A method for generating aeration and dosing strategies for sewage treatment plants and a control system

Abstract

The present invention belongs to the technical field of sewage treatment, and discloses a method and control system for generating aeration and dosing strategies of a sewage treatment plant, including collecting operating parameters of a sewage treatment plant, and constructing a first optimization model for aeration and dosing of the sewage treatment plant according to the operating parameters; constructing an ASM‑OLWPSO solution framework, and solving the first optimization model according to the ASM‑OLWPSO solution framework to obtain the predicted dosage of aeration and dosing of the sewage treatment plant; and generating the aeration and dosing strategy of the sewage treatment plant according to the predicted dosage. The present invention adopts an ASM‑OLWPSO solution framework that combines an ASM2d mechanism model, a WPSO algorithm, and an LSTM prediction model. In order to realize the precise optimization control of aeration and dosing of a sewage treatment plant, the ASM2d mechanism model is specifically used as a driver to make up for the deficiencies caused by the lack of measurement data; the WPSO algorithm is used as an optimization process to solve the problem that aeration and dosing cannot be accurately optimized; and the LSTM prediction model is used to perform prediction calculations, reduce unnecessary real evaluations, and improve the optimization calculation efficiency.

Description

Technical Field

[0001] The invention discloses a method for generating aeration and dosing strategies for a sewage treatment plant and a control system, belonging to the technical field of sewage treatment. Background Art

[0002] With the rapid advancement of urbanization, urban sewage treatment systems have become the cornerstone of ensuring environmental quality and public health. This process includes multiple complex processes such as pretreatment, biological treatment, deep purification and sludge treatment. Each link is closely connected and the operating conditions are flexible and changeable, which poses unprecedented challenges to automatic control. In particular, the accuracy of aeration control and the addition of chemicals (such as carbon sources, phosphorus removal agents, flocculants, etc.) is directly related to process efficiency and the stability of effluent water quality, and has become a core problem in the industry.

[0003] Although a huge amount of data has been accumulated during the sewage treatment process, the quality of this data is not satisfactory, with frequent missing and outlier values, which greatly weakens the reliability of the data. Traditional control methods mostly rely on static empirical formulas or preset parameters, lacking a deep understanding of the dynamic characteristics of the system and the ability to flexibly respond. In the real environment where water quality and water volume frequently fluctuate, these methods seem to be unable to cope with the situation and it is difficult to maintain the efficient and stable operation of the system.

[0004] What is more serious is that, because sewage treatment systems are usually relatively closed, their information level is low, data flow is not smooth, and real-time updates are delayed. This not only hinders the management's comprehensive grasp of the system's operating status, but also limits the application of optimization control strategies based on real-time data. The lack and lag of data make it difficult to make immediate adjustments to the control strategy according to actual conditions, which in turn affects the stability and compliance rate of the effluent water quality.

[0005] In actual operation, the above problems directly lead to a significant reduction in control accuracy, significant fluctuations in effluent quality, and difficulty in maintaining discharge standards for a long time. This not only poses a potential threat to the environment, but also increases the operating costs and management difficulties of sewage treatment plants. Summary of the invention

[0006] The purpose of the present invention is to provide a method and control system for generating aeration and dosing strategies for sewage treatment plants to solve the technical problem in the prior art that aeration and dosing amounts are difficult to accurately control when data is missing. To achieve the above purpose, the present invention proposes a method and control system for generating aeration and dosing strategies for sewage treatment plants, and the specific scheme is as follows:

[0007] First aspect:

[0008] A method for generating an aeration and dosing strategy for a sewage treatment plant, comprising:

[0009] Step 1: Collecting the operating parameters of the sewage treatment plant, and constructing a first optimization model of aeration and dosing of the sewage treatment plant according to the operating parameters;

[0010] Step 2: construct an ASM-OLWPSO solution framework, and solve the first optimization model according to the ASM-OLWPSO solution framework to obtain the predicted dosage of aeration and dosing of the sewage treatment plant;

[0011] Step 3: generating an aeration and dosing strategy for the sewage treatment plant according to the predicted dosage.

[0012] Preferably, an ASM-OLWPSO solution framework is constructed, specifically including:

[0013] Step 2.1, constructing an ASM2d mechanism model according to the operating parameters, for simulating the parameters of the sewage treatment plant during the sewage treatment process;

[0014] Step 2.2, embedding the ASM2d mechanism model into the WPSO algorithm;

[0015] Step 2.3, construct a training set, and obtain an LSTM prediction model based on the training set;

[0016] Step 2.4, coupling the LSTM prediction model and the WPSO algorithm to obtain the ASM-OLWPSO solution framework.

[0017] Preferably, the WPSO algorithm includes a particle swarm optimization algorithm and a grey wolf optimization algorithm with a mutation mechanism introduced.

