Control method, device and equipment for oxygen-rich aeration sewage treatment and medium

Abstract

This invention belongs to the field of wastewater treatment technology and provides a control method, device, equipment, and medium for oxygen-enriched aerated wastewater treatment, aimed at reducing energy consumption and operating costs. This application inputs preprocessed real-time influent data into a target prediction model to obtain target prediction information indicating predicted effluent treatment performance indicators. Based on this target prediction information, a multi-objective evolutionary algorithm is used to calculate and obtain the target oxygen concentration and target dissolved oxygen. Based on the target oxygen concentration and target dissolved oxygen, the required total aeration flow rate, air flow rate, and pure oxygen flow rate are determined. Thus, by predicting multi-dimensional effluent treatment performance indicators through the target prediction model, accurate prediction of actual water quality indicators and energy consumption is achieved. Furthermore, by simultaneously controlling the oxygen concentration and dissolved oxygen through the predicted water quality indicators and energy consumption, a multi-objective balance between energy consumption and effluent water quality is achieved, thereby reducing energy consumption and operating costs.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, and more specifically, to a control method, apparatus, equipment, and medium for oxygen-enriched aerated wastewater treatment. Background Technology

[0002] Urban wastewater treatment plants widely employ the activated sludge process, in which aerated wastewater treatment is the core step in achieving pollutant degradation. Aerated wastewater treatment refers to the use of microorganisms to degrade organic matter and complete nitrification under aerobic conditions, and then to denitrify under anoxic conditions, thereby breaking down organic matter into harmless carbon dioxide, water, and their own cellular material.

[0003] However, existing aerated wastewater treatment schemes have not incorporated oxygen concentration as a continuous dynamic control variable into the multi-objective optimization system and have not characterized the strong coupling characteristics of multiple variables. They also lack the ability to adaptively screen key features such as various water quality parameters, which leads to large prediction errors in energy consumption and effluent water quality. It is difficult to achieve the best balance between energy consumption and effluent water quality, thus increasing energy consumption and operating costs. Summary of the Invention

[0004] This application aims to provide a control method, device, equipment, and medium for oxygen-enriched aerated wastewater treatment, in order to solve the problems of high energy consumption and operating costs in existing aerated wastewater treatment solutions.

[0005] Firstly, this application provides a control method for oxygen-enriched aerated wastewater treatment, comprising: Acquire real-time influent data after data preprocessing; the real-time influent data is used to indicate the water quality indicators of the influent. Real-time influent data is input into the target prediction model to obtain the target prediction information output by the target prediction model. The target prediction information is used to indicate the predicted effluent treatment performance indicators. The target prediction model is a hybrid neural network based on SE-ResNet and iTransformer. Based on the target prediction information, the target oxygen enrichment concentration and target dissolved oxygen are calculated and obtained through a multi-objective evolutionary algorithm; Based on the target oxygen enrichment concentration and target dissolved oxygen, determine the target gas flow rate to be output. The target gas flow rate includes the total aeration flow rate, air flow rate, and pure oxygen flow rate.

[0006] The technical solution provided in this application offers at least the following beneficial effects: By inputting preprocessed real-time influent data into a target prediction model to obtain target prediction information indicating the predicted effluent treatment performance indicators, this application calculates and obtains the target oxygen concentration and target dissolved oxygen using a multi-objective evolutionary algorithm based on the target prediction information. Then, based on the target oxygen concentration and target dissolved oxygen, it determines the total aeration flow rate, air flow rate, and pure oxygen flow rate to be output. Thus, by predicting multi-dimensional effluent treatment performance indicators through the target prediction model, accurate prediction of actual water quality indicators and energy consumption is achieved. Furthermore, by simultaneously controlling the oxygen concentration and dissolved oxygen through the predicted water quality indicators and energy consumption, the optimal balance between energy consumption and effluent water quality is achieved, thereby reducing energy consumption and operating costs.

[0007] One possible implementation involves acquiring pre-processed real-time water inflow data, including: By improving the frost-ice optimization algorithm, the parameters of variational mode decomposition are optimized. After determining the optimal decomposition parameters, the original water inflow data is filtered and denoised to obtain the modal components of real-time water inflow data. The variational mode decomposition parameters include the number of modes and the penalty factor. The real-time influent data includes at least one of the following: total suspended solids, five-day biochemical oxygen demand, chemical oxygen demand, total nitrogen, and dissolved ammonia nitrogen.

[0008] One possible implementation includes, before inputting real-time inflow data into the target prediction model: Obtain the effluent tags corresponding to each group of historical influent data; the effluent tags are used to indicate the energy consumption data and effluent data corresponding to the historical influent data. Establish a training dataset, which includes historical influent data with added effluent labels and corresponding operational variables; the operational variables are used to indicate the oxygen enrichment concentration and dissolved oxygen under the corresponding historical influent data. An initial hybrid neural network based on the combination of SE-ResNet and iTransformer is constructed. The initial hybrid neural network is trained based on the training dataset to obtain a trained target prediction model.

[0009] One possible implementation is to obtain the effluent tags corresponding to each set of historical influent data, including: Acquire multiple sets of historical water inflow data, along with the corresponding runtime variables; Input the historical influent data and corresponding operating variables into the oxygen-enriched aeration simulation model to obtain the energy consumption data and effluent data output by the oxygen-enriched aeration simulation model. Among them, energy consumption data is used to indicate the energy consumption of blower aeration and pumping, effluent data is used to indicate the water quality indicators of the effluent after sewage treatment, and the oxygen-enriched aeration simulation model is established based on the BSM1 model.

[0010] One possible implementation involves inputting real-time water inflow data into a target prediction model to obtain target prediction information output by the model, including: Get the current running variables, which indicate the current oxygen enrichment concentration and current dissolved oxygen; Input the current operating variables and real-time water inflow data into the target prediction model to obtain the output target prediction information; The target prediction information includes at least one of the following: predicted energy consumption, predicted effluent quality, effluent ammonia nitrogen, and chemical oxygen demand.

