Water quality prediction system based on VMD-DBO

Abstract

This invention discloses a water quality prediction system based on VMD-DBO, belonging to the field of water quality prediction technology. It includes a water quality data acquisition unit, a feature selection unit, a network parameter setting unit, and a model prediction unit. This invention constructs sample data and decomposes the time series of raw wastewater water quality indicators to obtain the actual effluent chemical oxygen demand (COD), total nitrogen (TN), and total phosphorus (TP) corresponding to the water quality indicator data. This method reduces data instability and enhances the model's predictive performance. It also correlates all water quality indicators with the time series, making the subsequent model prediction more accurate. The fusion of Transformer and LSTM neural network models better handles long-short-term dependencies in the sequence, while also improving parallel processing capabilities and model robustness. Furthermore, the system optimizes network parameters, giving the model good adaptability and accuracy.

Description

Technical Field

[0001] This invention relates to the field of water quality prediction technology, specifically a water quality prediction system based on VMD-DBO. Background Technology

[0002] The water quality prediction system utilizes network models and statistical analysis to monitor and predict time-series data of wastewater effluent quality in real time, providing a scientific basis for water resource management and pollution prevention and control. The invention patent application number 202210108885.1 discloses a "water quality prediction method based on parameter migration," which establishes a target monitoring station water quality prediction model and multiple neighboring monitoring station water quality prediction models based on RVFL (Real-Time Flow Mapping) based on water quality information from a target monitoring station and water quality training sets from neighboring monitoring stations. The method trains models separately for the target monitoring station and neighboring monitoring stations. Based on the trained model parameters, parameter transfer is performed between the models to obtain a transferred water quality prediction model for the target monitoring station. This transferred model is then used to predict water quality, and a weighted average is calculated to obtain the final water quality prediction result. This method applies parameter transfer to water quality prediction, transferring shared parameters between the target monitoring station and neighboring monitoring station models. This effectively utilizes the nonlinear correlation between water quality information from the target and neighboring monitoring stations, improving the accuracy of water quality prediction.

[0003] The aforementioned existing technologies have solved the problem of not being able to effectively utilize the nonlinear correlation between water quality information of the target monitoring station and nearby monitoring stations. However, during operation, they often fail to accurately capture the subtle details of water quality changes, cannot effectively analyze the nonlinear structure of large-scale water quality time series, and the prediction ability of a single model is very limited, making it difficult to provide relatively accurate prediction results. At the same time, the network parameters of the model have not been optimized, resulting in the model not having good adaptability during operation. Summary of the Invention

[0004] The purpose of this invention is to provide a water quality prediction system based on VMD-DBO to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a water quality prediction system based on VMD-DBO, including a model prediction unit;

[0006] The water quality data acquisition unit acquires water quality index data of the sewage treatment plant at different time periods, constructs multiple sample data based on the water quality index data at different time periods, and stores them in the original dataset.

[0007] The feature selection unit uses different water quality indicators as initial feature vectors, analyzes them using a correlation determination algorithm to obtain the correlation coefficient of the feature vectors, selects the final feature vectors based on the correlation coefficient, sets the number of modes to be decomposed, the initial modal components and the center frequency, and constructs an unconstrained variational model. After statistically analyzing the frequency domain representation corresponding to the initial center frequency, the initial center frequency, the frequency domain representation and the initial modal components under different modes are analyzed to obtain new center frequencies and modal components, which are then transmitted to the unconstrained variational model. This process is repeated until the optimal modal components and center frequency are determined. Based on the output current modal components and center frequency, the effluent chemical oxygen demand, effluent total nitrogen and effluent total phosphorus under different modes are analyzed.

[0008] The network parameter setting unit determines the network parameters in the current neural network model, constructs a corresponding dung beetle population based on each network parameter, initializes the population size, maximum number of iterations, and upper and lower bounds of the search space, obtains the actual values ​​of the sample data and the predicted values ​​output by the current neural network model, calculates the fitness value based on the difference between the actual and predicted values, iteratively updates the corresponding dung beetle positions, obtains the optimal fitness value and optimal position for the current stage, and determines the optimal network parameters based on the dung beetle individuals with the optimal fitness value and optimal position within different populations.

