An intelligent aeration control method for sewage treatment based on a weighted fusion model

Abstract

The present application relates to the technical field of sewage treatment, and particularly relates to a sewage treatment intelligent aeration control method based on a weighted fusion model. Inflow water quality monitoring data set, process water quality monitoring data set, effluent water quality monitoring data set, aeration quantity monitoring data set and equipment working condition monitoring data set are acquired. The inflow water quality monitoring data set, the process water quality monitoring data set, the effluent water quality monitoring data set and the aeration quantity monitoring data set are fused into a water quality aeration fusion data set, and the equipment working condition monitoring data set and the aeration quantity monitoring data set are fused into a working condition aeration fusion data set. The water quality aeration fusion data set is input into a deep artificial neural network sub-model with adjustable parameters, and the working condition aeration fusion data set is input into a long short-term memory neural network sub-model with adjustable parameters. The results of the two models are fused into a long short-term memory neural network fusion model with adjustable parameters, and are fused according to the weight coefficients after percentage error normalization.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, and specifically to an intelligent aeration control method for wastewater treatment based on a weighted fusion model. Background Technology

[0002] The direct discharge of untreated industrial and domestic wastewater poses serious environmental pollution, damages aquatic ecosystems, and poses public health risks. Untreated wastewater contains large amounts of harmful substances, and its direct discharge pollutes rivers, lakes, and groundwater, leading to water quality deterioration. Pollutants also damage aquatic ecosystems, endangering the survival of aquatic life. Furthermore, polluted water can spread diseases, threatening the health and quality of life of nearby residents. This practice not only causes long-term environmental damage but also incurs enormous social and economic costs.

[0003] Biological treatment and aeration play crucial roles in wastewater treatment. Biological treatment utilizes microorganisms to degrade organic pollutants in wastewater, converting them into harmless substances and thus purifying the water. Aeration, on the other hand, introduces oxygen into the wastewater, enhancing the activity and reproduction rate of microorganisms and improving their efficiency in decomposing organic pollutants. Aeration not only promotes the biological treatment process but also effectively prevents the formation of anaerobic environments in wastewater, reducing the generation of odors and harmful gases.

[0004] Currently, the control of the aeration stage in wastewater biological treatment processes in China mainly relies on manual control. The disadvantages of manual aeration control include low efficiency, slow response speed, and insufficient precision. Manual control requires operators to constantly monitor and adjust, resulting in high workload and difficulty in responding to real-time changes in water quality, leading to unstable aeration effects. Furthermore, precise manual adjustments can easily cause over- or under-aeration, affecting wastewater treatment efficiency. High labor costs and the risk of operational errors also increase management complexity and operating costs.

[0005] In summary, existing technologies suffer from several problems: the aeration stage relies on manual control, which makes it impossible to ensure that the aeration volume can accurately meet the biochemical process; there are also issues such as excessive wastewater discharge causing environmental pollution and the need for adaptive adjustment under abnormal environmental conditions. Therefore, it is necessary to utilize intelligent models to improve the accuracy of control. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent aeration control method for wastewater treatment based on a weighted fusion model, in order to solve the problem that the aeration control in the prior art relies solely on manual control and ignores the guiding role of the intelligent model in the changes of aeration state.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a smart aeration control method for wastewater treatment based on a weighted fusion model, comprising the following steps:

[0008] S100. Obtain influent water quality monitoring dataset by sampling water quality indicators from influent monitoring; obtain process water quality monitoring dataset by sampling water quality indicators from process monitoring; obtain effluent water quality monitoring dataset by sampling water quality indicators from effluent monitoring; obtain aeration volume monitoring dataset by collecting equipment data from the aeration equipment in the aerobic tank of the biological treatment section of the sewage treatment plant; obtain equipment operating condition monitoring dataset by collecting equipment data from the sewage treatment equipment.

[0009] S200: Merge the influent water quality monitoring dataset, process water quality monitoring dataset, effluent water quality monitoring dataset, and aeration volume monitoring dataset into a water quality aeration fusion dataset; merge the equipment operating condition monitoring dataset and aeration volume monitoring dataset into an operating condition aeration fusion dataset.