[0018] Preferably, solving the first optimization model according to the ASM-OLWPSO solution framework to obtain the predicted dosage of aeration and dosing of the sewage treatment plant specifically includes:

[0019] Determine the fitness value of the candidate parameter set of the WPSO algorithm according to the simulation effect data of the ASM2d mechanism model, and iterate the candidate parameter set of the WPSO algorithm according to the fitness value until the sewage treatment effect corresponding to the candidate parameter set meets the preset conditions, and the ASM2d mechanism model is the solver of the WPSO algorithm;

[0020] The candidate parameter set that meets the preset conditions is recorded as the optimal parameter set, and the predicted dosage of aeration and dosing of the sewage treatment plant is determined according to the optimal parameter set.

[0021] Preferably, solving the first optimization model according to the ASM-OLWPSO solution framework further includes:

[0022] In the iteration of the WPSO algorithm, the candidate parameter set is screened using the LSTM prediction model;

[0023] The screened candidate parameter set is fed back to the WPSO algorithm.

[0024] Preferably, constructing a training set specifically includes:

[0025] The influent water quality data of the sewage treatment plant is used as input data;

[0026] The simulation effect data of the ASM2d mechanism model and the effluent water quality data corresponding to the influent water quality data are used as output data;

[0027] The input data and the output data are stored in a database in a preset structure, and a training set is constructed according to the database.

[0028] Preferably, after step 2, the method further comprises:

[0029] The training set is updated according to the candidate parameter set of each generation of the WPSO algorithm and its corresponding simulation effect data.

[0030] Preferably, the operating parameters specifically include: inlet water quality parameters, outlet water quality parameters and equipment operation parameters.

[0031] Preferably, the objective function of the first optimization model is to minimize the sum of the total energy consumption of the aeration device and the dosing device and the total drug consumption of the dosing device.

[0032] Second aspect:

[0033] A control system for aeration and dosing in a sewage treatment plant, comprising: a collection module, a strategy generation module and a control module;

[0034] The acquisition module is used to collect the operating parameters of the sewage treatment plant and transmit the operating parameters to the strategy generation module;

[0035] The strategy generation module is used to execute the generation method as described in the first aspect according to the operation parameters;

[0036] The control module is used to issue control instructions according to the aeration and dosing strategies obtained by the strategy generation module to regulate the aeration and dosing of the sewage treatment plant.

[0037] Beneficial effects: The present invention uses a machine learning model to couple the optimization algorithm with the mechanism model online, and innovatively proposes a method and system for optimizing the precise aeration and dosing of sewage treatment plants. This method is driven by the ASM2d mechanism model to make up for the shortcomings caused by the lack of measurement data; the WPSO algorithm is used as the optimization process to solve the problem that aeration and dosing cannot be accurately optimized; the LSTM model is used for predictive calculations to reduce unnecessary real evaluations and improve optimization calculation efficiency. The present invention effectively realizes the precise optimization control of aeration and dosing in sewage treatment plants, can guide operators to control sewage processes, and also leaves decision-making time for operators' regulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a graph of effluent ammonia nitrogen concentration in an actual case of a method and system for optimizing control of precise aeration and dosing in a sewage treatment plant according to the present invention;

[0039] Figure 2 It is an optimization convergence curve diagram of a sewage treatment plant precise aeration and dosing optimization control method and system of the present invention;

[0040] Figure 3 This is a flowchart of an optimization solution framework of a method and system for optimizing control of precise aeration and dosing in a sewage treatment plant according to the present invention;

[0041] Figure 4 This is a loss convergence curve diagram of the LSTM model of a sewage treatment plant precise aeration and dosing optimization control method and system of the present invention. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with specific implementation methods. It should be understood that the specific implementation methods described here are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0043] Embodiment 1:

[0044] In one embodiment, a method for generating an aeration and dosing strategy for a sewage treatment plant is disclosed, comprising the following steps:

[0045] Step 1: Collect the operating parameters of the sewage treatment plant, and build a first optimization model of aeration and dosing of the sewage treatment plant according to the operating parameters;

[0046] In response to the common problems of high aeration energy consumption and excessive consumption of carbon source dosing agents when sewage treatment plants adopt the A2 / O process, this embodiment constructs the first optimization model of aeration efficiency and carbon source dosage in a sewage treatment plant. The model has carefully designed multiple objective functions to comprehensively optimize system performance, and incorporates a variety of constraints to ensure the feasibility and effectiveness of actual operations. It should be understood that the present invention is not limited to a specific type of aeration equipment or type of dosing agent. In engineering practice, professional and technical personnel can flexibly choose according to actual conditions, such as coagulants, different types of carbon sources, etc. as the dosing objects. In this specific case, we focus on the optimization strategy of carbon source addition to demonstrate the application potential and actual effect of the model.