[0009] Preferably, the water quality data acquisition unit includes a water quality acquisition module and a sample generation module. The water quality acquisition module acquires water quality index data of the wastewater treatment plant at different time periods, including influent flow rate, influent chemical oxygen demand, influent total nitrogen, influent total phosphorus, influent ammonia nitrogen, suspended solids, and pH value. The sample generation module constructs multiple sample data using the water quality index data from different time periods, stores them in the original dataset, divides the original dataset into a training set, a validation set, and a test set according to a 7:1:2 ratio, and processes each sample data in the training set, validation set, and test set using a sample analysis algorithm. The specific sample analysis algorithm is as follows:

[0010]

[0011] in, This represents the minimum value of the variable in the sample. This represents the maximum value of the variable in the sample; This is the processed sample data.

[0012] Preferably, the feature selection unit includes a correlation coefficient calculation module and a feature vector determination module. The correlation coefficient calculation module uses influent flow rate, influent chemical oxygen demand (COD), influent total nitrogen (TNO), influent total phosphorus (TP), influent ammonia nitrogen (NHN), suspended solids (SSI), and pH value as initial feature vectors. It then analyzes these initial feature vectors using a correlation determination algorithm to obtain correlation coefficients for different initial feature vectors. The feature vector determination module selects initial feature vectors based on these correlation coefficients, using influent ammonia nitrogen (NHN), influent total nitrogen (TNO), influent total phosphorus (TP), influent COD (COD), and pH value as the final feature vectors. The correlation determination algorithm specifically includes:

[0013]

[0014] in, Represents two eigenvectors and The difference in rank, This represents the length of each feature vector. Represents two eigenvectors and The correlation coefficient between them and Indicates the feature vector number. The range is ,when This indicates a completely negative correlation, while Indicates a perfect positive correlation. 0 indicates that there is no correlation between the feature vectors.

[0015] Preferably, the feature selection unit further includes a model building module and an initial parameter analysis module. The model building module extracts sample data and recording time from the original dataset according to the feature vectors, and then sets the number of modes to be decomposed. Initial mode components corresponding to different modes and initial center frequency By introducing penalty term coefficients and Lagrange multipliers, an unconstrained variational model is constructed. The initial parameter analysis module statistically analyzes the initial center frequency. Corresponding frequency domain representation Then, the initial center frequency , and the initial modal components corresponding to different modes Analysis was performed to obtain a new center frequency. ,in ,Will and initial modal components By combining them, new modal components are obtained. ,in , and Indicates a parameter.

[0016] Preferably, the feature selection unit further includes a parameter output module and an index extraction module. The parameter output module transmits the new modal components and center frequencies to the unconstrained variational model and determines whether the convergence condition is met. If the condition is met, the current modal components and center frequencies are output. If the condition is not met, the new modal components and center frequencies are recalculated. The index extraction module analyzes the effluent chemical oxygen demand, effluent total nitrogen, and effluent total phosphorus under different modes based on the output current modal components and center frequencies.

[0017] Preferably, the network parameter setting unit includes a population construction module and a parameter determination module. The population construction module determines the network parameters in the current Transformer-LSTM neural network model, including the head and keys of the attention mechanism, the number of neurons, the Dropout rate, and the learning rate. A corresponding dung beetle population is constructed based on each network parameter. The parameter determination module divides the dung beetle population into rolling behavior, reproductive behavior, foraging behavior, and stealing behavior, and initializes the population size, maximum number of iterations, upper and lower bounds of the search space, dimension of the solution vector, and percentage of the producer population size to the total population size.

[0018] Preferably, the network parameter setting unit further includes a fitness calculation module and a network parameter analysis module. The fitness calculation module obtains the actual effluent chemical oxygen demand, total nitrogen, and total phosphorus corresponding to the sample data and the predicted values ​​output by the current neural network model. It then calculates the initial fitness value based on the difference between the actual and predicted values. The network parameter analysis module updates the corresponding dung beetle positions for different individuals in the population to obtain the optimal fitness value and optimal position for the current stage. This process continues until the iteration ends, outputting the dung beetle individuals with the optimal fitness value and optimal position. The optimal network parameters are determined based on the dung beetle individuals with the optimal fitness value and optimal position in different populations.

