate diagnostic indicators driven by computation based on the obtained feature vectors, including carbon-nitrogen ratio, sludge load, and mismatch between dissolved oxygen (DO) and removal efficiency. Based on the diagnostic indicators, the module judges the wastewater treatment status, including microbial inhibition state, high load shock state, and carbon source deficiency state. Based on the status judgment, different judgment information is generated and transmitted to the corresponding analysis unit. The microbial inhibition state classification response module is used to identify and deal with existing microbial inhibition states. It implements differentiated and safe emergency strategies for different inhibition causes. For low temperature inhibition, it locks the fan frequency and sets the sludge discharge rate. For pH abnormality, it adds materials with opposite properties according to different pH values. For toxic shock, it reduces the influent flow and suspends the discharge of excess sludge. For sludge aging, it starts the sludge pump and calculates the sludge pump discharge rate, generates corresponding classification response information, and transmits it to the intelligent decision optimization information output module. The high-load shock classification control and recovery module is used to adjust the sludge under high-load shock conditions. It calculates the influent change rate and sludge load, and judges the degree of shock by combining the two. For mild shocks, it controls the fan frequency and external return ratio. For moderate shocks, it reduces the influent valve opening and increases the fan frequency. For severe shocks, it sets the influent valve opening and fan frequency, and simultaneously starts the addition of powdered activated carbon. It generates classification control information and transmits it to the intelligent decision optimization information output module. The intelligent carbon source addition and denitrification optimization module is used to deal with the situation of insufficient carbon source. It dynamically adjusts the target carbon-nitrogen ratio according to the effluent standard, and introduces a correction factor to calculate the final addition rate. At the same time, it calculates and judges the denitrification efficiency index intermittently. If the denitrification efficiency index does not meet the requirements, it will be corrected. It also combines dissolved oxygen for linkage control to generate carbon source addition information and transmits it to the intelligent decision optimization information output module. The intelligent decision optimization information output module is used to display the acquired graded response information, graded control information, and carbon source addition information to the corresponding management personnel.

Images