[0011] One possible implementation involves calculating the target oxygen enrichment concentration and target dissolved oxygen based on target prediction information using a multi-objective evolutionary algorithm, including: The predicted energy consumption and predicted effluent quality are used as the dual objectives of optimization, and dissolved oxygen and oxygen enrichment concentration are used as decision variables. The Pareto front of the dual objectives is solved by a multi-objective evolutionary algorithm, and the objective solution is selected at the inflection point. The objective solution is used to indicate the optimal compromise solution. Based on the target solution, the target oxygen enrichment concentration and target dissolved oxygen are determined.

[0012] One possible implementation involves determining the target gas flow rate to be output based on the target oxygen enrichment concentration and the target dissolved oxygen, including: Based on the deviation between the target dissolved oxygen and the current dissolved oxygen, the fuzzy PID parameters are tuned, and the total aeration flow rate is determined based on the tuned fuzzy PID parameters. Based on the target oxygen enrichment concentration and the current air oxygen concentration, determine the proportionality coefficient that satisfies the principle of mass conservation. Based on the proportionality coefficient, the air flow rate and pure oxygen flow rate are determined.

[0013] Secondly, this application provides a control device for oxygen-enriched aeration wastewater treatment, comprising: The acquisition module is used to acquire real-time influent data after data preprocessing; the real-time influent data is used to indicate the water quality indicators of the influent. The processing module is used to input real-time influent data into the target prediction model and obtain the target prediction information output by the target prediction model. The target prediction information is used to indicate the predicted effluent treatment performance indicators. The target prediction model is a hybrid neural network based on SE-ResNet and iTransformer. The processing module is also used to calculate and obtain the target oxygen enrichment concentration and target dissolved oxygen based on the target prediction information and through a multi-target evolutionary algorithm; The processing module is also used to determine the target gas flow rate to be output based on the target oxygen concentration and target dissolved oxygen. The target gas flow rate includes the total aeration flow rate, air flow rate, and pure oxygen flow rate.

[0014] Thirdly, this application provides an electronic device comprising: a processor and a memory; the memory storing processor-executable instructions; when the processor is configured to execute the instructions, causing the electronic device to implement the method of the first aspect described above.

[0015] Fourthly, this application provides a computer-readable storage medium comprising: computer software instructions; when the computer software instructions are executed in an electronic device, they cause the electronic device to implement the method described in the first aspect.

[0016] The beneficial effects of the second to fourth aspects mentioned above are described in the corresponding description of the first aspect and will not be repeated here. Attached Figure Description

[0017] Figure 1 A schematic flowchart of a control method for oxygen-enriched aerated wastewater treatment provided in an embodiment of this application; Figure 2 A scatter plot comparing the simulation results and experimental measurement results of the improved BSM1 model provided in the embodiments of this application; Figure 3 Convergence curves of the IRIME algorithm and the comparison algorithm on the benchmark function provided in the embodiments of this application; Figure 4 A comparison chart of the results of the IVRRi model and the comparative model provided in this application embodiment on the prediction of effluent water quality and energy consumption; Figure 5 The Pareto optimal frontier diagram provided for embodiments of this application; Figure 6 A comparison diagram of setpoint tracking control provided in the embodiments of this application; Figure 7 This is a schematic diagram of real-time effluent water quality monitoring results provided in the embodiments of this application; Figure 8 A schematic diagram of the composition of a control device for oxygen-enriched aeration wastewater treatment provided in this application embodiment; Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0019] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0020] It should be noted that in the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0021] In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0022] A typical wastewater treatment aeration control system consists of a biological reaction tank (usually a multi-stage anoxic / aerobic tank), a secondary sedimentation tank, an air aeration device (such as a blower + aeration disc), a dissolved oxygen sensor, and a proportional-integral controller (PIC). These components are connected via electrical and information flow: the influent flows sequentially through the anoxic and aerobic tanks; the aeration device in the aerobic tank blows air (approximately 21% oxygen by volume) into the mixed liquor, providing oxygen for the microorganisms; the dissolved oxygen sensor monitors the dissolved oxygen concentration in the tank in real time and feeds it back to the PIC; based on the deviation between the setpoint and the measured dissolved oxygen value, the PIC outputs an adjustment signal to control the blower speed or valve opening, thereby changing the aeration intensity and maintaining the dissolved oxygen within the target range.

[0023] This allows microorganisms to degrade organic matter and complete nitrification based on the oxygen content within the target range, thus breaking down organic matter into harmless carbon dioxide, water, and their own cellular material.

[0024] However, existing air aeration control schemes in wastewater treatment have the following problems: First, existing air aeration control schemes have low oxygen mass transfer efficiency and high energy consumption: the partial pressure of oxygen in air aeration is low (only about 21%), and the driving force for oxygen mass transfer is limited, resulting in aeration energy consumption accounting for 50% to 70% of the total energy consumption of wastewater treatment. In order to meet increasingly stringent emission standards, excessive aeration is often required, which further aggravates energy waste.

[0025] Secondly, existing air aeration control schemes lack synergistic optimization of oxygen transfer efficiency (OE) and dissolved oxygen (DO). Current schemes treat dissolved oxygen and nitrate nitrogen as control variables, while OE is typically set to a fixed value (e.g., using only pure oxygen or air), failing to incorporate dissolved oxygen (OE) as a continuously adjustable dynamic optimization variable into the control system. In reality, OE and DO are coupled; increasing OE can enhance oxygen transfer efficiency and reduce the required aeration volume, but excessively high OE can inhibit denitrification and affect total nitrogen removal. Due to the lack of a synergistic optimization mechanism, existing schemes struggle to achieve a refined multi-objective trade-off between energy consumption and effluent quality.

[0026] Secondly, existing air aeration control schemes suffer from insufficient data-driven modeling accuracy, making it difficult to cope with the high noise and strong coupling characteristics of influent. While existing schemes attempt to use models such as Long Short-Term Memory (LSTM), Backpropagation (BP) neural networks, or Transformers to predict effluent quality or energy consumption to assist in optimized control, wastewater treatment influent data is characterized by high noise, non-stationarity, and strong coupling of multiple variables. A single neural network struggles to simultaneously capture local temporal fluctuations and global variable dependencies, leading to significant prediction errors. Furthermore, although some schemes introduce Variational Mode Decomposition (VMD) for noise reduction preprocessing, the decomposition quality of VMD is highly dependent on the number of modes and the penalty factor. Traditional empirical or trial-and-error methods cannot adaptively set these two parameters according to actual influent conditions, limiting the preprocessing effect and the accuracy of subsequent models.