[0019] Preferably, the model prediction unit includes a parameter setting module and a prediction value determination module. The parameter setting module transmits the optimal network parameters to the Transformer-LSTM neural network model, wherein the attention mechanism has 4 heads, 8 keys, 3 time steps, 16 neurons in the LSTM hidden layer, 2 LSTM stacked layers, a minimum batch size of 328, and a maximum training iterations of 100. The prediction value determination module transmits the sample data from the training set to the neural network model for analysis, thereby completing the training of the network model. After determining the predicted values ​​of ammonia nitrogen, total phosphorus, and total nitrogen in the effluent based on the sample data in the validation and test sets, the prediction results are used to calculate the error. Four evaluation indicators, namely mean absolute percentage error, root mean square error, mean absolute error, and coefficient of determination, are selected to evaluate the prediction results.

[0020] Compared with the prior art, the beneficial effects of the present invention are:

[0021] This invention constructs sample data and decomposes the time series of original wastewater quality indicators to obtain the actual effluent chemical oxygen demand, total nitrogen, and total phosphorus corresponding to the water quality indicator data. This approach reduces data instability and enhances the model's predictive performance. It also correlates all water quality indicators with the time series, making subsequent model predictions more accurate. The invention integrates Transformer and LSTM neural network models. Since LSTM networks excel at capturing long-term dependencies and are suitable for handling long-distance dependencies in time series data, while Transformers consider all positions in the sequence and help capture global dependencies, combining the two can better handle long- and short-term dependencies in the sequence. It also improves parallel processing capabilities and model robustness. Furthermore, the system optimizes network parameters, giving the model good adaptability and accuracy. Attached Figure Description

[0022] Figure 1 A schematic diagram of the overall system flow is provided for embodiments of the present invention;

[0023] Figure 2 This is an internal module block diagram of the water quality data acquisition unit provided in an embodiment of the present invention;

[0024] Figure 3 This is an internal module block diagram of the network parameter setting unit provided in an embodiment of the present invention;

[0025] Figure 4 This invention provides a wastewater effluent prediction framework based on Transformer-LSTM for embodiments of the present invention.

[0026] In the diagram: 1. Water quality data acquisition unit; 101. Water quality acquisition module; 102. Sample generation module; 2. Feature selection unit; 201. Correlation coefficient calculation module; 202. Feature vector determination module; 203. Model construction module; 204. Initial parameter analysis module; 205. Parameter output module; 206. Index extraction module; 3. Network parameter setting unit; 301. Population construction module; 302. Parameter determination module; 303. Fitness calculation module; 304. Network parameter analysis module; 4. Model prediction unit; 401. Parameter setting module; 402. Predicted value determination module. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Please see Figure 1 - Figure 4 The present invention provides a technical solution: a water quality prediction system based on VMD-DBO, including a model prediction unit 4;

[0029] Water quality data acquisition unit 1 acquires water quality index data of the sewage treatment plant at different time periods, constructs multiple sample data based on the water quality index data at different time periods, and stores them in the original dataset;

[0030] Feature selection unit 2 uses different water quality indicators as initial feature vectors, analyzes them using a correlation determination algorithm to obtain the correlation coefficient of the feature vectors, selects the final feature vectors based on the correlation coefficient, sets the number of modes to be decomposed, the initial modal components and the center frequency, and constructs an unconstrained variational model. After statistically analyzing the frequency domain representation corresponding to the initial center frequency, the initial center frequency, the frequency domain representation and the initial modal components under different modes are analyzed to obtain new center frequencies and modal components, which are then transmitted to the unconstrained variational model. This operation is repeated until the optimal modal components and center frequency are determined. Based on the output current modal components and center frequency, the effluent chemical oxygen demand, effluent total nitrogen and effluent total phosphorus under different modes are analyzed.

[0031] The network parameter setting unit 3 determines the network parameters in the current neural network model, constructs a corresponding dung beetle population based on each network parameter, initializes the population size, maximum number of iterations, and upper and lower bounds of the search space, obtains the actual values ​​of the sample data and the predicted values ​​output by the current neural network model, calculates the fitness value based on the difference between the actual value and the predicted value, iteratively updates the corresponding dung beetle positions, obtains the best fitness value and best position at the current stage, and determines the optimal network parameters based on the dung beetle individuals with the best fitness value and best position in different populations.

