A Feedforward-Feedback Dual-Loop Control Method for Phosphorus Removal Dosing Based on Anomaly Diagnosis
By constructing a feedforward-feedback dual-loop control method, combined with multiple linear regression and anomaly diagnosis, the causes of excessive total phosphorus in water can be accurately identified, solving the problem of inaccurate calculation of reagent dosage in existing technologies, and realizing efficient use of reagents and cost control.
Patent Information
- Application Number
- CN202611112559.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-24
- Publication Date
- 2026-11-10
AI Technical Summary
Existing intelligent chemical phosphorus removal models cannot distinguish the causes of excessive total phosphorus in water, resulting in conservative or wasteful calculations of reagent dosage. Furthermore, they fail to fully consider the contribution of biological phosphorus removal systems, leading to high reagent consumption and persistently high operating costs.
A feedforward-feedback dual-loop control method for phosphorus removal dosing based on anomaly diagnosis is adopted. The initial dosage is predicted by a multiple linear regression model, and combined with the amount of residual sludge, a dynamic historical optimal K1 value range is constructed. The total phosphorus and suspended solids concentrations in the effluent are diagnosed in real time, and multi-dimensional verification and feedback correction are performed to achieve precise dosing control.
It enables precise dosing control, reduces reagent consumption and operating costs, identifies abnormal operating conditions in advance, avoids excessive or insufficient dosing, and improves the intelligence level and operational reliability of wastewater treatment plants.
Smart Images

Figure CN123062016A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of water pollution control technology, specifically relating to a feedforward-feedback dual-loop control method for phosphorus removal dosing based on anomaly diagnosis. Background Technology
[0002] Water pollution control is a major strategic task in the construction of ecological civilization. Urban wastewater treatment, as a core infrastructure for controlling water pollution, plays an irreplaceable role in improving water quality and ensuring the sustainable use of water resources. Chemical phosphorus removal, a key process in achieving compliant discharge from urban wastewater treatment plants, involves adding chemical phosphorus removal agents such as aluminum, iron, or calcium salts to wastewater. These agents react with soluble orthophosphates in the wastewater to form insoluble phosphate precipitates, thereby reducing the total phosphorus concentration in the effluent to below the discharge standard limits. According to industry statistics, the cost of chemical phosphorus removal agents accounts for 60% to 80% of the total cost of chemicals used in urban wastewater treatment plants, making it a crucial aspect of wastewater treatment plant operating cost control. With the implementation of the Class A standard of the "Discharge Standard of Pollutants for Urban Wastewater Treatment Plants" (GB 18918-2002) and even stricter local standards, higher requirements have been placed on the precision of controlling the total phosphorus concentration in the effluent. Traditional methods relying on manual experience for agent dosing are no longer sufficient to meet the needs of refined operation and management. The research and application of intelligent dosing control technology has become an inevitable trend in the industry.
[0003] Intelligent dosing control for chemical phosphorus removal involves a deep integration of reaction chemistry principles, wastewater treatment process theory, and automatic control theory. Its core objective is to accurately determine the dosage of chemicals based on the dynamic changes in influent water quality and quantity, minimizing chemical consumption and operating costs while ensuring stable effluent total phosphorus compliance. Currently, scholars and engineers both domestically and internationally have proposed various intelligent dosing models for chemical phosphorus removal. These can be categorized into two types based on their modeling methods: mechanistic model-based dosing methods and data-driven dosing methods. Mechanism-based methods establish a quantitative relationship between phosphorus removal agent dosage and phosphorus removal rate based on the kinetic principles of coagulation and sedimentation reactions. This requires precise measurement of process parameters such as reaction rate constants and sedimentation conversion coefficients, leading to problems such as complex parameter calibration and poor model adaptability in practical applications. Data-driven methods, on the other hand, employ machine learning algorithms such as multiple linear regression, artificial neural networks, and support vector machines. Using variables such as influent flow rate, influent total phosphorus concentration, and effluent total phosphorus concentration as input features, they train a chemical dosage prediction model. This approach does not require in-depth understanding of the reaction mechanism and offers better operability and flexibility in engineering applications.
[0004] However, existing intelligent chemical phosphorus removal dosing models, regardless of the modeling method used, suffer from a fundamental technical flaw: the models have a singular control objective, focusing only on the mapping relationship between influent total phosphorus and reagent dosage, failing to establish a mechanism for identifying the causes of excessive effluent total phosphorus. Effluent total phosphorus is the sum of soluble orthophosphate and suspended phosphorus-containing particulate matter (i.e., phosphorus-containing sludge flocs) in wastewater. Dissolved orthophosphate needs to be removed through the coagulation and sedimentation of chemical phosphorus removal agents, while suspended phosphorus-containing particulate matter needs to be retained in the sludge through effective sludge-water separation. When effluent total phosphorus exceeds the standard, the source of the excess may be two distinct aspects: first, the concentration of soluble orthophosphate is too high, meaning the dosage of chemical phosphorus removal agents is insufficient to fully precipitate and remove phosphate from the influent; second, the concentration of suspended phosphorus-containing particulate matter is too high, meaning the residual sludge discharged from the biological treatment system carries a large amount of phosphorus-containing flocs into the effluent. The control strategies required for the two causes of phosphorus exceedance are completely different: for the first cause, the dosage of phosphorus removal agent should be increased to enhance the chemical phosphorus removal effect; for the second cause, increasing the dosage not only fails to solve the fundamental problem, but also leads to waste of agents and increased sludge disposal costs due to excessive dosage, and may also affect the effluent quality due to agent residue. Existing intelligent dosing models indiscriminately use excessive total phosphorus in the effluent as a trigger condition for increasing the dosage, ignoring the heterogeneity of the causes of exceedance. This results in the control system being unable to respond correctly to exceedances not caused by insufficient agents, leading to high chemical consumption costs and operational decisions heavily reliant on human experience. Furthermore, existing models typically use influent total phosphorus as the main input variable, neglecting the contribution of biological phosphorus removal systems to total phosphorus removal. For systems employing A... 2 Wastewater treatment plants employing biological phosphorus removal processes such as oxidation ditches have failed to fully consider the intrinsic relationship between excess sludge discharge and biological phosphorus removal capacity. This leads to overly conservative calculations of reagent dosage, resulting in unnecessary reagent waste. These technical problems severely restrict the control accuracy and economic efficiency of intelligent chemical phosphorus removal dosing control systems, necessitating the development of a novel intelligent dosing control method capable of intelligently identifying the causes of phosphorus exceedances and precisely matching control strategies. Summary of the Invention
[0005] The purpose of this invention is to provide a feedforward-feedback dual-loop control method for phosphorus removal dosing based on abnormal diagnosis, which can effectively solve the problems in the background art mentioned above.
