A control method and system based on DO model and time-delay feedforward model

Abstract

The present invention relates to the technical field of biochemical system data processing, and in particular to a control method and system based on a DO model and a time-delay feedforward model. The method comprises the following steps: obtaining biochemical wastewater treatment data and external shock load data; constructing a DO model based on DO concentration data and external shock load data in the biochemical wastewater treatment data to obtain a DO model; processing time lag effects based on the DO model to obtain a time lag effect model; constructing an external shock load model based on system state parameter data and external shock load data in the biochemical wastewater treatment data to obtain an external shock load model; and coupling the time lag effect model and the external shock load model to obtain a DO model and a time-delay feedforward model to perform auxiliary biochemical wastewater treatment operations. The present invention can better adapt to changes in the external environment, reduce unnecessary resource waste, and improve the economic efficiency and sustainability of the biochemical wastewater treatment system.

Description

Technical Field

[0001] The present invention relates to the technical field of biochemical system data processing, and in particular to a control method and system based on a DO model and a time-delay feedforward model. Background Art

[0002] In traditional sewage treatment plants, the operation of the biochemical system is judged by the test data of on-site online DO, ORP and other instruments, combined with the experience of process technicians. When the system is subjected to hydraulic shock load and organic matter shock load, the cause of the system crash cannot be diagnosed in a timely and accurate manner. The system operation is roughly judged only by the operating parameters of the online data and the apparent characteristics of the system, such as the size and burstability of bubbles in the aeration tank, the color of the sludge in the biochemical tank, and whether there is mud floating in the subsequent sedimentation tank. The technical measures implemented deviate from the correct adjustment direction, resulting in continued system instability and even unstable effluent that does not meet the standards. The activated sludge biological phase can quickly reflect the activity of the sludge and is directly correlated with parameters such as organic matter and water volume in the influent. Therefore, careful analysis of the type, number, and activity of indicator organisms can determine whether the three phases of organic matter, bacterial count (sludge concentration), and DO flowing into the aeration tank are in a good balance, facilitating timely adjustment of process operations to ensure normal production.

[0003] Traditional wastewater treatment plants typically rely on online instrumentation and the experience of process technicians to assess the operation of biochemical systems. While these methods are quite effective in identifying general issues, they still pose the risk of inaccurate diagnosis when the system is subjected to hydraulic and organic shock loads. For example, judging system status solely based on online data and superficial characteristics fails to capture underlying changes, leading to technical measures being implemented that deviate from the correct direction, potentially causing system instability and even substandard effluent.

[0004] To address this challenge, model-based control methods have gained increasing attention in recent years. These methods utilize mathematical models to describe the dynamic characteristics of biochemical wastewater treatment systems, achieving more effective control through modeling, simulation, and optimization. However, existing model-based control methods cannot accurately predict changes in biochemical wastewater treatment systems and are easily affected by external factors, resulting in reduced treatment effectiveness. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention proposes a control method and system based on a DO model and a time-delay feedforward model to solve at least one of the above technical problems.

[0006] The present application provides a control method based on a DO model and a time-delay feedforward model, the method comprising:

[0007] S1. Obtain biochemical wastewater treatment data and external shock load data;

[0008] S2, constructing a DO model based on DO concentration data in the biochemical sewage treatment data and external shock load data to obtain a DO model;

[0009] S3, processing the time lag effect according to the DO model to obtain a time lag effect model;

[0010] S4. constructing an external shock load model based on the system state parameter data and the external shock load data in the biochemical sewage treatment data to obtain an external shock load model;

[0011] S5. Couple the time lag effect model and the external shock load model to obtain a DO model and a time delay feedforward model to perform auxiliary operations for biochemical sewage treatment.

[0012] By constructing a DO model and a time-lag effect model in the present invention, the changing trend of the DO concentration in the biochemical sewage treatment system can be predicted more accurately, which helps to adjust the operating parameters in a timely manner, thereby improving the treatment efficiency. The time-lag effect processing takes into account the delay in the system response, making the model closer to the actual situation, and can more accurately predict the dynamic changes of the system, which helps to reduce system fluctuations and improve treatment stability. The construction of an external shock load model can take into account the impact of external environmental factors on the system, such as fluctuations in the influent water quality or changes in temperature, etc., which helps the system to better adapt to external changes and maintain stable treatment effects. Combining the time-lag effect model and the external shock load model, the obtained DO model and time-delay feedforward model can achieve precise control of the biochemical sewage treatment system, making the system operation more stable and efficient. By improving treatment efficiency and stability, energy consumption and chemical consumption in system operation can be reduced, operating costs can be reduced, and economic benefits can be improved.

[0013] Optionally, S1 includes the following steps:

[0014] The biochemical sewage treatment data is collected by using sensors or monitoring equipment preset in the biochemical sewage treatment system to obtain biochemical sewage treatment data;

[0015] The external impact load data is collected by using an external environmental monitoring device preset outside or a monitoring device in an input channel to obtain the external impact load data.

[0016] The present invention uses preset sensors or monitoring equipment to collect data from the biochemical sewage treatment system in real time, including key parameters such as DO concentration, water intake, and temperature, so that the system's operating status can be monitored and recorded in a timely manner. The collection of biochemical sewage treatment data can fully understand the system's operating conditions, including sewage treatment efficiency, water quality trends, etc., which helps to promptly identify problems and take appropriate measures to adjust them. The collection of external shock load data can monitor the impact of external environmental factors on the biochemical sewage treatment system, such as temperature changes, rainfall conditions, etc., providing a more comprehensive reference basis for system operation.

[0017] Optionally, S2 includes the following steps:

[0018] S21, performing feature extraction on the DO concentration data and the external shock load data in the biochemical sewage treatment data to obtain DO concentration feature data and external shock load feature data;

[0019] S22. Perform Gaussian kernel mapping on the DO concentration characteristic data and the external shock load characteristic data to obtain DO high-dimensional space data; wherein the Gaussian kernel mapping is specifically:

[0020]

[0021] K(x) is the DO high-dimensional space data, exp is the natural exponential term, x i is the DO concentration characteristic data, x y is the characteristic data of external shock load, σ is the bandwidth data of Gaussian kernel;

[0022] S23. Parameter optimization and iteration are performed on the DO high-dimensional spatial data to obtain a DO model; wherein the parameter optimization is specifically as follows:

[0023]

[0024] L ∈ (y,f(x)) is the model parameter optimization data, y is the real DO concentration data corresponding to the DO high-dimensional space data, f(x) is the DO concentration prediction data in the DO high-dimensional space data, and ∈ is the model prediction tolerance data.

[0025] In the present invention, by extracting features from the DO concentration data and external shock load data in the biochemical sewage treatment data, the obtained DO concentration feature data and external shock load feature data can accurately reflect the system operation status and external influences, providing an accurate data basis for subsequent model construction. The application of Gaussian kernel mapping can map the feature data from the original space to the high-dimensional space, effectively expanding the feature space, improving the feature discrimination and expression ability, and helping to better capture the complex relationship between the data. Through parameter optimization and iteration, the model's fitting ability and prediction accuracy can be further improved, so that the DO model can more accurately reflect the actual situation, enhance the model's generalization ability and robustness, and improve the system's reliability and stability. The use of model prediction tolerance control can effectively adjust the model's tolerance, making the model's prediction more accurate and reliable, improving the model's adaptability and generalization ability, and reducing prediction errors. The DO model obtained by integrating the above steps can more accurately control and adjust the biochemical sewage treatment system, improve the accuracy and reliability of the control method, and make the system's operation more stable and efficient.

[0026] Optionally, S21 includes the following steps:

[0027] S211, performing primary feature extraction on the DO concentration data and the external shock load data in the biochemical sewage treatment data to obtain primary DO concentration feature data and primary external shock load feature data;

[0028] S212, performing frequency domain mapping on the primary DO concentration characteristic data and the primary external impact load characteristic data to obtain DO concentration characteristic frequency domain mapping data and external impact load characteristic frequency domain mapping data, respectively;

[0029] S213, clustering the DO concentration characteristic frequency domain mapping data and the external impact load characteristic frequency domain mapping data to obtain DO concentration characteristic frequency domain clustering data and external impact load characteristic frequency domain clustering data respectively;

[0030] S214, performing feature selection on the DO concentration characteristic frequency domain clustering data and the external shock load characteristic frequency domain clustering data to obtain DO concentration feature selection data and external shock load feature selection data, respectively;

[0031] S215 , performing feature scaling on the DO concentration feature selection data and the external impact load feature selection data to obtain DO concentration feature data and external impact load feature data.