[0027] Furthermore, conventional controllers in existing solutions exhibit poor tracking performance due to their nonlinear and large time delay characteristics: wastewater treatment processes are characterized by strong nonlinearity, large time delay, and multivariable coupling. Traditional PI / PID controllers have fixed parameters and cannot be tuned online according to changes in operating conditions, resulting in significant overshoot or steady-state errors in tracking dissolved oxygen setpoints. This is especially problematic when influent load fluctuates drastically, which can easily lead to effluent water quality exceeding standards or increased energy consumption.

[0028] To address the aforementioned technical problems, this application provides a control method, apparatus, equipment, and medium for oxygen-enriched aerated wastewater treatment. This application inputs preprocessed real-time influent data into a target prediction model to obtain target prediction information for the predicted effluent treatment performance indicators. Based on this target prediction information, a multi-objective evolutionary algorithm is used to calculate and obtain the target oxygen concentration and target dissolved oxygen. Based on the target oxygen concentration and target dissolved oxygen, the required total aeration flow rate, air flow rate, and pure oxygen flow rate are determined. Thus, by predicting multi-dimensional effluent treatment performance indicators through the target prediction model, accurate prediction of actual water quality indicators and energy consumption is achieved. Furthermore, by simultaneously controlling the oxygen concentration and dissolved oxygen through the predicted water quality indicators and energy consumption, a multi-objective balance between energy consumption and effluent water quality is achieved, thereby reducing energy consumption and operating costs.

[0029] It should be noted that the control method for oxygen-enriched aeration wastewater treatment provided in this application embodiment is applied to an oxygen-enriched aeration wastewater treatment system, which includes a wastewater treatment unit, an oxygen-enriched aeration unit, a water quality detection unit, and an intelligent control unit.

[0030] The wastewater treatment unit includes an anoxic tank, an aerobic tank, and a secondary sedimentation tank connected in sequence. The oxygen-enriched aeration unit includes an air pump, a pure oxygen pump, and an aeration disc. The aeration disc is located at the bottom of the aerobic tank, and the air outlets of the air pump and the pure oxygen pump are connected to the aeration disc to introduce oxygen-enriched gas with a continuously adjustable oxygen volume fraction into the aerobic tank. The water quality detection unit includes an influent water quality sensor, a dissolved oxygen sensor in the tank, and an effluent water quality sensor, which are used to collect influent water quality indicators, real-time dissolved oxygen concentration in the aerobic tank, and effluent water quality indicators, respectively. The intelligent control unit is electrically connected to the water quality detection unit and the oxygen-enriched aeration unit, respectively, and is used to execute the control method for oxygen-enriched aeration wastewater treatment described in this invention. It outputs control signals to adjust the operating power of the air pump and the pure oxygen pump and the valve opening, thereby achieving optimized control of the aeration process.

[0031] In this embodiment, an oxygen-enriched aeration simulation model was constructed based on the International Water Association standard BSM1 model, serving as the foundational platform for algorithm verification and data generation. The original five-tank bioreactor of the BSM1 model was simplified and its functions expanded, resulting in an A / O process consisting of one anoxic tank and one aerobic tank. The reactor volumes were configured according to the principle of equivalent hydraulic retention time: the daily influent was set to approximately 1 m³, the anoxic tank volume to approximately 0.32 m³, the aerobic tank volume to approximately 0.48 m³, and the secondary sedimentation tank volume to approximately 0.36 m³. Simultaneously, an oxygen-enriched aeration module was introduced into the aerobic tank, allowing the oxygen concentration (OE) to be continuously adjustable within the range of 21% to 90%. Gas-liquid mass transfer and biodynamic models under oxygen-enriched conditions were established, including a linear mapping relationship between saturated dissolved oxygen concentration and OE based on Henry's Law, an oxygen-enriched correction model for the oxygen half-saturation constant of autotrophic bacteria, and a model for the inhibition of anoxic denitrification caused by oxygen-enriched recirculation. This provides a high-fidelity simulation environment for the implementation of subsequent control methods.

[0032] The following is a detailed description of a control method for oxygen-enriched aerated wastewater treatment provided by an embodiment of this application, with reference to the accompanying drawings.

[0033] Figure 1 This is a schematic flowchart illustrating a control method for oxygen-enriched aerated wastewater treatment, provided as an embodiment of this application. (Combined with...) Figure 1 The control method for oxygen-enriched aerated wastewater treatment provided in the embodiments of this application will be described below: S101. Obtain real-time water inflow data after data preprocessing.

[0034] In one possible implementation, the parameters of variational mode decomposition are optimized by improving the frost-ice optimization algorithm. After determining the optimal decomposition parameters, the original water inflow data is filtered and denoised to obtain the modal components of the real-time water inflow data.

[0035] The variational mode decomposition parameters include the number of modes and a penalty factor. Real-time influent data is used to indicate the water quality indicators of the influent. The real-time influent data includes at least total suspended solids (TSS), five-day biochemical oxygen demand (BOD5), chemical oxygen demand (COD), total nitrogen (TN), and dissolved ammonia nitrogen (S). NH One of the items in Soluble Ammonia Nitrogen.

[0036] For example, the raw influent data is collected by a water quality detection unit installed in the influent pipe at a sampling interval of 15 minutes. Considering that the wastewater treatment influent data has the characteristics of high noise, non-stationarity, and strong coupling of multiple variables, this step uses the improved frost-ice optimization algorithm IRIME (Improved Rime Optimization Algorithm) to filter and denoise the raw influent data to obtain the above-mentioned real-time influent data.

