Intelligent recommendation method for dosing amount of biogas slurry carbon source in sewage treatment plant

Abstract

The present application relates to sewage treatment and intelligent dosing control technical field, and disclose a kind of intelligent recommendation method for the dosing amount of biogas slurry carbon source used in sewage treatment plant.Schemes are laid out water inflow and outflow COD and flow monitoring point, unified sampling cycle, time sequence, total duration and number, set target effluent, carbon source maximum dosing and periodic recommended rate;Obtain biochemical pool volume and minimum credible flow, calculate residence time, do two groups of equal volume standard addition conversion biogas slurry equivalent and concentration conversion coefficient, establish discrete prediction model;Accordingly, the next cycle deviation, sensitivity and equivalent dosing error are obtained, a unified dimension double target is constructed, upper and lower bounds and multiplier are introduced into Lagrange constraint, and unconstrained solution and closed expression satisfying the constraint are given by setting stationary point condition, and branching and multiplier strategy are given for interval, lower bound, upper bound and low flow;According to the period rolling collection-calculation-determination-output full sequence, the deviation sum of squares is calculated and archived.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment and intelligent dosing control technology, specifically to an intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants. Background Technology

[0002] In actual operation, wastewater treatment plants often need to dynamically adjust the carbon source dosage based on influent water quality, flow rate, and other operating conditions to ensure stable effluent quality that meets standards. The precise addition of carbon sources directly affects not only the nitrogen and phosphorus removal efficiency of the biochemical reaction system and key indicators such as the chemical oxygen demand (COD) of the final effluent, but also operational economics, reagent consumption, and the risk of secondary pollution. This is especially true in processes using renewable organic matter such as biogas slurry as an external carbon source, where large fluctuations in water quality and uneven dosage can easily lead to system overload, effluent fluctuations, or reagent waste.

[0003] In existing technologies, carbon source dosing methods in wastewater treatment plants can be broadly categorized into three types. One type is manual adjustment based on operator experience. Operators manually adjust the dosage based on influent and effluent COD test results, daily process observations, and experience with the production line. This method is slow to respond, highly subjective, and lacks precision, making it unable to adapt to real-time changes in operating conditions. Another type uses simple proportional-integral-derivative (PID) and other traditional automatic control methods, using the deviation between the effluent concentration and the target value as the input signal to drive the opening or frequency of the carbon source dosing pump. While this achieves some degree of automatic adjustment, it is slow to respond to water quality disturbances, easily affected by extreme changes in flow rate and load, and struggles to balance stable dosage with cost-effectiveness. Some systems introduce model-based carbon source dosing optimization strategies. By establishing a kinetic model of the aeration tank's biochemical process and combining it with historical data for parameter identification and predictive control, these methods generally rely on large amounts of historical operating data, have complex model structures, poor parameter availability, and limited adaptability to on-site processes, thus limiting their practical application. Furthermore, existing intelligent dosing systems often suffer from problems such as closed algorithm structures, singular adjustment targets, large dosing fluctuations, and the inability to optimize reagent dosage while ensuring effluent meets standards. For example, most automatic carbon source dosing systems prioritize instantaneous achievement of effluent concentration standards, neglecting the economics of reagent consumption and the stability of the dosing process. This leads to drastic fluctuations in dosage, high reagent consumption, or ineffective handling of high-frequency water quality fluctuations in actual operation. Some model-based prediction systems may also experience engineering problems such as control instability due to parameter drift and model mismatch.

[0004] Therefore, this case aims to propose an intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants. First, chemical oxygen demand (COD) sampling points and flow measurement points are set up in the inlet and outlet pipelines. Combined with the on-site hydraulic retention time and the quantitative conversion coefficient of the spiking test, a discrete prediction model is constructed. Then, the prediction deviation and the stability of the dosage are incorporated into the same objective function. Through Lagrangian optimization with boundary constraints, the closed-loop recommended dosage for each cycle is obtained. Finally, the full-cycle dosage sequence is generated in a rolling iteration manner, and the operation records are fully archived. Summary of the Invention

[0005] This invention provides an intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants, which helps to solve the problems mentioned in the background art.

[0006] This invention provides the following technical solution: an intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants, comprising:

[0007] Set up sampling points for chemical oxygen demand in influent and effluent and measuring points for influent flow rate, and set the sampling cycle, sampling time sequence, total sampling duration and total number of samplings, target chemical oxygen demand in effluent, maximum allowable carbon source dosage and recommended dosing rate per cycle.

[0008] The available volume and minimum reliable flow threshold of the biochemical tank were obtained, the hydraulic retention time of each cycle was calculated, two sets of equal-volume spiked tests were carried out, and the biogas slurry equivalent coefficient and the concentration conversion coefficient per unit volume were converted to establish a discrete prediction model.

[0009] Calculate the deviation of the next cycle, construct the local sensitivity and equivalent acceleration error, and establish a dual objective with unified dimensions;

[0010] By introducing upper and lower bounds and setting corresponding multipliers, a constrained Lagrangian function is formed;

[0011] A quadratic objective for the recommended quantity is established based on the unperturbed natural prediction term and the added impact term. Stationary conditions are set, and the unconstrained solution and the closed-form expression when the constraints are satisfied are given.

[0012] The branch and multiplier solution strategies are given for the following cases: within the interval, below the lower bound, above the upper bound, and low flow threshold.