[0010] S300: Input the water quality aeration fusion dataset into the adjustable parameter deep artificial neural network sub-model; input the operating condition aeration fusion dataset into the adjustable parameter long short-term memory neural network sub-model;

[0011] S400: The results of the two models are fused into a fusion model of an artificial neural network and a long short-term memory network with adjustable parameters. The fusion is performed according to the weight coefficients after the percentage error is normalized. The adjustable parameter adaptive model is embedded in the control system terminal of the sewage treatment aeration equipment.

[0012] S500 controls the wastewater treatment aeration process based on the output results of a fusion model of artificial neural network and long short-term memory network.

[0013] Furthermore, in S100, the influent water quality monitoring dataset, the process water quality monitoring dataset, and the effluent water quality monitoring dataset are historical datasets of the target wastewater treatment plant; the aeration volume monitoring dataset includes data on the aeration volume of the wastewater treatment equipment in the aerobic tank of the process section; and the equipment operating condition monitoring dataset includes data on the anaerobic hydraulic retention time, anoxic hydraulic retention time, and aerobic hydraulic retention time.

[0014] Furthermore, in S200, the method for dataset fusion is to align the data rows and columns of the dataset, and set the data interval to the minimum sampling time.

[0015] Furthermore, in S300, the method for constructing a deep artificial neural network is as follows: First, prepare and preprocess 20% of the dataset, dividing it into a training set, a validation set, and a test set; then, randomly initialize the weights and bias parameters of the network; next, calculate the output of each layer through forward propagation, and then calculate the error between the network output and the actual result according to the loss function; calculate the gradient through the backpropagation algorithm, and update the network parameters using an optimization algorithm; repeat the above process until the percentage error of the loss function is less than 10% and convergence is achieved, or the predetermined number of training rounds of 500 is reached, while outputting adjustable parameters.

[0016] Furthermore, in S300, the method for constructing a long short-term memory neural network is as follows: First, prepare and preprocess 20% of the dataset, dividing it into a training set, a validation set, and a test set; then, randomly initialize the network's weights and bias parameters; next, calculate the output of each cell unit through forward propagation, and then calculate the error between the network output and the actual result according to the loss function; calculate the gradient through the backpropagation algorithm, and update the network parameters using an optimization algorithm; repeat the above process until the percentage error of the loss function converges to less than 10% or reaches the predetermined number of training rounds of 500, while simultaneously outputting adjustable parameters.

[0017] Furthermore, the S500 includes the following steps:

[0018] S510. Set the parameters of the neural network and long short-term memory network fusion model to temporary adjustable parameters, input 100% of the fusion data into the neural network and long short-term memory network fusion model, and the model outputs the recommended aeration volume for the aeration device.

[0019] S520: The control module controls the wastewater treatment aeration equipment in the aerobic tank of the biochemical section of the wastewater treatment aeration process according to the aeration rate recommended by the model.

[0020] S530. Monitor the water quality of the water sample collected through the real-time effluent sampling tube, obtain the effluent water quality, and obtain the effluent water quality status monitoring dataset at time T.

[0021] S540. Determine whether the water quality indicators meet the standards. If the water quality indicators meet the standards, obtain a wastewater discharge instruction to allow discharge.

[0022] The beneficial effects of this invention are:

[0023] 1. The method of the present invention solves the problems of low efficiency, slow response speed and high labor cost of manual control in the prior art, and the need for operators to constantly monitor and adjust, making it difficult to achieve real-time response and precise adjustment, resulting in unstable aeration effect;

[0024] 2. The method of the present invention solves the problems of high labor costs, risks of operational errors and mistakes, and difficulty in achieving 24-hour uninterrupted monitoring, and realizes efficient, intelligent, precise and stable control of the aeration process of sewage treatment. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0026] Figure 2 This is a schematic diagram of the monitoring sites in the method of the present invention;

[0027] Figure 3 This is a structural diagram of the control module of the method of the present invention. Detailed Implementation

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0029] Embodiments of the present invention:

[0030] A smart aeration control method for wastewater treatment based on a weighted fusion model, such as Figure 1 As shown, it includes the following steps:

[0031] S100 monitors indicators such as water quality, equipment operating conditions, and aeration volume.