[0032] The water quality data acquisition unit 1 includes a water quality acquisition module 101 and a sample generation module 102. The water quality acquisition module 101 acquires water quality index data of the wastewater treatment plant at different time periods, including influent flow rate, influent chemical oxygen demand, influent total nitrogen, influent total phosphorus, influent ammonia nitrogen, suspended solids, and pH value. The sample generation module 102 constructs multiple sample data using the water quality index data from different time periods and stores them in the original dataset. The original dataset is divided into a training set, a validation set, and a test set according to a 7:1:2 ratio. A sample analysis algorithm is then used to process each sample data in the training set, validation set, and test set. The specific sample analysis algorithm is as follows:

[0033]

[0034] in, This represents the minimum value of the variable in the sample. This represents the maximum value of the variable in the sample; The processed sample data;

[0035] The feature selection unit 2 includes a correlation coefficient calculation module 201 and a feature vector determination module 202. The correlation coefficient calculation module 201 uses influent flow rate, influent chemical oxygen demand (COD), influent total nitrogen (TNO), influent total phosphorus (TP), influent ammonia nitrogen (AMN), suspended solids (SSI), and pH value as initial feature vectors. It then analyzes these initial feature vectors using a correlation determination algorithm to obtain correlation coefficients for different initial feature vectors. The feature vector determination module 202 selects initial feature vectors based on these correlation coefficients, using ANO, TNO, TPI, and pH value as the final feature vectors. The correlation determination algorithm specifically includes:

[0036]

[0037] in, Represents two eigenvectors and The difference in rank, This represents the length of each feature vector. Represents two eigenvectors and The correlation coefficient between them and Indicates the feature vector number. The range is ,when This indicates a completely negative correlation, while Indicates a perfect positive correlation. 0 indicates that there is no correlation between the eigenvectors;

[0038] The feature selection unit 2 further includes a model building module 203 and an initial parameter analysis module 204. The model building module 203 extracts sample data and recording time from the original dataset according to the feature vectors, and then sets the number of modes to be decomposed. Initial mode components corresponding to different modes and initial center frequency By introducing penalty term coefficients and Lagrange multipliers, an unconstrained variational model is constructed. The initial parameter analysis module 204 statistically analyzes the initial central frequency. Corresponding frequency domain representation Then, the initial center frequency , and the initial modal components corresponding to different modes Analysis was performed to obtain a new center frequency. ,in ,Will and initial modal components By combining them, new modal components are obtained. ,in , and Indicates parameters;

[0039] The feature selection unit 2 further includes a parameter output module 205 and an index extraction module 206. The parameter output module 205 transmits the new modal components and center frequencies to the unconstrained variational model, determines whether the convergence condition is met, and outputs the current modal components and center frequencies if the condition is met; otherwise, it recalculates the new modal components and center frequencies. The index extraction module 206 analyzes the effluent chemical oxygen demand, total nitrogen, and total phosphorus under different modes based on the output current modal components and center frequencies. The unconstrained variational model is specifically as follows:

[0040]

[0041]

[0042] in, The first part represents the result of data decomposition. One modal component, Indicates the first The center frequency corresponding to each modal component Represents the Lagrange multiplier. This represents the coefficient of the penalty term. This represents the offset function. Represents the imaginary unit. Indicates a unit of time. This represents the convolution operator. Representing a time series, This indicates the defined function. Represents a time function. Represents the natural constant. Represents an unconstrained variational function. Indicates the total number of modes. Indicates the modal number;

[0043] The network parameter setting unit 3 includes a population construction module 301 and a parameter determination module 302. The population construction module 301 determines the network parameters in the current Transformer-LSTM neural network model, including the head and keys of the attention mechanism, the number of neurons, the Dropout rate, and the learning rate. A corresponding dung beetle population is constructed based on each network parameter. The parameter determination module 302 divides the dung beetle population into rolling behavior, reproductive behavior, foraging behavior, and stealing behavior, and initializes the population size, maximum number of iterations, upper and lower bounds of the search space, dimension of the solution vector, and percentage of the producer population size to the total population size.

[0044] The network parameter setting unit 3 further includes a fitness calculation module 303 and a network parameter analysis module 304. The fitness calculation module 303 obtains the actual effluent chemical oxygen demand, effluent total nitrogen, effluent total phosphorus and the predicted value output by the current neural network model corresponding to the sample data, and calculates the initial fitness value based on the difference between the actual value and the predicted value. The network parameter analysis module 304 updates the corresponding dung beetle position for different individuals in the population to obtain the best fitness value and best position at the current stage. Until the iteration ends, it outputs the dung beetle individual with the best fitness value and best position. The optimal network parameters are determined based on the dung beetle individuals with the best fitness value and best position in different populations.