[0047] Furthermore, the objective function of the first optimization model is to minimize the sum of the total energy consumption of the aeration device and the dosing device and the total drug consumption of the dosing device. Specifically, the aeration device in this embodiment is a blower, and the dosing device is a dosing pump; the objective function of the first optimization model in this embodiment also includes the penalty item caused by substandard water quality to ensure that the effluent water quality standard is not sacrificed while saving energy and reducing carbon emissions. The setting of this comprehensive objective function reflects the dual pursuit of economic and environmental benefits.

[0048] Furthermore, the operating parameters specifically include: inlet water quality parameters, outlet water quality parameters and equipment operation parameters.

[0049] In the process of model construction in this embodiment, the operating parameters of the sewage treatment plant are carefully considered, including but not limited to the inlet water quality parameters (such as COD, BOD, ammonia nitrogen, etc.), the outlet water quality parameters (must meet national or local emission standards) and the operating parameters of key equipment. In particular, for blowers and dosing pumps, key indicators such as their quantity, operating time, operating efficiency, operating head and flow rate are refined to accurately control energy consumption and chemical consumption.

[0050] To achieve the above goals, this embodiment uses MATLAB language as a development tool, and uses its powerful mathematical operation and optimization algorithm library to write a special program for aeration and dosing optimization of sewage treatment plants. The specific construction steps are as follows:

[0051] Step 1.1: Construct an objective function with energy conservation and carbon reduction as the goal, as follows:

[0052] In order to optimize aeration and carbon source addition in sewage treatment plant operation, we first need to define a comprehensive objective function, which aims to minimize the total energy consumption of the blower and dosing pump and the total carbon source consumption, while taking into account the possible penalties for substandard water quality. Aeration blower and The dosing pumps are in constant operation, controlling the aeration flow and carbon source dosage respectively within a period of time. The decision variables are Considering the aeration energy consumption and carbon source consumption, the specific objective function can be expressed as:

[0053] (1)

[0054] Where:

[0055] The sum of the total energy consumption of the blower and the dosing pump and the drug consumption of the dosing pump is the smallest;

[0056] The total energy consumption generated by the blower and dosing pump;

[0057] is the total consumption of carbon source;

[0058] is the inertia weight, which is used to adjust the relative importance between energy consumption and drug consumption;

[0059] It is a water quality penalty item. When the effluent quality does not meet the standard, this item will generate a non-negative penalty value.

[0060] Next, define how to calculate energy and drug consumption.

[0061] Energy consumption and drug consumption can be calculated by the following formula:

[0062] (2)

[0063] (3)

[0064] Where:

[0065] is the number of dosing pumps, is the number of blowers;

[0066] and Respectively The dosing pump and The operating time of each blower;

[0067] is the fluid density, kg / m 3 ;

[0068] is the acceleration due to gravity, m / s 2 ;

[0069] For the The head of the dosing pump during this operation period, m;

[0070] For the The flow rate of the dosing pump during this operation period, m 3 / s;

[0071] For the The overall efficiency of each dosing pump;

[0072] For the The total pressure of the blower during this operation period, m;

[0073] For the The flow rate of the blower during this operation period, m 3 / s;

[0074] For the The overall efficiency of the blower.

[0075] The water quality penalty item is used to constrain the effluent water quality to meet the standard conditions, which is expressed by the following formula:

[0076] (4)

[0077] Where:

[0078] is the penalty coefficient;

[0079] and Respectively represent the total concentration of ammonia nitrogen, chemical oxygen demand and total phosphorus in effluent quality;

[0080] and They respectively represent the maximum allowable limits of ammonia nitrogen, chemical oxygen demand (measured as maximum value) and total phosphorus content in the effluent quality standards of each sewage treatment plant. The upper limits of these indicators are an important basis for evaluating the treatment effect of sewage treatment plants. The upper limits of indicators vary from place to place and from sewage treatment plant to sewage plant.

[0081] Step 1.2, establish the constraints of the first optimization model for aeration and dosing.

[0082] Constraints include flow limits for the blower and dosing pump:

[0083] (5)

[0084] Where:

[0085] and For the The minimum and maximum flow rate allowed by a dosing pump, m 3 / s;

[0086] and For the The minimum and maximum flow rates allowed by the blower, m 3 / s.

[0087] These constraints ensure that dosing pumps and blowers are not operated beyond their design specifications and operating limits, thus avoiding equipment damage or inefficient operation.