[0013] The entire cycle of injection sequence is output through a process of periodic rolling data collection, calculation, branch determination, and sequence generation.

[0014] Output the final sequence, calculate the sum of squared deviations and archive the time, recommended dosage, effluent concentration and deviation record.

[0015] Optionally, the step of setting up sampling points for influent and effluent chemical oxygen demand (COD) and influent flow rate, and setting the sampling period, sampling time sequence, total sampling duration and total number of samplings, target effluent COD, maximum allowable carbon source dosage, and recommended periodic dosing rate specifically includes:

[0016] Install a chemical oxygen demand (COD) sampler in the inlet pipeline and record the values ​​measured in each cycle.

[0017] A chemical oxygen demand (COD) sampler was installed in the outlet pipe and the measured values ​​were recorded for each cycle.

[0018] An ultrasonic flow meter was installed on the main inlet pipe to record the instantaneous flow rate in each cycle;

[0019] Set a fixed sampling period and construct a sampling time sequence;

[0020] Set a target chemical oxygen demand (COD) for the effluent;

[0021] Obtain the maximum allowable amount of carbon source to be added;

[0022] The recommended investment acceleration rate for each cycle will be used as a control factor.

[0023] The total number of samples is obtained based on the sampling period and the total duration.

[0024] Optionally, the steps of obtaining the available volume and minimum reliable flow threshold of the biochemical tank, calculating the hydraulic retention time for each cycle, conducting two sets of equal-volume spiking experiments and converting the biogas slurry equivalent coefficient to the concentration conversion coefficient per unit dosage volume, and establishing a discrete prediction model specifically include:

[0025] Obtain the usable volume of the biological treatment tank;

[0026] Calculate the hydraulic retention time based on the inflow rate and the minimum reliable flow threshold for each cycle;

[0027] Two sets of equal-volume spiked tests were conducted, and the spiked mass and steady-state concentration increment of each set were recorded. The chemical oxygen demand corresponding to a unit mass of biogas slurry was calculated and converted into a concentration conversion factor per unit spiked volume.

[0028] The predicted effluent concentration for the next cycle is calculated based on two parts: the transport dilution factor and the dosing effect factor.

[0029] Optionally, the calculation of the next cycle deviation, the construction of local sensitivity and equivalent acceleration error, and the establishment of a dual objective with unified dimensions specifically include:

[0030] Calculate the water discharge deviation for the next cycle;

[0031] Local sensitivity is constructed by adjusting the flow rate based on the sampling period length, concentration conversion coefficient, and threshold.

[0032] The water discharge deviation is converted into an equivalent injection acceleration rate error;

[0033] The dual objectives are composed of a water quality satisfaction term and an addition stability term, forming a unified dimension.

[0034] Optionally, the step of introducing upper and lower bounds and setting corresponding multipliers to form a constrained Lagrangian function specifically includes:

[0035] Set the lower bound of the injection amount for each cycle to zero and the upper bound to the maximum injection amount;

[0036] Multipliers are introduced for the lower and upper bounds respectively, and a Lagrange function with upper and lower bound constraints is established.

[0037] Optionally, the step of establishing a quadratic objective with respect to the recommended quantity based on the unperturbed natural prediction term and the added influence term, setting stationary conditions, and providing the unconstrained solution and the closed-form expression when the constraints are satisfied specifically includes:

[0038] The predicted effluent concentration for the next cycle is expressed as a linear form of the difference between the undisturbed natural prediction term and the dosing effect term;

[0039] The equivalent dosing error is given, and a secondary target is established by comparing it with the target effluent concentration.

[0040] By setting stationary conditions for the decision variables and performing partial derivative calculations, unconstrained stationary solutions can be obtained.

[0041] When the upper and lower bounds are satisfied, the multiplier is set to zero and the unconstrained solution is used as the closed-form recommended value.

[0042] Optionally, the branch and multiplier solution strategies for the following cases are provided: within the interval, below the lower bound, above the upper bound, and at low flow thresholds. Specifically, these include:

[0043] Take this value and set the two multipliers to zero when the unconstrained solution is between the upper and lower bounds;

[0044] When the unconstrained solution is lower than the lower bound, the lower bound is taken and the lower bound multiplier is obtained by the stationary point equation;

[0045] When the unconstrained solution is higher than the upper bound, the upper bound is taken and the upper bound multiplier is obtained by the stationary point equation;

[0046] When the periodic flow rate is below the minimum confidence threshold, the sensitivity calculation is performed using the threshold instead, and the above branch continues.

[0047] Optionally, the process of periodically rolling data acquisition, calculation, branch determination, and sequence generation outputs a full-cycle addition sequence, specifically including:

[0048] Initialize the initial injection amount to zero.

[0049] In each sampling period, the influent concentration, effluent concentration and influent flow rate are collected sequentially, and the threshold correction flow rate, hydraulic residence time, concentration conversion coefficient, undisturbed natural prediction term and local sensitivity are calculated.

[0050] Calculate the unconstrained solution and obtain the recommended value according to the branching rule;

[0051] Generate the final dosage sequence in cyclical order.

[0052] Optionally, the output final sequence, the time for calculating and archiving the sum of squared deviations, the recommended dosage, and the records of effluent concentration and deviation indicators specifically include:

[0053] Output the final dosage sequence;

[0054] Calculate and record the sum of squares of the outflow deviation index periodically;

[0055] Save the complete record file, including the time, recommended dosage, effluent concentration, and deviation index fields.

