Sewage treatment aeration optimization control system and method

Abstract

The invention discloses a sewage treatment aeration optimization control system and a sewage treatment aeration optimization control method, which are used for solving the technical problem of non-ideal sewage treatment effect caused by the existing sewage treatment aeration optimization control system. The system comprises a PLC control cabinet, edge computing equipment, an inflow water quality detection probe, a dissolved oxygen probe, a fan and an aeration valve, the PLC control cabinet is electrically connected with the dissolved oxygen probe, the fan and the aeration valve respectively; the PLC control cabinet is in communication connection with the edge computing device, and the edge computing device is electrically connected with the inlet water quality detection probe; the edge computing device is used for generating an aeration tank dissolved oxygen predicted value according to the preprocessed initial inlet water quality sequence data by adopting a preset long-short-term network; the PLC control cabinet is used for determining target aeration control parameters according to the aeration tank dissolved oxygen predicted value and the aeration tank dissolved oxygen true value; the target aeration control parameters are used for adjusting the blower motor frequency of the fan and the aeration valve opening of the aeration valve.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage treatment, and in particular to a sewage treatment aeration optimization control system and method. Background Art

[0002] In recent years, people's living standards have improved rapidly, and they have higher requirements for water quality treatment. In addition, a large number of sewage treatment plants have been built, which has also made sewage discharge standards more stringent. The improvement of discharge standards has further increased the energy consumption in the sewage treatment process, making the energy consumption of sewage treatment plants more and more important.

[0003] Sewage treatment refers to the process of purifying sewage by physical, chemical, biological and other methods. The pollutants contained in the sewage are separated, removed, recycled, or converted into harmless substances. According to the principle, it is mainly divided into physical treatment method, chemical treatment method and biological treatment method. At present, the common aeration control method in the field of sewage treatment is to set the dissolved oxygen concentration DO target vector element (Dissolved Oxygen Concentration) by adjusting the air volume of the fan, the opening of the aeration valve and the aeration time.

[0004] Most existing sewage treatment aeration optimization control systems ensure water quality through manual adjustment and setting of process parameters. However, this process relies too much on manual adjustment and requires operators to invest a lot of time and energy in monitoring and adjustment, making it difficult for the system to respond quickly to changes in water quality, resulting in unsatisfactory sewage treatment results. Summary of the invention

[0005] The present invention provides a sewage treatment aeration optimization control system and method, which are used to solve the technical problem that the existing sewage treatment aeration optimization control system leads to unsatisfactory sewage treatment effect.

[0006] A first aspect of the present invention provides a sewage treatment aeration optimization control system, the system comprising a PLC control cabinet, an edge computing device, an inlet water quality detection probe, a dissolved oxygen probe, a fan and an aeration valve;

[0007] The PLC control cabinet is electrically connected to the dissolved oxygen probe, the fan and the aeration valve respectively;

[0008] The PLC control cabinet is communicatively connected to the edge computing device, and the edge computing device is electrically connected to the inlet water quality detection probe;

[0009] The inlet water quality detection probe is used to obtain initial inlet water quality sequence data and transmit it to the edge computing device;

[0010] The dissolved oxygen probe is used to obtain the actual value of dissolved oxygen in the aeration tank and transmit it to the PLC control cabinet;

[0011] The edge computing device is used to preprocess the initial influent water quality sequence data, output the target influent water quality sequence data, and use a preset long-term and short-term network to generate a predicted value of dissolved oxygen in the aeration tank according to the target influent water quality sequence data and transmit it to the PLC control cabinet;

[0012] The PLC control cabinet is used to determine the target aeration control parameter according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the target aeration control parameter is used to adjust the blower motor frequency of the fan and the aeration valve opening of the aeration valve.

[0013] Furthermore, the edge computing device is specifically used for:

[0014] Performing data cleaning on the initial influent water quality sequence data, and outputting intermediate influent water quality sequence data;

[0015] Performing outlier detection on the intermediate inlet water quality sequence data to generate smoothed inlet water quality sequence data;

[0016] Performing missing value processing and normalization on the smoothed influent water quality sequence data to generate normalized influent water quality sequence data;

[0017] Performing data conversion on the normalized influent water quality sequence data based on a preset sliding window method to generate target influent water quality sequence data;

[0018] The target influent water quality sequence data is input into the preset long-term and short-term network, and the predicted value of dissolved oxygen in the aeration tank is generated and transmitted to the PLC control cabinet.

[0019] Furthermore, the preset long-term and short-term network includes an input layer, a plurality of hidden layers and an output layer; the data processing process of the preset long-term and short-term network includes:

[0020] Performing feature extraction on the target influent water quality sequence data received by the input layer through the multiple hidden layers to generate hidden layer states;

[0021] The output layer is used to predict the target hidden state in the hidden layer state to generate a predicted value of dissolved oxygen in the aeration tank.

[0022] Furthermore, the hidden layer includes an input gate, a forget gate, a candidate memory unit, an update memory unit, and an output gate; and the data processing process of the hidden layer includes:

[0023] Update the input water quality sequence data input by the input gate to generate an input state;

[0024] The forget gate is used to select information from the input water quality sequence data, and a forget state is output;

[0025] Performing candidate memory construction on the input inlet water quality sequence data through the candidate memory unit to generate a candidate memory state;

[0026] Extracting information from the input inlet water quality sequence data through the output gate to generate an output state;

[0027] Using the update memory unit to update the forgetting state, the candidate memory state and the input state, and output a target memory state;

[0028] An output hidden state is generated according to the target memory state and the output state.

[0029] Furthermore, the model training process of the preset long-term and short-term network includes:

[0030] Acquire the inlet water quality sequence data of the source facility to be trained and the inlet water quality sequence data of the target facility to be trained, pre-process the inlet water quality sequence data of the source facility to be trained, and output the inlet water quality sequence data of the target facility to be trained;

[0031] Inputting the target influent water quality sequence data to be trained into the initial long-short term network, the hidden layer state to be trained and the predicted value of dissolved oxygen in the aeration tank to be trained;

[0032] Determine the hidden layer gradient and the output layer gradient based on the hidden layer state to be trained and the predicted value of dissolved oxygen in the aeration tank to be trained;

[0033] Based on a preset adaptive moment estimation optimizer, the model parameters of the hidden layer and the model parameters of the output layer in the initial long-short term network are updated using the hidden layer gradient and the output layer gradient, an intermediate long-short term network is determined, and the number of iterations is counted in real time;

[0034] Using the intermediate long-short term network and the improved mean square error loss function, the target loss value is calculated according to the target inlet water quality sequence data to be trained;

[0035] Determine whether the number of iterations reaches a preset iteration threshold and whether the target loss value converges;

[0036] If the number of iterations reaches the preset iteration threshold or the target loss value converges, the intermediate long-short-term network is used as the long-short-term network to be fine-tuned;

[0037] Based on the preset target domain loss function, the water quality sequence data of the target facility to be trained is used to perform local model training on the output layer in the long-term and short-term network to be fine-tuned, and the trained preset long-term and short-term network is determined.