[0088] Step 2: construct an ASM-OLWPSO solution framework, and solve the first optimization model according to the ASM-OLWPSO solution framework to obtain the predicted dosage of aeration and dosing of the sewage treatment plant;

[0089] Furthermore, the ASM-OLWPSO solution framework is constructed, which specifically includes:

[0090] Step 2.1, construct the ASM2d mechanism model according to the operation parameters to simulate the parameters of the sewage treatment plant during the sewage treatment process;

[0091] Step 2.2, embed the ASM2d mechanism model into the WPSO algorithm;

[0092] In order to reduce the computational cost and time consumption of the ASM2d mechanism model when simulating candidate parameter sets, we introduced the LSTM prediction model. Through its prediction ability, the LSTM prediction model selects some potential individuals in the offspring population of the WPSO algorithm for real evaluation, reduces the number of mechanism model calls, and improves the optimization calculation efficiency.

[0093] Step 2.3, construct a training set, and obtain the LSTM prediction model based on the training set training;

[0094] Step 2.4: Couple the LSTM prediction model and the WPSO algorithm to obtain the ASM-OLWPSO solution framework.

[0095] Furthermore, after step 2, the method further includes: updating the training set according to the candidate parameter set of each generation of the WPSO algorithm and its corresponding simulation effect data.

[0096] Furthermore, a training set is constructed, specifically including: taking the influent water quality data of the sewage treatment plant as input data; taking the simulation effect data of the ASM2d mechanism model and the effluent water quality data corresponding to the influent water quality data as output data; storing the input data and the output data in a database in a preset structure, and constructing a training set based on the database.

[0097] According to the above content, in this embodiment, the ASM-OLWPSO solution framework includes the WPSO algorithm, the ASM2d mechanism model and the LSTM machine learning model.

[0098] Among them, the WPSO algorithm is used to optimize nonlinear problems such as aeration and dosing; the ASM2d mechanism model characterizes the sewage treatment process and calculates key parameters as the solver of the objective function and constraints of the WPSO algorithm; the LSTM prediction model uses its powerful prediction ability to couple the ASM2d mechanism model with the WPSO algorithm online, realize the update and application of real-time optimization data, and reduce the number of calls to the ASM2d mechanism model.

[0099] Specifically, in this embodiment, the optimization algorithm program is written in MATLAB language, and the optimization function is written according to the ASM-OLWPSO solution framework.

[0100] First, the ASM2d mechanism model was constructed using WEST software. The ASM2d mechanism model was constructed based on the operating parameters of the sewage treatment plant, including process flow, inlet and outlet water characteristics (flow, COD, NH4+, TP, TSS, etc.), aeration and dosing data, etc. The ASM2d mechanism model was used to solve the objective function of the first optimization model and the water quality and energy consumption parameters required by the constraints.

[0101] Secondly, construct the WPSO algorithm, which includes the particle swarm optimization algorithm (PSO) and the grey wolf optimization algorithm (MGWO) with the introduction of mutation mechanism. The specific construction steps are as follows:

[0102] (1) Construct a particle swarm optimization algorithm. This algorithm is a well-known evolutionary algorithm that imitates the foraging behavior of bird groups. In each iteration, the particle's velocity and position are updated:

[0103] (6)

[0104] (7)

[0105] Where:

[0106] For the The weight of the iteration;

[0107] and They are the individual learning and social learning factors of particles;

[0108] and A random number between 0 and 1;

[0109] is the individual optimal solution of the particle;

[0110] is the global optimal solution for all particles;

[0111] and are the position and velocity of the particle respectively;

[0112] For the The position of the particle in the iteration;

[0113] For the The velocity of the particle at this iteration.

[0114] (2) Construct a gray wolf optimization algorithm that introduces mutation, namely the mutation gray wolf operator. The gray wolf group searches for the optimal solution by simulating the hunting behavior of gray wolves in nature. When the gray wolf group finds prey, it will surround the prey. The behavior of gray wolves surrounding prey can be simulated by the following formula:

[0115] (8)

[0116] Where:

[0117] For the The prey position of the iteration is the global optimal solution;

[0118] For the The position of the current gray wolf individual in this iteration, For the The position of the current gray wolf individual in this iteration;

[0119] is a random coefficient between [0,2];

[0120] is the mutant gray wolf encirclement operator, and its expression is:

[0121] (9)

[0122] (10)

[0123] Where:

[0124] is a random number in (0,1);

[0125] is the maximum number of iterations;

[0126] is the control factor;

[0127] is the mutation probability.

[0128] When evolution is in its early stages, , The value is large and decays slowly, the algorithm has better global search ability; in the later stage, that is, , The value decay is accelerated, and the algorithm has better local search ability, but it is also easy to fall into the local optimal trap. Therefore, a random mutation value is introduced to enable the algorithm to jump out of the local optimal state. The mutation probability is .

[0129] According to the hierarchy of the gray wolf pack, individuals and exist The other individuals in the wolf pack follow the above three wolves. , and Update their respective positions:

[0130] (11)

[0131] (12)

[0132] Where:

[0133] , , Respectively represent individuals , and In the The position of the iteration;

[0134] Indicates The location of the individual gray wolf in the next iteration;

[0135] , , Respectively , and The distance from other individuals.

