Method for controlling dissolved oxygen concentration in sewage treatment process based on RBF neural network
Through the control method based on RBF neural network, the problems of low control accuracy of dissolved oxygen concentration and poor anti-interference ability in the sewage treatment system are solved, and efficient and accurate control of dissolved oxygen concentration is achieved, reducing the difficulty of model dependence and parameter setting.
Patent Information
- Application Number
- CN202411153442.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-11-22
AI Technical Summary
The existing sewage treatment systems have low control accuracy, high model dependence, poor anti-interference ability and difficult parameter setting in terms of dissolved oxygen concentration control, making it difficult to achieve precise control and stability.
Using the control method based on RBF neural network, the radial basis function neural network model is constructed by collecting and normalizing the initial dissolved oxygen concentration data, and the model parameters are optimized using genetic algorithms, and the aeration valve opening is controlled in combination with DSP to achieve dissolved oxygen. Precise control of concentration.
It improves the accuracy and robustness of dissolved oxygen concentration control, reduces dependence on accurate models, enhances anti-interference ability, simplifies the parameter setting process, and realizes efficient and precise control of the sewage treatment process.
Smart Images

Figure CN119002571A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of sewage treatment, and in particular relates to a method for controlling dissolved oxygen concentration in a sewage treatment process based on a RBF neural network. Background Art
[0002] With the acceleration of industrial development and urbanization, the efficiency and cost-effectiveness of sewage treatment have become important research topics. The aeration link in the sewage treatment system is particularly critical. It injects air into the sewage to provide the necessary dissolved oxygen, thereby promoting the growth and metabolism of microorganisms and accelerating the degradation of organic matter in the sewage. Given that the energy consumption of the aeration link accounts for 40%-50% of the energy consumption of the entire sewage treatment system, accurately controlling the dissolved oxygen concentration in the sewage treatment process is of great significance for reducing energy consumption and increasing the organic matter treatment rate. Therefore, achieving precise control of the dissolved oxygen concentration can not only improve the efficiency of sewage treatment, but also significantly reduce the production cost of the sewage treatment system, which plays an important role in achieving sustainable sewage treatment.
[0003] Traditional methods of dissolved oxygen concentration control mainly rely on manual experience and PID controllers, which are usually based on fixed parameter settings and simple mathematical models to control the aeration process in the sewage treatment process. Due to the significant time lag, time-varying and nonlinear characteristics of the dissolved oxygen concentration control process, these traditional methods have limitations in accurately predicting and controlling changes in dissolved oxygen concentration. In addition, they often require frequent manual adjustments and corrections in actual sewage treatment applications, which is not only inefficient but also difficult to maintain the stability of control accuracy.
[0004] The main disadvantages of existing traditional control technologies for controlling dissolved oxygen concentration in sewage treatment processes include: (1) Low control accuracy: When a simple PID control method is applied to the time-varying dissolved oxygen concentration control process, the steady-state error cannot be completely eliminated, thus affecting the control accuracy of the dissolved oxygen concentration.
[0005] (1) High degree of model dependence: Existing dissolved oxygen concentration control methods usually rely on accurate mathematical models; however, in actual sewage treatment processes, these models may be difficult to obtain or lack accuracy, which in turn affects the accuracy of dissolved oxygen concentration control.
[0006] (3) Poor anti-interference ability: When dealing with dissolved oxygen concentration control processes with significant hysteresis and nonlinear characteristics, existing control methods are difficult to deal with effectively, especially when the sewage treatment system encounters large external interference, these methods often show poor anti-interference ability.
[0007] (4) Difficulty in parameter tuning: Traditional PID controllers require manual parameter tuning, which not only requires professional knowledge and practical experience, but also increases the difficulty of parameter tuning of the PID controller due to the significant nonlinearity and time-varying characteristics of the dissolved oxygen concentration control process.
[0008] In summary, the existing control technology has challenges and limitations in the precise control of dissolved oxygen concentration, model establishment, system anti-interference ability and parameter setting in the sewage treatment process. In order to improve control accuracy, reduce the difficulty of model establishment, enhance anti-interference ability and reduce the difficulty of parameter setting, it is particularly necessary to conduct in-depth research and innovative improvements on existing methods. Summary of the invention
[0009] The invention provides a method for controlling dissolved oxygen concentration in a sewage treatment process based on RBF neural network, which solves the problems in the prior art.