[0037] Furthermore, regarding the optimization of variational mode decomposition parameters using the improved frost-ice optimization algorithm, and the determination of the optimal decomposition parameters, the original influent data is then filtered and denoised, as explained below: The improved frost-ice optimization algorithm refers to a three-stage enhancement of the standard RIME algorithm, and the specific implementation process is as follows: a. In the population initialization stage, a piecewise linear chaotic map (PWLCM) is introduced. Its good ergodicity, uniformity, and initial value sensitivity are used to generate chaotic sequences to replace the original pseudo-random numbers for initializing the frost grain population. This improves the coverage of the initial population in the search space and enhances population diversity. The piecewise linear chaotic map can be expressed as equation (1): (1); Where p is a control parameter, and 0 < p <0.5, in this embodiment p=0.3; x i Let be the chaotic sequence value obtained in the i-th iteration. Map the generated chaotic sequence to the upper and lower bounds of each variable dimension to construct the initial frost grain population.

[0038] b. In the soft frost ice search stage, a Cauchy perturbation search strategy is introduced. As the frost particles approach the current dominant individual, a random step size following a Cauchy distribution is superimposed to enhance the algorithm's long-distance jump capability and improve its global exploration performance in complex multi-peak regions, thereby reducing the probability of getting trapped in local optima. The improved soft frost ice search update formula is shown in equation (2) below: (2); Among them, R ij new R represents the position of the updated frost particles. best,j The j-th particle of the best frost ice in the frost ice group R; R ij Let r1 be the position of the current frost grain in the j-th dimension; r1 is a random number in [0,1]; α is the step size adjustment factor of the Cauchy perturbation; casuchy(0,1) represents a random variable that follows a standard Cauchy distribution.

[0039] c. In the hard frost ice piercing stage, a lens reverse learning strategy is introduced. Based on the symmetric relationship between the current individual and the search boundary, a reverse candidate solution is constructed to expand the search range near the local high-quality region and improve the efficiency of the solution set update in the later stage of the algorithm. The updated formula of the improved hard frost ice piercing mechanism is shown in the following formula (3): (3); Among them, a j With b j represents the lower and upper bounds of the j-th dimension variable; k is the lens scaling factor.

[0040] Based on the above IRIME algorithm, the fitness function is used to adaptively optimize the number of modes k and the penalty factor α of variational mode decomposition (VMD). The fitness function expression is shown in the following equation (4): (4); Among them, P j Let x(j) be the normalized form of x(j); a(j) is the envelope signal obtained by demodulating the signal x(j) using Hilbert.

[0041] After optimization, the original influent data is decomposed into k intrinsic mode functions (IMFs) using the optimal number of modes k and the penalty factor α. Then, based on the minimum energy criterion (MEC), the decomposed modal components are screened to remove high-frequency components dominated by noise and retain the effective components to reconstruct the denoised real-time influent data, thus completing the data preprocessing.

[0042] In an optional implementation, particle swarm optimization algorithm or gray wolf optimization algorithm can be used instead of IRIME algorithm for VMD parameter optimization. The search range of mode number k is set to 3~12, and the search range of penalty factor α is set to 100~3000. The search range can be adjusted according to the fluctuation characteristics of the actual influent water quality.

[0043] S102. Input the real-time water inflow data into the target prediction model to obtain the target prediction information output by the target prediction model.

[0044] The target prediction information is used to indicate the predicted effluent treatment performance indicators, and includes at least the predicted energy consumption (EC), predicted effluent water quality (EQ), and effluent ammonia nitrogen (S). NH It is one of the chemical oxygen demand (COD); the target prediction model is a hybrid neural network based on SE-ResNet and iTransformer, hereinafter referred to as IVSRi hybrid neural network.

[0045] Before performing this step, the target prediction model must be built and trained in advance. The specific implementation process is as follows: S201. Obtain the effluent labels corresponding to each group of historical influent data.

[0046] In this step, the effluent label is used to indicate the energy consumption data and effluent data corresponding to the historical influent data; wherein, the energy consumption data is used to indicate the energy consumption of the blower aeration and the pumping energy consumption, and the effluent data is used to indicate the water quality indicators of the effluent after sewage treatment.

[0047] In practice, firstly, multiple sets of historical influent data and corresponding operating variables are acquired. The operating variables are used to indicate the oxygen enrichment concentration (OE) and dissolved oxygen (DO) under the corresponding historical influent data. Then, the historical influent data and the corresponding operating variables are input into the aforementioned oxygen-enriched aeration simulation model to obtain the energy consumption data and effluent data output by the oxygen-enriched aeration simulation model, thus completing the construction of the effluent label.

[0048] Among them, the energy consumption data EC consists of aeration energy consumption AE and pumping energy consumption PE, and the calculation formula is shown in the following formula (5): (5); Among them, V i Let Q be the volume of the i-th unit; T be the optimization period; Q be the volume of the i-th unit. a For internal reflux; Q r For sludge discharge; Q w For sludge recirculation.

[0049] The effluent water quality EQ is the penalty value for exceeding the standard of pollutants in the effluent. The calculation formula is shown in the following formula (6): (6); Where SS is the suspended solids concentration; COD is the chemical oxygen demand; S NKj S is Kjeldahl nitrogen; NO Nitrate nitrogen; BOD5 is the five-day biochemical oxygen demand; Q e This refers to the outflow rate.

[0050] S202. Establish the training dataset.

[0051] The training dataset includes the historical influent data with the effluent label added and the corresponding runtime variables. The preprocessed 672 sets of historical data are divided into a training set and a test set, with the first 618 sets serving as the training set and the last 54 sets as the test set. Both the training and test sets have a 7-dimensional input dimension, including 5 influent water quality indicators (TSS, BOD5, COD, TN, S). NH The system consists of two runtime variables (DO and OE), and the output dimension is four-dimensional, including EC, EQ, and S. NH COD.

[0052] S203. Construct and train the target prediction model.

[0053] In this step, an initial hybrid neural network based on the combination of SE-ResNet and iTransformer is constructed. Based on the training dataset, the initial hybrid neural network is trained to obtain the trained target prediction model.

[0054] The specific structure of the IVRRi hybrid neural network is as follows: The SE-ResNet feature extraction branch contains four sequentially connected residual blocks. Each residual block consists of two convolutional layers, a batch normalization layer, and a ReLU activation function. Each residual block is followed by an SE channel attention module with a compression ratio of 8. The SE module first compresses the feature map into channel descriptors through global average pooling, and then outputs the weight coefficients of each channel through two fully connected layers to recalibrate the channels of the original feature map, thereby realizing local temporal feature extraction and adaptive calibration of the weights of key variables.