[0056] The present invention has the following beneficial effects:

[0057] 1. High-frequency automated online monitoring equipment is deployed on key pipelines for influent, effluent, and flow rates. This system systematically collects and archives data such as COD and flow rate for each cycle. By scientifically setting key parameters such as sampling period, target effluent value, and maximum dosage, high-resolution, end-to-end data support is provided for subsequent model calculations and decision-making. Compared to traditional methods relying on manual sampling and dealing with delayed or incomplete data, this method improves the real-time nature, comprehensiveness, and objectivity of the data. It facilitates the timely detection of water quality fluctuations and operational anomalies, preventing model decision failures due to incomplete data and laying a solid foundation for subsequent precise control.

[0058] 2. By obtaining the effective volume of the biochemical tank and the hydraulic retention time during the sampling cycle, and combining two sets of spiked test data from the standard digestion experiment, the equivalent coefficient of the biogas slurry carbon source and the concentration conversion coefficient corresponding to the unit addition volume are scientifically derived, thereby establishing an accurate discrete process prediction model. Compared with existing traditional control methods that rely on empirical estimation or coarse linear relationships, this invention fully utilizes experimental data for model parameter calibration, realizing the engineering quantification and process traceability of the impact of biogas slurry carbon source on effluent water quality. This improves the model's prediction accuracy, facilitates adaptive correction of the addition amount under dynamic operating conditions, and reduces resource waste and the risk of exceeding water quality standards caused by manual settings or fuzzy decisions.

[0059] 3. This invention not only addresses the single objective of achieving COD compliance in effluent but also simultaneously incorporates the stability of the dosing operation into a dual-objective framework. It employs local sensitivity indicators to unify the dimensions of each objective, thereby establishing a quantifiable cost indicator for each dosing cycle. Compared to traditional simplistic models that only focus on compliance rates and ignore the operational risks caused by dosing fluctuations, this invention achieves multi-objective synergy and dimensional unification. It can effectively control drastic changes in dosing dosage while ensuring effluent compliance, reducing fluctuations and shocks in the biochemical system load. This strategy embodies a comprehensive balance between process control and operational stability and energy economy, contributing to the achievement of the triple objectives of "compliance, energy saving, and stability" in practical engineering applications.

[0060] 4. By setting upper and lower bounds for the recommended dosage and introducing constraints into the Lagrangian function, an engineering optimization solution framework with KKT multipliers is constructed, achieving standardized handling of extreme conditions and actual operational boundaries. Compared with existing methods that simply truncate the dosage based on experience or rules, this approach seeks a mathematically consistent closed-form solution between the theoretical optimum and the actual process boundary, preventing both over- and under-dosing and smooth switching of process states, ensuring that the model output always meets on-site safety and operational constraints.

[0061] 5. Complex process models and objective functions are mathematically derived and reduced to directly calculable linear and closed-form expressions, enabling efficient engineering solutions for recommended periodic dosages. Unlike traditional methods that require multiple rounds of manual trial calculations or black-box parameter tuning, this method is highly practical and transparent, facilitating rapid on-site integration and automated operation. The derivation process organically combines target water quality, historical dosage data, and current process parameters, helping to systematically improve the accuracy of prediction and decision-making, reducing the risks associated with parameter uncertainty, and ensuring the long-term stability and maintainability of the model.

[0062] 6. By employing explicit branching and multiplier calculations for different scenarios, such as the recommended dosage within the constraint interval, lower bound, upper bound, and flow rate below the confidence threshold, this invention ensures that the dosage recommendation under each operating condition has both mathematical basis and physical rationality. Compared to existing technologies that handle extreme conditions in a simplistic and crude manner, resulting in model "freezing" or unreasonable outputs, this invention innovatively achieves automatic adaptation across multiple scenarios, improving the model's robustness and engineering fault tolerance. This branching solution system can cover various abnormal and extreme situations in actual wastewater treatment operations, providing strong safety assurance for long-term automatic operation.

[0063] 7. Through rolling data collection, parameter calculation, branch determination, and sequence generation, a full-cycle dosage sequence is formed, ensuring that decisions for each cycle are based on the latest monitoring data and operating parameters. Compared with traditional "fixed-point dosing" or "manual batch setting" methods, this rolling calculation mechanism has stronger adaptability and dynamic response capabilities. Faced with the variability of actual wastewater quality and flow, it can effectively avoid lag, omissions, or error accumulation, achieving real-time, closed-loop, and fully automatic process control, reducing the frequency of manual intervention and operational difficulty.

[0064] 8. Ultimately, this invention achieves full archiving of dosage, effluent concentration, and deviation indicators for each sampling cycle, and outputs performance indicators such as the total sum of squared deviations, providing a detailed data foundation for process optimization, operation and maintenance assessment, and historical review. Unlike traditional systems that only output single results and lack historical analysis capabilities, this solution strengthens data traceability, process evaluation, and continuous optimization capabilities. It can provide full-process raw data for subsequent machine learning, intelligent operation and maintenance, and management decisions, comprehensively improving the informatization and lean management level of wastewater treatment plants. Attached Figure Description

[0065] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0066] 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.