[0032] SA100 monitors water quality indicators for influent, process section and effluent;

[0033] SA1001: Monitor the water quality of water samples collected through the inlet sampling tube and obtain an inlet water quality monitoring dataset;

[0034] SA10011. Obtain historical influent sampling datasets for the target wastewater treatment plant.

[0035] SA10012. Based on the historical water inflow sampling dataset, perform data preprocessing, including filling missing values ​​and deleting outliers;

[0036] SA10013. Configure the parameters of the water quality monitor according to the fluctuation characteristics of water quality indicators.

[0037] SA1002. Water quality monitoring is performed on water samples collected through the sampling tubes of the biochemical section to obtain process water quality monitoring datasets.

[0038] SA10021. Obtain the historical water quality sampling dataset of the process biochemical section of the target wastewater treatment plant.

[0039] SA10022. Based on the sampled dataset from the past process segment, perform data preprocessing, including filling in missing values ​​and deleting outliers;

[0040] SA10023. Configure the parameters of the water quality monitor according to the fluctuation characteristics of water quality indicators.

[0041] SA1003: Perform water quality monitoring on water samples collected through the effluent sampling tube to obtain an effluent water quality monitoring dataset;

[0042] SA10031. Obtain historical effluent sampling datasets from the target wastewater treatment plant;

[0043] SA10032. Based on the past water outlet section sampling dataset, perform data preprocessing, fill in missing values, and delete outliers;

[0044] SA10033. Configure the parameters of the water quality monitor according to the fluctuation characteristics of water quality indicators.

[0045] Data acquisition requires corresponding data acquisition devices. The water quality data acquisition devices at each site should include at least three water quality monitors to take an average value and reduce errors.

[0046] Specifically, in these embodiments SA1001, SA1002 and SA1003, the target wastewater treatment plant is a professional wastewater treatment institution that uses biochemical treatment methods to treat wastewater. The wastewater undergoes wastewater biochemical treatment processes such as wastewater influent regulation, effluent regulation, anaerobic reaction, anoxic reaction and aerobic reaction in the wastewater treatment equipment of the target wastewater treatment plant.

[0047] SB100: Monitor the aeration rate in the aerobic tank of the biological treatment section;

[0048] Data collection is performed on the aeration volume of wastewater treatment equipment to obtain an aeration volume monitoring dataset. The data types in the aeration volume monitoring dataset include the aeration volume of the wastewater treatment equipment in the aerobic tank of the process section.

[0049] SC100 monitors equipment operating conditions.

[0050] By collecting equipment data from the water treatment equipment, a data set of equipment operating condition monitoring is obtained. Wastewater undergoes various biological treatment processes within the wastewater treatment equipment of the target wastewater treatment plant, including influent regulation, effluent regulation, anaerobic reaction, anoxic reaction, and aerobic reaction. Based on the control system, the anaerobic hydraulic retention time, anoxic hydraulic retention time, aerobic hydraulic retention time, and the timing of adding advanced treatment agents are adjusted and controlled. This achieves a reasonable spatial arrangement of anaerobic, anoxic, and aerobic conditions within the wastewater treatment equipment, ensuring the normal operation of advanced treatment and enabling the effective implementation of these processes.

[0051] Data is collected from the control system used to control the sewage treatment equipment for sewage treatment, and the actual operating data of the sewage treatment equipment is collected to obtain the equipment condition monitoring dataset. The data types in the equipment condition monitoring dataset include anaerobic hydraulic retention time data, anoxic hydraulic retention time data, and aerobic hydraulic retention time data.

[0052] S200. The datasets used for building multiple models are merged into a water quality aeration fusion dataset and an operating condition aeration fusion dataset, respectively.