[0045] The model prediction unit 4 includes a parameter setting module 401 and a prediction value determination module 402. The parameter setting module 401 transmits the optimal network parameters to the Transformer-LSTM neural network model, wherein the attention mechanism has 4 heads, 8 keys, 3 time steps, 16 neurons in the LSTM hidden layer, 2 LSTM stacked layers, a minimum batch size of 328, and a maximum training iterations of 100. The prediction value determination module 402 transmits the sample data from the training set to the neural network model for analysis, thereby completing the training of the network model. After determining the predicted values ​​of ammonia nitrogen, total phosphorus, and total nitrogen at the next time step based on the sample data in the validation set and test set, the error of the prediction results is calculated, and four evaluation indicators are selected to evaluate the prediction results: mean absolute percentage error, root mean square error, mean absolute error, and coefficient of determination.

[0046] Working Principle: This invention acquires water quality index data of a wastewater treatment plant at different time periods through the water quality acquisition module 101 in the water quality data acquisition unit 1. The sample generation module 102 constructs multiple sample data. The correlation coefficient calculation module 201 in the feature selection unit 2 determines the correlation coefficient of different initial feature vectors. The feature vector determination module 202 selects the initial feature vectors according to the correlation coefficient. The model construction module 203 initializes the modal components and center frequencies under different modes. The initial parameter analysis module 204 iteratively updates the modal components and center frequencies. The parameter output module 205 outputs the current modal components and center frequencies. The index extraction module 206 analyzes the effluent chemical oxygen demand and other parameters under different modes. Total nitrogen in the water and total phosphorus in the effluent are determined by the population construction module 301 in the network parameter setting unit 3, which constructs a corresponding dung beetle population based on each network parameter. After the parameter determination module 302 initializes the population size and the maximum number of iterations, the fitness calculation module 303 calculates the initial fitness value based on the difference between the actual and predicted values ​​of the sample data. The network parameter analysis module 304 determines the optimal network parameters using the best fitness value and the best position of the dung beetle individuals in different populations. The optimal network parameters are then transmitted to the neural network model through the parameter setting module 401 in the model prediction unit 4. The prediction value determination module 402 selects four evaluation indicators—mean absolute percentage error, root mean square error, mean absolute error, and coefficient of determination—to evaluate the prediction results.

[0047] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0048] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A water quality prediction system based on VMD-DBO, comprising a model prediction unit (4), characterized in that: Water quality data acquisition unit (1) acquires water quality index data of the sewage treatment plant at different time periods, constructs multiple sample data based on the water quality index data at different time periods, and stores them in the original dataset; Feature selection unit (2) takes different water quality indicators as initial feature vectors, analyzes them using a correlation determination algorithm, obtains the correlation coefficient of the feature vectors, selects the final feature vectors based on the correlation coefficient, sets the number of modes to be decomposed, the initial modal components and the center frequency, and constructs an unconstrained variational model. After statistically analyzing the frequency domain representation corresponding to the initial center frequency, the initial center frequency, the frequency domain representation and the initial modal components under different modes are analyzed to obtain new center frequencies and modal components, which are then transmitted to the unconstrained variational model. The operation is repeated until the optimal modal components and center frequency are determined. Based on the current modal components and center frequency output, the chemical oxygen demand, total nitrogen and total phosphorus of the effluent under different modes are analyzed. The network parameter setting unit (3) determines the network parameters in the current neural network model, constructs a corresponding dung beetle population based on each network parameter, initializes the population size, maximum number of iterations and upper and lower bounds of the search space, obtains the actual value of the sample data and the predicted value output by the current neural network model, calculates the fitness value based on the difference between the actual value and the predicted value, iteratively updates the corresponding dung beetle position, obtains the best fitness value and best position at the current stage, and determines the optimal network parameters based on the dung beetle individuals with the best fitness value and best position in different populations.

2. The water quality prediction system based on VMD-DBO according to claim 1, characterized in that: The water quality data acquisition unit (1) includes a water quality acquisition module (101) and a sample generation module (102). The water quality acquisition module (101) acquires water quality index data of the sewage treatment plant at different time periods. The water quality data includes influent flow rate, influent chemical oxygen demand, influent total nitrogen, influent total phosphorus, influent ammonia nitrogen, suspended solids and pH value. The sample generation module (102) uses the water quality index data at different time periods to construct multiple sample data, stores them in the original dataset, divides the original dataset into training set, validation set and test set according to 7:1:2, and uses the sample analysis algorithm to process each sample data in the training set, validation set and test set.