[0032] The primary feature extraction stage of the present invention can extract primary DO concentration feature data and primary external shock load feature data from the biochemical sewage treatment data. These features can comprehensively reflect the operating status of the system and external influences, and provide diverse features for subsequent analysis. Frequency domain mapping can map the original feature data to the frequency domain space, which improves the dimension and expression ability of the data, helps to capture more complex relationships between data, and makes the features more representative and discriminative. Clustering processing can divide the data into groups and cluster data with similar features, which helps to discover the inherent structure and laws in the data and improves the interpretability and discriminability of the features. Through feature selection, the features that are most sensitive and important to the system status and external influences can be screened out, reducing the data dimension, reducing the model complexity, and improving the generalization ability and prediction accuracy of the model. Feature scaling can unify the numerical ranges of different features to the same scale, eliminating the dimensional influence between features, making the feature data more comparable and interpretable, and helping to improve the stability and convergence speed of the model.

[0033] Optionally, the feature selection step in S214 is specifically as follows:

[0034] S2141. Extract cluster label data based on the DO concentration characteristic frequency domain cluster data and the external impact load characteristic frequency domain cluster data to obtain cluster label data;

[0035] S2142, performing variance calculation based on the cluster label data, the DO concentration characteristic frequency domain cluster data, and the external impact load characteristic frequency domain cluster data to obtain characteristic variance data;

[0036] S2143, performing time series division on the DO concentration data in the biochemical sewage treatment data to obtain DO concentration time series division data;

[0037] S2144, calculating the coefficient of variation based on the DO concentration time series data to obtain DO concentration coefficient of variation data;

[0038] S2145, performing stability screening on the DO concentration time series data according to the DO concentration coefficient of variation data to obtain DO concentration stability screening data;

[0039] S2146. Use the DO concentration stability screening data and the characteristic variance data to screen and process the DO concentration characteristic frequency domain clustering data and the external shock load characteristic frequency domain clustering data to obtain DO concentration characteristic selection data and external shock load characteristic selection data, respectively.

[0040] In the present invention, by extracting cluster label data, the original data can be divided into different groups, and a basis is provided for subsequent feature selection, making feature selection more targeted and effective. Combined with the calculation of feature variance data and time series stability data, the degree of change and stability of the features can be comprehensively considered, thereby more accurately evaluating the impact of the features on the system state, and improving the precision and accuracy of feature selection. By calculating the coefficient of variation and screening the stability of the DO concentration time series data in the biochemical sewage treatment data, abnormal fluctuations and noise in the time series data can be eliminated, and the stability and reliability of the data can be improved. By using feature variance data and time series stability screening data to screen and process feature data, representative and stable features can be selected more carefully, and the accuracy and reliability of feature selection can be improved. Through the screening process of feature selection, redundant information and invalid features can be eliminated, the dimension of the data and the complexity of the model are reduced, and the generalization ability and prediction accuracy of the model are improved.

[0041] Optionally, the time lag effect model includes a first time lag effect model and a second time lag effect model, and S3 includes the following steps:

[0042] S31. Performing time-lag processing on the biochemical sewage treatment data and the external impact load data to obtain time-lag processed data;

[0043] S32, constructing a differential equation model based on the DO model and the time-lag processing data to obtain a first time-lag effect model;

[0044] S33. Construct a state space model based on the DO model and the time-delay processing data to obtain a second time-delay effect model.

[0045] In the present invention, by performing time-lag processing on the biochemical sewage treatment data and the external impact load data, data reflecting the time-lag characteristics of the system response can be obtained, and the dynamic response process of the system can be accurately captured. Based on the time-lag processing data and the existing DO model, a differential equation model is used to construct a first time-lag effect model, which can better describe the time-lag characteristics of the system and further improve the accuracy and fidelity of the model. Based on the time-lag processing data and the existing DO model, a state-space model is used to construct a second time-lag effect model, which can more comprehensively describe the dynamic characteristics of the system, including time-lag effects and state change laws, and help to more accurately predict the future state of the system. Based on the description of the time-lag characteristics of the system, the first time-lag effect model and the second time-lag effect model respectively adopt different modeling methods, making the model more comprehensive and flexible, and applicable to different systems and application scenarios, thereby improving the versatility and applicability of the model. By constructing a time-lag effect model, the dynamic response process of the biochemical sewage treatment system can be more deeply understood, which helps to discover problems and bottlenecks in the system, and further optimize and improve it to improve the performance and efficiency of the system.

[0046] Optionally, S31 includes the following steps:

[0047] Time series analysis is performed based on biochemical sewage treatment data and external shock load data to obtain time series analysis data;

[0048] According to the time series analysis data, the time lag calculation is performed on the biochemical sewage treatment data and the external impact load data to obtain the time lag calculation data;

[0049] The time-delay processing data is subjected to discrete event simulation through a distributed computing platform to obtain discrete event time-delay data;

[0050] The discrete event time-delay data are organized and integrated to obtain time-delay processed data.

[0051] Through time series analysis in the present invention, it is possible to gain an in-depth understanding of the changing trends and laws of the biochemical sewage treatment system and the external shock load, provide basic data for subsequent time-lag calculations, and contribute to a comprehensive understanding of the dynamic characteristics of the system. By performing time-lag calculations on the biochemical sewage treatment data and the external shock load data, the time-lag effect of the system can be accurately evaluated, and the degree of lag in the system response can be determined, providing an important reference basis for subsequent modeling and control. By performing discrete event simulation on a distributed computing platform, the response process of the system under different external shock conditions, including the influence of the time-lag effect, can be simulated, thereby better understanding the dynamic behavior of the system. Organizing and integrating the time-lag data obtained from the discrete event simulation can systematically manage and analyze the time-lag data, providing a reliable data basis for subsequent model establishment and analysis. By fully understanding the dynamic characteristics of the system and accurately evaluating the time-lag effect, the time-lag model of the system can be established more accurately, the accuracy and reliability of the model can be improved, and more effective support can be provided for the control and optimization of the system.

[0052] Optionally, S4 includes the following steps:

[0053] Feature extraction is performed based on system state parameter data and external impact load data in the biochemical sewage treatment data to obtain system state parameter feature data and external impact load feature data;

[0054] Perform correlation calculation based on system state parameter characteristic data and external impact load parameter data to obtain characteristic correlation data;

[0055] Screening the system state parameter characteristic data and the external impact load characteristic data according to the characteristic correlation data to obtain system state parameter characteristic screening data and external impact load characteristic screening data;

[0056] A linear model is constructed for the system state parameter characteristic screening data and the external shock load characteristic screening data to obtain the external shock load model.

[0057] Through feature extraction in the present invention, it is possible to gain an in-depth understanding of the state parameters of the biochemical sewage treatment system and the key features of the external shock load, thereby fully understanding the operating state of the system and the impact of the external environment. Through correlation calculation, the correlation between the system state parameters and the external shock load can be accurately evaluated, and the key features that affect the operation of the system can be determined, providing an important basis for feature screening. By screening the system state parameter characteristic data and the external shock load characteristic data according to the feature correlation data, it is possible to optimize the feature selection process and improve the representativeness and discrimination of the selected features. Through linear model construction, a linear relationship between the system state parameter characteristic data and the external shock load characteristic data can be established, thereby constructing an external shock load model for predicting the impact of external influences on the system. By modeling the external shock load, the impact of external factors on the system can be more accurately predicted, thereby improving the accuracy and reliability of system control and providing more effective support for the operation optimization of the biochemical sewage treatment system.

[0058] Optionally, S5 includes the following steps:

[0059] Obtain current biochemical system status data;

[0060] Generate system state change data based on current biochemical system state data and time lag effect model to obtain system state change data;

[0061] The external impact load model is used to correct the system state change data to obtain the external impact change data;

[0062] The time-lag effect model is modified using external shock change data to obtain the DO model and the time-delay feedforward model for auxiliary operations in biochemical wastewater treatment.