[0055] The iTransformer global dependency modeling branch inputs the multidimensional feature sequence output by SE-ResNet into the iTransformer module. The iTransformer adopts an inverted structure, embedding the complete time series of each variable independently as a token. Feature interaction is performed through 3 encoder layers, with 8 multi-head attention heads set in each encoder layer and a hidden layer dimension of 128. After layer normalization and feedforward network to complete cross-variable information interaction, the prediction result is output through linear projection layer.

[0056] During model training, the Adam optimizer was used with an initial learning rate of 0.001, a batch size of 32, and 800 training epochs. The mean squared error (MSE) loss function was used. After training, the model performance was verified using a test set. In this embodiment, the IVSRi model achieved a determination coefficient R² of 99.808% and a root mean squared error (RMSE) of 0.505 for EQ prediction, and an R² of 96.367% and an RMSE of 0.569 for EC prediction, meeting the accuracy requirements for subsequent optimization control.

[0057] In optional implementations, the number of residual blocks in SE-ResNet, the compression ratio of SE modules, and the number of encoder layers, attention heads, and hidden layer dimensions of iTransformer can be adjusted according to actual computing resources and data scale. For example, the number of residual blocks can be adjusted between 2 and 6, the number of encoder layers can be adjusted between 2 and 4, and the number of attention heads can be adjusted between 4 and 16.

[0058] After the model training is completed, the specific process of step S102 is as follows: obtain the current operating variables, which are used to indicate the current oxygen enrichment concentration and the current dissolved oxygen; input the current operating variables and real-time influent data into the target prediction model to obtain the output target prediction information.

[0059] S103. Based on the target prediction information, the target oxygen concentration and target dissolved oxygen are calculated and obtained through a multi-target evolutionary algorithm.

[0060] In one possible implementation, the predicted energy consumption (EC) and the predicted effluent water quality (EQ) are used as dual objectives for optimization. The Pareto front of the dual objectives is solved using the decomposition-based multi-objective evolutionary algorithm MOEA / D, and the objective solution is selected at the inflection point. The objective solution is used to indicate the optimal compromise solution. Based on the objective solution, the target oxygen enrichment concentration and the target dissolved oxygen are determined.

[0061] The specific implementation process is as follows: S301. Optimization Problem Definition: The decision variables are dissolved oxygen (DO) and oxygen enrichment concentration (OE), where DO ranges from 1 to 4 mg / L and OE ranges from 0.21 to 0.9; the optimization objective is bi-objective minimization, i.e., minimizing EC and EQ; the constraint is: effluent ammonia nitrogen S... NH The effluent concentrations are ≤5 mg / L, COD ≤50 mg / L, TN ≤15 mg / L, and BOD5 ≤10 mg / L. The constrained optimization problem is transformed into an unconstrained optimization problem using the penalty function method.

[0062] S302, MOEA / D Algorithm Configuration: The Chebyshev aggregation method is used to transform the multi-objective optimization problem into a single-objective sub-problem, as shown in the following equation (7): (7); Among them, f i (x) represents the i-th maximization objective; λ i For the i-th target f i The nonnegative weights of (x); z It is an ideal vector.

[0063] The core parameters of the algorithm are set as follows: population size 30, neighborhood size 4, maximum number of iterations 50, crossover parameter 0.5, mutation probability 0.4, and distribution index 15.

[0064] S303. Selection of the optimal compromise solution: The Pareto optimal frontier with EC and EQ in an "L" shape distribution is obtained by solving the solution. The solution located in the inflection point region of the curve is selected as the optimal compromise solution. This solution is the best balance state that can reduce energy consumption to the greatest extent without causing serious deterioration of the effluent water quality. In this embodiment, the target dissolved oxygen corresponding to the typical optimal compromise solution is about 2.67 mg / L, and the target oxygen enrichment concentration is about 0.47.

[0065] In optional implementations, other multi-objective evolutionary algorithms such as NSGA-II and MOPSO can be used to replace MOEA / D for solving the Pareto front. Alternatively, algorithm parameters such as population size and number of iterations can be adjusted according to the actual operational needs of the wastewater treatment plant to balance optimization accuracy and computational real-time performance.

[0066] S104. Determine the target gas flow rate to be output based on the target oxygen concentration and target dissolved oxygen.

[0067] The target gas flow rate includes the total aeration flow rate, air flow rate, and pure oxygen flow rate. The specific implementation process is divided into two stages: setting the total aeration flow rate and the gas flow rate ratio.

[0068] The tuning process for the total aeration flow rate is explained below: Based on the deviation between the target dissolved oxygen and the current dissolved oxygen, the PID parameters are tuned, and the total aeration flow rate is determined based on the tuned PID parameters.

[0069] In practice, an adaptive fuzzy PID controller is used for closed-loop control of DO, with the deviation e(t) between the measured DO value and the setpoint and its rate of change as the basis of measurement. (t) is the input, and the proportional coefficient K of the PID is tuned online through fuzzy inference. p Integral coefficient K i Differential coefficient K d The controller outputs the total aeration flow rate.

[0070] Among them, the fuzzy inference engine will input variable e(t), (t) and output variable ΔK p ΔK i ΔK d The universe of discourse is divided into 7 fuzzy language subsets, namely {NB, NM, NS, ZO, PS, PM, PB}, corresponding to {negative large, negative medium, negative small, zero, positive small, positive medium, positive large}. The membership function adopts the triangular membership function. Based on the experience of experts in the field of wastewater treatment, 49 fuzzy rules are constructed to form a parameter tuning rule base.

[0071] The formula for online tuning of PID parameters is shown in the following formula (8): (8); Among them, K p0 K is the initial value of the proportional coefficient. i0 K represents the initial value of the integral coefficients. d0 This is the initial value of the differential coefficient. In this embodiment, K... p0 Take 1.2, K i0 Take 0.05, K d0 Take 0.02.

[0072] Based on the tuned PID parameters, the control output formula for the total aeration flow rate is shown in the following formula (9): (9); Among them, Q total (t) represents the total aeration flow rate at time t.