[0053] The data fusion method involves aligning the data rows and columns of the dataset, and setting the data interval to the minimum sampling time.

[0054] SA200 integrates the influent water quality monitoring dataset, process water quality monitoring dataset, effluent water quality monitoring dataset, and aeration volume monitoring dataset into a water quality aeration fusion dataset.

[0055] SB200 merges the equipment condition monitoring dataset and the aeration volume monitoring dataset into a condition aeration fusion dataset.

[0056] S300: Build an adaptive model, perform model training and loss optimization.

[0057] This includes the SA300 deep artificial neural network sub-model and the SB300 long short-term memory neural network sub-model.

[0058] SA3001. Construct a sub-model of a deep artificial neural network with adjustable parameters. The adjustable parameter adaptive model includes the number of data input batches of the deep artificial neural network, the size of the data input batches of the deep artificial neural network, the number of hidden layers of the neural network, and the dimension of the input data of the neural network.

[0059] SA3002 uses 20% of the water quality aeration fusion dataset as input variables to train a deep artificial neural network and obtain temporary adjustable parameters. This includes inputting the water quality aeration fusion dataset into a deep artificial neural network sub-model, adapting the model based on the adjustable parameters, and outputting adaptive parameters.

[0060] The core structure of a deep artificial neural network (DNN) includes an input layer, hidden layers, an output layer, and activation functions. A typical DNN sub-model consists of one input layer, six hidden layers, one output layer, and six activation functions. Each layer comprises multiple neurons interconnected via weight matrices and bias parameters, enabling information transfer and transformation. The input layer receives raw data and passes it to the hidden layers. Hidden layers typically contain multiple layers, with neurons in each layer using non-linear activation functions to perform complex feature extraction and transformation on the input data. Through non-linear transformations, DNNs can capture complex patterns and relationships within the data. The output layer synthesizes the features extracted by the hidden layers to generate the final prediction or classification label. The layer structure of the DNN sub-model is distributed as follows: first, one input layer; then, a combination of six identical hidden layers and activation functions; and finally, one output layer.

[0061] The construction method of a deep artificial neural network includes the following steps: First, prepare and preprocess 20% of the dataset, dividing it into training, validation, and test sets. Then, randomly initialize the network's weights and bias parameters. Next, compute the output of each layer through forward propagation, and then calculate the error between the network output and the actual result based on the loss function. Calculate the gradient using the backpropagation algorithm, and update the network parameters using an optimization algorithm. Repeat the above process until the percentage error of the loss function converges to less than 10% or a predetermined number of training epochs (500 epochs) is reached, while simultaneously outputting adjustable parameters.

[0062] The deep artificial neural network sub-model is as follows:

[0063]

[0064]

[0065]

[0066] in:

[0067] x i These are the input values, including the influent water quality monitoring dataset, the process water quality monitoring dataset, and the effluent water quality monitoring dataset.

[0068] y i These are the output values, including predicted values ​​from the aeration monitoring dataset;

[0069] p i These are the actual values ​​corresponding to the output values, including the actual values ​​corresponding to the predicted values ​​in the aeration monitoring dataset.

[0070] w i It is the weight of each layer;

[0071] b i It is the offset of each layer;

[0072] g i It is the activation function of each layer;

[0073] m is the number of sample points;

[0074] J(θ) is the loss function;

[0075] θ is the loss;

[0076] α is the learning rate.

[0077] SB3001. Construct a sub-model of a Long Short-Term Memory (LSTM) neural network with adjustable parameters. The adjustable parameter adaptive model includes the number of LSM data input batches, the size of the LSM data input batches, the number of hidden layers in the LSM, the dimension of the input data, and the time step of the LSM.

[0078] SB3002 uses 20% of the water quality operating condition aeration fusion dataset as input variables to train a long short-term memory neural network and obtain temporary adjustable parameters. This includes inputting the operating condition aeration fusion dataset into a long short-term memory neural network sub-model, adapting the model based on the adjustable parameters, and outputting adaptive parameters.