[0038] Furthermore, the target aeration control parameters include a target blower motor frequency and a target aeration valve opening; the PLC control cabinet is specifically used for:

[0039] Performing a difference operation on the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank, and outputting a deviation value of dissolved oxygen in the aeration tank;

[0040] Performing proportional-integral-differential adjustment on the dissolved oxygen deviation value of the aeration tank to determine the proportional-integral-differential control value;

[0041] Based on the proportional-integral-differential control value, determining the blower motor frequency control value and the aeration valve opening control value;

[0042] The blower motor frequency control value and the aeration valve opening control value are used to update the blower motor frequency of the fan and the aeration valve opening of the aeration valve, and determine a target blower motor frequency and a target aeration valve opening.

[0043] A second aspect of the present invention provides a sewage treatment aeration optimization control method, which is applied to the above-mentioned sewage treatment aeration optimization control system, comprising:

[0044] Obtain the initial influent water quality sequence data and the actual value of dissolved oxygen in the aeration tank;

[0045] Preprocessing the initial influent water quality sequence data, outputting target influent water quality sequence data, and using a preset long-term and short-term network to generate a predicted value of dissolved oxygen in an aeration tank according to the target influent water quality sequence data;

[0046] A target aeration control parameter is determined according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the target aeration control parameter is used to adjust the blower motor frequency of the fan and the aeration valve opening of the aeration valve.

[0047] A third aspect of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the sewage treatment aeration optimization control method as described above.

[0048] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed, the steps of the wastewater treatment aeration optimization control method as described above are implemented.

[0049] A fifth aspect of the present invention provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer executes the steps of the sewage treatment aeration optimization control method as described above.

[0050] It can be seen from the above technical solutions that the present invention has the following advantages:

[0051] The first aspect of the above technical solution of the present invention provides a sewage treatment aeration optimization control system, which includes a PLC control cabinet, an edge computing device, an inlet water quality detection probe, a dissolved oxygen probe, a fan and an aeration valve; wherein the PLC control cabinet is electrically connected to the dissolved oxygen probe, the fan and the aeration valve respectively; the PLC control cabinet is communicatively connected to the edge computing device, and the edge computing device is electrically connected to the inlet water quality detection probe; the inlet water quality detection probe is used to obtain initial inlet water quality sequence data and transmit it to the edge computing device; the dissolved oxygen probe is used to obtain the true value of dissolved oxygen in the aeration tank and transmit it to the PLC control cabinet; the edge computing device is used to pre-process the initial inlet water quality sequence data, output the target inlet water quality sequence data, and use a preset long-term and short-term network to generate the aeration tank dissolved oxygen prediction value according to the target inlet water quality sequence data and transmit it to the PLC control cabinet; The LC control cabinet is used to determine the target aeration control parameters according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the target aeration control parameters are used to adjust the blower motor frequency of the fan and the aeration valve opening of the aeration valve; based on the above scheme, the edge computing device uses a preset long-term and short-term network to process the initial influent water quality sequence data after preprocessing to generate the predicted value of dissolved oxygen in the aeration tank, and combines with the PLC control cabinet to determine the target aeration control parameters according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the blower motor frequency of the fan and the aeration valve opening of the aeration valve are adjusted by the target aeration control parameters to realize the aeration optimization process. The present invention automatically adjusts the process parameters according to the acquired influent water quality data and the actual value of dissolved oxygen in the aeration tank, which can reduce manual intervention and enable the system to quickly respond to water quality changes, thereby improving the sewage treatment effect.

[0052] The second aspect of the above technical solution of the present invention provides a sewage treatment aeration optimization control method, which is applied to the sewage treatment aeration optimization control system proposed by the present invention. First, the initial influent water quality sequence data and the actual value of dissolved oxygen in the aeration tank are obtained; then, the initial influent water quality sequence data are preprocessed, and the target influent water quality sequence data are output, and a preset long-term and short-term network is used to generate an aeration tank dissolved oxygen prediction value according to the target influent water quality sequence data; finally, the target aeration control parameter is determined according to the aeration tank dissolved oxygen prediction value and the aeration tank dissolved oxygen actual value; the target aeration control parameter is used to adjust the blower motor frequency and aeration valve opening of the aeration valve; based on the above scheme, a preset long-term and short-term network is used to process the initial influent water quality sequence data after preprocessing to generate an aeration tank dissolved oxygen prediction value, and according to the aeration tank dissolved oxygen prediction value and the aeration tank dissolved oxygen actual value, a target aeration control parameter is determined; the blower motor frequency of the fan and the aeration valve opening of the aeration valve are adjusted by the target aeration control parameter to realize the aeration optimization process. The present invention automatically adjusts the process parameters according to the acquired influent water quality data and the aeration tank dissolved oxygen actual value, which can reduce manual intervention and enable the system to quickly respond to water quality changes, thereby improving the sewage treatment effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0054] Figure 1 A schematic diagram of the structure of a sewage treatment aeration optimization control system provided in Embodiment 1 of the present invention;

[0055] Figure 2 This is an application framework diagram of the sewage treatment aeration optimization control system provided in the first embodiment of the present invention;

[0056] Figure 3 A flowchart of a sewage treatment aeration optimization control method provided in Embodiment 2 of the present invention;

[0057] The meanings of the symbols in the accompanying drawings are as follows:

[0058] 1. PLC control cabinet; 2. Edge computing equipment; 3. Inlet water quality detection probe; 4. Dissolved oxygen probe; 5. Fan; 6. Aeration valve. DETAILED DESCRIPTION

[0059] The embodiment of the present invention provides a sewage treatment aeration optimization control system and method, which are used to solve the technical problem that the existing sewage treatment aeration optimization control system leads to unsatisfactory sewage treatment effect.