[0073] The process of mixing air flow rate and pure oxygen flow rate is explained: Based on the target oxygen concentration and the current air oxygen concentration, determine the proportionality coefficient that satisfies the principle of mass conservation; based on the proportionality coefficient, determine the air flow rate and the pure oxygen flow rate.

[0074] In practice, based on the principle of mass conservation, the total aeration flow rate is allocated to air flow rate and pure oxygen flow rate according to the target oxygen concentration setting. First, the proportionality coefficient K is defined, and the calculation formula is shown in the following formula (10): (10); Where K is the proportionality coefficient; C air The oxygen concentration in the air is 0.21; C oxy The concentration of pure oxygen is 1.

[0075] Based on the proportionality coefficient K, the air flow rate Q is obtained. air With pure oxygen flow rate Q oxy The allocation relationship is shown by the following formula (11): (11); The intelligent control unit converts the calculated air flow and pure oxygen flow into control signals, which are then sent to the air pump and pure oxygen pump, respectively. This adjusts the operating power of the pumps and the valve opening, thereby achieving precise control of the total aeration flow and oxygen concentration, and completing the entire optimized control closed loop.

[0076] In optional implementations, the fuzzy rule base and initial PID parameters of the fuzzy PID controller can be customized according to the process characteristics of a specific wastewater treatment plant to adapt to different influent conditions and treatment scales. Alternatively, a closed-loop feedback can be added to the oxygen enrichment concentration control stage, using a gas phase oxygen concentration sensor to detect the oxygen volume fraction of the aeration gas in real time, and fine-tuning the pure oxygen flow rate to further improve the control accuracy of the oxygen enrichment concentration.

[0077] In this embodiment, real-time influent data, after data preprocessing, is input into the target prediction model to obtain target prediction information indicating the predicted effluent treatment performance indicators. Based on this target prediction information, a multi-objective evolutionary algorithm is used to calculate and obtain the target oxygen concentration and target dissolved oxygen. Based on the target oxygen concentration and target dissolved oxygen, the target gas flow rate to be output is determined. Thus, by predicting multi-dimensional effluent treatment performance indicators through the target prediction model, accurate prediction of actual water quality indicators and energy consumption is achieved. Furthermore, by simultaneously controlling the oxygen concentration and dissolved oxygen based on the predicted water quality indicators and energy consumption, the optimal balance between energy consumption and effluent water quality is achieved, thereby reducing energy consumption and operating costs.

[0078] Figure 2 A scatter plot comparing the simulation results and experimental results of the improved BSM1 model provided in the embodiments of this application.

[0079] like Figure 2 As shown, the horizontal axis represents the experimentally measured values ​​(mg / L) of effluent ammonia nitrogen or chemical oxygen demand (COD), and the vertical axis represents the model simulation output values ​​(mg / L) at the corresponding time points. The figure contains 72 sampling points, with each data point closely distributed around the diagonal (y=x line), showing no significant systematic deviation, indicating a high degree of consistency between the simulated and measured values. Statistical analysis shows that the overall root mean square error (RMSE) of the model for effluent ammonia nitrogen is 0.42 mg / L, and the overall RMSE for effluent COD is 0.51 mg / L. This figure visually verifies the accuracy and generalization ability of the oxygen-enriched aeration A / O simulation model constructed in this invention under different operating ranges, providing a reliable platform foundation for subsequent simulation verification of optimized control strategies.

[0080] Figure 3 Convergence curves of the IRIME algorithm and the comparison algorithm provided in the embodiments of this application on the benchmark function.

[0081] like Figure 3 As shown, four CEC2005 benchmark functions with different characteristics were selected for testing. The population size was 30, the maximum number of iterations was 500, the dimension was 30, and each function was run independently 50 times. The comparison algorithms included Artificial Hummingbird Algorithm (AHA), Eagle Optimization Algorithm (AO), Osprey Optimization Algorithm (OOA), and Improved Frost Ice Optimization Algorithm (IRIME). The horizontal axis in the figure represents the number of iterations, and the vertical axis represents the fitness value (logarithmic scale). Figure 3 It can be seen that the IRIME algorithm has the fastest convergence curve descent speed and the highest final convergence accuracy on the four benchmark functions. Moreover, the average value and standard deviation of the operation are lower than those of other comparative algorithms, indicating that the IRIME algorithm proposed in this invention has stronger global optimization ability and convergence stability.

[0082] Figure 4A comparison chart showing the results of the IVRRi model and the comparative model provided in this application for predicting effluent water quality and energy consumption.

[0083] like Figure 4 As shown in the figure, the horizontal axis represents the time series (sampling points), and the vertical axes represent energy consumption (EC), effluent water quality (EQ), and effluent ammonia nitrogen (S), respectively. NH The actual and predicted values ​​of chemical oxygen demand (COD) were compared. The comparison models included IVRST (Transformer), BP neural network (BPNN), and long short-term memory network (LSTM). Figure 4 As can be seen, the predicted curves of the IVRRi model closely match the actual curves, accurately capturing the overall evolution trend of each indicator and effectively tracking and fitting the transient peak-valley changes caused by influent impact. In contrast, other comparative models exhibit certain prediction lag and amplitude deviation in the range of drastic fluctuations, further validating the superiority of the IVRRi hybrid neural network proposed in this invention for modeling complex wastewater treatment data.

[0084] Figure 5 The Pareto optimal frontier diagram provided for embodiments of this application.

[0085] like Figure 5 As shown, Figure 5 (a) shows the mapping relationship between energy consumption (EC) and effluent water quality (EQ) in the target space. The two are distributed in a significantly conflicting "L" shape, and the pentagram marks the optimal compromise solution at the inflection point of the curve. Figure 5 (b) shows the interaction between the decision variables dissolved oxygen (DO) and oxygen enrichment concentration (OE). The scatter plot color maps the EC gradient: energy consumption increases dramatically when DO is high (>2.7 mg / L) and OE is low (<0.4), and energy consumption decreases significantly when OE is low. The red star corresponds to the optimal solution (DO≈2.67 mg / L, OE≈0.47).

[0086] Figure 6 A comparison diagram of setpoint tracking control provided in the embodiments of this application.