[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A feedforward-feedback dual-loop control method for phosphorus removal dosing based on anomaly diagnosis includes the following specific steps: Step 1: Construct a feedforward loop: Using historical operating data, train an initial dosage prediction model using a multiple linear regression model. This model takes influent flow rate, total influent phosphorus, and excess sludge discharge as input features and outputs a basic dosage ratio coefficient. Calculate the phosphorus removal agent dosage based on the input features. Step 2: Constructing the historical best K1 value range: The system automatically records the K1base value and the corresponding total phosphorus and residual sludge concentrations in each dosing cycle. When the total phosphorus and suspended solids in the effluent meet the standards, it is marked as an effective K1 value. The system maintains a first-in-first-out queue to store the most recent effective K1 value, calculates the queue mean and standard deviation, and constructs a dynamic historical best K1 value range. Step 3: Perform multi-dimensional verification of K1 value: After each feedforward loop calculates the K1base value, it is compared with the historical best K1 interval, and the corresponding strategy is executed based on the comparison result. Step 4: Construct a feedback diagnosis and correction loop: Real-time acquisition of total phosphorus and residual sludge concentrations in the effluent, receiving the root cause analysis results and their confidence levels output by the host computer's anomaly diagnosis module, and performing a three-layer judgment and corresponding correction strategy based on the information. Step 5: Implement data closed-loop optimization: Data from the target period is automatically included in the historical database and the optimal range of K1 is continuously updated.
[0007] Preferably, the input features of the multiple linear regression model in step 1 specifically include: influent flow rate Qin, in cubic meters per hour; total phosphorus influent TPin, in milligrams per liter; and excess sludge discharge Qwaste, in cubic meters per hour. The output feature is the baseline dosage ratio coefficient K1base, in milligrams of phosphorus removal agent per milligram of phosphorus. The dosage calculation formula is: , in : Dosage of phosphorus removal agent (unit: mg / L or engineering unit matched with metering pump); : Total phosphorus concentration in influent (mg / L); The concentration of residual sludge (g / L) can be taken as an empirical value of 5-10. : Dimensionless biological phosphorus removal efficiency coefficient, initially 0.5, adaptively corrected; Excess sludge discharge (m³) 3 / h); Inlet flow rate (m³) 3 / h); : Basic dosage ratio coefficient (unit: mg phosphorus remover / mg phosphorus); : Constant correction term (mg / L). The amount of excess sludge discharged is introduced as an input variable to achieve the concept of precise dosing, primarily using biochemical methods and secondarily using chemical methods.
[0008] Preferably, in step 2, the length T of the first-in-first-out queue is set to a preset time length, storing only the most recent valid K1 value, i.e., periodic data where both total phosphorus and suspended solids in the effluent meet the standards. The formula for calculating the dynamic historical optimal K1 value interval is: The historical best K1 value range = [μ - nσ, μ + nσ]; Where μ is the mean of the effective K1 values, σ is the standard deviation, and n is set based on actual operating experience, with a preset range. In the initial stage of system operation, when the number of effective K1 value samples is less than the preset minimum sample size, a fixed initial interval based on expert experience is used. Once the sample size reaches the minimum, dynamic interval calculation [μ - nσ, μ + nσ] is switched. The interval coefficient n is dynamically adjusted according to the queue length; a smaller preset value is used when the sample size is small, and the upper limit of the preset range is used when the sample size is sufficient, to balance sensitivity and stability. Non-operating days are not included in the queue.
[0009] Preferably, the multi-dimensional verification rule for the K1 value in step 3 specifically includes three comparison results and corresponding execution strategies: when When the dosage falls below the lower limit of the range, the system determines that there may be a risk of insufficient dosage, issues a mild warning, and performs step-by-step corrections according to the following formula: ,in, This is the step factor, ranging from 0.3 to 0.5, with a default value of 0.5. This represents the lower limit of the historical best K1 value range. After correction, the system pushes a warning message to the central control system: "Feedforward calculation..." The water level is below the lower limit of the historical optimal range, and the dosage has been automatically increased step by step. Please monitor the water quality at the outlet. when When the dosage exceeds the upper limit of the range, the system determines that there may be excessive and wasteful dosage, issues a mild warning, and performs step-by-step corrections according to the following formula: ,in, This represents the upper limit of the historical best K1 value range. After correction, the system pushes a warning message to the central control system: "Feedforward calculation..." The dosage is above the upper limit of the historical optimal range, and the dosage has been automatically adjusted downwards. Please monitor the dosage. The magnitude of a single correction shall not exceed the current The value is capped at 10%; if it exceeds this, a limit will be applied. (If corrected...) It has not yet entered the historical optimal range. The next cycle will continue to adjust according to the above rules until... It has entered its historical optimal range. If it corrects for three consecutive cycles... It still cannot enter the historical best range, triggering the "model mismatch" advanced alarm, indicating that the feedforward model needs to be recalibrated.
[0010] Preferably, the multi-source data fusion mechanism of the feedback diagnosis and correction loop in step 4 includes two levels: an instrument layer and a diagnostic layer. The instrument layer acquires the total phosphorus and residual sludge concentrations in the effluent in real time. The diagnostic layer receives the root cause analysis results and their confidence levels output by the host computer's abnormal diagnosis module, which uses the patented "Probabilistic Diagnosis Method for Abnormal Operating Conditions in Wastewater Treatment Based on Multi-level Decision Tree Fusion and Bayesian Inference" for diagnosis.