[0010] The technical solution of the present invention is implemented as follows: a method for controlling dissolved oxygen concentration in a sewage treatment process based on an RBF neural network, the method comprising: S1: Collection of initial dissolved oxygen concentration control time series data set, including initial characteristic information of aeration volume, temperature, sewage treatment volume, chemical oxygen demand, biological oxygen demand, and corresponding optimal dissolved oxygen concentration information. Normalization processing is performed on the initial characteristic information to eliminate the influence of data scale differences, and the input reference characteristic data of the prediction model is obtained in combination with the dissolved oxygen concentration information; S2: Construct a dissolved oxygen concentration prediction model based on radial basis function (RBF) neural network to predict the nonlinear change process of dissolved oxygen concentration; S3: Use genetic algorithm to optimize the prediction model constructed in S2, improve the global optimization ability and generalization ability of the constructed model, and then improve the accuracy of dissolved oxygen concentration prediction, and obtain the final dissolved oxygen concentration prediction model; S4: Use DSP to collect the prediction data output by the optimization model, and control the opening of the aeration valve in the aeration link through DSP to control the dissolved oxygen concentration; S5: Deploy the optimized dissolved oxygen concentration prediction model and adjust the aeration valve opening of the aeration link in real time, ultimately achieving efficient and accurate control of the dissolved oxygen concentration in the sewage treatment process.
[0011] As a preferred embodiment of the method, the collection and processing process of the initial dissolved oxygen concentration control data set includes: S11 initial data collection, using a combined high-precision sensor to collect and save dissolved oxygen concentration Relevant data, including aeration volume ( Indicates no aeration. Indicates the maximum opening of the aeration valve), temperature , sewage treatment volume , Chemical Oxygen Demand , Biological Oxygen Demand ; S12 time series data division, collects sewage treatment information data from sewage treatment plants within a year, and analyzes the dissolved oxygen concentration within each hour. Information, 10 uniform time samplings are performed to obtain the dissolved oxygen concentration time series data corresponding to the time period for: (1); among which, Used to represent the dissolved oxygen concentration time series data collected in the first hour of a year; similarly, Represents the dissolved oxygen concentration time series data collected every second hour in a year; For the The dissolved oxygen concentration at each sampling time point, and ; Similarly, the aeration volume collected in the first hour of a year can be obtained ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand The time series data are: (2); (3); (4); (5); (6); S13 is a data processing method based on normalization. In order to eliminate the influence of differences in units and scales of sewage treatment data indicators, it is necessary to normalize the sampled data and use the maximum and minimum normalization methods to transform all sampled data into intervals. In the formula (7), the normalization formula is shown as: (7); among which, Used to represent the key information collected during the wastewater treatment process, namely dissolved oxygen concentration , Aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand ; Therefore, formula (1) can be rewritten as: (8); among them, After normalized data processing The dissolved oxygen concentration at each sampling time point, and ; Similarly, the same data processing is performed on equations (2)-(6) to obtain the corresponding time series data: (9); (10); (11); (12); (13); among which , , , and They are the first aeration volume, temperature, wastewater treatment volume, chemical oxygen demand and biological oxygen demand at each sampling time point, and ; S14 is based on the time series data division method of S12, and the sewage treatment information data (dissolved oxygen concentration , Aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand ) to obtain a total Group sewage treatment data: (14); (15); (16); (17); (18); (19); the aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand As input reference characteristic data, dissolved oxygen concentration As the actual data of dissolved oxygen concentration, it is used to compare with the subsequent predicted values of dissolved oxygen concentration.
[0012] As a preferred implementation method, the process of constructing a dissolved oxygen concentration prediction model based on a radial basis function (RBF) neural network includes: designing a network structure, selecting an activation function, initializing network parameters, and optimizing the network: The input layer S21 is used to connect the external input data with the RBF neural network and the sewage treatment data described in S14 As input information of a dissolved oxygen concentration prediction model, as input information of a dissolved oxygen concentration prediction model, and as output information of the prediction model; The S22 hidden layer is used to process the nonlinear mapping process from the input layer to the hidden layer. In order to improve the local response capability, computational efficiency and multi-dimensional problem processing capability of the dissolved oxygen concentration prediction model, the Gaussian function is selected as the basis function of the hidden layer, as shown in formula (20): (20); among them is the number of hidden layer nodes; Used to represent sewage treatment input data ; is the center of the Gaussian function, and have the same dimensions; is the Gaussian function variance; Euclidean norm To express arrive The distance between exist has a unique maximum value at The S23 output layer is used to process the linear mapping process from the hidden layer to the output to obtain the output prediction information of the dissolved oxygen concentration. It is expressed by formula (16): (21); among them For the The weights of the hidden layer nodes to the output layer.