[0136] The mutant gray wolf hunting operator is coupled into the particle swarm optimization algorithm speed update formula to improve the local search ability and convergence performance of the particle swarm optimization algorithm when solving optimization problems, that is, formula (6) is updated as:

[0137] (13)

[0138] Where:

[0139] For the The location of the individual gray wolf in the next iteration;

[0140] For the The position of the particle in the iteration;

[0141] The particle swarm optimization algorithm that integrates the mutated gray wolf operator not only retains the individual historical optimal information and the search guidance of the global historical optimal particles, but also significantly improves the global search ability and convergence speed of the algorithm by introducing the gray wolf's position update strategy and mutation mechanism, effectively solving the problems of poor local search ability and weak convergence ability that the particle swarm optimization algorithm may encounter when solving complex optimization problems.

[0142] Finally, MATLAB is used to call the interface provided by the WEST software and the ASM2d mechanism model is used as the solver of the objective function and constraints of the WPSO algorithm.

[0143] After successfully building the ASM-OLWPSO solution framework, in order to deeply analyze its internal operation and fully verify its practical value, the working mechanism of the ASM-OLWPSO solution framework is explained in detail according to the construction process of the ASM-OLWPSO solution framework.

[0144] The first optimization model is solved according to the ASM-OLWPSO solution framework to obtain the predicted aeration and dosing dosage of the sewage treatment plant, including:

[0145] Determine the fitness value of the candidate parameter set of the WPSO algorithm according to the simulation effect data of the ASM2d mechanism model, and iterate the candidate parameter set of the WPSO algorithm according to the fitness value until the sewage treatment effect corresponding to the candidate parameter set meets the preset conditions. The ASM2d mechanism model is the solver of the WPSO algorithm;

[0146] The candidate parameter set that meets the preset conditions is recorded as the optimal parameter set, and the predicted aeration and dosing amounts of the sewage treatment plant are determined based on the optimal parameter set.

[0147] Specifically, Figure 3 As shown, in this embodiment, the first optimization model is solved according to the ASM-OLWPSO solution framework to obtain the predicted dosage of aeration and dosing of the sewage treatment plant, specifically including:

[0148] First, set the parameters of the WPSO algorithm, including the size of the particle swarm, the maximum number of iterations, the learning factor, the inertia weight, the step size, etc.; define the input parameter range of the ASM2d mechanism model: these parameters usually include aeration volume, dosage, sludge return volume, etc., which have a significant impact on the sewage treatment effect.

[0149] Population initialization, that is, generating initial candidate set parameters, randomly generates a certain number of candidate parameter sets as the initial particle positions of the WPSO algorithm within the feasible domain of the ASM2d mechanism model parameters.

[0150] The ASM2d mechanism model evaluates the fitness value of particles, that is, for each candidate parameter set, the ASM2d mechanism model is used for simulation to obtain the corresponding sewage treatment effect data, such as the removal rate or concentration of effluent water quality indicators (such as BOD, COD, NH3-N, etc.). The fitness value is calculated based on the sewage treatment effect data, and the simulation effect of each candidate parameter set is evaluated based on the fitness value.

[0151] Subsequently, the global optimum and the individual optimum are updated, that is, according to the update formula of the WPSO algorithm, the speed and position of the particle are updated, and the particle is moved to the new position using the updated speed, that is, a new set of candidate parameters is generated iteratively.

[0152] The ASM2d mechanism model calculates and evaluates the fitness value of particles, that is, the ASM2d mechanism model is simulated and the effect is evaluated for the new candidate parameter set.

[0153] Update the global optimum and individual optimum, that is, the WPSO algorithm updates the position of the particle swarm according to the evaluation results of the ASM2d mechanism model, and gradually approaches the global optimal solution.

[0154] Determine whether the goal is achieved, that is, until the preset conditions are met (such as reaching the preset maximum number of iterations, there is no significant improvement in the results of multiple consecutive iterations, etc.), the algorithm terminates and outputs the final candidate parameter set. The final candidate parameter set is recorded as the optimal parameter set as the optimal solution. According to the optimal parameter set, the predicted aeration and dosing dosage of the sewage treatment plant is determined.

[0155] Specifically, Figure 3 As shown, it also includes constructing a database, that is, in this embodiment, a deep learning model database is constructed, which includes input data and output data, wherein the input data is the historical inlet water quality, the output data is the measured water quality, and the calculation results of the ASM2d mechanism model.

[0156] Adaptively select sub-datasets. It should be understood that in this embodiment, after building the database, all data in the database are finely screened. Combined with the idea of ​​clustering, the real-time water quality inlet flow and inlet water quality are used as feature vectors to adaptively screen out sub-datasets with strong correlation as the training set of the LSTM prediction model.