[0011] The three-layer determination includes: Scenario 1 is defined as the effluent meeting the standards, that is, the total phosphorus concentration in the effluent is less than or equal to 0.5 mg / L and the suspended solids concentration in the effluent is less than or equal to 10 mg / L. The execution strategy for Scenario 1 is: (1) If the current (1) If the current range is within the historical optimal range, maintain the current range and directly implement the drug addition; (2) If the current range is within the historical optimal range, maintain the current range and directly implement the drug addition; Deviation from the historical best range but less than 5% (i.e. Adjust the step correction according to the step correction rule described in step 3 in the interval direction, and take the step coefficient as... (3) If the current If the deviation from the historical best range is ≥5%, adjust according to the normal stepping rules described in step 3. The actual value of this period The actual execution value (before or after correction) is included as a valid K1 value in the historical database, updating the historical best K1 value range. Data from Scenario 1 is only included in the statistics under natural compliance conditions, that is, only the periodic data of water effluent meeting the standards after dosing is calculated directly by the feedforward loop without feedback correction from Scenario 2 or Scenario 3, excluding data that barely meets the standards after feedback correction, thus ensuring the reliability of the historical best K1 value range.
[0012] Scenario 2 is defined as a situation where the total phosphorus in the effluent exceeds the standard and the suspended solids in the effluent are too high, that is, the total phosphorus concentration in the effluent is greater than 0.5 mg / L and meets any of the following conditions: (a) the suspended solids concentration in the effluent exceeds twice the standard deviation of the average value of the historical compliance period; (b) the suspended solids concentration in the effluent is greater than 30 mg / L (this threshold can be used as an on-site setting parameter, and the recommended range is 20~40 mg / L).
[0013] Scenario 3 is defined as effluent total phosphorus exceeding the standard but not belonging to the exceeding situation of Scenario 2 (i.e., effluent total phosphorus concentration is greater than 0.5 mg / L, but effluent suspended solids concentration does not meet any of the high conditions of Scenario 2), and the diagnosis results of the upper-level diagnostic module do not include root causes of mud-water separation failure or the confidence level is lower than the preset threshold.
[0014] Preferably, the quantitative determination criteria for elevated suspended solids in the effluent in scenario two include: The effluent suspended solids exceed the preset suspended solids threshold, or exceed the preset standard deviation of the historical average for compliance periods. The root causes of sludge-water separation failures output by the upper-level diagnostic module include secondary sedimentation tank scraper malfunctions and excessively high SVI (Self-Identity Index), with confidence levels exceeding a preset confidence threshold. This threshold can be adjusted within a preset range based on the on-site tolerance for false alarm rates. Correction strategies also include marking the current cycle's operational data as anomalous and not using it to update the historical K1 database to avoid model contamination.
[0015] Preferably, the control parameters of the incremental PI controller in scenario three include: The input is the deviation between the total phosphorus in the effluent and the preset target value. The output is Correction amount The sampling period is each dosing cycle or a preset time interval; the incremental PI control formula is: , In the formula, For the t-th sampling time Correction amount; Kp is the proportional coefficient, initially set to 0.5; Ki is the integral coefficient, initially set to 0.1; e(t) is the deviation at sampling time t, defined as e(t) = TP out(t) -TP target , of which TP target The target total phosphorus concentration in the effluent; e(t-1) is the deviation at the previous sampling time. Corrected Used for the current cycle, not added to the historical queue, and can be used for feedforward calculations in the next cycle as a new benchmark.
[0016] Preferably, the data closed-loop optimization mechanism in step 5 further includes the following: The system continuously records operational data for each dosing cycle, including influent water quality parameters, effluent water quality parameters, dosage, and K1base value. The K1base value is only included in the historical database as a valid sample when the total phosphorus and suspended solids in the effluent meet the standards. Valid samples are managed in a first-in, first-out manner, with the queue length fixed at a preset length. The system calculates the mean and standard deviation of valid samples in real time and dynamically updates the historical best K1 value range, enabling the model to have self-learning and adaptive capabilities.
[0017] Preferably, the method of the present invention also includes a collaborative mechanism with the superior anomaly diagnosis system: the system is deeply integrated with the patent "Probabilistic Diagnosis Method for Abnormal Operating Conditions of Wastewater Treatment Based on Multi-level Decision Tree Fusion and Bayesian Inference"; the root cause analysis results and their confidence levels output by the superior diagnosis system serve as key inputs for scenario determination in this method; the collaborative mechanism forms a complete closed loop of perception-diagnosis-decision-control.
[0018] Preferably, the method of the present invention is applied to urban wastewater treatment plants employing biological phosphorus removal processes, including but not limited to A. 2 The system can receive operational data from multiple parallel processing structures, independently perform feedforward calculations and feedback corrections for each structure, and use the corrected parameters for feedforward calculations in subsequent cycles. These processes include O / O processes, oxidation ditch processes, and sequencing batch reactor (SBR) activated sludge processes.
[0019] Compared with the prior art, the present invention has the following beneficial effects: 1. Precision dosing control: This invention introduces the amount of residual sludge discharge as an input feature to establish a quantitative relationship between biological phosphorus removal capacity and reagent dosage, realizing the concept of precision dosing with biochemical dosing as the main method and chemical dosing as a supplement, effectively reducing reagent consumption and operating costs.
[0020] 2. Proactive early warning: This invention constructs a dynamic historical optimal K1 value range and performs multi-dimensional verification after each feedforward loop calculation and before dosing. It can identify abnormal operating conditions and issue early warnings before the effluent exceeds the standard, thus avoiding over-dosing or under-dosing.
[0021] 3. Intelligent identification of causes of exceeding standards: This invention analyzes the contribution of suspended solids and dissolved phosphorus in the effluent to the total phosphorus exceeding the standard online. Combined with the root cause analysis results of the upper-level diagnostic system, it automatically distinguishes between two causes of exceeding the standard: insufficient reagent and process failure. Based on this, completely different control strategies are implemented, effectively avoiding the waste of resources caused by blindly adding reagents.
[0022] 4. Self-learning and adaptive capability: This invention uses a closed-loop optimization mechanism that automatically incorporates compliance period data into the historical database and continuously updates the optimal range of K1, enabling the model to automatically adapt to seasonal changes in water quality and changes in sludge characteristics without the need for manual recalibration, and the control accuracy continues to improve over time.
[0023] 5. System Coordination and Integrity: This invention is deeply integrated with the superior anomaly diagnosis patent to form a complete intelligent closed loop of perception-diagnosis-decision-control, which significantly improves the intelligence level and operational reliability of the chemical phosphorus removal system in the wastewater treatment plant. Attached Figure Description
[0024] Figure 1 This is a flowchart of the overall process of the present invention; Figure 2 This is a flowchart of the feedforward control method of the present invention. Detailed Implementation
[0025] Example 1
[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.