[0063] The present invention obtains and analyzes the current biochemical system status data, which can realize real-time monitoring of the biochemical sewage treatment system and timely understand the system operation status. Based on the time-lag effect model, combined with the current system status data, the state change trend of the system can be accurately predicted. By using the external impact load model to correct the system state change data, the influence of external factors on the system operation can be considered, and the adaptability and stability to external changes can be improved. By using the external impact change data to correct the time-lag effect model, the precision and accuracy of the model can be improved, and the actual operation of the biochemical sewage treatment system can be better reflected. After obtaining the corrected DO model and time-delay feedforward model, auxiliary operations of biochemical sewage treatment can be carried out, and the efficiency and effect of biochemical sewage treatment can be improved through precise control and optimization of the system.

[0064] Optionally, the present application further provides a control system based on a DO model and a time-delay feedforward model, for executing the control method based on the DO model and the time-delay feedforward model as described above, wherein the control system based on the DO model and the time-delay feedforward model includes:

[0065] Biochemical system basic data acquisition module, used to obtain biochemical sewage treatment data and external impact load data;

[0066] A DO model building module is used to build a DO model based on DO concentration data in biochemical sewage treatment data and external shock load data to obtain a DO model;

[0067] A time lag effect model building module is used to process the time lag effect according to the DO model to obtain a time lag effect model;

[0068] An external shock load model construction module is used to construct an external shock load model based on system state parameter data and external shock load data in the biochemical sewage treatment data to obtain an external shock load model;

[0069] The DO model and time-delay feedforward model construction module is used to couple the time-delay effect model and the external impact load model to obtain the DO model and the time-delay feedforward model for auxiliary operations of biochemical sewage treatment.

[0070] The present invention aims to construct a DO model with a time-lag effect and an external shock load model by comprehensively considering the DO concentration data within the biochemical wastewater treatment system and the external shock load data. This allows for more precise control of the biochemical system's operation and fully accounts for the impact of the external environment on the system. When establishing the DO model, a time-lag effect is incorporated to account for the delayed nature of the system's response, making the model more accurate and enabling more precise prediction of system dynamics. By coupling the time-lag effect model with the external shock load model, a time-delay feedforward model is derived, enabling the control strategy to be dynamically adjusted based on the system's current state and changes in the external environment, improving the real-time and adaptability of the control. The time-delay feedforward model can more accurately predict the system's response. Combined with the external shock load model, it can optimize the biochemical wastewater treatment operation strategy, improve treatment efficiency and water quality stability, and thus achieve better treatment results. Through accurate model prediction and dynamic adjustment of the control strategy, operating costs can be effectively reduced. The system can better adapt to changes in the external environment, reduce unnecessary resource waste, and improve the economic efficiency and sustainability of the biochemical wastewater treatment system. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0072] Figure 1 A flowchart showing a control method based on a DO model and a time-delay feedforward model according to an embodiment is shown;

[0073] Figure 2 A flowchart showing the steps of a DO model construction method according to an embodiment is shown;

[0074] Figure 3 A flowchart showing the steps of a method for extracting basic features of biochemical wastewater according to an embodiment is shown;

[0075] Figure 4 A flowchart showing the steps of a method for selecting basic features of biochemical wastewater according to an embodiment is shown;

[0076] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0077] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are 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 those skilled in the art without making any creative work are within the scope of protection of the present invention.

[0078] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0079] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0080] See also Figures 1 to 4 The present application provides a control method based on a DO model and a time-delay feedforward model, the method comprising:

[0081] S1. Obtain biochemical wastewater treatment data and external shock load data;

[0082] The present invention provides an embodiment for deploying a sensor network covering the entire sewage treatment system, including water quality sensors, flow sensors, temperature sensors, etc. Modern sensor technologies, such as electrochemical sensors and fiber optic sensors, are used to monitor various parameters in real time. Environmental monitoring stations are set up around the sewage treatment plant, and meteorological sensors, rainfall sensors and other equipment are used to monitor changes in the external environment in real time. Biochemical sewage treatment data, such as DO concentration, pH value, sludge concentration, etc., are expressed in standard units. External shock load data, such as rainfall and temperature changes, are expressed in corresponding units.

[0083] S2, constructing a DO model based on DO concentration data in the biochemical sewage treatment data and external shock load data to obtain a DO model;

[0084] This invention provides an embodiment that utilizes DO concentration data from biochemical wastewater treatment and external shock load data to construct a DO concentration prediction model using a machine learning algorithm. The external shock load data serves as the input variable, and the DO concentration from the biochemical wastewater treatment data serves as the predicted output variable. The model is trained using historical data using machine learning algorithms such as regression analysis, neural networks, and support vector machines to predict changes in DO concentration.

[0085] S3, processing the time lag effect according to the DO model to obtain a time lag effect model;

[0086] The present invention provides an embodiment that utilizes methods such as a Kalman filter or a state-space model to perform time-delay processing on the DO model, taking into account the dynamic response characteristics of the biochemical system, and obtaining a DO model that accounts for time delay. The Kalman filter processes the system state through state estimation and measurement updates, obtaining a real-time state estimate. The state-space model represents the DO model in state-space form, describing the system's dynamic behavior through state equations and observation equations, and accounting for time delay effects.

[0087] The DO model is converted into a state-space form to facilitate time-delay processing using a Kalman filter. A time-delay term is introduced into the state-space model and the Kalman filter is used to correct the time-delay and obtain a real-time state estimate.

[0088] The present invention provides an embodiment that incorporates time-delay effects into an established DO model. Time-delay effects are phenomena whereby the system response lags behind the input signal due to factors such as biochemical reactions and mass transfer. Time-delay effects are simulated by introducing a delay parameter. For example, in a difference equation model, a certain time-delay term is introduced; in a state-space model, a state delay term can be added.

[0089] DO concentration data (unit: mg / L) was collected from the biochemical wastewater treatment system over a period of time, totaling 10 data points:

[0090] Time (hours) DO concentration (mg / L) 1 5.6 2 5.2 3 5.8 4 6.1 5 5.9 6 5.5 7 5.3 8 5.7 9 6.0 10 6.2

[0091] The change of DO concentration is described by a difference equation model, using a first-order inertial model: DO(t) is the DO concentration at time t, t is the time data corresponding to the DO concentration data, DO in (t-τ) is the DO concentration before the time lag parameter τ, and k1 is the weighted data of the time lag parameter.

[0092] S4. constructing an external shock load model based on the system state parameter data and the external shock load data in the biochemical sewage treatment data to obtain an external shock load model;

[0093] The present invention provides an embodiment for extracting features from system state parameter data and external shock load data, such as statistics such as maximum value, minimum value, mean value, variance, or spectral features, time series features, etc. Feature selection methods such as correlation analysis, variance analysis, principal component analysis (PCA), etc. are used to select features related to the operating state of the biochemical sewage treatment system. Based on the selected features, an external shock load model is constructed. Various modeling methods, such as linear regression, support vector machine (SVM), neural network, etc., can be used to fit the relationship between system state parameters and external shock loads. For example, a linear regression model is used to describe the relationship between the DO concentration and the influent organic matter concentration of the biochemical sewage treatment system.

[0094] DO concentration data (unit: mg / L) and influent organic matter concentration data (unit: mg / L) were collected from the biochemical wastewater treatment system over a period of time, totaling 10 data points:

[0095] Time (hours) DO concentration (mg / L) Influent organic matter concentration (mg / L) 1 5.6 10 2 5.2 9.5 3 5.8 9 4 6.1 8.7 5 5.9 8.2 6 5.5 8.6 7 5.3 9.1 8 5.7 9.4 9 6.0 9.7 10 6.2 10.1

[0096] Statistics such as the mean and variance of DO concentration and influent organic matter concentration were calculated as features. Feature selection focused on features most closely related to DO concentration changes, such as influent organic matter concentration. A linear regression model was chosen to describe the relationship between DO concentration and influent organic matter concentration. The model is expressed as: DO = β0 + β1 × C, where DO is the dissolved oxygen concentration data from the biochemical wastewater treatment system, β0 is the initial dissolved oxygen concentration data, β1 is the model parameter data for the influent organic matter concentration data, and C is the influent organic matter concentration data.

[0097] S5. Couple the time lag effect model and the external shock load model to obtain a DO model and a time delay feedforward model to perform auxiliary operations for biochemical sewage treatment.

[0098] The present invention provides an embodiment of coupling an established time lag effect model with an external shock load model, and mathematically describing the relationship between the two models, for example, using the output of the external shock load model as the input of the time lag effect model.