[0087] like Figure 6 As shown in the figure, the tracking effect after 12 optimized runs on the sixth day is illustrated. The adaptive fuzzy PID controller responds quickly and with small error to changes in the DO setpoint, while traditional PID controllers suffer from significant overshoot. The OE precisely maintains the target value through open-loop proportioning control, verifying the effectiveness of the control strategy of this invention.

[0088] Figure 7 This is a schematic diagram of the real-time effluent water quality monitoring results provided in the embodiments of this application.

[0089] like Figure 7As shown, BOD5, TN, and TSS all consistently meet emission standards; SNH and COD, which are prone to exceeding standards, also basically meet environmental quality requirements. Furthermore, the strategy of this invention effectively suppresses the peak concentrations of SNH and COD, further reducing the risk of effluent exceeding standards, and verifying the comprehensive advantages of the proposed method in ensuring effluent quality while reducing energy consumption.

[0090] In some embodiments, this application also provides a control device for oxygen-enriched aeration wastewater treatment. This control device may include one or more functional modules for implementing the control method for oxygen-enriched aeration wastewater treatment described in the above embodiments.

[0091] For example, Figure 8 This is a schematic diagram illustrating the composition of a control device for oxygen-enriched aeration wastewater treatment, provided as an embodiment of this application. Figure 8 As shown, the control device for oxygen-enriched aerated wastewater treatment includes: an acquisition module 401 and a processing module 402.

[0092] The acquisition module 401 is used to acquire real-time water intake data after data preprocessing; wherein, the real-time water intake data is used to indicate the water quality indicators of the incoming water.

[0093] The processing module 402 is used to input real-time influent data into the target prediction model to obtain the target prediction information output by the target prediction model; wherein, the target prediction information is used to indicate the predicted effluent treatment performance index, and the target prediction model is a hybrid neural network based on SE-ResNet and iTransformer.

[0094] The processing module 402 is also used to calculate and obtain the target oxygen concentration and target dissolved oxygen based on the target prediction information and through a multi-target evolutionary algorithm.

[0095] The processing module 402 is also used to determine the target gas flow rate to be output based on the target oxygen enrichment concentration and the target dissolved oxygen. The target gas flow rate includes the total aeration flow rate, the air flow rate, and the pure oxygen flow rate.

[0096] In some embodiments, the acquisition module 401 is specifically used to filter and denoise the original water inflow data using a frost-ice optimization algorithm to obtain real-time water inflow data; The real-time influent data must include at least one of the following: total suspended solids, five-day biochemical oxygen demand, chemical oxygen demand, total nitrogen, and dissolved ammonia nitrogen.

[0097] In some embodiments, the processing module 402 is specifically used to obtain the effluent tags corresponding to each set of historical influent data before inputting the real-time influent data into the target prediction model; wherein, the effluent tags are used to indicate the energy consumption data and effluent data corresponding to the historical influent data; Establish a training dataset, which includes historical influent data with added effluent labels and corresponding operational variables; the operational variables are used to indicate the oxygen enrichment concentration and dissolved oxygen under the corresponding historical influent data. An initial hybrid neural network based on the combination of SE-ResNet and iTransformer is constructed. The initial hybrid neural network is trained based on the training dataset to obtain a trained target prediction model.

[0098] In some embodiments, the processing module 402 is specifically used to acquire multiple sets of historical water inflow data and corresponding operating variables; Input the historical influent data and corresponding operating variables into the oxygen-enriched aeration simulation model to obtain the energy consumption data and effluent data output by the oxygen-enriched aeration simulation model. Among them, energy consumption data is used to indicate the energy consumption of blower aeration, effluent data is used to indicate the water quality indicators of the effluent after sewage treatment, and the oxygen-enriched aeration simulation model is established based on the BSM1 model.

[0099] In some embodiments, the processing module 402 is specifically used to obtain the current operating variables, which are used to indicate the current oxygen enrichment concentration and the current dissolved oxygen. Input the current operating variables and real-time water inflow data into the target prediction model to obtain the output target prediction information; The target prediction information includes at least one of the following: predicted energy consumption, predicted effluent quality, effluent ammonia nitrogen, and chemical oxygen demand.

[0100] In some embodiments, the processing module 402 is specifically used to take the predicted energy consumption and the predicted effluent water quality as optimization dual objectives, solve the optimization dual objective Pareto front through a multi-objective evolutionary algorithm, and select the objective solution at the inflection point. The objective solution is used to indicate the optimal compromise solution. Based on the target solution, the target oxygen enrichment concentration and target dissolved oxygen are determined.

[0101] In some embodiments, the processing module 402 is specifically used to tune the PID parameters according to the deviation between the target dissolved oxygen and the current dissolved oxygen, and to determine the total aeration flow rate based on the tuned PID parameters; Based on the target oxygen enrichment concentration and the current air oxygen concentration, determine the proportionality coefficient that satisfies the principle of mass conservation. Based on the proportionality coefficient, the air flow rate and pure oxygen flow rate are determined.

[0102] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device includes: a processor 502, a communication interface 503, and a bus 504. Optionally, the electronic device may also include a memory 501.

[0103] Processor 502 may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 502 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 502 may also be a combination of functions implementing computing capabilities, such as a combination including CPU0 and CPU1, a DSP, and a microprocessor.

[0104] The communication interface 503 includes a receiving unit and a transmitting unit, and is used to connect with other devices via a communication network. This communication network can be Ethernet, a wireless access network, a wireless local area network (WLAN), etc.

[0105] The memory 501 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0106] As one possible implementation, the memory 501 can exist independently of the processor 502. The memory 501 can be connected to the processor 502 via a bus 504 and is used to store instructions or program code. When the processor 502 calls the instructions or program code stored in the memory 501, it can implement the control method for oxygen-enriched aeration wastewater treatment provided in this embodiment of the invention.

[0107] In another possible implementation, the memory 501 can also be integrated with the processor 502.

[0108] Bus 504 can be an extended industry standard architecture (EISA) bus, etc. Bus 504 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0109] Through the above description of the implementation methods, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the service calling device can be divided into different functional modules to complete all or part of the functions described above.