[0067] Example, refer to Figure 1 A smart recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants, comprising:

[0068] Set up sampling points for chemical oxygen demand in influent and effluent and measuring points for influent flow rate, and set the sampling cycle, sampling time sequence, total sampling duration and total number of samplings, target chemical oxygen demand in effluent, maximum allowable carbon source dosage and recommended dosing rate per cycle.

[0069] The available volume and minimum reliable flow threshold of the biochemical tank were obtained, the hydraulic retention time of each cycle was calculated, two sets of equal-volume spiked tests were carried out, and the biogas slurry equivalent coefficient and the concentration conversion coefficient per unit volume were converted to establish a discrete prediction model.

[0070] Calculate the deviation of the next cycle, construct the local sensitivity and equivalent acceleration error, and establish a dual objective with unified dimensions;

[0071] By introducing upper and lower bounds and setting corresponding multipliers, a constrained Lagrangian function is formed;

[0072] A quadratic objective for the recommended quantity is established based on the unperturbed natural prediction term and the added impact term. Stationary conditions are set, and the unconstrained solution and the closed-form expression when the constraints are satisfied are given.

[0073] The branch and multiplier solution strategies are given for the following cases: within the interval, below the lower bound, above the upper bound, and low flow threshold.

[0074] The entire cycle of injection sequence is output through a process of periodic rolling data collection, calculation, branch determination, and sequence generation.

[0075] Output the final sequence, calculate the sum of squared deviations and archive the time, recommended dosage, effluent concentration and deviation record.

[0076] First, online monitoring points for influent and effluent chemical oxygen demand (COD) and flow rate were deployed, and basic parameters such as sampling time and cycle were standardized to achieve real-time perception of the entire wastewater treatment process's operational status. Next, the volume of the biological treatment tank was acquired and experimental calibration was conducted. The experimental data was then transformed into a process prediction model, breaking the limitations of relying on empirical coefficients or single linear relationships and improving model accuracy. Third, a cost function with dual objectives of achieving water quality standards and stable dosing was constructed. This method, using a unified dimension, reconciled the contradiction between water quality stability and dosing fluctuations, filling the gap in existing technologies that only focused on compliance rates while neglecting stability. Subsequently, upper and lower bounds for dosing and corresponding multipliers were introduced. Combined with optimization theory, a closed-loop dosing solution was given analytically, avoiding the uncertainties brought about by traditional trial-and-error or black-box optimization. Furthermore, branch solution strategies were designed for different extreme cases to ensure reasonable suggestions are still provided when dosing is too low, too high, or the flow rate is abnormal, improving system robustness. Finally, a full-cycle sequence was formed through rolling calculations, and deviation indicators were archived to help engineers compare historical data and evaluate performance. Overall, this solution organically integrates four major modules: monitoring, modeling, optimization, and archiving, providing an efficient, intelligent, and robust overall solution for carbon source addition in wastewater treatment plants. It offers higher real-time performance, accuracy, and operability compared to traditional manual adjustment or simple closed-loop control.

[0077] The establishment of sampling points for influent and effluent chemical oxygen demand (COD) and influent flow rate measurement points, along with the setting of sampling period, sampling time sequence, total sampling duration and total number of samplings, target effluent COD, maximum allowable carbon source dosage, and recommended periodic dosing rate, specifically includes:

[0078] Install a chemical oxygen demand (COD) sampler in the inlet pipeline and record the values ​​measured in each cycle.

[0079] A chemical oxygen demand (COD) sampler was installed in the outlet pipe and the measured values ​​were recorded for each cycle.

[0080] An ultrasonic flow meter was installed on the main inlet pipe to record the instantaneous flow rate in each cycle;

[0081] Set a fixed sampling period and construct a sampling time sequence;

[0082] Set a target chemical oxygen demand (COD) for the effluent;

[0083] Obtain the maximum allowable amount of carbon source to be added;

[0084] The recommended investment acceleration rate for each cycle will be used as a control factor.

[0085] The total number of samples is obtained based on the sampling period and the total duration.

[0086] Further specific implementation steps include:

[0087] Install water inlet on the inlet pipe of the sewage treatment plant The concentration sampler collects data and records it as follows: ;in, For the first Periodically measured influent Concentration, in units of ; For discrete sampling sequence numbers;

[0088] Install water outlet on the effluent pipe of the sewage treatment plant The concentration sampler collects data and records it as follows: ;in, For the first Periodic measurement of water discharge Concentration, in units of ;

[0089] An ultrasonic flow meter was installed on the main inlet pipe to record the flow rate. The instantaneous flow rate of the first cycle will be the first The influent flow rate measured in each cycle is recorded as follows: The unit is ;

[0090] Set a fixed sampling period as The sampling time sequence is set as follows: , ;in, For the first The time corresponding to the next sample; This represents the total number of sampling periods;

[0091] Set target water output Concentration, denoted as The unit is ;

[0092] The maximum allowable amount of carbon source to be added is denoted as: The unit is ;

[0093] The first The recommended carbon source input rate for each cycle is denoted as . The unit is ;

[0094] Set the total sampling duration to The unit is and order .