[0060] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] Terminology explanation:

[0062] Transfer learning: Transfer learning is a machine learning method that improves learning efficiency and model performance by applying knowledge learned in one domain (source domain) to another domain (target domain).

[0063] LSTM: Long Short Term Memory Network (LSTM), a special type of Recurrent Neural Network (RNN), is designed to solve the gradient vanishing and gradient exploding problems faced by traditional RNN when processing long sequence data. It can effectively capture long-term dependencies in the sequence and is suitable for time series prediction.

[0064] Aeration: It is a technology commonly used in water treatment and wastewater treatment. Its main purpose is to increase the dissolved oxygen concentration in the water by introducing air or oxygen into the water body, allowing microorganisms in the water to treat pollutants.

[0065] DO: The abbreviation of “Dissolved Oxygen” refers to the oxygen content dissolved in water. Dissolved oxygen is an important water quality indicator in aquatic ecosystems and is essential to the survival of aquatic organisms and the health of water bodies.

[0066] Feedback control: Feedback control is a fundamental concept of control systems and is widely used in engineering, automation, robotics, aerospace, etc. It monitors the output of the system and compares it with the desired target, thereby adjusting the input of the system to achieve the desired performance or behavior.

[0067] PID control: PID control (Proportional-Integral-Derivative Control) is a feedback control strategy widely used in industrial control systems. It achieves precise control of system output by weighting different components of the error. The PID controller is designed to improve the stability, response speed and accuracy of the system.

[0068] PLC: PLC (Programmable Logic Controller) is an electronic device specially used for industrial automation control. It can perform various control tasks through programming and is widely used in production lines, mechanical equipment, building automation and other fields.

[0069] Edge computing devices: Edge computing devices are hardware devices that process and analyze data closer to the data source or user. These devices are able to perform calculations at or near the location where the data is generated, rather than sending all data to a remote data center or cloud for processing. The main purpose of edge computing is to reduce latency, improve response speed, reduce bandwidth requirements, and enhance data security.

[0070] Source Tasks and Target Tasks: Source tasks are tasks performed in an existing or newly built small-scale sewage treatment facility or system, and are usually used to train and optimize models. In this stage, the system collects a large amount of water quality data and operating parameters in order to establish an effective prediction model. Target tasks refer to applying the trained model to a new sewage treatment facility or system. In this stage, the model needs to be fine-tuned and applied in different environments and conditions to achieve practical control and optimization.

[0071] See also Figure 1 , Figure 1 A schematic diagram of the structure of a sewage treatment aeration optimization control system provided in Example 1 of the present invention.

[0072] The present invention provides a sewage treatment aeration optimization control system, the system includes a PLC control cabinet 1, an edge computing device 2, an inlet water quality detection probe 3, a dissolved oxygen probe 4, a fan 5 and an aeration valve 6;

[0073] The PLC control cabinet 1 is electrically connected to the dissolved oxygen probe 4, the fan 5 and the aeration valve 6 respectively;

[0074] The PLC control cabinet 1 is communicatively connected to the edge computing device 2, and the edge computing device 2 is electrically connected to the inlet water quality detection probe 3;

[0075] The inlet water quality detection probe 3 is used to obtain the initial inlet water quality sequence data and transmit it to the edge computing device 2;

[0076] The dissolved oxygen probe 4 is used to obtain the actual value of dissolved oxygen in the aeration tank and transmit it to the PLC control cabinet 1;

[0077] The edge computing device 2 is used to pre-process the initial influent water quality sequence data, output the target influent water quality sequence data, and use the preset long-term and short-term network to generate the aeration tank dissolved oxygen prediction value according to the target influent water quality sequence data and transmit it to the PLC control cabinet 1;

[0078] The PLC control cabinet 1 is used to determine the target aeration control parameter according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the target aeration control parameter is used to adjust the blower motor frequency of the fan 5 and the aeration valve opening of the aeration valve 6.

[0079] It should be noted that in the target task, the input data collected in real time by the inlet water quality detection probe 3, that is, the initial inlet water quality sequence data The characteristic types of the initial influent water quality sequence data include water quality indicators COD, ammonia nitrogen, TN, TP, DO, pH, flow rate, and temperature; the actual value of dissolved oxygen in the aeration tank obtained by the dissolved oxygen probe 4 is transmitted to the PLC control cabinet 1.

[0080] As a further improvement, the edge computing device 2 is specifically used for:

[0081] Perform data cleaning on the initial influent water quality sequence data and output the intermediate influent water quality sequence data;

[0082] Perform outlier detection on the intermediate influent water quality sequence data to generate smoothed influent water quality sequence data;

[0083] Process missing values ​​and normalize the smoothed influent water quality series data to generate normalized influent water quality series data;

[0084] Based on the preset sliding window method, the normalized influent water quality sequence data is converted to generate the target influent water quality sequence data;

[0085] The target influent water quality sequence data is input into the preset long-term and short-term network to generate the predicted dissolved oxygen value of the aeration tank and transmit it to the PLC control cabinet.

[0086] It should be noted that for data preprocessing, the data are cleaned, outlier detected, missing value processed and standardized / normalized in turn to eliminate noise and maintain data consistency. Specifically, after the initial influent water quality sequence data is cleaned, 3 Principle detection of outliers, upper and lower limits = ,in is the mean, The standard deviation is used to replace outliers with values ​​at adjacent time points, and duplicate records are deleted to ensure the independence of the data. Next, a binary function is used to mark the position of missing values ​​in the data set. The principle of the missing value processing steps can be expressed as follows:

[0087] ;

[0088] in, A matrix to identify whether a data item is missing; For data items.

[0089] Furthermore, missing values ​​are filled forward, and the process can be expressed as: .

[0090] Furthermore, the data processed with missing values ​​are normalized to obtain normalized influent water quality series data; the normalization formula can be expressed as:

[0091] ;

[0092] in, is the data after normalization; x is the data before normalization; is the smallest data item in the original data set involved in normalization; is the largest data item in the original data set involved in normalization.

[0093] Furthermore, the sliding window method (preset sliding window method) is used to convert the time series data (normalized inlet water quality series data) into a format suitable for LSTM (preset long short-term network) input. For the given time series data (normalized inlet water quality series data) , set the window size, and then get the target inlet water quality sequence data, which includes multiple input sequences and target vector elements corresponding to each input sequence:

[0094] ;

[0095] As a further improvement, the preset long-short term network includes an input layer, a plurality of hidden layers and an output layer; the data processing process of the preset long-short term network includes:

[0096] S11, extracting features of the target influent water quality sequence data received by the input layer through multiple hidden layers to generate hidden layer states;

[0097] S12. Use the output layer to predict the target hidden state in the hidden layer state to generate a predicted value of dissolved oxygen in the aeration tank.