[0027] like Figure 1 , 2 As shown, this invention provides a feedforward-feedback dual-loop control method for phosphorus removal dosing based on anomaly diagnosis. This method achieves preliminary calculation of phosphorus removal agent dosage by constructing a feedforward loop, performs pre-emptive verification and proactive early warning by constructing a historical optimal K1 value range, accurately identifies and classifies effluent exceeding standards by constructing a feedback diagnosis and correction loop, and achieves self-learning and adaptive modeling through a data closed-loop optimization mechanism. This method is applicable to urban wastewater treatment plants employing biological phosphorus removal processes, including but not limited to A… 2 / O process, oxidation ditch process, sequencing batch activated sludge process, etc.
[0028] In the above method, step 1 constructs a feedforward loop and uses a multiple linear regression model to train the initial dosage prediction model, thereby achieving a preliminary calculation of the phosphorus removal agent dosage. The training process of the multiple linear regression model uses historical operating data as the sample set, and the sample data includes influent flow rate (…). ), total phosphorus in the influent ( ), residual sludge discharge ( The three input feature variables and their corresponding measured K1 values are used as output labels.
[0029] The model training employs least squares for parameter estimation, establishing a linear mapping between input features and output variables. After training, the model coefficient matrix and intercept term are saved for real-time inference calculations. The specific input features of the model include: inflow rate. The unit is cubic meters per hour, and the cycle water inflow is collected in real time and accumulated by an electromagnetic flow meter installed in the inlet channel; total phosphorus in the inlet water The unit is milligrams per liter, measured using an online total phosphorus analyzer with ammonium molybdate spectrophotometry; residual sludge discharge... The unit is cubic meters per hour, measured in real time by an electromagnetic flowmeter installed on the excess sludge discharge pipe. The model's output characteristics are based on the dosage ratio coefficient. The unit is milligrams of phosphorus removal agent per milligram of phosphorus. This coefficient represents the mass of phosphorus removal agent required per unit mass of total phosphorus in the influent. The dosage calculation formula is as follows: ; In the formula, : Dosage of phosphorus removal agent (unit: mg / L or engineering unit matched with metering pump); : Total phosphorus concentration in influent (mg / L); The concentration of residual sludge (g / L) can be taken as an empirical value of 5-10. : Dimensionless biological phosphorus removal efficiency coefficient, initially 0.5, adaptively corrected; Excess sludge discharge (m³) 3 / h); Inlet flow rate (m³) 3 / h); : Basic dosage ratio coefficient (unit: mg phosphorus remover / mg phosphorus); : Constant correction term (mg / L). The design principle of this formula is that the system prioritizes predicting the amount of phosphorus that can be removed by biological phosphorus removal based on the amount of residual sludge discharged. The reagent is only added to remove the remaining part that the biological system has failed to absorb, thereby realizing the concept of precise dosing with biochemical treatment as the main method and chemical treatment as a supplement.
[0030] The calculation process for the feedforward loop includes the following steps: First, the controller reads the influent flow rate, total influent phosphorus, and excess sludge discharge for the current cycle through the data acquisition interface, and queries the currently effective model parameters K1base, γ, and β. Then, it substitutes these parameters into the dosage calculation formula to solve for the phosphorus removal agent dosage Dt for the current cycle. Finally, based on Dt and the metering pump calibration curve, the controller converts the dosage into the real-time flow setpoint of the metering pump and issues it for execution. The output of the feedforward loop serves as the input benchmark for the feedback diagnostic loop, and is compared and verified with the historical best K1 interval to form the final dosing command.
[0031] In the above method, step 2 constructs the historical best K1 value interval. The system automatically records the K1base value and the corresponding effluent total phosphorus and residual sludge concentrations for each dosing cycle, thus constructing a dynamic statistical baseline. The system maintains a first-in, first-out queue of length T to store valid K1 value samples. The queue length T is set to 30 days, storing only the most recent valid K1 values, i.e., the cycle data where both effluent total phosphorus and suspended solids meet the standards. The criteria for valid samples are: effluent total phosphorus concentration less than or equal to 0.5 mg / L and effluent suspended solids concentration less than or equal to 10 mg / L. In the initial stage of system operation, when the number of valid K1 value samples is less than the preset minimum sample size of 10, a fixed initial interval based on expert experience is used as the historical best K1 value interval (e.g., ...). Once the sample size reaches the minimum, the calculation switches to dynamic interval calculation. The formula for calculating the dynamic historical best K1 value interval is: The historical best K1 value range = [μ - nσ, μ + nσ]; In the formula, μ is the mean of the effective K1 values, σ is the standard deviation, and n is the confidence coefficient, set based on actual operating experience, with a value range of 1.5 to 2.0. The interval coefficient n is dynamically adjusted according to the queue length: when the number of effective samples is less than 20, n is 1.0 to improve sensitivity; when the number of effective samples is between 20 and 50, n is 1.5 to balance sensitivity and stability; when the number of effective samples is greater than 50, n is 2.0 to improve anti-interference capability. Non-operating days are not included in the queue, that is, when the system detects that the biological treatment tank is shut down or production is stopped, data samples for that period are not collected. This sliding window mechanism ensures that the historical best K1 value range can adapt to seasonal changes in water quality.
[0032] The dynamic update process for the historical best K1 value range includes the following steps: First, at the end of each dosing cycle, the system determines whether the cycle contains valid samples. If valid, its K1base value is added to the tail of the FIFO queue, and the oldest sample at the head of the queue is removed to maintain a constant queue length. If invalid, no action is taken. Next, the system calculates the mean μ and standard deviation σ of all valid samples in the current queue. Finally, the system calculates the lower and upper limits of the interval based on μ, σ, and the dynamically adjusted n value, updating the historical best K1 value interval. The updated interval serves as the benchmark reference for the multi-dimensional verification of K1 values in step 3.
[0033] In the above method, step 3 performs multi-dimensional verification of the K1 value, and the feedforward loop calculates the K1 value each time. After determining the value, it is compared with the historical best K1 interval, and the corresponding pre-verification strategy is executed based on the comparison result. The verification process includes the following three comparison results and corresponding execution strategies: when When the dosing ratio is within the historical optimal range, the system determines that the dosing ratio is reliable and directly executes the dosing based on the dosage Dt calculated by the feedforward loop without any additional adjustments. At this time, the controller converts Dt into the metering pump flow setpoint and issues it for execution, while simultaneously recording the basic parameters of this dosing cycle for subsequent feedback and correction.