[0095] By installing chemical oxygen demand (COD) samplers and ultrasonic flow meters in the inlet, outlet, and main pipelines respectively, and synchronously collecting concentration and flow data at fixed intervals, the consistency of data volume and sampling sequence is ensured, eliminating data drift caused by asynchronous sampling from multiple devices. After clearly defining parameters such as sampling sequence number, time sequence, and total number of cycles, real-time monitoring values ​​can be directly mapped to discrete cycles, eliminating the workload of manual annotation and data cleaning. Furthermore, by pre-setting the target effluent concentration and the maximum allowable dosing rate, the system can acquire core control targets and operating boundaries in the initial stage, avoiding blind spots or risks of exceeding limits during later dosing processes. Compared to traditional methods that only monitor a few key points or rely on human experience to guess monitoring frequency, this method provides a standardized and quantifiable monitoring framework, improving the accuracy of subsequent model building and dosing decisions, providing a reliable data foundation for subsequent automated control, and reducing human error and maintenance costs.

[0096] The process involves obtaining the available volume and minimum reliable flow threshold of the biochemical tank, calculating the hydraulic retention time for each cycle, conducting two sets of equal-volume spiking experiments, converting the biogas slurry equivalent coefficient to the concentration conversion coefficient per unit dosage volume, and establishing a discrete prediction model. Specifically, this includes:

[0097] Obtain the usable volume of the biological treatment tank;

[0098] Calculate the hydraulic retention time based on the inflow rate and the minimum reliable flow threshold for each cycle;

[0099] Two sets of equal-volume spiked tests were conducted, and the spiked mass and steady-state concentration increment of each set were recorded. The chemical oxygen demand corresponding to a unit mass of biogas slurry was calculated and converted into a concentration conversion factor per unit spiked volume.

[0100] The predicted effluent concentration for the next cycle is calculated based on two parts: the transport dilution factor and the dosing effect factor.

[0101] Further specific implementation steps include:

[0102] Obtain the usable volume of the biological treatment tank, denoted as . The unit is ;

[0103] The periodic hydraulic residence time is set as follows: , ;in, For the first The hydraulic residence time of the cycle, in units of ; For the first The period is used to calculate the threshold correction flow rate, in units of ; Minimum trusted traffic threshold, in units of ;

[0104] Constructing biogas slurry Equivalent coefficient Specifically:

[0105] ;

[0106] The two sets of spiked test data were obtained by standard digestion method. and In equal volume The calculation formula is given below:

[0107] ;in, For the first The mass of the group experiment, in units of ; Index for experimental groups; For the first steady state measured by the group of experiments Concentration increment, in units ; The volume of the experimental system is expressed in units of 1000 liters. ;

[0108] The standard digestion method described is existing technology.

[0109] The conversion factor from unit dosage volume to concentration is obtained from the volume conversion. The unit is Used to Equivalent to each Caused by addition Increment;

[0110] Construct discrete prediction equations: ;in, For the first Predicted water discharge in cycles The unit is ; For the transfer dilution term, the unit is... To add the impact item, the unit is .

[0111] First, the volume of the biochemical tank was obtained, and the periodic residence time was calculated based on the corrected flow rate, avoiding calculation deviations caused by abnormal flow rates. Combining two equal-volume spiking experiments, the actual usable equivalent coefficient of the biogas slurry carbon source was extracted and converted into a unit volume influence conversion coefficient, quantifying the chemical reactions in the experimental system for engineering operation and overcoming the limitations of rough empirical coefficients. Next, the transport and dilution terms were separated from the dosing influence terms, forming a process model suitable for discrete prediction. This model considers both the water body's own dilution and the carbon source compensation effect in the prediction results. Compared to traditional methods that only use fixed conversion coefficients or simply linearly fit the influent and effluent concentrations, this method seamlessly integrates experimental measurements with the on-site process, achieving closed-loop verification of the model from the experimental platform to the on-site engineering, improving prediction accuracy and model stability, and providing a reliable predictive basis for subsequent dynamic optimization of dosing.

[0112] The calculation of the next cycle deviation, the construction of local sensitivity and equivalent acceleration error, and the establishment of a dual objective with unified dimensions specifically include:

[0113] Calculate the water discharge deviation for the next cycle;

[0114] Local sensitivity is constructed by adjusting the flow rate based on the sampling period length, concentration conversion coefficient, and threshold.

[0115] The water discharge deviation is converted into an equivalent injection acceleration rate error;

[0116] The dual objectives are composed of a water quality satisfaction term and an addition stability term, forming a unified dimension.

[0117] Further specific implementation steps include:

[0118] Calculate the deviation for the next cycle: ;in, For the first Periodic water output deviation, in units of ;

[0119] Constructing local sensitivity: ;in, For the first Local sensitivity of the period, in units of ;

[0120] Convert the deviation into an equivalent acceleration error: ;in, For the first Equivalent acceleration error of the period, unit ;

[0121] Constructing a dual objective with unified dimensions:

[0122] ;in, Objective function, the first The cost of the cycle, in units of ; This refers to the dosage amount added in the previous cycle, in units of... .