[0099] When the system is affected by an external shock load, the parameters of the time-lag effect model can be dynamically adjusted based on the predictions of the external load model to reflect the system's faster or slower response to the external influence. The predictions of the external shock load model can be used to correct the predictions of the time-lag effect model to make them more consistent with reality. For example, when predicting changes in DO concentration, the impact of the external shock load is taken into account to adjust the prediction results. The time-lag effect model and the external shock load model are treated as a joint optimization problem, and the control strategy and system operating status are optimized by considering the system's dynamic response and changes in the external environment.

[0100] By constructing a DO model and a time-lag effect model in the present invention, the changing trend of the DO concentration in the biochemical sewage treatment system can be predicted more accurately, which helps to adjust the operating parameters in a timely manner, thereby improving the treatment efficiency. The time-lag effect processing takes into account the delay in the system response, making the model closer to the actual situation, and can more accurately predict the dynamic changes of the system, which helps to reduce system fluctuations and improve treatment stability. The construction of an external shock load model can take into account the impact of external environmental factors on the system, such as fluctuations in the influent water quality or changes in temperature, etc., which helps the system to better adapt to external changes and maintain stable treatment effects. Combining the time-lag effect model and the external shock load model, the obtained DO model and time-delay feedforward model can achieve precise control of the biochemical sewage treatment system, making the system operation more stable and efficient. By improving treatment efficiency and stability, energy consumption and chemical consumption in system operation can be reduced, operating costs can be reduced, and economic benefits can be improved.

[0101] Optionally, S1 includes the following steps:

[0102] The biochemical sewage treatment data is collected by using sensors or monitoring equipment preset in the biochemical sewage treatment system to obtain biochemical sewage treatment data;

[0103] The present invention provides an embodiment in which sensors, such as DO sensors, pH sensors, temperature sensors, etc., are installed in a biochemical sewage treatment system to monitor various parameters during the treatment process in real time.

[0104] The external impact load data is collected by using an external environmental monitoring device preset outside or a monitoring device in an input channel to obtain the external impact load data.

[0105] The present invention provides an embodiment in which an environmental monitoring device, such as a weather station or a rainfall sensor, is provided outside the processing system to monitor changes in the external environment, such as temperature, humidity, and rainfall, in real time.

[0106] The present invention uses preset sensors or monitoring equipment to collect data from the biochemical sewage treatment system in real time, including key parameters such as DO concentration, water intake, and temperature, so that the system's operating status can be monitored and recorded in a timely manner. The collection of biochemical sewage treatment data can fully understand the system's operating conditions, including sewage treatment efficiency, water quality trends, etc., which helps to promptly identify problems and take appropriate measures to adjust them. The collection of external shock load data can monitor the impact of external environmental factors on the biochemical sewage treatment system, such as temperature changes, rainfall conditions, etc., providing a more comprehensive reference basis for system operation.

[0107] Optionally, S2 includes the following steps:

[0108] S21, performing feature extraction on the DO concentration data and the external shock load data in the biochemical sewage treatment data to obtain DO concentration feature data and external shock load feature data;

[0109] The present invention provides an embodiment for extracting DO concentration feature data, which extracts relevant features of DO concentration, such as mean, variance, kurtosis, and skewness, as well as time series features such as autocorrelation and lagged correlation, from biochemical wastewater treatment data. Furthermore, the present invention also provides an embodiment for extracting external shock load feature data, which extracts features of various parameters from external shock load data, such as rainfall duration, intensity, and frequency.

[0110] Calculate the mean of the DO concentration data to determine the overall level of DO concentration. Calculate the variance of the DO concentration data to measure the degree of fluctuation in DO concentration, i.e., the variability of DO concentration. Calculate the kurtosis of the DO concentration data to describe the sharpness of the DO concentration data distribution. Calculate the skewness of the DO concentration data to describe the symmetry of the DO concentration data distribution. At the same time, extract time series features, including: Calculate the autocorrelation coefficient of the DO concentration data to describe the correlation between DO concentration data, i.e., the relationship between the DO concentration at the current moment and the DO concentration at previous moments. Calculate the lagged correlation coefficient of the DO concentration data to describe the correlation of DO concentration data at different time lags.

[0111] External shock load feature data extraction involves extracting the characteristics of various parameters from external shock load data. For example, for rainfall data, the following features can be extracted: duration, which is the duration of a rainfall event, i.e., the length of time from the start to the end; intensity, which is the intensity of rainfall, expressed as the amount of precipitation per unit time; and frequency, which is the frequency of rainfall events, i.e., the number of rainfall events per unit time.

[0112] Suppose we have biochemical wastewater treatment data for a period of time, including DO concentration data and external rainfall data. We calculated the DO concentration data to have a mean of 5 mg / L, a variance of 0.8 mg / L, a kurtosis of 0.5, and a skewness of 0.2. We also calculated the DO concentration data to have an autocorrelation coefficient of 0.7 and a lagged correlation coefficient of 0.5. For the external rainfall data, we extracted a single rainfall event with a duration of 2 hours, an intensity of 10 mm / h, and a frequency of 1 event per day.

[0113] S22. Perform Gaussian kernel mapping on the DO concentration characteristic data and the external shock load characteristic data to obtain DO high-dimensional space data; wherein the Gaussian kernel mapping is specifically:

[0114]

[0115] K(x) is the DO high-dimensional space data, exp is the natural exponential term, x i is the DO concentration characteristic data, x y is the characteristic data of external shock load, σ is the bandwidth data of Gaussian kernel;

[0116] The present invention provides an embodiment of performing Gaussian kernel mapping calculation on extracted DO concentration characteristic data and external impact load characteristic data to obtain DO high-dimensional spatial data. A Gaussian kernel function is used to measure the similarity between the data, wherein the bandwidth parameter σ of the Gaussian kernel function affects the decay rate of the kernel function.

[0117] Calculate a similarity matrix between the DO concentration feature data and the external shock load feature data. Convert the calculated similarity matrix into a Gaussian kernel matrix by applying a Gaussian kernel function to each element in the similarity matrix. Use a mapping function to map the original feature data into a high-dimensional space. The mapping function is the result of the Gaussian kernel function. After mapping the DO concentration feature data and the external shock load feature data into the high-dimensional space, merge them to obtain DO high-dimensional space data.

[0118] The following DO concentration feature data and external shock load feature data are provided: DO concentration feature data: [6, 7, 8, 9, 10], and external shock load feature data: [20, 25, 30, 35, 40]. The bandwidth parameter σ of the Gaussian kernel function is preset to 1, and the calculation is performed according to the above steps. The calculated results are as follows: [1, 0.88, 0.61, 0.37, 0.14], [0.88, 1, 0.88, 0.61, 0.37], [0.61, 0.88, 1, 0.88, 0.61], [0.37, 0.61, 0.88, 1, 0.88], [0.14, 0.37, 0.61, 0.88, 1]. The Gaussian kernel function is applied to each element in the similarity matrix to obtain the Gaussian kernel matrix. The original feature data is mapped to a high-dimensional space. In this example, the data in the mapped high-dimensional space is identical to the original data. The mapped DO concentration characteristic data and the external shock load characteristic data are merged to obtain the DO high-dimensional space data.

[0119] S23. Parameter optimization and iteration are performed on the DO high-dimensional spatial data to obtain a DO model; wherein the parameter optimization is specifically as follows:

[0120]

[0121] L ∈ (y,f(x)) is the model parameter optimization data, y is the real DO concentration data corresponding to the DO high-dimensional space data, f(x) is the DO concentration prediction data in the DO high-dimensional space data, and ∈ is the model prediction tolerance data.

[0122] The present invention provides an embodiment that utilizes a model parameter optimization method to adjust the parameters of a DO model based on the obtained DO high-dimensional spatial data to improve the model's fitting ability and prediction accuracy. The optimization goal is to minimize the prediction error, i.e., the difference between the model's predicted value and the true value. Based on the optimized parameters, the DO model is iteratively optimized until the model converges or a preset number of iterations is reached. During this iterative process, the model parameters are adjusted based on the actual situation to achieve better fitting and prediction performance.

[0123] Consider a high-dimensional space of DO data containing five samples, represented as [0.1, 0.2, 0.3, 0.4, 0.5]. Initialize the model parameters, for example, by selecting an initial weight of 1 and a bias of 0. Then define the mean squared error as the loss function. Next, use gradient descent to optimize the model parameters to minimize the loss function. During the optimization process, perform multiple iterations, updating the model parameters each time to minimize the loss function. The loss function is calculated after each iteration until the loss function converges or the set number of iterations is reached.