[0013] As a preferred embodiment of the method, the process of optimizing the prediction model constructed by S2 using a genetic algorithm includes: using a genetic algorithm to optimize the center of the Gaussian function of the RBF neural network , Gaussian function variance , weight To optimize: S31 sets the parameter set of the RBF neural network (the center of the Gaussian function , Gaussian function variance , weight ) is encoded as a chromosome , used to represent the solution of the required parameter set, and its encoding method adopts binary encoding, that is, the solution of each chromosome is represented by a binary string composed of 0 and 1; The chromosome encoding is shown in formula (22): (22); The set of chromosomes is expressed by formula (23): (twenty three); S32 randomly generates the initial population ,in is the population size; and calculate the fitness of each chromosome in S31 , , fitness is used to indicate the quality of the parameter set solution. The higher the fitness, the closer the solution represented by the chromosome is to the optimal solution. is the mean square error of the chromosome set in S3-1; S33 performs crossover and mutation operations on the initial population in S32. In order to improve the robustness and global search capability of the optimization model, the population crossover rate is set according to the empirical value. , mutation rate , and generate a new generation of population ; If the number of crossover and mutation reaches the maximum number of iterations of 1000 in S34, the prediction model optimization is stopped, otherwise, it returns to S32 to continue iteration; and the optimal solution of the RBF neural network parameter set is selected from the crossover and compiled population, that is, a set of parameters with the smallest error between the actual value of dissolved oxygen concentration and the predicted value of dissolved oxygen concentration. At this time, the optimal Gaussian center is , the optimal Gaussian function variance is , the optimal weight is , and substitute the optimal solution into the prediction model built by S21 to obtain the output prediction optimization information of dissolved oxygen concentration .
[0014] As a preferred embodiment of the method, the DSP is used to collect the prediction data output by the optimization model and control the opening information of the aeration tank to adjust the aeration amount. The process includes: S41 uses the DSP's analog-to-digital converter to collect the predicted dissolved oxygen concentration data output by the optimization model and the actual dissolved oxygen concentration ; Secondly, calculate and Finally, the aeration volume of the aeration tank is controlled in real time according to the difference. , and then the dissolved oxygen concentration in the sewage pool, the high-speed signal processing capability of DSP can ensure that the dissolved oxygen concentration in the sewage treatment process is kept in the most ideal state in real time.
[0015] As a preferred implementation method, the process of deploying the optimization prediction model to an actual dissolved oxygen concentration control system includes: S51 is based on aeration volume sensor, temperature sensor, sewage treatment volume sensor, chemical oxygen demand sensor, biological oxygen demand sensor and dissolved oxygen concentration sensor, and collects key information of sewage treatment process at 6-minute intervals, including aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand , Biological Oxygen Demand and dissolved oxygen concentration , a total of 100 sets of data were collected, and the data were processed using the normalization method to obtain the input reference feature data of the prediction model; S52 uses the genetic algorithm described in S3 to optimize the prediction model built by S2 to obtain the optimized prediction output data of dissolved oxygen concentration in the sewage treatment process ; S53 outputs data based on DSP optimization prediction of dissolved oxygen concentration With actual data Collect and calculate the difference between the two, and use the difference as the control signal of the aeration tank valve opening to achieve precise control of dissolved oxygen concentration; S54 changes the sampling time of the sewage treatment data so that the time series data sample obtained by collecting the sewage treatment data is large enough to meet the real-time control requirements of the system, and gradually replaces the historical time data with the newly collected time series data information, and repeats steps S52 to S53 to achieve automatic adjustment of the aeration tank opening and efficient and precise control of the dissolved oxygen concentration.