[0123] By calculating the effluent deviation for the next cycle, constructing local sensitivity, and converting the equivalent dosing rate error, and merging these two into a unified-dimensional bi-objective cost function, this method resolves the contradiction that traditional single-objective optimization easily leads to large water quality fluctuations or unstable dosing. The specific steps include first calculating the deviation between the target and the predicted effluent, then combining the periodic sampling frequency, conversion coefficient, and corrected flow rate to derive sensitivity, and finally quantifying the deviation as an equivalent dosing rate error, allowing direct comparison of water quality deviation and dosing rate under the same dimension. One objective measures water quality satisfaction, while the other measures dosing stability. This approach comprehensively considers both compliance requirements and system stability requirements during each dosing cycle. Compared to existing methods that often only focus on effluent compliance while ignoring dosing fluctuations, this method is significantly innovative in multi-objective collaborative optimization, enabling the system to respond quickly to water quality anomalies while avoiding secondary shocks caused by large-scale dosing, thus improving the safety and energy efficiency of the biological treatment tank operation.

[0124] The process of introducing upper and lower bounds and setting corresponding multipliers to form a constrained Lagrangian function specifically includes:

[0125] Set the lower bound of the injection amount for each cycle to zero and the upper bound to the maximum injection amount;

[0126] Multipliers are introduced for the lower and upper bounds respectively, and a Lagrange function with upper and lower bound constraints is established.

[0127] Further specific implementation steps include:

[0128] Set the constraints as follows: ;

[0129] Construct the Lagrangian function: ;in, For the first The periodic value of the Lagrange function; for Chengzi, For the first The multiplier corresponding to the lower periodicity constraint For the first The multipliers corresponding to the periodic upper bound constraint are all in units. .

[0130] First, the legal range from zero to the maximum dosage is determined. Then, multipliers are configured for the upper and lower limits respectively, which are used to automatically adjust the numerical compensation when the boundary is reached during the optimization process. This method integrates the dosage constraint with the cost function, ensuring that the mathematical optimization results do not give illegal or unreasonable dosage suggestions. Unlike the common ex-post truncation processing or simple empirical constraints in existing technologies, this optimization form that combines constraints enables the system to adaptively adapt to boundary limits in actual operation and continue to maintain the optimal solution through multiplier feedback. This improves the safety and feasibility of the dosage suggestions and provides a more reliable automated control strategy for engineering sites.

[0131] The process of establishing a quadratic objective for the recommended quantity based on the unperturbed natural prediction term and the added influence term, setting stationary conditions, and providing an unconstrained solution and a closed-form expression satisfying the constraints specifically includes:

[0132] The predicted effluent concentration for the next cycle is expressed as a linear form of the difference between the undisturbed natural prediction term and the dosing effect term;

[0133] The equivalent dosing error is given, and a secondary target is established by comparing it with the target effluent concentration.

[0134] By setting stationary conditions for the decision variables and performing partial derivative calculations, unconstrained stationary solutions can be obtained.

[0135] When the upper and lower bounds are satisfied, the multiplier is set to zero and the unconstrained solution is used as the closed-form recommended value.

[0136] Further specific implementation steps include:

[0137] Will Written about Linear form:

[0138] ;in, For the first The period is in the absence of added disturbance (i.e. Naturally predicted water discharge under the given conditions, in units of ;

[0139] Equivalent dosing error: ;in, The equivalent acceleration rate required to achieve the target, in units of ;

[0140] Objective function: ;

[0141] The conditions for setting up a base are as follows:

[0142] ;in, For decision variables The partial derivative;

[0143] The general solution at the stationary point is: ;in, For the first Unconstrained stationary solutions of the periodicity, in units of ;

[0144] Under the condition that the constraints are satisfied ,at this time:

[0145] This is a closed-form analytic expression for the recommended quantity.

[0146] By jointly expressing the predicted water output and the effects of water addition in a linear form, and setting stationary conditions to derive closed-form solutions for both unconstrained and constrained conditions, this method solves the problems of long optimization time or unstable solutions in large-scale online calculations. First, the predicted term and the effects of water addition are decomposed into a linear function of the current cycle's addition amount. Then, using the equivalent addition rate error and the addition amount of the previous cycle as inputs, a quadratic objective is constructed and stationary conditions are set. This directly yields the unconstrained adjustment expression, and when the boundary conditions are met, the final recommended value can be output without additional iterations. This makes the calculation of the addition amount per cycle simple and efficient, completing at the microsecond level, improving the system's online real-time performance and computational efficiency. Compared to traditional methods that require numerical iteration or third-party optimization tools, this method simplifies the calculation process and ensures the interpretability and stability of the results, providing applicability for field controllers with limited equipment resources.

[0147] The given branch and multiplier solution strategies for the given interval, below the lower bound, above the upper bound, and low flow threshold cases specifically include:

[0148] Take this value and set the two multipliers to zero when the unconstrained solution is between the upper and lower bounds;

[0149] When the unconstrained solution is lower than the lower bound, the lower bound is taken and the lower bound multiplier is obtained by the stationary point equation;

[0150] When the unconstrained solution is higher than the upper bound, the upper bound is taken and the upper bound multiplier is obtained by the stationary point equation;

[0151] When the periodic flow rate is below the minimum confidence threshold, the sensitivity calculation is performed using the threshold instead, and the above branch continues.

[0152] Further specific implementation steps include:

[0153] like Then take ;

[0154] like ,make And substitute the stationary point equations into Find:

[0155] ;in, It is a non-negative truncation operator function. As the independent variable;

[0156] like ,make And substitute the stationary point equations into , Seek ;

[0157] when At that time, adopt ;in, The sensitivity is obtained by replacing the threshold. This is the lower limit threshold for sensitivity.