[0124] In the present invention, by extracting features from the DO concentration data and external shock load data in the biochemical sewage treatment data, the obtained DO concentration feature data and external shock load feature data can accurately reflect the system operation status and external influences, providing an accurate data basis for subsequent model construction. The application of Gaussian kernel mapping can map the feature data from the original space to the high-dimensional space, effectively expanding the feature space, improving the feature discrimination and expression ability, and helping to better capture the complex relationship between the data. Through parameter optimization and iteration, the model's fitting ability and prediction accuracy can be further improved, so that the DO model can more accurately reflect the actual situation, enhance the model's generalization ability and robustness, and improve the system's reliability and stability. The use of model prediction tolerance control can effectively adjust the model's tolerance, making the model's prediction more accurate and reliable, improving the model's adaptability and generalization ability, and reducing prediction errors. The DO model obtained by integrating the above steps can more accurately control and adjust the biochemical sewage treatment system, improve the accuracy and reliability of the control method, and make the system's operation more stable and efficient.

[0125] Optionally, S21 includes the following steps:

[0126] S211, performing primary feature extraction on the DO concentration data and the external shock load data in the biochemical sewage treatment data to obtain primary DO concentration feature data and primary external shock load feature data;

[0127] The present invention provides an embodiment for extracting primary features of DO concentration. This method uses statistical analysis methods to extract primary features of DO concentration from biochemical wastewater treatment data, such as mean, variance, kurtosis, and skewness, reflecting the overall distribution of DO concentration data. Furthermore, this method extracts primary features of external shock load data, such as rainfall and temperature changes, to reflect the impact of the external environment on the biochemical system. Statistical methods are used for feature extraction.

[0128] Primary feature extraction for DO concentration: Mean: Calculates the mean value of the DO concentration data, reflecting the overall level of DO concentration. Variance: Calculates the variance of the DO concentration data, reflecting the degree of dispersion of the DO concentration data. Kurtosis: Calculates the kurtosis of the DO concentration data, reflecting the peak state characteristics of the DO concentration data. Skewness: Calculates the skewness of the DO concentration data, reflecting the degree of skewness of the DO concentration data. Primary feature extraction for external shock load: Rainfall: Calculates the total rainfall of the rainfall event. Temperature change: Calculates the range or rate of change of the temperature data.

[0129] The DO concentration data and external shock load data over a period of time are as follows: DO concentration data: [2.3, 2.1, 2.5, 2.4, 2.2], external shock load data (rainfall): [10, 15, 8, 12, 11]. The following primary characteristics are calculated: Primary characteristics of DO concentration data: mean: 2.3, variance: 0.04, kurtosis of DO concentration data: -0.367, skewness of DO concentration data: 0.0, primary characteristics of external shock load data: total rainfall 56, temperature change: 16.

[0130] S212, performing frequency domain mapping on the primary DO concentration characteristic data and the primary external impact load characteristic data to obtain DO concentration characteristic frequency domain mapping data and external impact load characteristic frequency domain mapping data, respectively;

[0131] The present invention provides an embodiment of DO concentration frequency domain mapping, which performs a frequency domain transformation, such as a Fourier transform, on primary DO concentration characteristic data to convert the time domain data into frequency domain data, generating DO concentration frequency domain mapping data for analyzing the periodicity and spectral distribution of the DO concentration data. External shock load frequency domain mapping also performs the same frequency domain transformation on primary external shock load characteristic data to generate external shock load frequency domain mapping data for analyzing the periodicity and spectral distribution of the external shock load. Frequency domain analysis methods, such as Fourier transform, are used to transform the data.

[0132] A Fourier transform is performed on the primary DO concentration signature data, converting the time-domain data into frequency-domain data. The resulting frequency-domain mapping data can be used to analyze the periodicity and spectral distribution of the DO concentration data, for example, to detect periodic fluctuations or frequency components. The same Fourier transform is performed on the primary external shock load signature data. The resulting frequency-domain mapping data can be used to analyze the periodicity and spectral distribution of external shock loads, for example, to detect the frequency components of rainfall events or other external fluctuations.

[0133] S213, clustering the DO concentration characteristic frequency domain mapping data and the external impact load characteristic frequency domain mapping data to obtain DO concentration characteristic frequency domain clustering data and external impact load characteristic frequency domain clustering data respectively;

[0134] This invention provides an embodiment of frequency-domain clustering of DO concentration characteristics. This method performs cluster analysis on the frequency-domain mapping data of DO concentration, dividing the dataset into several categories. This ensures that data within the same category has high similarity, while data between different categories exhibits significant differences. Frequency-domain clustering of external shock load characteristics also performs similar clustering on the frequency-domain mapping data of external shock loads to distinguish different types of external shock load data. Clustering algorithms such as K-means clustering and hierarchical clustering are used for data classification and grouping.

[0135] S214, performing feature selection on the DO concentration characteristic frequency domain clustering data and the external shock load characteristic frequency domain clustering data to obtain DO concentration feature selection data and external shock load feature selection data, respectively;

[0136] The present invention provides an embodiment that selects representative DO concentration feature frequency domain mapping data based on clustering results. Specifically, the DO concentration data frequency domain mapping results are screened to retain the most discriminative and representative features. Similarly, the external shock load feature frequency domain mapping data are selected to retain the data that best reflects the external shock load characteristics.

[0137] The present invention provides an embodiment in which information gain feature selection is performed on DO concentration feature frequency domain clustering data and external shock load feature frequency domain clustering data to obtain first DO concentration feature selected data and first external shock load feature selected data, respectively. Variance analysis feature selection is performed on the DO concentration feature frequency domain clustering data and the external shock load feature frequency domain clustering data to obtain second DO concentration feature selected data and second external shock load feature selected data, respectively. Similarity screening is performed on the first DO concentration feature selected data and the second DO concentration feature selected data to obtain DO concentration feature selected data, and similarity screening is performed on the first external shock load feature selected data and the second external shock load feature selected data to obtain external shock load feature selected data.

[0138] During the information gain feature selection process, the mutual_info_classif function is used to calculate the information gain between each feature and the target variable. Information gain measures the correlation between the feature and the target variable, with higher scores indicating greater feature importance. A threshold is set, and feature selection is performed based on the information gain score. For example, empirical values ​​or cross-validation can be used to determine the threshold. Based on the information gain score and the set threshold, important features are selected.

[0139] During the ANOVA feature selection process, the f_classif function is used to calculate the ANOVA score between each feature and the target variable. ANOVA measures the explanatory power of a feature for the target variable, with higher scores indicating greater feature importance. A threshold is set for feature selection based on the ANOVA score. Similar to information gain, the threshold is determined empirically or through cross-validation. Based on the ANOVA score and the set threshold, important features are selected.

[0140] S215 , performing feature scaling on the DO concentration feature selection data and the external impact load feature selection data to obtain DO concentration feature data and external impact load feature data.

[0141] The present invention provides an embodiment for standardizing or normalizing selected DO concentration feature data to eliminate dimensional effects between different features and ensure relatively balanced weighting of each feature. The same processing is performed on selected external shock load feature data to facilitate subsequent model construction and analysis. This includes data scaling methods such as maximum-minimum scaling and normalization.

[0142] The primary feature extraction stage of the present invention can extract primary DO concentration feature data and primary external shock load feature data from the biochemical sewage treatment data. These features can comprehensively reflect the operating status of the system and external influences, and provide diverse features for subsequent analysis. Frequency domain mapping can map the original feature data to the frequency domain space, which improves the dimension and expression ability of the data, helps to capture more complex relationships between data, and makes the features more representative and discriminative. Clustering processing can divide the data into groups and cluster data with similar features, which helps to discover the inherent structure and laws in the data and improves the interpretability and discriminability of the features. Through feature selection, the features that are most sensitive and important to the system status and external influences can be screened out, reducing the data dimension, reducing the model complexity, and improving the generalization ability and prediction accuracy of the model. Feature scaling can unify the numerical ranges of different features to the same scale, eliminating the dimensional influence between features, making the feature data more comparable and interpretable, and helping to improve the stability and convergence speed of the model.