[0034] when When the dosage is below the lower limit of the range, the system determines that there may be a risk of insufficient dosage, issues a mild warning, and finely adjusts the dosage with reference to the lower limit of the range.
[0035] Specifically, system computing The deviation from the lower limit ΔK=K best,lowe r- Add 50% of this deviation as an adjustment amount to The adjusted drug dosage ratio coefficient K1 is obtained. adj = +0.5×ΔK. The controller is based on K1.adj Recalculate the dosage and execute the dosing, while simultaneously sending an early warning message to the central control system: "Feedforward calculation" The water quality is below the lower limit of the historical optimal range. The dosage has been automatically adjusted. Please pay attention to the effluent water quality.
[0036] when When the dosage exceeds the upper limit of the range, the system determines that there may be excessive and wasteful dosage, issues a mild warning, and finely adjusts the dosage with reference to the upper limit of the range.
[0037] Specifically, system computing The deviation from the upper limit ΔK= -K best,upper Use 50% of this deviation as the adjustment amount from After deducting from the middle, the adjusted drug dosage ratio coefficient K1 is obtained. adj = -0.5×ΔK. The controller is based on K1. adj Recalculate the dosage and execute the dosing, while simultaneously sending an early warning message to the central control system: "Feedforward calculation" The dosage is above the upper limit of the historical optimal range. The dosage has been automatically adjusted. Please pay attention to the drug consumption.
[0038] This verification mechanism enables proactive early warning, allowing for the identification of abnormal operating conditions and the implementation of preventative measures before effluent exceeds standards. This effectively avoids reduced treatment effectiveness and waste of chemicals caused by excessive or insufficient chemical dosing.
[0039] In the above method, step 4 constructs a feedback diagnosis and correction loop, acquires the total phosphorus and residual sludge concentrations in the effluent in real time, receives the root cause analysis results and their confidence levels output by the upper computer's anomaly diagnosis module, and performs a three-layer judgment and corresponding correction strategy based on the above information. The multi-source data fusion mechanism of the feedback diagnosis and correction loop includes two levels: the instrumentation layer and the diagnosis layer.
[0040] The instrumentation layer is responsible for acquiring real-time effluent water quality data, including total phosphorus concentration and suspended solids concentration. Total phosphorus concentration is measured using an online total phosphorus analyzer installed at the effluent outlet, with a sampling cycle set to 5 minutes, and analyzed using the potassium persulfate digestion-ammonium molybdate spectrophotometric method. Suspended solids concentration is measured using an online suspended solids analyzer, employing either infrared scattering or filtration-weighing methods. The instrumentation layer also performs preliminary data validation, eliminating obviously abnormal jumps in data, and compensating for instrument zero-point drift and range drift.
[0041] The diagnostic layer is responsible for receiving the root cause analysis results and confidence levels from the host computer's anomaly diagnosis module. The host computer module employs a multi-level decision tree fusion and Bayesian inference method for probabilistic diagnosis of abnormal operating conditions, corresponding to the patent "Probabilistic Diagnosis Method for Abnormal Operating Conditions in Wastewater Treatment Based on Multi-level Decision Tree Fusion and Bayesian Inference." The diagnostic layer establishes a communication connection with the host computer module via TCP / IP protocol or OPC interface, receives diagnostic messages, and parses the diagnostic result field and confidence level field. The diagnostic result field identifies the anomaly type, including sludge-water separation failure, secondary sedimentation tank scraper failure, excessively high SVI, influent shock load, insufficient reagent dosage, etc.; the confidence level field is an integer value from 0 to 100, indicating the reliability of the diagnostic result.
[0042] Based on the above information, the system performs a three-level decision-making process, with the following rules: Scenario 1 is defined as effluent meeting standards, i.e., total phosphorus concentration in the effluent is less than or equal to 0.5 mg / L and suspended solids concentration is less than or equal to 10 mg / L. The implementation strategy for Scenario 1 is: maintain or fine-tune. This period's The value is included as a valid K1 value in the historical database, updating the historical best K1 value range. At the end of this period, the system will... Parameters such as influent flow rate, total phosphorus influent, and residual sludge discharge are packaged into valid sample records and added to the tail of the FIFO queue. Data from Scenario 1 are only included in the statistics under natural compliance conditions, excluding data that barely meets the standards after feedback correction, thus ensuring the reliability of the historical best K1 value range.
[0043] Scenario 2 is defined as a situation where the effluent total phosphorus exceeds the standard and the effluent suspended solids are too high, specifically, the effluent total phosphorus concentration is greater than 0.5 mg / L and the effluent suspended solids concentration is greater than 30 mg / L, or the SS exceeds twice the standard deviation of the historical average for compliance periods; simultaneously, the upper-level diagnostic module outputs a root cause of "sludge-water separation failure" (such as secondary sedimentation tank scraper failure, excessively high SVI, etc.) with a confidence level exceeding a preset threshold (e.g., 70%, which can be adjusted between 60% and 80% based on the on-site tolerance for false alarms). The diagnostic result for Scenario 2 is sludge-water separation failure. The execution strategy for Scenario 2 is: do not add phosphorus removal agents, issue a process alarm to the central control system to "check sludge settling / filtration system", and suspend the use of this abnormal data for correction. .
[0044] Specifically, the control command is for the system to immediately send a command to the dosing pump to maintain the current dosage of the reagent and even slightly reduce it to prevent overdose; the alarm command is for the system to push a high-level alarm to the central control room, and the alarm information includes fields such as the diagnostic result "sludge-water separation failure", confidence level "85%", suggested measures "please check the operation status of the secondary sedimentation tank sludge scraper immediately" and urgency level "severe"; the data processing is for the system to mark the operating data of this cycle as abnormal data and not use it to update the historical K1 database to avoid contaminating the model.