[0098] The hidden layer state also includes the input state, forgotten state, candidate memory state, output state and target memory state outputted by the last hidden layer in all hidden layers at the last moment.

[0099] It should be noted that the output layer is used to predict the target hidden state to generate the predicted value of dissolved oxygen in the aeration tank, where the target hidden state is the hidden state output by the last hidden layer in all hidden layers at the last moment. The prediction process of the output layer can be expressed as:

[0100] ;

[0101] in, is the predicted value of dissolved oxygen in the aeration tank; is the weight matrix of the output layer; is the target hidden state at time t; is the bias of the output layer.

[0102] As a further improvement, the hidden layer includes an input gate, a forget gate, a candidate memory unit, an update memory unit, and an output gate; the data processing process of the hidden layer includes:

[0103] S21, updating the input water quality sequence data through the input gate to generate an input state;

[0104] It should be noted that the data processing process of the input gate can be expressed as:

[0105] ;

[0106] in, is the input state at time t; is the sigmoid activation function; is the weight matrix of the input gate; is the hidden state at time t-1, indicating the target hidden state at the previous moment; is the current input vector, i.e. the target influent water quality sequence data; is the bias of the input gate.

[0107] S22, using a forget gate to select information from the input water quality sequence data, and outputting a forget state;

[0108] It should be noted that the data processing process of the forget gate can be expressed as:

[0109] ;

[0110] in, is the forgetting state at time t; is the weight matrix of the forget gate; is the bias of the forget gate; is the sigmoid activation function; is the hidden state at time t-1, indicating the target hidden state at the previous moment.

[0111] S23, constructing candidate memory for the input inlet water quality sequence data through the candidate memory unit to generate a candidate memory state;

[0112] It should be noted that the data processing process of the candidate memory unit can be expressed as:

[0113] ;

[0114] in, is the candidate memory state at time t; is the weight matrix of the candidate memory unit; is the bias of the candidate memory unit; is the hidden state at time t-1, indicating the target hidden state at the previous moment; tanh is the hyperbolic tangent activation function.

[0115] S24, extracting information from the input inlet water quality sequence data through the output gate to generate an output state;

[0116] It should be noted that the data processing process of the output gate can be expressed as:

[0117] ;

[0118] in, is the output state at time t; is the weight matrix of the output gate; is the bias of the output gate; is the sigmoid activation function; is the hidden state at time t-1, indicating the target hidden state at the previous moment.

[0119] S25, using the update memory unit to update the forgetting state, the candidate memory state and the input state, and outputting the target memory state;

[0120] It should be noted that the data processing process of updating the memory unit can be expressed as:

[0121] ;

[0122] in, is the target memory state at time t; is the forgetting state at time t; is the target memory state at time t-1; is the input state at time t; is the candidate memory state at time t.

[0123] S26. Generate an output hidden state according to the target memory state and the output state.

[0124] It should be noted that the process of outputting the hidden state can be expressed as:

[0125] ;

[0126] in, is the output hidden state at time t; is the target memory state at time t; is the output state at time t; tanh is the hyperbolic tangent activation function.

[0127] It is worth mentioning that the input inlet water quality sequence data is the sequence data input into the hidden layer. It can be understood that the input inlet water quality sequence data can correspond to any sequence data input into the hidden layer for data processing during the model training or prediction process; the output hidden state is the sequence data output by the hidden layer. It can be understood that it can correspond to any sequence data output after the hidden layer performs data processing during the model training or prediction process, wherein the output hidden state output by the last hidden layer among all hidden layers at the last moment is used as the target hidden state.

[0128] As a further improvement, the model training process of the preset long-term and short-term networks includes:

[0129] S31, obtaining the inlet water quality sequence data of the source facility to be trained and the inlet water quality sequence data of the target facility to be trained, preprocessing the inlet water quality sequence data of the source facility to be trained, and outputting the inlet water quality sequence data of the target facility to be trained;

[0130] It should be noted that in the source sewage treatment facility, sensors are used to collect input features for model training, that is, the influent water quality sequence data of the source facility to be trained, including the influent water quality indicators COD, ammonia nitrogen, TN, TP, DO, pH, flow rate, and temperature, recorded as , where N is the number of samples and S is the characteristic type, including the influent water quality indicators COD, ammonia nitrogen, TN, TP, DO, pH, flow rate and temperature.

[0131] Furthermore, a small amount of operation data (sequence data of water quality of the target facility to be trained) is collected. ,in, is the first inlet water quality data in the inlet water quality sequence data of the target facility to be trained at time t, is the second inlet water quality data in the inlet water quality sequence data of the target facility to be trained at time t, is the Mth influent water quality data in the influent water quality sequence data of the target facility to be trained at time t, M is the total number of influent water quality data, M≤N.

[0132] S32, inputting the target influent water quality sequence data to be trained into the initial long-short term network, the hidden layer state to be trained and the predicted value of dissolved oxygen in the aeration tank to be trained;

[0133] It should be noted that the hidden layer state to be trained is the output of the last hidden layer in the initial long-short term network at the last moment, and the output at the last moment includes the input state, the forgotten state, the candidate memory state, the output state, the target memory state and the target hidden state.

[0134] S33, determining the hidden layer gradient and the output layer gradient based on the hidden layer state to be trained and the predicted value of dissolved oxygen in the aeration tank to be trained;

[0135] Hidden layer gradients include output gate gradients, unit state gradients, forget gate gradients, input gate gradients, and candidate unit state gradients.

[0136] It should be noted that the calculation of the gradient of the loss function for each model parameter based on back propagation involves the calculation of the partial derivatives of the parameters of each layer in the LSTM network in order to update the parameters; among them, the calculation formula of the gradient of the hidden layer when back propagating to the hidden layer can be expressed as:

[0137] ;

[0138] ;

[0139] ;

[0140] ;

[0141] ;

[0142] in, is the output gate gradient; is the unit state gradient; is the forget gate gradient; is the input gate gradient; is the candidate unit state gradient; To represent the improved mean square error loss function for the target hidden state h of the LSTM model at time step t t The partial derivative of .