[0079] The core structure of a Long Short-Term Memory (LSTM) neural network consists of memory cell units, each composed of a forget gate, an input gate, and an output gate. Information is transmitted and updated in an orderly manner through these memory cell units. First, the forget gate determines how much of the previous memory to forget based on the current input and the previous hidden state. Then, the input gate combines the current input and the previous hidden state to generate new candidate memories and determines the update ratio of this new information. Next, the memory cell state is updated by combining the forgotten old memories and the newly added memories. Finally, the output gate generates a new hidden state based on the current memory cell state and the input and outputs it, thus achieving efficient transmission and retention of information between time steps. The main structure of a deep LSM neural network includes 128 memory cell units.

[0080] The construction method of a Long Short-Term Memory (LSTM) neural network includes the following steps: First, prepare and preprocess 20% of the dataset, dividing it into training, validation, and test sets. Then, randomly initialize the network's weights and bias parameters. Next, compute the output of each cell layer through forward propagation, and calculate the error between the network output and the actual result based on the loss function. Calculate the gradient using the backpropagation algorithm, and update the network parameters using an optimization algorithm. Repeat the above process until the percentage error of the loss function converges to less than 10% or the predetermined number of training epochs (500 epochs) is reached, while simultaneously outputting adjustable parameters.

[0081] The Long Short-Term Memory (LSTM) neural network sub-model is as follows:

[0082]

[0083] F t =σ(w F x t +w F H t-1 +b F )

[0084] I t =σ(w I x t +w I H t-1 +b I )

[0085]

[0086] C t =F t *C t-1 +I t *tanh(w C x t +w C H t-1 +b C )

[0087] O t =σ(w O x t +w O H t-1 +b O )

[0088] H t =O t *tanh(C t-1 )

[0089] in:

[0090] σ(x) = sigmoid(x) is the sigmoid activation function;

[0091] x t These are the input values, including the equipment condition monitoring dataset;

[0092] F t It is the Gate of Oblivion;

[0093] w F It is the weight matrix of the forget gate, applied to the current input x. t ;

[0094] w F H t-1It is the weight matrix of the forget gate, applied to the hidden state H of the previous time step. t-1 ;

[0095] b F It is the bias term of the forget gate;

[0096] I t It is an input gate;

[0097] w I It is the weight matrix of the input gate, applied to the current input x. t ;

[0098] w I H t-1 It is the weight matrix of the input gate, applied to the hidden state H of the previous time step. t-1 ;

[0099] b I It is the bias term of the input gate;

[0100] tanh(x) is the hyperbolic tangent activation function;

[0101] C t C t-1 It is a memory cell state update;

[0102] w C It is a weight matrix representing the states of memory cells, applied to the current input x. t And the hidden state H of the previous moment t-1 ;

[0103] b C It is the bias term for the state of the memory unit;

[0104] O t It is the output gate, which includes the aeration volume monitoring dataset;

[0105] w O It is the weight matrix of the output gate, applied to the current input x. t ;

[0106] w O H t-1 It is the weight matrix of the output gate, applied to the hidden state H of the previous time step. t-1 ;

[0107] b O It is the bias term of the output gate;

[0108] H t It is in a hidden state.

[0109] The results of the two models are fused into a fusion model of an artificial neural network and a long short-term memory network with adjustable parameters. The fusion is performed using weight coefficients normalized to percentage error. The adjustable parameter adaptive model is embedded in the control system terminal of the wastewater treatment aeration equipment. Based on the output results of the fusion model of the artificial neural network and the long short-term memory network, the wastewater treatment aeration process is controlled.

[0110] S400: The deep artificial neural network sub-model and the long short-term memory network sub-model are fused into a fusion model of artificial neural network and long short-term memory network. The fusion is performed using weight coefficients normalized to percentage error. The adjustable parameter adaptive model is embedded in the control system terminal of the wastewater treatment aeration equipment. The intelligent aeration control system module for wastewater treatment based on the fusion of neural network and long short-term memory network is as follows: Figure 3 As shown.