[0158] This method employs a clear branching and multiplier-solving process to address different operating scenarios—recommended flow rates within the legal range, below the lower limit, above the upper limit, and below the threshold. This resolves the issues of dosing failure or unreasonable oscillations caused by improper handling of extreme cases in existing technologies. For normal operating conditions within the range, the recommended value is directly output. For cases exceeding the upper and lower limits, the multiplier is calculated sequentially by substituting back from the stationary point equation, ensuring a legal and smooth adjustment amount is still provided when constraints are triggered. For cases where the influent flow rate is below the reliable threshold, the sensitivity is readjusted to avoid amplifying dosing errors due to calculation anomalies caused by low flow rates. This branching strategy can cover various sudden and extreme scenarios in field operation, achieving robust adaptation to extremely low flow rates or abnormal operating conditions. This contrasts sharply with traditional approaches that simply truncate or ignore extreme conditions, reducing the probability of dosing errors and improving the system's fault tolerance and reliability.

[0159] The process of periodic rolling acquisition, calculation, branch determination, and sequence generation outputs the full-cycle addition sequence, specifically including:

[0160] Initialize the initial injection amount to zero.

[0161] In each sampling period, the influent concentration, effluent concentration and influent flow rate are collected sequentially, and the threshold correction flow rate, hydraulic residence time, concentration conversion coefficient, undisturbed natural prediction term and local sensitivity are calculated.

[0162] Calculate the unconstrained solution and obtain the recommended value according to the branching rule;

[0163] Generate the final dosage sequence in cyclical order.

[0164] Further specific implementation steps include:

[0165] initialization ;

[0166] In each sampling period ,collection ;calculate , ;calculate Determine the branch and give ;

[0167] Final dosage sequence of the process .

[0168] Through a rolling iterative process, influent and effluent data are collected sequentially in each cycle, and flow corrections, hydraulic retention times, conversion coefficients, sensitivity, and unconstrained solutions are updated. A recommended dosage is then provided based on a comprehensive branch decision, creating a closed-loop automated sequence for the entire dosing process. This solves the drawbacks of data silos, static model operation, and manual batch command issuance in wastewater treatment processes. This method ensures that dosing decisions are always based on the latest operating conditions, enabling rapid response to fluctuations in water quality or flow, and avoiding the lag and cumulative bias of feedforward control. Simultaneously, the rolling approach simplifies the overall system calculation logic, balancing real-time performance and continuity, and reducing human intervention. Unlike traditional one-time settings or semi-automatic adjustments, this method continuously corrects model biases through rolling iterations, improving long-term operational stability and accuracy, and providing a solid guarantee for achieving truly fully automated, refined carbon source control in wastewater treatment plants.

[0169] The output final sequence, the time for calculating and archiving the sum of squared deviations, the recommended dosage, and the records of effluent concentration and deviation indicators specifically include:

[0170] Output the final dosage sequence;

[0171] Calculate and record the sum of squares of the outflow deviation index periodically;

[0172] Save the complete record file, including the time, recommended dosage, effluent concentration, and deviation index fields.

[0173] Further specific implementation steps include:

[0174] Output the final dosing sequence ;

[0175] Calculate the sum of squares of the outflow deviation index: ;in, This is the sum of squares of the total deviations, in units of ;

[0176] Save the complete record file, including the following fields: Time Recommended dosage effluent concentration Deviation indicators .

[0177] By outputting the full-cycle dosing sequence, calculating the total deviation index, and generating a complete archive file, this method solves the problems of missing historical data and difficulty in evaluating water quality and dosing effectiveness in wastewater treatment plant operations. Specific steps include outputting the dosing amount and corresponding effluent concentration periodically, calculating the sum of squared deviations to measure the overall effectiveness of the dosing strategy, and recording information such as time, recommended dosage, water quality, and deviations in the project archive. This process not only provides visualized indicators for daily operations but also provides reliable data support for subsequent process optimization, equipment maintenance, and management decisions. Compared to existing technologies that only provide single-time outputs or lack systematic archiving, this method improves operational transparency and traceability, creates conditions for performance evaluation and continuous improvement, and promotes the informatization and refinement of wastewater treatment plant management.

[0178] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0179] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligently recommending the dosage of biogas slurry carbon source in wastewater treatment plants, characterized in that, include: Set up sampling points for chemical oxygen demand in influent and effluent and measuring points for influent flow rate, and set the sampling cycle, sampling time sequence, total sampling duration and total number of samplings, target chemical oxygen demand in effluent, maximum allowable carbon source dosage and recommended dosing rate per cycle. The available volume and minimum reliable flow threshold of the biochemical tank were obtained, the hydraulic retention time of each cycle was calculated, two sets of equal-volume spiking tests were carried out, and the biogas slurry equivalent coefficient and the concentration conversion coefficient per unit volume were converted to establish a discrete prediction model. Calculate the deviation of the next cycle, construct the local sensitivity and equivalent acceleration error, and establish a dual objective with unified dimensions; By introducing upper and lower bounds and setting corresponding multipliers, a constrained Lagrangian function is formed; A quadratic objective for the recommended quantity is established based on the unperturbed natural prediction term and the added impact term. Stationary conditions are set, and the unconstrained solution and the closed-form expression when the constraints are satisfied are given. The branch and multiplier solution strategies are given for the following cases: within the interval, below the lower bound, above the upper bound, and low flow threshold. The entire cycle of injection sequence is output through a process of periodic rolling data collection, calculation, branch determination, and sequence generation. Output the final sequence, calculate the sum of squared deviations and archive the time, recommended dosage, effluent concentration and deviation record.