[0045] Scenario 3 is defined as a situation where the effluent total phosphorus exceeds the standard, the effluent suspended solids are normal or low, and the upper-level diagnostic does not point to a sludge-water separation fault. Specifically, the effluent total phosphorus concentration is greater than 0.5 mg / L and the effluent suspended solids concentration is less than or equal to 30 mg / L (or SS does not exceed twice the standard deviation of the historical average for compliance periods), and the diagnostic results from the upper-level diagnostic module do not include a sludge-water separation fault as a root cause or the confidence level is below the preset threshold. The diagnostic result for Scenario 3 is "insufficient phosphorus removal agent dosage" or "influent shock load." The execution strategy for Scenario 3 is to correct through an incremental PI controller. The control formula is as follows: ; In the formula, For the t-th sampling time Correction amount; Kp is the proportional coefficient, initially set to 0.5; Ki is the integral coefficient, initially set to 0.1; e(t) is the deviation at sampling time t, defined as e(t) = TP out(t) -TP target , of which TP target The target total phosphorus concentration in the effluent is set to 0.5 mg / L; e(t-1) represents the deviation from the previous sampling time. The sampling period is set to each dosing cycle or every 5 minutes. The single correction range is limited to ±10% to prevent over-adjustment from causing system oscillation. The corrected K1base is used for the dosage calculation of the current cycle and is not added to the historical queue for statistical updates, but it can be used as a new baseline parameter for the feedforward calculation of the next cycle.
[0046] The execution process of the three-layer decision-making includes the following steps: First, the system reads real-time data of total phosphorus and suspended solids in the effluent at each sampling time. Then, the system checks whether it has received a diagnostic message from the host computer diagnostic module. If so, it parses the diagnostic results and confidence levels. Next, the system performs condition checks in the order of Scenario 1, Scenario 2, and Scenario 3. Once all the conditions of a scenario are met, the system immediately exits the check and executes the corresponding strategy. Finally, the system generates control commands and alarm information based on the check results and sends them out for execution through the communication interface.
[0047] The alarm generation and push mechanism, as a crucial component of the feedback diagnostic loop, ensures that maintenance personnel can promptly be aware of system anomalies and take appropriate countermeasures. Alarm information includes the following fields: Alarm Number, uniquely identified using a timestamp and serial number; Alarm Type, indicating the anomaly category, such as "Process Fault Warning," "Dosing Deviation Warning," or "Model Parameter Anomaly"; Alarm Content, describing the specific anomaly and diagnostic results; Confidence Level, displaying the confidence level value output by the higher-level diagnostic module; Recommended Actions, providing handling suggestions for this anomaly type; Urgency Level, categorized into three levels: Informational, Warning, and Severe; and Timestamp, recording the alarm generation time. Alarm information is pushed to the monitoring interface and mobile terminals via the central control system communication interface, ensuring timely response from relevant personnel.
[0048] In the above method, step 5 achieves data closed-loop optimization. Data from the target period is automatically included in the historical database, and the optimal K1 interval is continuously updated, enabling the model to have self-learning and adaptive capabilities. The data closed-loop optimization mechanism includes the following: The system continuously records operational data for each dosing cycle, including but not limited to influent flow rate Qin, total phosphorus TPin, ammonia nitrogen, COD, total nitrogen TNin, excess sludge discharge Qwaste, total phosphorus TPout, suspended solids SS, COD, ammonia nitrogen, dissolved oxygen DO, sludge concentration MLSS, sludge volume index SVI, real-time dosage Dt, baseline dosage coefficient K1base, corrected dosage coefficient K1adj, sampling time, and operational status indicators. Data recording frequency is consistent with the sampling cycle, and data is stored in a time-series database format, supporting historical data querying and statistical analysis.
[0049] Valid samples are determined using a strict dual-compliance standard: total phosphorus concentration in the effluent must be less than or equal to 0.5 mg / L and suspended solids concentration must be less than or equal to 10 mg / L. The system will only process a sample for the current cycle if both conditions are met simultaneously. Values are included in the historical database as valid samples. Valid samples are managed using a first-in, first-out (FIFO) approach, with a fixed queue length of 30 days. When the queue is full, the oldest sample record is automatically removed to maintain a constant queue length.
[0050] The self-learning and adaptive capability is achieved using an exponentially weighted moving average method, with the system dynamically updating based on historical valid samples. The target reference value.
[0051] Specifically, the system calculates the weighted average of all valid samples in the current FIFO queue. The weighting factors decay exponentially based on the sample collection time, with higher weights for recent samples and lower weights for older samples, thus enabling the model to automatically adapt to seasonal changes in water temperature and sludge characteristics. (Updated version) The reference value is used to set the initial value for feedforward calculation in subsequent cycles, eliminating the need for manual recalibration, and the control accuracy continues to improve over time.
[0052] In the above method, steps 1 to 5 constitute a complete feedforward-feedback dual-loop control system architecture for phosphorus removal dosing. The feedforward loop performs preliminary calculations of the phosphorus removal agent dosage, using influent flow rate, total phosphorus influent, and excess sludge discharge as input features, and outputs a basic dosage ratio coefficient through a multiple linear regression model. The dosage of phosphorus removal agent is calculated using the dosage formula. The feedback loop enables accurate identification and classification when effluent exceeds standards. Total phosphorus and suspended solids in the effluent are used as instrument layer inputs, and the root cause analysis results from the upper-level computer diagnostic module are used as diagnostic layer inputs. Through three-layer judgment logic, different causes of effluent exceeding standards are distinguished, and two completely different control strategies are executed accordingly: agent dosage correction or process fault alarm.
[0053] The workflow of dual-loop collaborative control includes the following stages: When the dosing cycle starts, the controller performs feedforward loop calculation to obtain the initial dosage; then it performs historical best K1 value interval verification, and decides whether to fine-tune the dosage based on the verification result; then it sends the dosing command to the metering pump for execution; during the dosing process, the controller continuously collects effluent water quality data and receives diagnostic results from the host computer; at the end of the dosing cycle, the controller performs a three-level judgment of the feedback loop based on the effluent data and diagnostic results, executes the corresponding strategy based on the judgment result, and updates the model parameters.
[0054] System safety constraints and anomaly handling mechanisms ensure safe and stable operation under various working conditions. The minimum dosing cycle is set to 5 minutes to prevent frequent start-stop of the metering pump from affecting equipment lifespan; the maximum dosing cycle is set to 4 hours, after which the current cycle is forcibly terminated and an anomaly record is generated. Upper limit constraints are set for model parameters. The adjustment range of γ is limited to 0.5 to 2.0 times the initial value, the adjustment range of γ is limited to 0.3 to 0.7, and the adjustment range of β is limited to ±50% of the initial value, to prevent parameter runaway caused by extreme errors.