[0157] Construct an LSTM prediction model, that is, train a deep learning model. The machine learning model uses an LSTM prediction model that can describe the sewage treatment time series data. Figure 4 As shown in the figure, the loss curves of the LSTM model training set and test set show that the loss gradually decreases and finally stabilizes at a smaller value, reflecting that the recognition effect of the constructed model is good.

[0158] Further, solving the first optimization model according to the ASM-OLWPSO solution framework also includes: in the iteration of the WPSO algorithm, using the LSTM prediction model to screen the candidate parameter set;

[0159] Specifically, in each iteration of the WPSO algorithm, we first use the LSTM prediction model to quickly predict all particles in the contemporary population (i.e., candidate parameter sets). The prediction result is the estimated value of the sewage treatment effect corresponding to each particle calculated by the LSTM prediction model. In order to select particles that are likely to produce excellent results from these candidate parameter sets for further precise evaluation, we introduced a screening criterion as shown in the following formula:

[0160] (14)

[0161] Where:

[0162] For the The LSTM prediction model prediction value of each particle;

[0163] For particles The individual optimal solution of

[0164] is the elastic coefficient.

[0165] The LSTM prediction model is used to select a certain number of potential particles according to the screening criteria, and feed the candidate parameter set that has been screened and evaluated by the ASM2d mechanism model back to the WPSO algorithm, and the ASM2d mechanism model is used to calculate the true fitness value of these particles (i.e., the accurate evaluation of the sewage treatment effect). This step reduces the number of calls to the ASM2d mechanism model and ensures the authenticity of the evaluation.

[0166] The candidate parameter set that has been screened and evaluated by the ASM2d mechanism model is fed back to the WPSO algorithm to update the position and velocity of the particle swarm and the global optimal solution.

[0167] Update the model database, that is, we also dynamically update the training set of the LSTM prediction model according to each generation of candidate parameter sets and their corresponding simulation effect data to continuously improve the prediction ability and accuracy of the model. This process forms a closed-loop iterative optimization framework, which enables the ASM-OLWPSO solution framework to obtain high-quality optimization solutions while ensuring solution efficiency.

[0168] Specifically, in this embodiment, the optimization results are added to the database in an online update manner according to the optimization calculation results of each generation of the WPSO algorithm. All the data in the database are finely screened, and combined with the idea of ​​clustering, the inlet flow and inlet water quality of real-time water quality are used as feature vectors to adaptively screen out sub-datasets with strong correlation as the training set of the LSTM prediction model, so as to realize dynamic online update of the data set and improve the prediction ability of the model.

[0169] The LSTM prediction model database is updated online. Combined with the characteristics of the WPSO population update position, the samples in the database are adaptively determined in a space according to the following formula:

[0170] (15)

[0171] Where:

[0172] and They are The maximum and minimum values ​​of the dimensional search space;

[0173] and are the maximum and minimum coordinates of the particles in the current population respectively;

[0174] and They are The maximum and minimum allowed values ​​of the dimensional search space;

[0175] is the diffusion coefficient between (0,1).

[0176] As the WPSO population continues to evolve, only some sample data close to the current population in the sample database are selected as subsets to further train the LSTM prediction model.

[0177] Specifically, the WPSO algorithm generates a series of particles representing candidate parameters in each iteration. These particles serve as potential optimization solutions. In each iteration, the LSTM prediction model first quickly traverses and predicts all particles in the current population, and calculates the estimated value of the sewage treatment effect represented by each particle based on its internal algorithm. Subsequently, through the preset screening criteria, those particles with better prediction effects and greater potential will be selected and submitted to the ASM2d mechanism model as candidate solutions for more in-depth and accurate evaluation, that is, the real evaluation process.

[0178] In a further solution, after each iteration, the potential particles verified by the ASM2d mechanism model will be added to the database as new sample data. At this time, the database starts the adaptive screening mechanism to identify and select some sample data with strong correlation with the current optimal solution from the new sample data as a sub-dataset. This sub-dataset is then used as a training set for the LSTM prediction model to ensure that the model can continuously absorb the latest and highly relevant data, thereby continuously improving its prediction performance and accuracy.

[0179] In practical applications, after completing the writing of all functions of the ASM-OLWPSO solution framework, set the parameters of the main function of the first optimization algorithm program. Run the optimization program, and after reaching the set number of generations, output the predicted dosage of aeration and dosing in the sewage treatment plant.

[0180] Step 3: Generate the aeration and dosing strategy of the sewage treatment plant based on the predicted dosage.

[0181] The sewage treatment plant generates the aeration and dosing strategy of the sewage treatment plant according to the predicted dosage, realizes the precise optimization control of the aeration and dosing amount of the sewage treatment plant, and can also guide the operators to control the sewage process according to the aeration and dosing strategy of the sewage treatment plant.