[0143] Furthermore, the calculation process of the output layer gradient can be expressed as:

[0144] ;

[0145] in, is the output layer gradient; L is the improved mean square error loss function; is the predicted value of dissolved oxygen in the aeration tank to be trained; y is the actual value of dissolved oxygen in the aeration tank.

[0146] Furthermore, the improved mean square error loss function (Huber loss function) is as follows:

[0147] ;

[0148] in, is the target loss value; is the true value of dissolved oxygen in the i-th aeration tank; is the predicted value of dissolved oxygen in the i-th aeration tank to be trained; is a hyperparameter, The value is selected as 5% to 10% of the target variable value range, taking 0.1-0.4, and selecting The values ​​are [0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4], train the model, evaluate the model performance at each value, and choose the value that performs best on the validation set.

[0149] It is worth mentioning that since LSTM processes sequence data, the gradients at the time steps need to be accumulated. This means that when calculating the gradient, the gradient of the current time step needs to be added to the gradient passed back from the future time step.

[0150] S34, based on the preset adaptive moment estimation optimizer, using the hidden layer gradient and the output layer gradient to update the model parameters of the hidden layer and the model parameters of the output layer in the initial long-short term network, determine the intermediate long-short term network, and count the number of iterations in real time;

[0151] It should be noted that the calculated gradient is used to update the model parameters according to the rules of the Adam optimizer (preset adaptive moment estimation optimizer). For each parameter, the update rule of Adam is as follows:

[0152] First moment estimate (mean), ;

[0153] Second moment estimate (variance), ;

[0154] bias correction, , ;

[0155] Update parameters, ;

[0156] in, is the learning rate, the value is 0.001; is the first hyperparameter, with a value of 0.9; is the second hyperparameter, with a value of 0.999; It is a constant to prevent division by zero error, and its value is 10 -8 ; is the gradient, including the hidden layer gradient and the output layer gradient; are the updated model parameters, including weight matrix and bias; are the model parameters before updating; is the bias-corrected mean; is the bias-corrected variance.

[0157] It is worth mentioning that the hidden layer gradient and the output layer gradient update the model parameters of the corresponding network gates in turn according to the Adam update rule. Among them, the update process of the model parameters of the output layer is: based on the output layer gradient, the first-order moment estimate (mean value) and the second-order moment estimate (variance) are calculated, and the bias-corrected mean value and the bias-corrected variance are calculated according to the first-order moment estimate (mean value) and the second-order moment estimate (variance). Then, the model parameters (weight matrix and bias) of the output layer are updated according to the bias-corrected mean value and the bias-corrected variance to determine the updated output layer. The hidden layer gradient updates the corresponding model parameters of the input gate, forget gate, candidate memory unit, update memory unit, and output gate based on the above-mentioned update principle of the model parameters of the output layer, thereby obtaining the intermediate long-term and short-term network.

[0158] S35, using the intermediate long-short term network and the improved mean square error loss function to calculate the target loss value according to the target inlet water quality sequence data to be trained;

[0159] It should be noted that the target inlet water quality sequence data to be trained is input into the intermediate long-short term network to obtain a new predicted value of dissolved oxygen in the aeration tank to be trained, and then the new predicted value of dissolved oxygen in the aeration tank to be trained is substituted into the improved mean square error loss function to calculate the target loss value.

[0160] S36, judging whether the number of iterations reaches a preset iteration threshold and whether the target loss value converges;

[0161] S37, if the number of iterations reaches a preset iteration threshold or the target loss value converges, the intermediate long-short-term network is used as the long-short-term network to be fine-tuned;

[0162] It should be noted that if the number of iterations does not reach the preset iteration threshold and the target loss value does not converge, the intermediate long-short-term network is used as the new initial long-short-term network, and the process jumps to S32 until the number of iterations reaches the preset iteration threshold or the target loss value converges, and the intermediate long-short-term network determined when the number of iterations reaches the preset iteration threshold or the target loss value converges is used as the long-short-term network to be fine-tuned.

[0163] S38. Based on the preset target domain loss function, the water quality sequence data of the target facility to be trained is used to perform local model training on the output layer in the long-term and short-term network to be fine-tuned, and the trained preset long-term and short-term network is determined.

[0164] It should be noted that the model parameters of the input layer and the hidden layer in the fine-tuned long-term and short-term network are frozen, and only the output layer in the fine-tuned long-term and short-term network is subjected to local model training to adapt to the data distribution of the target facility.

[0165] Among them, the parameters of the intermediate long and short-term networks determined when the number of iterations reaches the preset iteration threshold or the target loss value converges As the initial parameters of the long-term and short-term network to be fine-tuned , minimize the following loss function in the target domain, that is, the preset target domain loss function, specifically:

[0166] ;

[0167] in, is the loss value corresponding to the preset target domain loss function; is the predicted value of dissolved oxygen in the i-th aeration tank at time t of the long-term and short-term network output to be fine-tuned; is the true value of dissolved oxygen in the ith aeration tank at time t; is the regularization coefficient, used to prevent overfitting; is the regularization term.

[0168] It is worth mentioning that the principles of the steps for local model training of the output layer in the long-term and short-term network to be fine-tuned are consistent with the principles of the steps for model training of the initial long-term and short-term network mentioned above, but the local model training of the output layer in the long-term and short-term network to be fine-tuned only updates the model parameters of the output layer, and the training stopping condition is that the number of iterations reaches a threshold or the loss value corresponding to the preset target domain loss function, which will not be elaborated in the present invention.

[0169] As a further improvement, the target aeration control parameters include the target blower motor frequency and the target aeration valve opening; the PLC control cabinet is specifically used for:

[0170] Perform difference calculation on the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank, and output the deviation value of dissolved oxygen in the aeration tank;

[0171] Perform proportional-integral-differential adjustment on the aeration tank dissolved oxygen deviation value to determine the proportional-integral-differential control value;

[0172] Based on the proportional-integral-differential control value, the blower motor frequency control value and the aeration valve opening control value are determined;

[0173] The blower motor frequency of the fan and the aeration valve opening of the aeration valve are updated using the blower motor frequency control value and the aeration valve opening control value to determine a target blower motor frequency and a target aeration valve opening.

[0174] It should be noted that the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank are differentially calculated to output the dissolved oxygen deviation value e(t) of the aeration tank. Then, the blower motor frequency and the aeration valve opening are controlled by the PID controller in the PLC control cabinet. The output u(t) (proportional integral differential control value) of the PID controller is expressed as:

[0175] ;

[0176] in, is the proportional-integral-derivative control value; is the proportionality coefficient; is the dissolved oxygen deviation value of the aeration tank; is the integration coefficient; is the differential coefficient.