[0055] Example 2
[0056] This embodiment provides a specific application example of the above-mentioned feedforward-feedback dual-loop control method for phosphorus removal and dosing based on anomaly diagnosis in a scenario involving multiple parallel processing structures.
[0057] A city's wastewater treatment plant uses A 2The / O process is designed to treat 50,000 cubic meters per day. The biological reactors are divided into four groups operating in parallel, each with an effective volume of 2,500 cubic meters. Aeration is achieved using blower aeration, with microporous aeration discs arranged at the bottom of the tanks. The aeration rate is controlled by an electric regulating valve based on the dissolved oxygen setpoint. Chemical phosphorus removal uses polyaluminum chloride as the phosphorus removal agent, with the addition point located in the mixed liquor pipeline at the aeration tank outlet. The phosphorus removal agent is precisely added via a metering pump. Online instrumentation includes: an electromagnetic flow meter and an online total phosphorus analyzer installed on the influent pipeline; an online total phosphorus analyzer, an online suspended solids analyzer, a dissolved oxygen meter, and a pH meter installed on the effluent from each group of biological reactors; and an electromagnetic flow meter installed on the excess sludge pipeline. The host computer system is equipped with an anomaly diagnosis module based on multi-level decision tree fusion and Bayesian inference, which communicates with the controller of this method via the OPC protocol.
[0058] During system operation, the specific execution process of a certain dosing cycle is as follows: the influent flow meter shows that the current influent flow rate Qin(t) is 1800 cubic meters per hour, the influent total phosphorus online analyzer shows that the influent total phosphorus concentration TP is 4.2 mg / L, and the residual sludge flow meter shows that the residual sludge discharge is 45 cubic meters per hour.
[0059] After reading the above parameters, the controller executes step 1, feedforward loop calculation: querying the currently effective model parameters. =1.75, γ=0.52, β=0.3, substituting into the dosage formula, we calculate Dt=1.75×4.2-0.52×(45 / (1800×8))+0.3=7.35-0.16+0.3=7.49 mg / L. Converting this to the metering pump flow setting: if the metering pump calibration curve shows 1% opening corresponds to a dosage of 0.5 mg / L, then the setting value is 14.98% opening.
[0060] The controller executes step 2 to query the historical best K1 value range: the current FIFO queue length is 30 days, the number of valid samples is 28, and the calculated mean μ=1.78, standard deviation σ=0.12, confidence coefficient n=1.5, and lower limit K of the interval. best,lower =1.78 - 1.5 × 0.12 = 1.60, Upper limit of interval K best,upper =1.78 + 1.5 × 0.12 = 1.96. Because =1.75 is within the range of 1.60 and 1.96, so the dosage ratio is considered reliable, and the dosage should be applied directly according to the calculated amount.
[0061] During the dosing process, the controller continuously collects effluent water quality data and checks for diagnostic messages from the host computer. The sampling cycle is 5 minutes, and after collecting 12 data points, the system enters the judgment phase. The online total phosphorus analyzer shows that the total phosphorus concentration fluctuates between 7.1 and 7.4 mg / L, with an average of 7.25 mg / L, exceeding the target value by 0.5 mg / L. The online suspended solids analyzer shows that the suspended solids concentration fluctuates between 8 and 11 mg / L, with an average of 9.5 mg / L, which is within the normal range. At this point, the controller determines that the total phosphorus concentration in the effluent exceeds the standard, but the suspended solids concentration is normal.
[0062] Meanwhile, the controller received a diagnostic message from the host computer diagnostic module: the diagnostic result was "influent shock load" with a confidence level of 65%. Since the confidence level was below the 70% threshold and the diagnostic result did not include the root cause of the mud-water separation failure, the controller determined it to be Scenario 3: the total phosphorus in the effluent exceeded the standard and the suspended solids in the effluent were normal, and the host computer diagnosis did not point to a mud-water separation failure.
[0063] The controller executes step 4, scenario 3, correction strategy: Calculate the current deviation e(t) = 7.25 - 0.5 = 6.75 mg / L, look up the deviation e(t-1) = 6.50 mg / L at the previous sampling time, and substitute it into the incremental PI control formula to calculate ΔK1(t) = 0.5 × (6.75 - 6.50) + 0.1 × 6.75 = 0.125 + 0.675 = 0.80. Single correction amplitude limit check: ΔK1 / K1base = 0.80 / 1.75 = 45.7%, which is greater than 10%, therefore amplitude limit correction is performed, and the actual correction amount ΔK1_lim = 1.75 × 10% = 0.175. The corrected drug dosage ratio coefficient K1 adj =1.75 + 0.175 = 1.925, recalculate the dosage Dt. adj =1.925×4.2-0.52×(45 / (1800×8))+0.3=8.09-0.16+0.3=8.23 mg / L. The controller will issue the adjusted dosing command to the metering pump for execution. The operating data of this cycle will not be included in the historical queue for statistical updates.
[0064] After the dosing cycle ends, the system generates a cycle operation report, which includes the following: cycle start and end time, influent flow rate, total influent phosphorus, excess sludge discharge, feedforward calculation K1base, historical best K1 value range, verification results, effluent total phosphorus, effluent suspended solids, higher-level diagnostic results, scenario judgment results, implementation status of correction strategies, adjusted K1base, and whether valid samples were included in this cycle. This report is saved to the historical database for subsequent analysis and pushed to the central control system monitoring interface for maintenance personnel to view.
[0065] After the system ran continuously for 30 days, the number of valid samples reached 30, and the FIFO queue was updated. At this point, the controller recalculated the historical optimal K1 value range: μ=1.82, σ=0.11, n=1.5 (taking the upper limit when the sample size is sufficient), and the lower limit of the range K. best,lower =1.82 - 1.5 × 0.11 = 1.655, Upper limit of interval K best,upper =1.82 + 1.5 × 0.11 = 1.985. This dynamically updated interval reflects the actual compliance level of the system in recent operation and can be used as a reference for feedforward verification in the next stage.
[0066] The engineering practice of this embodiment shows that the above-mentioned feedforward-feedback dual-loop control method for phosphorus removal based on anomaly diagnosis can effectively distinguish different causes of excessive total phosphorus in effluent, avoid ineffective dosing in the case of mud-water separation failure, increase the dosage in time when the dosage is insufficient, and achieve proactive early warning through the pre-verification mechanism of the historical best K1 value range, which effectively reduces the cost of chemical consumption and the risk of excessive effluent.