[0016] After adopting the above technical solution, the beneficial effects of the present invention are: (1) Eliminating the impact of data scale differences: A joint high-precision sensor is used to collect key data in the sewage treatment process, including aeration volume, temperature, sewage treatment volume, chemical oxygen demand, biological oxygen demand, and the corresponding optimal dissolved oxygen concentration, and the normalization method is used to eliminate the impact of data scale differences.
[0017] (2) Reduced dependence on precise models: By designing a dissolved oxygen concentration prediction model based on RBF neural network, RBF neural network has strong fitting ability and generalization, and can automatically learn the complex nonlinear relationship between input and output through training data, thereby reducing dependence on precise models. In addition, RBF neural network has strong adaptability and can efficiently handle the nonlinear change process of dissolved oxygen concentration.
[0018] (3) Improved the robustness of the dissolved oxygen concentration control process: The RBF neural network model was optimized through a genetic algorithm, and the parameters of the basis function used in the hidden layer of the RBF neural network were optimized to find an optimal solution, which improved the global optimization and generalization capabilities of the model, thereby improving the accuracy and robustness of dissolved oxygen prediction.
[0019] (4) Improved dissolved oxygen concentration control accuracy: DSP is used to collect the predicted data output by the optimization model. As a high-speed digital signal processing chip, DSP can effectively improve the control accuracy of the aeration process, thereby improving the control accuracy of the dissolved oxygen concentration. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0021] Figure 1 This is a schematic diagram of data collection in combination with a high-precision sensor in the present invention; Figure 2 It is a structural diagram of the dissolved oxygen concentration prediction model based on RBF neural network of the present invention; Figure 3 The schematic diagram of the principle of DSP collecting optimization prediction data and controlling the opening of the aeration tank in the present invention; Figure 4 This is an implementation block diagram of the dissolved oxygen concentration control system based on the RBF neural network of the present invention. DETAILED DESCRIPTION
[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example
[0023] like Figure 1 As shown, a method for controlling dissolved oxygen concentration in a sewage treatment process based on an RBF neural network comprises: S1: collecting an initial dissolved oxygen concentration control time series data set, including initial characteristic information of aeration volume, temperature, sewage treatment volume, chemical oxygen demand, biological oxygen demand, and corresponding optimal dissolved oxygen concentration information, performing normalization processing on the initial characteristic information to eliminate the influence of data scale differences, and combining the dissolved oxygen concentration information to obtain input reference characteristic data of the prediction model; S2: Construct a dissolved oxygen concentration prediction model based on radial basis function (RBF) neural network to predict the nonlinear change process of dissolved oxygen concentration; S3: Use genetic algorithm to optimize the prediction model constructed in S2, improve the global optimization ability and generalization ability of the constructed model, and then improve the accuracy of dissolved oxygen concentration prediction, and obtain the final dissolved oxygen concentration prediction model; S4: Use DSP to collect the prediction data output by the optimization model, and control the opening of the aeration valve in the aeration link through DSP to control the dissolved oxygen concentration; S5: Deploy the optimized dissolved oxygen concentration prediction model and adjust the aeration valve opening of the aeration link in real time, ultimately achieving efficient and accurate control of the dissolved oxygen concentration in the sewage treatment process.