[0143] Optionally, the feature selection step in S214 is specifically as follows:

[0144] S2141. Extract cluster label data based on the DO concentration characteristic frequency domain cluster data and the external impact load characteristic frequency domain cluster data to obtain cluster label data;

[0145] The present invention provides an embodiment for extracting cluster label data from DO concentration characteristic frequency domain cluster data and external impact load characteristic frequency domain cluster data. For each sample, the system assigns a unique label or category to it according to the cluster to which it belongs. These labels or categories are cluster label data. Cluster labels can be represented by numbers or other symbols (such as using symbols, using text symbols, such as text data mapped to specific numerical values ​​corresponding to cluster centers, such as text data using high, medium, and low to describe the degree) to distinguish different clusters.

[0146] S2142, performing variance calculation based on the cluster label data, the DO concentration characteristic frequency domain cluster data, and the external impact load characteristic frequency domain cluster data to obtain characteristic variance data;

[0147] The present invention provides an embodiment that uses clustered cluster label data in conjunction with frequency-domain cluster data of DO concentration characteristics and external impact load characteristics to calculate their variance. For each cluster, the variance of the frequency-domain cluster data of DO concentration characteristics and the frequency-domain cluster data of external impact load characteristics is calculated. By obtaining the variance of each characteristic, the degree of variation of the data can be understood.

[0148] S2143, performing time series division on the DO concentration data in the biochemical sewage treatment data to obtain DO concentration time series division data;

[0149] The present invention provides an embodiment for performing time series segmentation on DO concentration data in biochemical sewage treatment data, dividing the data into different time periods or time periods. For example, the DO concentration data in the biochemical sewage treatment data is subjected to a first time series segmentation to obtain first DO concentration time series segmentation data; the DO concentration data in the biochemical sewage treatment data is subjected to a second time series segmentation to obtain second DO concentration time series segmentation data; and the first DO concentration time series segmentation data and the second DO concentration time series segmentation data are segmented and combined to obtain DO concentration time series segmentation data, wherein the first time series segmentation and the second time series segmentation are different time series segmentation methods.

[0150] The first time series division method is specifically as follows: performing a fixed segment time series change degree calculation based on the DO concentration data in the biochemical sewage treatment data to obtain fixed segment time series change degree data; generating fixed segment parameter data based on the fixed segment time series change degree data; performing time series division on the DO concentration data in the biochemical sewage treatment data based on the fixed segment parameter data to obtain first DO concentration time series division data;

[0151] The second time series division method is specifically as follows: obtaining biochemical sewage treatment system event data based on DO concentration data in the biochemical sewage treatment data; performing time series division on the DO concentration data in the biochemical sewage treatment data based on the biochemical sewage treatment system event data to obtain second DO concentration time series division data.

[0152] The fixed segment time series change degree calculation in the first time series division is to use days, hours, and seconds to calculate the change degree and obtain the corresponding change degree data; based on the corresponding change degree data, parameters are generated, such as the determination of the time period length, start time, end time, etc., to reduce the problem of simply fixed segmentation resulting in the division parameters being too large to reflect small changes in the data, and the data granularity provided by the division parameters being too small to reflect changes at the overall level of the data. The DO concentration data is time-series divided according to the event data, and the data is divided into different time periods, each of which corresponds to a specific event or operation. The present invention provides data perspectives from different angles, which helps to more accurately understand the changing patterns of DO concentration and the operation of the system.

[0153] S2144, calculating the coefficient of variation based on the DO concentration time series data to obtain DO concentration coefficient of variation data;

[0154] The present invention provides an embodiment, which divides the DO concentration time series data, calculates the coefficient of variation of the DO concentration, and evaluates the fluctuation and stability of the DO concentration data. The coefficient of variation is a statistic that describes the degree of data variation. It is the ratio of the standard deviation to the mean, and is used to measure the relative variation of the data. It is not affected by the dimension of the data. Calculating the coefficient of variation of the DO concentration can help evaluate the fluctuation and stability of the DO concentration data, thereby better understanding the operation of the biochemical sewage treatment system. The DO concentration data over a period of time is as follows: DO = [2.3, 2.1, 2.5, 2.4, 2.2]. The coefficient of variation of the DO concentration calculated based on this set of data is approximately 6.13%, indicating that the relative variation of the DO concentration data is low and has a certain degree of stability.

[0155] S2145, performing stability screening on the DO concentration time series data according to the DO concentration coefficient of variation data to obtain DO concentration stability screening data;

[0156] The present invention provides an embodiment for using DO concentration coefficient of variation data for stability screening, selecting DO concentration data with small fluctuations and stability to ensure that the selected features have a certain degree of stability.

[0157] S2146. Use the DO concentration stability screening data and the characteristic variance data to screen and process the DO concentration characteristic frequency domain clustering data and the external shock load characteristic frequency domain clustering data to obtain DO concentration characteristic selection data and external shock load characteristic selection data, respectively.

[0158] The present invention provides an embodiment, which combines the stability screening results and characteristic variance data to screen and process the DO concentration characteristic frequency domain clustering data and the external shock load characteristic frequency domain clustering data, selects the most representative and discriminative features, and obtains DO concentration feature selection data and external shock load feature selection data.

[0159] In the present invention, by extracting cluster label data, the original data can be divided into different groups, and a basis is provided for subsequent feature selection, making feature selection more targeted and effective. Combined with the calculation of feature variance data and time series stability data, the degree of change and stability of the features can be comprehensively considered, thereby more accurately evaluating the impact of the features on the system state, and improving the precision and accuracy of feature selection. By calculating the coefficient of variation and screening the stability of the DO concentration time series data in the biochemical sewage treatment data, abnormal fluctuations and noise in the time series data can be eliminated, and the stability and reliability of the data can be improved. By using feature variance data and time series stability screening data to screen and process feature data, representative and stable features can be selected more carefully, and the accuracy and reliability of feature selection can be improved. Through the screening process of feature selection, redundant information and invalid features can be eliminated, the dimension of the data and the complexity of the model are reduced, and the generalization ability and prediction accuracy of the model are improved.

[0160] Optionally, the time lag effect model includes a first time lag effect model and a second time lag effect model, and S3 includes the following steps:

[0161] S31. Performing time-lag processing on the biochemical sewage treatment data and the external impact load data to obtain time-lag processed data;

[0162] The present invention provides an embodiment of the invention, which obtains time-delay processed data by performing time-delay processing on biochemical wastewater treatment data and external impact load data. The time-delay processing adopts various methods, such as hysteresis filtering and differential method, to identify and quantify the time-delay effect of the system.

[0163] S32, constructing a differential equation model based on the DO model and the time-lag processing data to obtain a first time-lag effect model;

[0164] The present invention provides an embodiment that constructs a first time-delay effect model using a difference equation model based on a DO model and time-delay processed data. This model considers the impact of a single time-delay effect on system dynamics and describes it using a difference equation. The difference equation used to describe the impact of time-delay effects on system dynamics includes first-order or higher-order difference equations.

[0165] S33. Construct a state space model based on the DO model and the time-delay processing data to obtain a second time-delay effect model.

[0166] The present invention provides an embodiment that constructs a second time-delay effect model using a state-space model based on a DO model and time-delayed data. Compared to the first time-delay effect model, the second time-delay effect model is more complex, taking into account the influence of more time-delay effects and using a state-space model to describe the dynamic characteristics of the system. The state-space model describes the dynamic characteristics of time-delay effects, and employs methods such as the Kalman filter for state estimation and prediction.

[0167] A state-space model is established, in which the DO concentration in the system is selected as the state variable, denoted as x(t). A state equation is established based on the DO model and time-delayed processing data to describe the change in DO concentration over time. When considering time-delay effects, the state equation includes a time-delay parameter, τ. The observation equation describes the relationship between the state variable and the observed value. For biochemical wastewater treatment systems, the observed value is the measured DO concentration data. Model parameter estimation is performed using parameter estimation methods, such as the least squares method. Determination of the time-delay parameter, τ, requires the use of a model fitting or optimization algorithm.