[0111] The mathematical model for the fusion model of artificial neural networks and long short-term memory networks is as follows:

[0112] 1. Calculate MAPE separately

[0113]

[0114]

[0115] 2. Calculate the weights

[0116]

[0117]

[0118] 3. Calculate the normalized weights

[0119]

[0120]

[0121] 4. Calculate the weighted fusion value

[0122] y i =w1y i1 +w2y i2

[0123] in:

[0124] x i1 These are the input values ​​of the artificial neural network sub-model;

[0125] y i1 It is the output value of the artificial neural network sub-model;

[0126] x i1 These are the input values ​​for the Long Short-Term Memory (LSTM) network sub-model.

[0127] y i1 It is the output value of the Long Short-Term Memory network sub-model;

[0128] p i It is the actual value corresponding to the input value;

[0129] MAPE(p i ,y i1 ) is the mean absolute percentage error of the artificial neural network sub-model;

[0130] MAPE(p i ,y i2 ) is the mean absolute percentage error of the long short-term memory network submodel;

[0131] These are the weights of the sub-model of the artificial neural network;

[0132] These are the weights of the Long Short-Term Memory (LSTM) network sub-model;

[0133] w1 is the normalized weight of the artificial neural network sub-model;

[0134] w2 is the normalized weight of the Long Short-Term Memory network sub-model;

[0135] y i Weighted fusion value.

[0136] S500: Input the fused data into the neural network and long short-term memory network fusion model. Based on the neural network and long short-term memory network fusion model, control the wastewater treatment aeration equipment in the aerobic tank of the biochemical section of the wastewater treatment aeration process according to adaptive parameters.

[0137] S510. Set the parameters of the neural network and long short-term memory network fusion model to temporary adjustable parameters, input 100% of the fusion data into the neural network and long short-term memory network fusion model, and the model outputs the recommended aeration volume for the aeration device.

[0138] S520: The control module controls the wastewater treatment aeration equipment in the aerobic tank of the biochemical section of the wastewater treatment aeration process according to the aeration rate recommended by the model.

[0139] S530. Based on the data acquisition device, water quality monitoring is performed on the water sample collected through the real-time effluent sampling tube to obtain the effluent water quality and obtain the effluent water quality status monitoring dataset at time T.

[0140] S540. Determine whether the water quality indicators meet the national standards for permitted wastewater quality indicators. If the water quality indicators meet the national standards, obtain a wastewater discharge instruction to allow discharge.

[0141] This invention is not limited to the preferred embodiments described above. Anyone can derive other forms of products under the guidance of this invention. However, regardless of any changes made in their shape or structure, any technical solution that is the same as or similar to this application falls within the protection scope of this invention.

Claims

1. A smart aeration control method for wastewater treatment based on a weighted fusion model, characterized by comprising the following steps: S100. Obtain influent water quality monitoring dataset by sampling water quality indicators from influent monitoring; obtain process water quality monitoring dataset by sampling water quality indicators from process monitoring; obtain effluent water quality monitoring dataset by sampling water quality indicators from effluent monitoring; obtain aeration volume monitoring dataset by collecting equipment data from the aeration equipment in the aerobic tank of the biological treatment section of the sewage treatment plant; obtain equipment operating condition monitoring dataset by collecting equipment data from the sewage treatment equipment. S200: Merge the influent water quality monitoring dataset, process water quality monitoring dataset, effluent water quality monitoring dataset, and aeration volume monitoring dataset into a water quality aeration fusion dataset; merge the equipment operating condition monitoring dataset and aeration volume monitoring dataset into an operating condition aeration fusion dataset. S300: Input the water quality aeration fusion dataset into the adjustable parameter deep artificial neural network sub-model; input the operating condition aeration fusion dataset into the adjustable parameter long short-term memory neural network sub-model; S400: The results of the two models are fused into a fusion model of an artificial neural network and a long short-term memory network with adjustable parameters, and the fusion is performed according to the weight coefficients after percentage error normalization. The mathematical model for the fusion model of artificial neural networks and long short-term memory networks is as follows: 1.1 Calculate MAPE separately 1.2 Calculate the weights 1.3 Calculate the normalized weights 1.4 Calculate the weighted fusion value in: These are the input values ​​of the artificial neural network sub-model; It is the output value of the artificial neural network sub-model; These are the input values ​​for the Long Short-Term Memory (LSTM) network sub-model. It is the output value of the Long Short-Term Memory network sub-model; It is the actual value corresponding to the input value; It is the mean absolute percentage error of the artificial neural network sub-model; It is the mean absolute percentage error of the long short-term memory network submodel; These are the weights of the sub-model of the artificial neural network; These are the weights of the Long Short-Term Memory (LSTM) network sub-model; These are the normalized weights of the artificial neural network sub-model; These are the normalized weights of the Long Short-Term Memory (LSTM) network sub-model; Weighted fusion value; Among them, the adjustable parameter artificial neural network and long short-term memory network fusion model is embedded in the control system terminal of the sewage treatment aeration equipment. S500 controls the wastewater treatment aeration process based on the output results of a fusion model of artificial neural network and long short-term memory network.