[0067] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the above embodiments are merely illustrative of the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.
Claims
1. A feedforward-feedback dual-loop control method for phosphorus removal dosing based on anomaly diagnosis, characterized in that, Includes the following steps: Step 1: Construct a feedforward loop and use historical operating data to train an initial dosage prediction model using a multiple linear regression model. This model uses the influent flow rate as the starting point. Total phosphorus in influent Excess sludge discharge Given the input features, output the baseline dosage ratio coefficient. The dosage of phosphorus removal agent is calculated based on the input features; Step 2: Construct a historical optimal K1 value range, record the basic dosing ratio coefficient value and the corresponding total phosphorus and residual sludge concentration in each dosing cycle, mark the effluent total phosphorus and suspended solids as effective K1 values when they meet the standards, maintain a first-in-first-out queue to store the most recent effective K1 values, calculate the queue mean and standard deviation, and construct a dynamic historical optimal K1 value range. Step 3: Perform multi-dimensional verification of K1 value. After each feedforward loop calculates the basic drug dosage ratio coefficient value, it is compared with the historical best K1 interval, and the corresponding strategy is executed based on the comparison results. Step 4: Construct a feedback diagnosis and correction loop, obtain the total phosphorus and residual sludge concentration in the effluent in real time, receive the root cause analysis results and their confidence levels output by the upper computer anomaly diagnosis module, perform three-level judgment based on the information and execute the corresponding correction strategy. Step 5: Data from the target period is incorporated into the historical database and the optimal range of K1 is continuously updated.
2. The phosphorus removal dosing feedforward-feedback dual-loop control method according to claim 1, characterized in that, The formula for calculating the dosage is as follows: , in, : Dosage of phosphorus removal agent; Total phosphorus concentration in influent; : Concentration of residual sludge; : Dimensionless biological phosphorus removal efficiency coefficient; : Excess sludge discharge; : Inlet water flow rate; : Basic drug dosage ratio coefficient; : Constant correction term.
3. The phosphorus removal dosing feedforward-feedback dual-loop control method according to claim 1, characterized in that, In step 2, the length T of the first-in-first-out queue is set to a preset time length, storing only the most recent valid K1 value, i.e., periodic data where both total phosphorus and suspended solids in the effluent meet the standards. The formula for calculating the dynamic historical optimal K1 value interval is: The historical best K1 value range = [μ - nσ, μ + nσ]; Where μ is the mean of the effective K1 values, σ is the standard deviation, and n is set according to actual operating experience, with a value range of the preset range.
4. The phosphorus removal dosing feedforward-feedback dual-loop control method according to claim 1, characterized in that, The multi-dimensional verification rules for the K1 value in step 3 specifically include three comparison results and corresponding execution strategies: when When the dosage falls below the lower limit of the range, the system determines that there is a risk of insufficient dosage, issues an early warning, and performs step-by-step corrections according to the following formula: ,in, This is the step factor; This represents the lower limit of the historical best K1 value range. when When the dosage exceeds the upper limit of the interval, the system determines that there is excessive waste due to excessive dosage, issues a warning, and performs step-by-step correction according to the following formula: ,in, This represents the upper limit of the historical best K1 value range; The magnitude of a single correction shall not exceed the current The value is 10%; if it exceeds this, a limit will be applied; if corrected... It has not yet entered the historical optimal range. The next cycle will continue to adjust according to the above rules until... Entering the historical best range; if it still cannot enter the range after 3 consecutive period corrections, a "model mismatch" alarm is triggered.
5. The feedforward-feedback dual-loop control method for phosphorus removal dosing according to claim 1, characterized in that, The multi-source data fusion mechanism of the feedback diagnosis and correction loop in step 4 includes two levels: the instrument layer and the diagnostic layer. The instrument layer acquires the total phosphorus and residual sludge concentrations in the effluent in real time. The diagnostic layer receives the root cause analysis results and their confidence levels output by the upper computer's abnormal diagnosis module.
6. The phosphorus removal dosing feedforward-feedback dual-loop control method according to claim 5, characterized in that, The three-layer determination includes: Scenario 1 assumes the water output meets the standards; maintain or make minor adjustments. and will this cycle's The value is included in the historical database as a valid K1 value, and the range of historical best K1 values is updated. Among them, the valid K1 values included in the historical database are limited to the data of the natural compliance period without feedback correction. Data that barely meets the standard after feedback correction is not included in the historical database. Scenario 2 involves excessive total phosphorus and high suspended solids in the effluent. The diagnosis is a sludge-water separation failure. The strategy is to not add phosphorus removal agents, issue a process alarm to the central control system to check sludge settling or the filtration system, and suspend the use of this abnormal data for correction. ; Scenario 3 involves effluent with excessive total phosphorus but normal or low suspended solids, and the upper-level diagnostics do not indicate a mud-water separation fault. The diagnostic result is insufficient phosphorus removal agent dosage or influent shock load. The execution strategy is to correct the issue using an incremental PI controller. .
7. The phosphorus removal dosing feedforward-feedback dual-loop control method according to claim 5, characterized in that, The control parameters of the incremental PI controller in scenario three include: The input is the deviation between the total phosphorus in the effluent and the preset target value. The output is Correction amount The sampling period is each dosing cycle or a preset time interval; the incremental PI control formula is: , In the formula, For the t-th sampling time Correction amount; Kp is the proportional coefficient, initially set to 0.5; Ki is the integral coefficient, initially set to 0.1; e(t) is the deviation at sampling time t, defined as e(t) = TP out(t) -TP target , of which TP target t is the target total phosphorus concentration in the effluent; e(t-1) is the deviation at the previous sampling time.
8. The phosphorus removal dosing feedforward-feedback dual-loop control method according to claim 1, characterized in that, The data closed-loop optimization mechanism in step 5 also includes the system continuously recording the operating data of each dosing cycle, including influent water quality parameters, effluent water quality parameters, drug dosage, and basic dosing ratio coefficient; the determination of valid samples adopts the dual compliance standard of effluent total phosphorus concentration being less than or equal to the preset total phosphorus compliance limit and effluent suspended solids concentration being less than or equal to the preset suspended solids compliance limit; the valid samples are managed in a first-in-first-out manner with the queue length fixed to the preset queue length; The system calculates the mean and standard deviation of the effective samples in real time and dynamically updates the historical best K1 value range.