[0177] Furthermore, the processing of the blower motor frequency control value and the aeration valve opening control value can be expressed as:

[0178] ;

[0179] in, is the blower motor frequency control value; is the proportional-integral-derivative control value; is a proportional coefficient in [0,1], used to distribute the control output to the blower and valve; is the aeration valve opening control value.

[0180] Furthermore, the update process of the target blower motor frequency and the target aeration valve opening is specifically as follows:

[0181] ;

[0182] in, is the target blower motor frequency at time t; is the blower motor frequency at time t-1, indicating the target blower motor frequency at the previous moment; is the aeration valve opening at time t-1, indicating the target aeration valve opening at the previous moment; is the target aeration valve opening at time t.

[0183] It is worth mentioning that for the adjustment of the PID parameters of the PID control, the present invention adopts the Ziegler-Nichols method, specifically:

[0184] ;

[0185] in, is the critical coefficient; is the oscillation period.

[0186] For example, see Figure 2 The present invention collects inlet water quality data through the inlet water quality detection probe 3, transmits the inlet water quality data to the edge computing device 2, uses the preset long-term and short-term network for prediction, outputs the predicted DO concentration value and transmits it to the PLC control cabinet, the PLC control cabinet receives the actual value of dissolved oxygen in the aeration tank obtained by the dissolved oxygen probe 4 and the predicted DO concentration value, and performs difference calculation, PID adjustment, and updates the working parameters, thereby obtaining the blower motor frequency and the aeration valve opening, and transmits the blower motor frequency to the blower inverter in the fan, thereby adjusting the fan, and adjusting the aeration flow valve by using the aeration valve opening to achieve optimal control of the aeration amount, and then dynamically adjusting the frequency of the fan motor and the opening of the aeration valve in real time, thereby achieving aeration energy saving.

[0187] As a comparison of technical effects, it can be combined with existing technologies for reference. With the global shortage of water resources and increasingly severe environmental protection requirements, energy conservation and efficient operation of sewage treatment facilities have become important issues in the industry. Traditional sewage treatment systems rely on manual adjustment and setting of process parameters to ensure qualified water quality, but this method has problems such as high energy consumption, low reaction efficiency, and complex management, which is particularly prominent in large-scale facilities. In the aeration stage, traditional sewage treatment processes often face the problem of excessive energy consumption and difficulty in precise control. In addition, sewage treatment facilities vary greatly, especially sewage treatment plants in remote areas, which often face the challenges of insufficient operating data and lack of real-time monitoring and debugging methods, making it difficult to accurately optimize operating parameters and effectively reduce energy consumption.

[0188] In this context, how to ensure that the effluent quality is qualified and achieve energy conservation and emission reduction has become the key to technological innovation. In recent years, deep learning technology, especially LSTM (Long Short-Term Memory Network), has been widely used in the field of sewage treatment due to its excellent time series prediction capabilities. These models can model and predict water quality changes, load fluctuations and equipment status, thereby optimizing the operation and energy saving of facilities. However, there are still some problems with existing technologies, especially when facing sewage treatment facilities of different sizes. How to effectively migrate existing optimization models and how to accurately train and adjust models when data is insufficient are still difficulties that need to be solved.

[0189] In existing technical solutions, the application of LSTM models is mainly concentrated in water quality prediction, load fluctuation modeling and equipment status monitoring. LSTM models can effectively capture complex patterns in time series data, helping operators to adjust process parameters in a timely manner and optimize the sewage treatment process. However, existing technologies still have shortcomings in model migration and training under insufficient data. Faced with sewage treatment facilities of different sizes, how to effectively migrate existing optimization models to new environments, and how to accurately train and adjust models in the absence of sufficient data are still challenges facing the industry.

[0190] Based on the above foundation, the traditional sewage treatment system relies on manual adjustment and setting of process parameters, resulting in excessive energy consumption in the aeration and reflow stages and low treatment efficiency. This high energy consumption not only increases operating costs, but also burdens the environment. At the same time, the reliance on manual adjustment makes it difficult for the system to respond quickly to changes in water quality, resulting in unstable treatment effects. Operators need to invest a lot of time and energy in monitoring and adjustment, which increases the complexity of management. Especially in large sewage treatment plants in remote areas, they often face problems such as insufficient operating data and lack of real-time monitoring methods. This makes it difficult for operators to obtain enough data to optimize operating parameters, affecting the overall performance of the system; the existing technology has poor adaptability of the model when facing sewage treatment facilities of different sizes, and it is difficult to effectively migrate the existing optimization model. This leads to a significant reduction in the model's predictive ability and optimization effect in the new environment; the existing technology lacks an intelligent decision support system, making it difficult to respond quickly based on real-time data, and unable to adjust the aeration volume and other process parameters in time, thus affecting the treatment effect.

[0191] In view of the above problems, the present invention proposes a sewage treatment aeration optimization control system. The present invention can be fine-tuned according to the specific operating conditions of different facilities through the LSTM model, has good adaptability, is seamlessly integrated with the existing PLC system, and is easy to implement in existing facilities. Compared with some existing technologies that require large-scale transformation or replacement of equipment, the present invention provides a more flexible solution. Secondly, the edge computing device can operate independently on site without relying on the central server or cloud computing platform. This independence improves the reliability of the system, especially when the network connection is unstable or interrupted, it can still maintain normal operation. By processing data on site, the need to transmit sensitive data to the cloud is reduced, thereby reducing the risk of data leakage. This is particularly important for processing sensitive data related to the environment and public health. Physical mobile edge computing devices can be moved between different sewage treatment facilities as needed to adapt to different operating environments and needs. This flexibility makes the use of equipment more efficient.