[0182] Embodiment 2:

[0183] A control system for aeration and dosing in a sewage treatment plant, comprising: a collection module, a strategy generation module and a control module;

[0184] The collection module is used to collect the operation parameters of the sewage treatment plant and transmit the operation parameters to the strategy generation module;

[0185] The strategy generation module is used to execute the generation method of the first aspect according to the operation parameters;

[0186] The control module is used to issue control instructions according to the aeration and dosing strategies obtained by the strategy generation module to regulate the aeration and dosing of the sewage treatment plant.

[0187] In this embodiment, the acquisition module is the front end of the control system, responsible for real-time acquisition of various operating parameters of the sewage plant. These parameters include but are not limited to: influent water quality parameters (such as COD, BOD, ammonia nitrogen, total phosphorus, etc.), sewage flow, sludge concentration, dissolved oxygen concentration in the aeration tank, the operating status of each treatment unit (such as the on / off status of the pump, blower, and agitator), and external environmental factors (such as temperature, humidity, weather conditions, etc., which may affect the treatment effect).

[0188] The strategy generation module is the core of the control system. It receives the operating parameters transmitted by the acquisition module and generates an optimized strategy for aeration and dosing according to the method for generating aeration and dosing strategies of the first aspect of the invention. These strategies should be able to provide the best aeration volume and dosage of the agent according to the current water quality conditions and treatment requirements.

[0189] The control module is the execution end of the control system. It receives the aeration and dosing strategies generated by the strategy generation module and converts them into specific control instructions and sends them to various control devices in the sewage treatment plant. These control instructions should be able to accurately control the aeration volume of the aeration system, the dosage of the dosing system, and the operating status of other related equipment.

[0190] The present invention comprehensively considers the mutual influence between the effluent quality and the operating cost of the sewage treatment plant, takes the minimum operating energy consumption and carbon source dosage under the condition that the effluent quality meets the standard as the objective function, and takes the blower flow rate, the dosing pump flow rate and the like as the constraint conditions to establish the aeration and dosing optimization model of the sewage treatment plant; the ASM2d mechanism model is used to calculate the parameters of each stage of the sewage treatment process, and the particle swarm optimization (WPSO) algorithm of the variant gray wolf operator is used as the evaluation solver of the objective function and the constraint conditions to solve the established optimization model. In the optimization process, the LSTM prediction model is used to couple the ASM2d mechanism model with the iterative optimization of the WPSO algorithm, and all individuals that are expected to become the optimal solution are selected in the offspring population in an online update manner through the prediction ability of the LSTM prediction model, and the ASM2d mechanism model is used for evaluation. At the same time, the optimization results of the offspring optimization calculation are added to the LSTM prediction model training database, and supplemented and updated in real time. Considering the disturbance caused by the model prediction error, the real-time outflow water quality is not all up to standard, so only the data that meets the water quality standard is selected as data expansion; and the data with similar inflow water quality conditions is selected as a sub-database for training to further improve the accuracy of the LSTM prediction model. Finally, an online LSTM-assisted optimization algorithm (ASM-OLWPSO) based on the ASM2d mechanism model is formed.

[0191] The present invention calculates the sewage treatment process through the ASM2d mechanism model, making up for the lack of data in the sewage treatment plant; finds the optimal solution that meets the effluent standard under complex constraints through the WPSO algorithm, and improves the search ability of the original particle swarm algorithm by using the improvement of the mutant gray wolf operator; realizes the rapid and accurate optimization of the WPSO algorithm in a data-driven evolutionary optimization manner. The proposed optimization algorithm finds a certain number of promising optimal solutions in the offspring population through the LSTM prediction model, thereby greatly reducing the number of calls to the ASM2d mechanism model and solving the time-consuming problem of conventional optimization methods. The real-time update of the machine learning model database makes up for the deficiency that the existing optimization methods cannot use real-time data in subsequent applications. In addition, by screening the expanded data, the machine learning model is trained in the form of a sub-dataset, which simultaneously solves the time-consuming problem caused by the large database training of the model after each update, and further improves the model accuracy. The efficient and accurate optimization solution of aeration and dosing in the sewage treatment plant is realized.

[0192] In the application of actual sewage treatment process, the present invention has demonstrated excellent performance and optimization effect. Based on the actual hourly operation data of a sewage treatment plant, by implementing the aeration and dosing strategy generation method and control system of the present invention, we have successfully achieved significant optimization of the effluent ammonia nitrogen concentration, and its concentration is stably maintained within the national Class A emission standard. Figure 1 As shown, this achievement directly reflects the important role of the present invention in improving the quality of sewage treatment.