3. The water quality prediction system based on VMD-DBO according to claim 1, characterized in that: The feature selection unit (2) includes a correlation coefficient calculation module (201) and a feature vector determination module (202). The correlation coefficient calculation module (201) takes the influent flow rate, influent chemical oxygen demand, influent total nitrogen, influent total phosphorus, influent ammonia nitrogen, suspended solids and pH value as initial feature vectors, and uses the correlation determination algorithm to analyze the initial feature vectors to obtain the correlation coefficients of different initial feature vectors. The feature vector determination module (202) selects the initial feature vectors according to the correlation coefficients, and takes the influent ammonia nitrogen, influent total nitrogen, influent total phosphorus, influent chemical oxygen demand and pH value as the final feature vectors.

4. The water quality prediction system based on VMD-DBO according to claim 3, characterized in that: The feature selection unit (2) further includes a model building module (203) and an initial parameter analysis module (204). The model building module (203) extracts sample data and recording time from the original dataset according to the feature vectors, and then sets the number of modes to be decomposed. Initial mode components corresponding to different modes and initial center frequency By introducing penalty term coefficients and Lagrange multipliers, an unconstrained variational model is constructed. The initial parameter analysis module (204) statistically analyzes the initial center frequency. Corresponding frequency domain representation Then, the initial center frequency , and the initial modal components corresponding to different modes Analysis was performed to obtain a new center frequency. ,in ,Will and initial modal components By combining them, new modal components are obtained. ,in , and Indicates a parameter.

5. The water quality prediction system based on VMD-DBO according to claim 4, characterized in that: The feature selection unit (2) further includes a parameter output module (205) and an index extraction module (206). The parameter output module (205) transmits the new modal components and center frequencies to the unconstrained variational model and determines whether the convergence conditions are met. If they are met, the current modal components and center frequencies are output. If they are not met, the new modal components and center frequencies are recalculated. The index extraction module (206) analyzes the effluent chemical oxygen demand, effluent total nitrogen, and effluent total phosphorus under different modes based on the output current modal components and center frequencies.

6. The water quality prediction system based on VMD-DBO according to claim 1, characterized in that: The network parameter setting unit (3) includes a population construction module (301) and a parameter determination module (302). The population construction module (301) determines the network parameters in the current Transformer-LSTM neural network model, including the head and key of the attention mechanism, the number of neurons, the Dropout rate and the learning rate. The corresponding dung beetle population is constructed according to each network parameter. The parameter determination module (302) divides the dung beetle population into rolling behavior, reproductive behavior, foraging behavior and stealing behavior, and initializes the population size, maximum number of iterations, upper and lower bounds of the search space, dimension of the solution vector and the percentage of the producer population size to the total population size.

7. The water quality prediction system based on VMD-DBO according to claim 6, characterized in that: The network parameter setting unit (3) further includes a fitness calculation module (303) and a network parameter analysis module (304). The fitness calculation module (303) obtains the actual effluent chemical oxygen demand, effluent total nitrogen, effluent total phosphorus and the predicted value output by the current neural network model corresponding to the sample data, and calculates the initial fitness value based on the difference between the actual value and the predicted value. The network parameter analysis module (304) updates the corresponding dung beetle position for different individuals in the population, obtains the best fitness value and best position for the current stage, and outputs the dung beetle individual with the best fitness value and best position until the iteration ends. The optimal network parameters are determined based on the dung beetle individuals with the best fitness value and best position in different populations.

8. The water quality prediction system based on VMD-DBO according to claim 1, characterized in that: The model prediction unit (4) includes a parameter setting module (401) and a prediction value determination module (402). The parameter setting module (401) transmits the optimal network parameters to the Transformer-LSTM neural network model, wherein the attention mechanism has 4 heads, 8 keys, 3 time steps, 16 neurons in the LSTM hidden layer, 2 stacked layers in the LSTM, a minimum batch size of 328, and a maximum number of training iterations of 100. The prediction value determination module (402) transmits the sample data from the training set to the neural network model for analysis, thereby completing the training of the network model. After determining the predicted values ​​of ammonia nitrogen, total phosphorus, and total nitrogen at the effluent based on the sample data in the validation set and test set, the error of the prediction results is calculated, and four evaluation indicators, namely mean absolute percentage error, root mean square error, mean absolute error, and coefficient of determination, are selected to evaluate the prediction results.

Images