[0192] Compared with the prior art, traditional sewage treatment systems often face the problem of excessive energy consumption and difficulty in precise control during the aeration stage. The present invention combines transfer learning with the LSTM model to achieve more accurate aeration control, thereby reducing energy consumption and improving treatment efficiency; the existing system relies on manual adjustment and setting of process parameters, resulting in complex management and low reaction efficiency. The present invention uses intelligent control methods, real-time data and deep learning technology to automatically optimize operating parameters, reduce manual intervention, and improve the automation level of the system; sewage treatment plants in remote areas or military sensitive areas often face the challenges of insufficient operating data and lack of real-time monitoring means. The present invention solves the problem of insufficient data of the target facility by training the optimization model in the source facility and migrating it to the target facility, ensuring the feasibility of real-time monitoring and debugging; the prior art has poor adaptability of the model when facing sewage treatment facilities of different scales, and it is difficult to effectively migrate the existing optimization model. The present invention can effectively migrate the optimization experience of the source facility to the target facility through transfer learning technology, significantly improving the adaptability and feasibility of the model; the management and maintenance of traditional sewage treatment facilities are complex, especially in large-scale facilities. The present invention simplifies the management process and reduces the difficulty of maintenance through an intelligent control system, allowing operators to manage facilities more easily; the existing technology lacks an intelligent decision support system and it is difficult to respond quickly based on real-time data. The present invention provides intelligent decision support through real-time data analysis and prediction to help operators adjust the aeration volume and other process parameters in a timely manner to optimize the treatment effect; traditional models, such as the ASM series models, respond slowly to environmental changes and are difficult to adapt to rapidly changing water quality and load fluctuations. The present invention utilizes the time series prediction capability of the LSTM model to quickly respond to environmental changes, adjust treatment strategies in a timely manner, and ensure that the effluent quality is qualified;

[0193] In summary, the present invention optimizes the operating parameters of sewage treatment facilities, reduces energy consumption, and improves processing efficiency by combining transfer learning and LSTM models, thereby achieving the goal of energy conservation and emission reduction. The present invention aims to reduce the reliance on manual adjustment through intelligent control methods, automatically optimize process parameters using real-time data, and improve the automation level and response speed of the system. At the same time, by training the optimization model in a small-scale sewage treatment facility with similar water quality and climatic conditions to the target facility, and migrating it to the target facility, the problem of insufficient data of the target facility is solved, and the feasibility of real-time monitoring and debugging is ensured. In addition, the present invention can effectively transfer the optimization experience of the source facility to the target facility through transfer learning technology, significantly improve the adaptability and feasibility of the model, so that it can play a role in sewage treatment facilities of different sizes, and through real-time data analysis and prediction, the present invention provides intelligent decision support to help operators adjust the aeration volume in time to optimize the treatment effect and ensure that the effluent water quality is qualified.

[0194] In an embodiment of the present invention, the present invention provides a sewage treatment aeration optimization control system, which includes a PLC control cabinet, an edge computing device, an inlet water quality detection probe, a dissolved oxygen probe, a fan and an aeration valve; wherein the PLC control cabinet is electrically connected to the dissolved oxygen probe, the fan and the aeration valve respectively; the PLC control cabinet is communicatively connected to the edge computing device, and the edge computing device is electrically connected to the inlet water quality detection probe; the inlet water quality detection probe is used to obtain initial inlet water quality sequence data and transmit it to the edge computing device; the dissolved oxygen probe is used to obtain the true value of dissolved oxygen in the aeration tank and transmit it to the PLC control cabinet; the edge computing device is used to pre-process the initial inlet water quality sequence data, output the target inlet water quality sequence data, and use a preset long-term and short-term network to generate a predicted value of dissolved oxygen in the aeration tank according to the target inlet water quality sequence data and transmit it to the PLC control cabinet; The C control cabinet is used to determine the target aeration control parameters according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the target aeration control parameters are used to adjust the blower motor frequency of the fan and the aeration valve opening of the aeration valve; based on the above scheme, the edge computing device uses a preset long-term and short-term network to process the initial influent water quality sequence data after preprocessing to generate the predicted value of dissolved oxygen in the aeration tank, and combines with the PLC control cabinet to determine the target aeration control parameters according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the blower motor frequency of the fan and the aeration valve opening of the aeration valve are adjusted by the target aeration control parameters to realize the aeration optimization process. The present invention automatically adjusts the process parameters according to the acquired influent water quality data and the actual value of dissolved oxygen in the aeration tank, which can reduce manual intervention and enable the system to quickly respond to water quality changes, thereby improving the sewage treatment effect.

[0195] See also Figure 3 , Figure 3A flowchart of the steps of a sewage treatment aeration optimization control method provided in Embodiment 2 of the present invention.

[0196] The present invention provides a sewage treatment aeration optimization control method, which is applied to the above-mentioned sewage treatment aeration optimization control system, comprising:

[0197] Step 301, obtaining the initial influent water quality sequence data and the actual value of dissolved oxygen in the aeration tank;

[0198] Step 302: pre-process the initial influent water quality sequence data, output the target influent water quality sequence data, and use the preset long-term and short-term network to generate the predicted value of dissolved oxygen in the aeration tank according to the target influent water quality sequence data;

[0199] Step 303: Determine target aeration control parameters according to the predicted dissolved oxygen value and the actual dissolved oxygen value in the aeration tank; the target aeration control parameters are used to adjust the blower motor frequency of the fan and the aeration valve opening of the aeration valve.

[0200] In an embodiment of the present invention, the present invention provides a sewage treatment aeration optimization control method, which is applied to the sewage treatment aeration optimization control system proposed by the present invention. First, the initial influent water quality sequence data and the actual value of dissolved oxygen in the aeration tank are obtained; then, the initial influent water quality sequence data are preprocessed, and the target influent water quality sequence data are output, and a preset long-term and short-term network is used to generate an aeration tank dissolved oxygen prediction value according to the target influent water quality sequence data; finally, the target aeration control parameter is determined according to the aeration tank dissolved oxygen prediction value and the aeration tank dissolved oxygen actual value; the target aeration control parameter is used to adjust the blower motor frequency of the fan and the aeration tank dissolved oxygen actual value. aeration valve opening of the air valve; based on the above scheme, a preset long-term and short-term network is used to process the initial influent water quality sequence data after preprocessing to generate an aeration tank dissolved oxygen prediction value, and according to the aeration tank dissolved oxygen prediction value and the aeration tank dissolved oxygen actual value, a target aeration control parameter is determined; the blower motor frequency of the fan and the aeration valve opening of the aeration valve are adjusted by the target aeration control parameter to realize the aeration optimization process. The present invention automatically adjusts the process parameters according to the acquired influent water quality data and the aeration tank dissolved oxygen actual value, which can reduce manual intervention and enable the system to quickly respond to water quality changes, thereby improving the sewage treatment effect.

[0201] An embodiment of the present invention further provides a computer device, including a memory and a processor, wherein a computer program is stored in the memory; when the computer program is executed by the processor, the processor executes the steps of the sewage treatment aeration optimization control method of the second embodiment.