[0168] In the present invention, by performing time-lag processing on the biochemical sewage treatment data and the external impact load data, data reflecting the time-lag characteristics of the system response can be obtained, and the dynamic response process of the system can be accurately captured. Based on the time-lag processing data and the existing DO model, a differential equation model is used to construct a first time-lag effect model, which can better describe the time-lag characteristics of the system and further improve the accuracy and fidelity of the model. Based on the time-lag processing data and the existing DO model, a state-space model is used to construct a second time-lag effect model, which can more comprehensively describe the dynamic characteristics of the system, including time-lag effects and state change laws, and help to more accurately predict the future state of the system. Based on the description of the time-lag characteristics of the system, the first time-lag effect model and the second time-lag effect model respectively adopt different modeling methods, making the model more comprehensive and flexible, and applicable to different systems and application scenarios, thereby improving the versatility and applicability of the model. By constructing a time-lag effect model, the dynamic response process of the biochemical sewage treatment system can be more deeply understood, which helps to discover problems and bottlenecks in the system, and further optimize and improve it to improve the performance and efficiency of the system.

[0169] Optionally, S31 includes the following steps:

[0170] Time series analysis is performed based on biochemical sewage treatment data and external shock load data to obtain time series analysis data;

[0171] The present invention provides an embodiment for performing time series analysis using biochemical sewage treatment data and external shock load data to obtain time series analysis data, including analysis of data trends, seasonality, periodicity and other characteristics, and calculation of relevant statistical indicators.

[0172] According to the time series analysis data, the time lag calculation is performed on the biochemical sewage treatment data and the external impact load data to obtain the time lag calculation data;

[0173] The present invention provides an embodiment for performing time lag calculation on biochemical wastewater treatment data and external impact load data based on the results of time series analysis. The time lag calculation uses the autocorrelation function method, the cross-correlation function method, etc. to determine the time lag relationship between the data.

[0174] The time-delay processing data is subjected to discrete event simulation through a distributed computing platform to obtain discrete event time-delay data;

[0175] The present invention provides an embodiment for performing discrete event simulation on time-delay processing data using a distributed computing platform to obtain discrete event time-delay data. For example, continuous time-delay data is converted into discrete event sequences for subsequent data integration and processing. Discrete event simulation is performed using a distributed computing platform, such as Hadoop and Spark, to process large-scale time-delay data.

[0176] The discrete event time-delay data are organized and integrated to obtain time-delay processed data.

[0177] The present invention provides an embodiment for organizing and integrating discrete event time-lag data to generate time-lag processing data. This step integrates discrete time-lag event sequences to facilitate subsequent time-lag effect model construction and analysis. The time-lag data is integrated and stored using data warehouse technology or distributed database technology.

[0178] Through time series analysis in the present invention, it is possible to gain an in-depth understanding of the changing trends and laws of the biochemical sewage treatment system and the external shock load, provide basic data for subsequent time-lag calculations, and contribute to a comprehensive understanding of the dynamic characteristics of the system. By performing time-lag calculations on the biochemical sewage treatment data and the external shock load data, the time-lag effect of the system can be accurately evaluated, and the degree of lag in the system response can be determined, providing an important reference basis for subsequent modeling and control. By performing discrete event simulation on a distributed computing platform, the response process of the system under different external shock conditions, including the influence of the time-lag effect, can be simulated, thereby better understanding the dynamic behavior of the system. Organizing and integrating the time-lag data obtained from the discrete event simulation can systematically manage and analyze the time-lag data, providing a reliable data basis for subsequent model establishment and analysis. By fully understanding the dynamic characteristics of the system and accurately evaluating the time-lag effect, the time-lag model of the system can be established more accurately, the accuracy and reliability of the model can be improved, and more effective support can be provided for the control and optimization of the system.

[0179] Optionally, S4 includes the following steps:

[0180] Feature extraction is performed based on system state parameter data and external impact load data in the biochemical sewage treatment data to obtain system state parameter feature data and external impact load feature data;

[0181] The present invention provides an embodiment for performing feature extraction based on system state parameter data and external impact load data in biochemical sewage treatment data to obtain system state parameter feature data and external impact load feature data, such as extracting features that have a greater impact on the system state and external load from the original data.

[0182] Perform correlation calculation based on system state parameter characteristic data and external impact load parameter data to obtain characteristic correlation data;

[0183] The present invention provides an embodiment for performing correlation calculation based on system state parameter characteristic data and external impact load parameter data to obtain characteristic correlation data, which is used, for example, to analyze the correlation between characteristics to determine which characteristics have a greater impact on the system state and external load. The correlation between characteristics is calculated using, for example, the Pearson correlation coefficient, the Spearman correlation coefficient, and the like.

[0184] Screening the system state parameter characteristic data and the external impact load characteristic data according to the characteristic correlation data to obtain system state parameter characteristic screening data and external impact load characteristic screening data;

[0185] The present invention provides an embodiment for filtering system state parameter feature data and external impact load feature data based on feature correlation data to obtain filtered system state parameter feature data and filtered external impact load feature data, thereby removing irrelevant or redundant features and improving model accuracy and efficiency. Feature filtering is performed using filtering, packaging, embedding, and other methods.

[0186] A linear model is constructed for the system state parameter characteristic screening data and the external shock load characteristic screening data to obtain the external shock load model.

[0187] The present invention provides an embodiment that constructs a linear model based on filtered system state parameter feature data and filtered external impact load feature data to obtain an external impact load model. The filtered feature data is then applied to the linear model to establish a relationship model between the system state parameters and the external impact load. A linear regression model, a ridge regression model, or the like is used to establish the linear relationship between the system state parameters and the external impact load.

[0188] Through feature extraction in the present invention, it is possible to gain an in-depth understanding of the state parameters of the biochemical sewage treatment system and the key features of the external shock load, thereby fully understanding the operating state of the system and the impact of the external environment. Through correlation calculation, the correlation between the system state parameters and the external shock load can be accurately evaluated, and the key features that affect the operation of the system can be determined, providing an important basis for feature screening. By screening the system state parameter characteristic data and the external shock load characteristic data according to the feature correlation data, it is possible to optimize the feature selection process and improve the representativeness and discrimination of the selected features. Through linear model construction, a linear relationship between the system state parameter characteristic data and the external shock load characteristic data can be established, thereby constructing an external shock load model for predicting the impact of external influences on the system. By modeling the external shock load, the impact of external factors on the system can be more accurately predicted, thereby improving the accuracy and reliability of system control and providing more effective support for the operation optimization of the biochemical sewage treatment system.

[0189] Optionally, S5 includes the following steps:

[0190] Obtain current biochemical system status data;

[0191] The present invention provides an embodiment for obtaining the status data of the current biochemical system through sensors or monitoring equipment, including but not limited to parameter data such as DO concentration, ORP (oxidation-reduction potential), and temperature.

[0192] Generate system state change data based on current biochemical system state data and time lag effect model to obtain system state change data;

[0193] The present invention provides an embodiment for generating system state change data based on current biochemical system state data and a time lag effect model. The time lag effect model is used in combination with the current system state data to predict the change trend of the system state at a future time.

[0194] The external impact load model is used to correct the system state change data to obtain the external impact change data;

[0195] The present invention provides an embodiment that uses an external impact load model to correct system state change data to obtain external impact change data. The external impact load model considers the impact of external environmental factors on the biochemical system and corrects the system state change data to more accurately reflect the actual situation.

[0196] The time-lag effect model is modified using external shock change data to obtain the DO model and the time-delay feedforward model for auxiliary operations in biochemical wastewater treatment.

[0197] The present invention provides an embodiment that uses external impact change data to modify a time-lag effect model, resulting in a DO model and a time-delay feedforward model. By accounting for the influence of external factors, the time-lag effect model is adjusted to more accurately describe the dynamic characteristics of a biochemical wastewater treatment system. A correction algorithm is used to modify the system state change data, accounting for the influence of external factors and reflecting them in the time-lag effect model.

[0198] The present invention obtains and analyzes the current biochemical system status data, which can realize real-time monitoring of the biochemical sewage treatment system and timely understand the system operation status. Based on the time-lag effect model, combined with the current system status data, the state change trend of the system can be accurately predicted. By using the external impact load model to correct the system state change data, the influence of external factors on the system operation can be considered, and the adaptability and stability to external changes can be improved. By using the external impact change data to correct the time-lag effect model, the precision and accuracy of the model can be improved, and the actual operation of the biochemical sewage treatment system can be better reflected. After obtaining the corrected DO model and time-delay feedforward model, auxiliary operations of biochemical sewage treatment can be carried out, and the efficiency and effect of biochemical sewage treatment can be improved through precise control and optimization of the system.