[0110] This application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be executed by computer instructions instructing related hardware. The program can be stored in the aforementioned computer-readable storage medium, and when executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be any of the foregoing embodiments or memory. The aforementioned computer-readable storage medium can also be an external storage device of the aforementioned service invocation device, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the aforementioned service invocation device. Further, the aforementioned computer-readable storage medium can include both internal storage units of the aforementioned service invocation device and external storage devices. The aforementioned computer-readable storage medium is used to store the aforementioned computer program and other programs and data required by the aforementioned service invocation device. The aforementioned computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0111] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A control method for oxygen-enriched aerated wastewater treatment, characterized in that, The method includes: Acquire real-time water inflow data after data preprocessing; wherein the real-time water inflow data is used to indicate the water quality indicators of the inflow. The real-time influent data is input into the target prediction model to obtain the target prediction information output by the target prediction model; wherein, the target prediction information is used to indicate the predicted effluent treatment performance index, and the target prediction model is a hybrid neural network based on SE-ResNet and iTransformer. Based on the target prediction information, the target oxygen concentration and target dissolved oxygen are calculated and obtained through a multi-target evolutionary algorithm. Based on the target oxygen concentration and the target dissolved oxygen, the target gas flow rate to be output is determined, including the total aeration flow rate, the air flow rate, and the pure oxygen flow rate.

2. The control method for oxygen-enriched aerated wastewater treatment according to claim 1, characterized in that, The acquisition of preprocessed real-time water inflow data includes: The parameters of variational mode decomposition are optimized by improving the frost-ice optimization algorithm. After determining the optimal decomposition parameters, the original water inflow data is filtered and denoised to obtain the modal components of the real-time water inflow data. The variational mode decomposition parameters include the number of modes and the penalty factor, and the real-time influent data includes at least one of the following: total suspended solids, five-day biochemical oxygen demand, chemical oxygen demand, total nitrogen, and dissolved ammonia nitrogen.

3. The control method for oxygen-enriched aerated wastewater treatment according to claim 1, characterized in that, Before inputting the real-time water inflow data into the target prediction model, the following steps are also included: Obtain the effluent tags corresponding to each group of historical influent data; wherein, the effluent tags are used to indicate the energy consumption data and effluent data corresponding to the historical influent data; A training dataset is established, which includes the historical influent data with the effluent label added and the corresponding operating variables; wherein the operating variables are used to indicate the oxygen enrichment concentration and dissolved oxygen under the corresponding historical influent data. An initial hybrid neural network based on the combination of SE-ResNet and iTransformer is constructed. The initial hybrid neural network is trained based on the training dataset to obtain the target prediction model after training.

4. The control method for oxygen-enriched aerated wastewater treatment according to claim 3, characterized in that, The step of obtaining the effluent tags corresponding to each group of historical influent data includes: Obtain multiple sets of the historical water inflow data, and the corresponding operational variables; The historical influent data and the corresponding operating variables are input into the oxygen-enriched aeration simulation model to obtain the energy consumption data and effluent data output by the oxygen-enriched aeration simulation model. The energy consumption data is used to indicate the energy consumption of the blower aeration and the pumping energy consumption, the effluent data is used to indicate the water quality indicators of the effluent after sewage treatment, and the oxygen-enriched aeration simulation model is established based on the BSM1 model.

5. The control method for oxygen-enriched aerated wastewater treatment according to claim 1, characterized in that, The real-time water inflow data is input into the target prediction model to obtain the target prediction information output by the target prediction model, including: Obtain the current operating variables, which indicate the current oxygen enrichment concentration and the current dissolved oxygen; The current operating variables and the real-time water inflow data are input into the target prediction model to obtain the output target prediction information; The target prediction information includes at least one of the following: predicted energy consumption, predicted effluent water quality, effluent ammonia nitrogen, and chemical oxygen demand.

6. The control method for oxygen-enriched aerated wastewater treatment according to claim 5, characterized in that, The step of calculating and obtaining the target oxygen enrichment concentration and target dissolved oxygen based on the target prediction information using a multi-objective evolutionary algorithm includes: The predicted energy consumption and predicted effluent quality are used as the dual objectives of optimization, and dissolved oxygen and oxygen enrichment concentration are used as decision variables. The Pareto front of the dual objectives is solved by the multi-objective evolutionary algorithm, and the objective solution is selected at the inflection point. The objective solution is used to indicate the optimal compromise solution. Based on the target solution, the target oxygen enrichment concentration and the target dissolved oxygen are determined.

7. The control method for oxygen-enriched aerated wastewater treatment according to claim 1, characterized in that, The step of determining the target gas flow rate to be output based on the target oxygen concentration and the target dissolved oxygen includes: Based on the deviation between the target dissolved oxygen and the current dissolved oxygen, the fuzzy PID parameters are tuned, and the total aeration flow rate is determined based on the tuned fuzzy PID parameters. Based on the target oxygen concentration and the current air oxygen concentration, determine the proportionality coefficient that satisfies the principle of mass conservation. Based on the aforementioned proportionality coefficient, the air flow rate and pure oxygen flow rate are determined.

8. A control device for oxygen-enriched aeration wastewater treatment, characterized in that, include: The acquisition module is used to acquire real-time water inflow data after data preprocessing; wherein the real-time water inflow data is used to indicate the water quality indicators of the inflow. The processing module is used to input the real-time influent data into the target prediction model to obtain the target prediction information output by the target prediction model; wherein, the target prediction information is used to indicate the predicted effluent treatment performance index, and the target prediction model is a hybrid neural network based on SE-ResNet and iTransformer. The processing module is also used to calculate and obtain the target oxygen concentration and target dissolved oxygen based on the target prediction information and through a multi-target evolutionary algorithm. The processing module is further configured to determine the target gas flow rate to be output based on the target oxygen concentration and the target dissolved oxygen, wherein the target gas flow rate includes the total aeration flow rate, the air flow rate, and the pure oxygen flow rate.

9. An electronic device, characterized in that, It includes a processor and a memory, the processor being coupled to the memory; the memory is used to store computer instructions, which are loaded and executed by the processor to enable the computer device to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes computer-executable instructions that, when executed on a computer, cause the computer to perform the method according to any one of claims 1 to 7.

Images