[0202] The embodiment of the present invention further provides a computer-readable storage medium on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the steps of the wastewater treatment aeration optimization control method as described in the above-mentioned embodiment 2 are implemented.

[0203] The embodiment of the present invention further provides a computer program product, including a computer program / instruction, which implements the steps of the sewage treatment aeration optimization control method of the above-mentioned embodiment 2 when the computer program / instruction is executed by a processor.

[0204] In the several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0205] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0206] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A sewage treatment aeration optimization control system, characterized in that: The system includes a PLC control cabinet, an edge computing device, an inlet water quality detection probe, a dissolved oxygen probe, a fan and an aeration valve; The PLC control cabinet is electrically connected to the dissolved oxygen probe, the fan and the aeration valve respectively; The PLC control cabinet is communicatively connected to the edge computing device, and the edge computing device is electrically connected to the inlet water quality detection probe; The inlet water quality detection probe is used to obtain initial inlet water quality sequence data and transmit it to the edge computing device; The dissolved oxygen probe is used to obtain the actual value of dissolved oxygen in the aeration tank and transmit it to the PLC control cabinet; The edge computing device is used to preprocess the initial influent water quality sequence data, output the target influent water quality sequence data, and use a preset long-term and short-term network to generate a predicted value of dissolved oxygen in the aeration tank according to the target influent water quality sequence data and transmit it to the PLC control cabinet; The PLC control cabinet is used to determine the target aeration control parameter according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the target aeration control parameter is used to adjust the blower motor frequency of the fan and the aeration valve opening of the aeration valve.

2. The sewage treatment aeration optimization control system according to claim 1, characterized in that: The edge computing device is specifically used for: Performing data cleaning on the initial influent water quality sequence data, and outputting intermediate influent water quality sequence data; Performing outlier detection on the intermediate inlet water quality sequence data to generate smoothed inlet water quality sequence data; Performing missing value processing and normalization on the smoothed influent water quality sequence data to generate normalized influent water quality sequence data; Performing data conversion on the normalized influent water quality sequence data based on a preset sliding window method to generate target influent water quality sequence data; The target influent water quality sequence data is input into the preset long-term and short-term network, and the predicted value of dissolved oxygen in the aeration tank is generated and transmitted to the PLC control cabinet.

3. The sewage treatment aeration optimization control system according to claim 2 is characterized in that: The preset long-term and short-term network includes an input layer, multiple hidden layers and an output layer; the data processing process of the preset long-term and short-term network includes: Performing feature extraction on the target influent water quality sequence data received by the input layer through the multiple hidden layers to generate hidden layer states; The output layer is used to predict the target hidden state in the hidden layer state to generate a predicted value of dissolved oxygen in the aeration tank.

4. The sewage treatment aeration optimization control system according to claim 3 is characterized in that: The hidden layer includes an input gate, a forget gate, a candidate memory unit, an update memory unit, and an output gate; the data processing process of the hidden layer includes: Update the input water quality sequence data input by the input gate to generate an input state; The forget gate is used to select information from the input water quality sequence data, and a forget state is output; Performing candidate memory construction on the input inlet water quality sequence data through the candidate memory unit to generate a candidate memory state; Extracting information from the input inlet water quality sequence data through the output gate to generate an output state; Using the update memory unit to update the forgetting state, the candidate memory state and the input state, and output a target memory state; An output hidden state is generated according to the target memory state and the output state.

5. The sewage treatment aeration optimization control system according to claim 2, characterized in that: The model training process of the preset long-term and short-term network includes: Acquire the inlet water quality sequence data of the source facility to be trained and the inlet water quality sequence data of the target facility to be trained, pre-process the inlet water quality sequence data of the source facility to be trained, and output the inlet water quality sequence data of the target facility to be trained; Inputting the target influent water quality sequence data to be trained into the initial long-short term network, the hidden layer state to be trained and the predicted value of dissolved oxygen in the aeration tank to be trained; Determine the hidden layer gradient and the output layer gradient based on the hidden layer state to be trained and the predicted value of dissolved oxygen in the aeration tank to be trained; Based on a preset adaptive moment estimation optimizer, the model parameters of the hidden layer and the model parameters of the output layer in the initial long-short term network are updated using the hidden layer gradient and the output layer gradient, an intermediate long-short term network is determined, and the number of iterations is counted in real time; Using the intermediate long-short term network and the improved mean square error loss function, the target loss value is calculated according to the target inlet water quality sequence data to be trained; Determine whether the number of iterations reaches a preset iteration threshold and whether the target loss value converges; If the number of iterations reaches the preset iteration threshold or the target loss value converges, the intermediate long-short-term network is used as the long-short-term network to be fine-tuned; Based on the preset target domain loss function, the water quality sequence data of the target facility to be trained is used to perform local model training on the output layer in the long-term and short-term network to be fine-tuned, and the trained preset long-term and short-term network is determined.

6. The sewage treatment aeration optimization control system according to claim 1, characterized in that: The target aeration control parameters include the target blower motor frequency and the target aeration valve opening; the PLC control cabinet is specifically used for: Performing a difference operation on the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank, and outputting a deviation value of dissolved oxygen in the aeration tank; Performing proportional-integral-differential adjustment on the dissolved oxygen deviation value of the aeration tank to determine the proportional-integral-differential control value; Based on the proportional-integral-differential control value, determining the blower motor frequency control value and the aeration valve opening control value; The blower motor frequency control value and the aeration valve opening control value are used to update the blower motor frequency of the fan and the aeration valve opening of the aeration valve, and determine a target blower motor frequency and a target aeration valve opening.

7. A method for optimizing aeration control in sewage treatment, characterized in that: The sewage treatment aeration optimization control system as claimed in claim 1 comprises: Obtain the initial influent water quality sequence data and the actual value of dissolved oxygen in the aeration tank; Preprocessing the initial influent water quality sequence data, outputting target influent water quality sequence data, and using a preset long-term and short-term network to generate a predicted value of dissolved oxygen in an aeration tank according to the target influent water quality sequence data; A target aeration control parameter is determined according to the predicted value of dissolved oxygen in the aeration tank and the actual value of dissolved oxygen in the aeration tank; the target aeration control parameter is used to adjust the blower motor frequency of the fan and the aeration valve opening of the aeration valve.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the sewage treatment aeration optimization control method as claimed in claim 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the wastewater treatment aeration optimization control method according to claim 7 is implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer is caused to execute the wastewater treatment aeration optimization control method according to claim 7.

Images