[0199] Optionally, the present application further provides a control system based on a DO model and a time-delay feedforward model, for executing the control method based on the DO model and the time-delay feedforward model as described above, wherein the control system based on the DO model and the time-delay feedforward model includes:

[0200] Biochemical system basic data acquisition module, used to obtain biochemical sewage treatment data and external impact load data;

[0201] A DO model building module is used to build a DO model based on DO concentration data in biochemical sewage treatment data and external shock load data to obtain a DO model;

[0202] A time lag effect model building module is used to process the time lag effect according to the DO model to obtain a time lag effect model;

[0203] An external shock load model construction module is used to construct an external shock load model based on system state parameter data and external shock load data in the biochemical sewage treatment data to obtain an external shock load model;

[0204] The DO model and time-delay feedforward model construction module is used to couple the time-delay effect model and the external impact load model to obtain the DO model and the time-delay feedforward model for auxiliary operations of biochemical sewage treatment.

[0205] The present invention aims to construct a DO model with a time-lag effect and an external shock load model by comprehensively considering the DO concentration data within the biochemical wastewater treatment system and the external shock load data. This allows for more precise control of the biochemical system's operation and fully accounts for the impact of the external environment on the system. When establishing the DO model, a time-lag effect is incorporated to account for the delayed nature of the system's response, making the model more accurate and enabling more precise prediction of system dynamics. By coupling the time-lag effect model with the external shock load model, a time-delay feedforward model is derived, enabling the control strategy to be dynamically adjusted based on the system's current state and changes in the external environment, improving the real-time and adaptability of the control. The time-delay feedforward model can more accurately predict the system's response. Combined with the external shock load model, it can optimize the biochemical wastewater treatment operation strategy, improve treatment efficiency and water quality stability, and thus achieve better treatment results. Through accurate model prediction and dynamic adjustment of the control strategy, operating costs can be effectively reduced. The system can better adapt to changes in the external environment, reduce unnecessary resource waste, and improve the economic efficiency and sustainability of the biochemical wastewater treatment system.

[0206] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes that fall within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0207] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A control method based on DO model and time-delay feedforward model, characterized in that: The method comprises: S1. Obtain biochemical wastewater treatment data and external shock load data; S2, constructing a DO model based on DO concentration data in the biochemical sewage treatment data and external shock load data to obtain a DO model; S3, processing the time lag effect according to the DO model to obtain a time lag effect model; S4. constructing an external shock load model based on the system state parameter data and the external shock load data in the biochemical sewage treatment data to obtain an external shock load model; S5. Couple the time lag effect model and the external shock load model to obtain the DO model and the time delay feedforward model to perform auxiliary operations for biochemical sewage treatment; S2 includes the following steps: S21, performing feature extraction on the DO concentration data and the external shock load data in the biochemical sewage treatment data to obtain DO concentration feature data and external shock load feature data; S22. Perform Gaussian kernel mapping on the DO concentration characteristic data and the external shock load characteristic data to obtain DO high-dimensional space data; wherein the Gaussian kernel mapping is specifically: ; is DO high-dimensional space data, is the natural exponential term, is the DO concentration characteristic data, is the characteristic data of external shock load, is the Gaussian kernel bandwidth data; S23. Parameter optimization and iteration are performed on the DO high-dimensional spatial data to obtain a DO model; wherein the parameter optimization is specifically as follows: ; Optimizing data for model parameters, is the real data of DO concentration corresponding to the DO high-dimensional space data, is the DO concentration prediction data in the DO high-dimensional space data, Predict tolerance data for the model.

2. The method according to claim 1, characterized in that S1 includes the following steps: The biochemical sewage treatment data is collected by using sensors or monitoring equipment preset in the biochemical sewage treatment system to obtain biochemical sewage treatment data; The external impact load data is collected by using an external environmental monitoring device preset outside or a monitoring device in an input channel to obtain the external impact load data.

3. The method according to claim 1, characterized in that S21 includes the following steps: S211, performing primary feature extraction on the DO concentration data and the external shock load data in the biochemical sewage treatment data to obtain primary DO concentration feature data and primary external shock load feature data; S212, performing frequency domain mapping on the primary DO concentration characteristic data and the primary external impact load characteristic data to obtain DO concentration characteristic frequency domain mapping data and external impact load characteristic frequency domain mapping data, respectively; S213, clustering the DO concentration characteristic frequency domain mapping data and the external impact load characteristic frequency domain mapping data to obtain DO concentration characteristic frequency domain clustering data and external impact load characteristic frequency domain clustering data respectively; S214, performing feature selection on the DO concentration characteristic frequency domain clustering data and the external shock load characteristic frequency domain clustering data to obtain DO concentration feature selection data and external shock load feature selection data, respectively; S215 , performing feature scaling on the DO concentration feature selection data and the external impact load feature selection data to obtain DO concentration feature data and external impact load feature data.

4. The method according to claim 3, characterized in that The specific steps of feature selection in S214 are: S2141. Extract cluster label data based on the DO concentration characteristic frequency domain cluster data and the external impact load characteristic frequency domain cluster data to obtain cluster label data; S2142, performing variance calculation based on the cluster label data, the DO concentration characteristic frequency domain cluster data, and the external impact load characteristic frequency domain cluster data to obtain characteristic variance data; S2143, performing time series division on the DO concentration data in the biochemical sewage treatment data to obtain DO concentration time series division data; S2144, calculating the coefficient of variation based on the DO concentration time series data to obtain DO concentration coefficient of variation data; S2145, performing stability screening on the DO concentration time series data according to the DO concentration coefficient of variation data to obtain DO concentration stability screening data; S2146. Use the DO concentration stability screening data and the characteristic variance data to screen and process the DO concentration characteristic frequency domain clustering data and the external shock load characteristic frequency domain clustering data to obtain DO concentration characteristic selection data and external shock load characteristic selection data, respectively.

5. The method according to claim 1, wherein The time lag effect model includes a first time lag effect model and a second time lag effect model, and S3 includes the following steps: S31. Performing time-lag processing on the biochemical sewage treatment data and the external impact load data to obtain time-lag processed data; S32, constructing a differential equation model based on the DO model and the time-lag processing data to obtain a first time-lag effect model; S33. Construct a state space model based on the DO model and the time-delay processing data to obtain a second time-delay effect model.

6. The method according to claim 5, characterized in that S31 includes the following steps: Time series analysis is performed based on biochemical sewage treatment data and external shock load data to obtain time series analysis data; According to the time series analysis data, the time lag calculation is performed on the biochemical sewage treatment data and the external impact load data to obtain the time lag calculation data; The time-delay processing data is subjected to discrete event simulation through a distributed computing platform to obtain discrete event time-delay data; The discrete event time-delay data are organized and integrated to obtain time-delay processed data.

7. The method according to claim 1, characterized in that S4 includes the following steps: Feature extraction is performed based on system state parameter data and external impact load data in the biochemical sewage treatment data to obtain system state parameter feature data and external impact load feature data; Perform correlation calculation based on system state parameter characteristic data and external impact load parameter data to obtain characteristic correlation data; Screening the system state parameter characteristic data and the external impact load characteristic data according to the characteristic correlation data to obtain system state parameter characteristic screening data and external impact load characteristic screening data; A linear model is constructed for the system state parameter characteristic screening data and the external shock load characteristic screening data to obtain the external shock load model.

8. The method according to claim 1, characterized in that S5 includes the following steps: Obtain current biochemical system status data; Generate system state change data based on current biochemical system state data and time lag effect model to obtain system state change data; The external impact load model is used to correct the system state change data to obtain the external impact change data; The time-lag effect model is modified using external shock change data to obtain the DO model and the time-delay feedforward model for auxiliary operations in biochemical wastewater treatment.

9. A control system based on DO model and time-delay feedforward model, characterized in that: For executing the control method based on the DO model and the time-delay feedforward model according to claim 1, the control system based on the DO model and the time-delay feedforward model comprises: Biochemical system basic data acquisition module, used to obtain biochemical sewage treatment data and external impact load data; A DO model building module is used to build a DO model based on DO concentration data in biochemical sewage treatment data and external shock load data to obtain a DO model; A time lag effect model building module is used to process the time lag effect according to the DO model to obtain a time lag effect model; An external shock load model construction module is used to construct an external shock load model based on system state parameter data and external shock load data in the biochemical sewage treatment data to obtain an external shock load model; The DO model and time-delay feedforward model construction module is used to couple the time-delay effect model and the external impact load model to obtain the DO model and the time-delay feedforward model for auxiliary operations of biochemical sewage treatment.

Images