[0193] at the same time, Figure 2 The optimization convergence curve presented clearly demonstrates the efficiency and stability of the method of the present invention. During the optimization process, the algorithm quickly converges to the optimal solution, which not only verifies the reliability of the ASM-OLWPSO solution framework, but also demonstrates its powerful ability in optimizing complex sewage treatment systems.

[0194] More importantly, from the perspective of economic benefits and environmental impact, the implementation of the present invention has brought significant resource savings and cost reduction to the sewage treatment plant. Specifically, by accurately controlling the aeration volume and the dosage of the reagents, the energy consumption of the sewage treatment plant was reduced by 15.6%, and the dosage of chemical reagents such as carbon sources was also greatly reduced by 21.3%. This data not only means direct savings in operating costs for sewage treatment plants, but also reflects its positive contribution to energy conservation, emission reduction, and the promotion of green and sustainable development.

[0195] The above are only several embodiments of the present invention and are not intended to limit the present invention in any form. Although the present invention is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any technician familiar with the profession, without departing from the scope of the technical solution of the present invention, using the above disclosed technical content to make slight changes or modifications are equivalent to equivalent implementation cases and fall within the scope of the technical solution.

Claims

1. A method for generating aeration and dosing strategies for a sewage treatment plant, characterized in that: include: Step 1: Collecting the operating parameters of the sewage treatment plant, and constructing a first optimization model of aeration and dosing of the sewage treatment plant according to the operating parameters; Step 2: construct an ASM-OLWPSO solution framework, and solve the first optimization model according to the ASM-OLWPSO solution framework to obtain the predicted dosage of aeration and dosing of the sewage treatment plant; Step 2.1, constructing an ASM2d mechanism model according to the operating parameters to simulate the parameters of the sewage treatment plant during the sewage treatment process; Step 2.2, embedding the ASM2d mechanism model into the WPSO algorithm; Step 2.3, construct a training set, and obtain an LSTM prediction model based on the training set; Step 2.4, coupling the LSTM prediction model and the WPSO algorithm to obtain the ASM-OLWPSO solution framework; Solving the first optimization model according to the ASM-OLWPSO solution framework to obtain the predicted dosage of aeration and dosing of the sewage treatment plant specifically includes: Determine the fitness value of the candidate parameter set of the WPSO algorithm according to the simulation effect data of the ASM2d mechanism model, and iterate the candidate parameter set of the WPSO algorithm according to the fitness value until the sewage treatment effect corresponding to the candidate parameter set meets the preset conditions, and the ASM2d mechanism model is the solver of the WPSO algorithm; Recording the candidate parameter set that meets the preset conditions as the optimal parameter set, and determining the predicted dosage of aeration and dosing of the sewage treatment plant according to the optimal parameter set; Step 3: generating an aeration and dosing strategy for the sewage treatment plant according to the predicted dosage.

2. The method for generating aeration and dosing strategies for a sewage treatment plant according to claim 1, characterized in that: The WPSO algorithm includes a particle swarm optimization algorithm and a grey wolf optimization algorithm with a mutation mechanism introduced.

3. The method for generating aeration and dosing strategies for a sewage treatment plant according to claim 1, characterized in that: Solving the first optimization model according to the ASM-OLWPSO solution framework also includes: In the iteration of the WPSO algorithm, the candidate parameter set is screened using the LSTM prediction model; The screened candidate parameter set is fed back to the WPSO algorithm.

4. The method for generating aeration and dosing strategies for a sewage treatment plant according to claim 1, characterized in that: Construct a training set, including: The influent water quality data of the sewage treatment plant is used as input data; The simulation effect data of the ASM2d mechanism model and the effluent water quality data corresponding to the influent water quality data are used as output data; The input data and the output data are stored in a database in a preset structure, and a training set is constructed according to the database.

5. The method for generating aeration and dosing strategies for a sewage treatment plant according to claim 1, characterized in that: After step 2, the method further includes: The training set is updated according to the candidate parameter set of each generation of the WPSO algorithm and its corresponding simulation effect data.

6. The method for generating aeration and dosing strategies for a sewage treatment plant according to claim 1, characterized in that: The operating parameters specifically include: inlet water quality parameters, outlet water quality parameters and equipment operation parameters.

7. The method for generating aeration and dosing strategies for a sewage treatment plant according to claim 1, characterized in that: The objective function of the first optimization model is to minimize the sum of the total energy consumption of the aeration device and the dosing device and the total drug consumption of the dosing device.

8. A control system for aeration and dosing in a sewage treatment plant, characterized in that: include: Acquisition module, strategy generation module and control module; The acquisition module is used to collect the operating parameters of the sewage treatment plant and transmit the operating parameters to the strategy generation module; The strategy generation module is used to execute the generation method according to any one of claims 1 to 7 according to the operation parameters; The control module is used to issue control instructions according to the aeration and dosing strategies obtained by the strategy generation module to regulate the aeration and dosing of the sewage treatment plant.

Images