2. The intelligent aeration control method for wastewater treatment based on a weighted fusion model according to claim 1, characterized in that: In S100, the influent water quality monitoring dataset, process water quality monitoring dataset, and effluent water quality monitoring dataset are historical datasets of the target wastewater treatment plant; the aeration volume monitoring dataset includes data on the aeration volume of the wastewater treatment equipment in the aerobic tank of the process section; and the equipment operating condition monitoring dataset includes data on the anaerobic hydraulic retention time, anoxic hydraulic retention time, and aerobic hydraulic retention time.

3. The intelligent aeration control method for wastewater treatment based on a weighted fusion model according to claim 1, characterized in that: In S200, the method for dataset fusion is to align the data rows and columns of the dataset, and set the data interval to the minimum sampling time.

4. A smart aeration control method for wastewater treatment based on a weighted fusion model according to claim 1, characterized in that: In the S300 described above, the method for constructing a deep artificial neural network is as follows: First, prepare and preprocess 20% of the dataset, dividing it into a training set, a validation set, and a test set; then, randomly initialize the weights and bias parameters of the network; next, calculate the output of each layer through forward propagation, and then calculate the error between the network output and the actual result according to the loss function; calculate the gradient through the backpropagation algorithm, and update the network parameters using an optimization algorithm; repeat the above process until the percentage error of the loss function converges to less than 10% or the predetermined number of training rounds of 500 is reached, while simultaneously outputting adjustable parameters.

5. A smart aeration control method for wastewater treatment based on a weighted fusion model according to claim 1, characterized in that: In the S300 described above, the method for constructing a long short-term memory neural network is as follows: First, prepare and preprocess 20% of the dataset, dividing it into a training set, a validation set, and a test set; then, randomly initialize the weights and bias parameters of the network; next, calculate the output of each cell unit through forward propagation, and then calculate the error between the network output and the actual result according to the loss function; calculate the gradient through the backpropagation algorithm, and update the network parameters using an optimization algorithm; repeat the above process until the percentage error of the loss function is less than 10% and convergence is achieved, or the predetermined number of training rounds of 500 is reached, while simultaneously outputting adjustable parameters.

6. A smart aeration control method for wastewater treatment based on a weighted fusion model according to claim 1, characterized in that, The S500 includes the following steps: S510. Set the parameters of the neural network and long short-term memory network fusion model to temporary adjustable parameters, input 100% of the fusion data into the neural network and long short-term memory network fusion model, and the model outputs the recommended aeration volume for the aeration device. S520: The control module controls the wastewater treatment aeration equipment in the aerobic tank of the biochemical section of the wastewater treatment aeration process according to the aeration rate recommended by the model. S530. Monitor the water quality of the water sample collected through the real-time effluent sampling tube, obtain the effluent water quality, and obtain the effluent water quality status monitoring dataset at time T. S540. Determine whether the water quality indicators meet the standards. If the water quality indicators meet the standards, obtain a wastewater discharge instruction to allow discharge.

Images