[0024] Furthermore, the collection and processing of the initial dissolved oxygen concentration control data set includes: S11 initial data collection, using a combined high-precision sensor to collect and save dissolved oxygen concentration Relevant data, including aeration volume ( Indicates no aeration. Indicates the maximum opening of the aeration valve), temperature , sewage treatment volume , Chemical Oxygen Demand , Biological Oxygen Demand ; S12 Time series data division: Since the sewage treatment process is dynamic and continuously changing, the change of dissolved oxygen concentration cannot be accurately predicted based on the sewage treatment characteristic data at a single moment. The time series characteristic data division method collects sewage treatment information data from the sewage treatment plant within a year, and predicts the dissolved oxygen concentration within each hour. Information, 10 uniform time samplings are performed to obtain the dissolved oxygen concentration time series data corresponding to the time period for: (1); among which, Used to represent the dissolved oxygen concentration time series data collected in the first hour of a year; similarly, Represents the dissolved oxygen concentration time series data collected every second hour in a year; For the The dissolved oxygen concentration at each sampling time point, and ; Similarly, the aeration volume collected in the first hour of a year can be obtained ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand The time series data are: (2); (3); (4); (5); (6); S13 is a data processing method based on normalization. In order to eliminate the influence of differences in units and scales of sewage treatment data indicators, it is necessary to normalize the sampled data. The maximum and minimum normalization methods are used to transform all sampled data into intervals. In the formula (7), the normalization formula is shown as: (7); among which, Used to represent the key information collected during the wastewater treatment process, namely dissolved oxygen concentration , Aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand ; Therefore, formula (1) can be rewritten as: (8); among them, After normalized data processing The dissolved oxygen concentration at each sampling time point, and ; Similarly, the same data processing is performed on equations (2)-(6) to obtain the corresponding time series data: (9); (10); (11); (12); (13); among which , , , and They are the first aeration volume, temperature, wastewater treatment volume, chemical oxygen demand and biological oxygen demand at each sampling time point, and ; S14 is based on the time series data division method of S12, and the sewage treatment information data (dissolved oxygen concentration , Aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand ) to obtain a total Group sewage treatment data: (14); (15); (16); (17); (18); (19); The aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand As input reference characteristic data, dissolved oxygen concentration As the actual data of dissolved oxygen concentration, it is used to compare with the subsequent predicted values of dissolved oxygen concentration.
[0025] Based on the above steps, the historical data collected within one year are analyzed to determine the optimal dissolved oxygen concentration in the sewage treatment process, that is, the dissolved oxygen concentration that can make the sewage treatment efficiency reach more than 85%.
[0026] Furthermore, the process of constructing a dissolved oxygen concentration prediction model based on radial basis function (RBF) neural network includes: designing network structure, selecting activation function, initializing network parameters, and optimizing network.
[0027] RBF neural network is a feedforward neural network with good generalization performance, simple network structure, strong local stability and other characteristics. It uses radial basis function as activation function to construct the output of hidden layer neurons. According to the typical RBF neural network model, Figure 2 The dissolved oxygen concentration prediction model based on RBF neural network is shown.
[0028] The input layer S21 is used to connect the external input data with the RBF neural network and the sewage treatment data described in S14 As input information for the dissolved oxygen concentration prediction model, As the output information of the prediction model; The S22 hidden layer is used to process the nonlinear mapping process from the input layer to the hidden layer. In order to improve the local response capability, computational efficiency and multi-dimensional problem processing capability of the dissolved oxygen concentration prediction model, the Gaussian function is selected as the basis function of the hidden layer, as shown in formula (20): (20); among them is the number of hidden layer nodes; Used to represent sewage treatment input data ; is the center of the Gaussian function, and have the same dimensions; is the Gaussian function variance; Euclidean norm To express arrive The distance between exist has a unique maximum value at The S23 output layer is used to process the linear mapping process from the hidden layer to the output to obtain the output prediction information of the dissolved oxygen concentration. It is expressed by formula (16): (21); among them For the The weights of the hidden layer nodes to the output layer.
[0029] Furthermore, in order to minimize the error between the predicted value of dissolved oxygen concentration and the actual value of dissolved oxygen concentration of the constructed model and improve the robustness and generalization ability of the prediction model, the prediction model constructed by S2 is optimized by genetic algorithm. The process includes: using genetic algorithm to optimize the Gaussian function center of RBF neural network. Gaussian function variance , weight To optimize: S31 sets the parameter set of the RBF neural network (the center of the Gaussian function , Gaussian function variance , weight ) is encoded as a chromosome , used to represent the solution of the required parameter set, and its encoding method adopts binary encoding, that is, the solution of each chromosome is represented by a binary string composed of 0 and 1; The chromosome encoding is shown in formula (22): (22); The set of chromosomes is expressed by formula (23): (twenty three); S32 randomly generates the initial population ,in is the population size; and calculate the fitness of each chromosome in S31 , , fitness is used to indicate the quality of the parameter set solution. The higher the fitness, the closer the solution represented by the chromosome is to the optimal solution. is the mean square error of the chromosome set in S3-1; S33 performs crossover and mutation operations on the initial population in S32. In order to improve the robustness and global search capability of the optimization model, the population crossover rate is set according to the empirical value. , mutation rate , and generate a new generation of population ; If the number of crossover and mutation reaches the maximum number of iterations of 1000 in S34, the prediction model optimization is stopped, otherwise, it returns to S32 to continue iteration; and the optimal solution of the RBF neural network parameter set is selected from the crossover and compiled population, that is, a set of parameters with the smallest error between the actual value of dissolved oxygen concentration and the predicted value of dissolved oxygen concentration. At this time, the optimal Gaussian center is , the optimal Gaussian function variance is , the optimal weight is , and substitute the optimal solution into the prediction model built by S21 to obtain the output prediction optimization information of dissolved oxygen concentration .