2. The intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants according to claim 1, characterized in that, The establishment of sampling points for influent and effluent chemical oxygen demand (COD) and influent flow rate measurement points, along with the setting of sampling period, sampling time sequence, total sampling duration and total number of samplings, target effluent COD, maximum allowable carbon source dosage, and recommended periodic dosing rate, specifically includes: Install a chemical oxygen demand (COD) sampler in the inlet pipeline and record the values ​​measured in each cycle. A chemical oxygen demand (COD) sampler was installed in the outlet pipe and the measured values ​​were recorded for each cycle. An ultrasonic flow meter was installed on the main inlet pipe to record the instantaneous flow rate in each cycle; Set a fixed sampling period and construct a sampling time sequence; Set a target chemical oxygen demand (COD) for the effluent; Obtain the maximum allowable amount of carbon source to be added; The recommended investment acceleration rate for each cycle will be used as a control factor. The total number of samples is obtained based on the sampling period and the total duration.

3. The intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants according to claim 2, characterized in that, The process involves obtaining the available volume and minimum reliable flow threshold of the biochemical tank, calculating the hydraulic retention time for each cycle, conducting two sets of equal-volume spiking experiments, converting the biogas slurry equivalent coefficient to the concentration conversion coefficient per unit dosage volume, and establishing a discrete prediction model. Specifically, this includes: Obtain the usable volume of the biological treatment tank; Calculate the hydraulic retention time based on the inflow rate and the minimum reliable flow threshold for each cycle; Two sets of equal-volume spiked tests were conducted, and the spiked mass and steady-state concentration increment of each set were recorded. The chemical oxygen demand corresponding to a unit mass of biogas slurry was calculated and converted into a concentration conversion factor per unit spiked volume. The predicted effluent concentration for the next cycle is calculated based on two parts: the transport dilution factor and the dosing effect factor.

4. The intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants according to claim 3, characterized in that, The calculation of the next cycle deviation, the construction of local sensitivity and equivalent acceleration error, and the establishment of a dual objective with unified dimensions specifically include: Calculate the water discharge deviation for the next cycle; Local sensitivity is constructed by adjusting the flow rate based on the sampling period length, concentration conversion coefficient, and threshold. The water discharge deviation is converted into an equivalent injection acceleration rate error; The dual objectives are composed of a water quality satisfaction term and an addition stability term, forming a unified dimension.

5. The intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants according to claim 4, characterized in that, The process of introducing upper and lower bounds and setting corresponding multipliers to form a constrained Lagrangian function specifically includes: Set the lower bound of the injection amount for each cycle to zero and the upper bound to the maximum injection amount; Multipliers are introduced for the lower and upper bounds respectively, and a Lagrange function with upper and lower bound constraints is established.

6. The intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants according to claim 5, characterized in that, The process of establishing a quadratic objective for the recommended quantity based on the unperturbed natural prediction term and the added influence term, setting stationary conditions, and providing an unconstrained solution and a closed-form expression satisfying the constraints specifically includes: The predicted effluent concentration for the next cycle is expressed as a linear form of the difference between the undisturbed natural prediction term and the dosing effect term; The equivalent dosing error is given, and a secondary target is established by comparing it with the target effluent concentration. By setting stationary conditions for the decision variables and performing partial derivative calculations, unconstrained stationary solutions can be obtained. When the upper and lower bounds are satisfied, the multiplier is set to zero and the unconstrained solution is used as the closed-form recommended value.

7. The intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants according to claim 6, characterized in that, The given branch and multiplier solution strategies for the given interval, below the lower bound, above the upper bound, and low flow threshold cases specifically include: When the unconstrained solution is between the upper and lower bounds, take this value and set the two multipliers to zero; When the unconstrained solution is lower than the lower bound, the lower bound is taken and the lower bound multiplier is obtained by the stationary point equation; When the unconstrained solution is higher than the upper bound, the upper bound is taken and the upper bound multiplier is obtained by the stationary point equation; When the periodic flow rate is below the minimum confidence threshold, the sensitivity calculation is performed using the threshold instead, and the above branch continues.

8. The intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants according to claim 7, characterized in that, The process of periodic rolling acquisition, calculation, branch determination, and sequence generation outputs the full-cycle addition sequence, specifically including: Initialize the initial injection amount to zero. In each sampling period, the influent concentration, effluent concentration and influent flow rate are collected sequentially, and the threshold correction flow rate, hydraulic residence time, concentration conversion coefficient, undisturbed natural prediction term and local sensitivity are calculated. Calculate the unconstrained solution and obtain the recommended value according to the branching rule; Generate the final dosage sequence in cyclical order.

9. The intelligent recommendation method for the dosage of biogas slurry carbon source in wastewater treatment plants according to claim 8, characterized in that, The output final sequence, the time for calculating and archiving the sum of squared deviations, the recommended dosage, and the records of effluent concentration and deviation indicators specifically include: Output the final dosage sequence; Calculate and record the sum of squares of the outflow deviation index periodically; Save the complete record file, including the time, recommended dosage, effluent concentration, and deviation index fields.

Images