[0030] like Figure 3 As shown, DSP is used to collect the prediction data output by the optimization model and control the opening information of the aeration tank to adjust the aeration volume. The process includes: S41 uses the DSP's analog-to-digital converter to collect the predicted dissolved oxygen concentration data output by the optimization model and the actual dissolved oxygen concentration ; Secondly, calculate and Finally, the aeration volume of the aeration tank is controlled in real time according to the difference. , and then the dissolved oxygen concentration in the sewage pool, the high-speed signal processing capability of DSP can ensure that the dissolved oxygen concentration in the sewage treatment process is kept in the most ideal state in real time.
[0031] like Figure 4 As shown in the figure, the process of deploying the optimization prediction model to the actual dissolved oxygen concentration control system includes: S51 is based on aeration volume sensor, temperature sensor, sewage treatment volume sensor, chemical oxygen demand sensor, biological oxygen demand sensor and dissolved oxygen concentration sensor, and collects key information of sewage treatment process at 6-minute intervals, including aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand , Biological Oxygen Demand and dissolved oxygen concentration , a total of 100 sets of data were collected, and the data were processed using the normalization method to obtain the input reference feature data of the prediction model; S52 uses the genetic algorithm described in S3 to optimize the prediction model built by S2 to obtain the optimized prediction output data of dissolved oxygen concentration in the sewage treatment process ; S53 outputs data based on DSP optimization prediction of dissolved oxygen concentration With actual data Collect and calculate the difference between the two, and use the difference as the control signal of the aeration tank valve opening to achieve precise control of dissolved oxygen concentration; S54 changes the sampling time of the sewage treatment data so that the time series data sample obtained by collecting the sewage treatment data is large enough to meet the real-time control requirements of the system, and gradually replaces the historical time data with the newly collected time series data information, and repeats steps S52 to S53 to achieve automatic adjustment of the aeration tank opening and efficient and precise control of the dissolved oxygen concentration.
[0032] In summary, the key data in the sewage treatment process are collected and processed, including the collection of dissolved oxygen concentration, aeration volume, sewage treatment volume, chemical oxygen demand, and biological oxygen demand, to provide initial data for the subsequent data processing process; secondly, a dissolved oxygen concentration prediction model based on RBF neural network is constructed to predict the nonlinear change of dissolved oxygen concentration; then the dissolved oxygen concentration prediction model based on RBF neural network is optimized by genetic algorithm to improve the global optimization ability of the dissolved oxygen concentration prediction model and avoid the model from falling into the local optimum, so as to improve the accuracy of the output results of the dissolved oxygen concentration prediction model; then the digital signal processing chip (Digital Signal Processing, DSP) is used to collect the predicted data output by the optimization model to control the aeration valve opening of the aeration link, and then control the dissolved oxygen concentration in the sewage treatment process; finally, the prediction model and data collection process are deployed in the actual dissolved oxygen concentration control system to achieve precise control of the dissolved oxygen concentration in the sewage treatment process.
[0033] In the description of the present invention, it should be understood that the terms "longitudinal", "lateral", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a mechanical connection or an electrical connection, or it can be the internal communication of two elements, it can be a direct connection, or it can be an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.
[0034] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for controlling dissolved oxygen concentration in a sewage treatment process based on RBF neural network, characterized in that: The method includes: S1: Collection of initial dissolved oxygen concentration control time series data set, including initial characteristic information of aeration volume, temperature, sewage treatment volume, chemical oxygen demand, biological oxygen demand, and corresponding optimal dissolved oxygen concentration information. Normalization processing is performed on the initial characteristic information to eliminate the influence of data scale differences, and the input reference characteristic data of the prediction model is obtained in combination with the dissolved oxygen concentration information; S2: Construct a dissolved oxygen concentration prediction model based on radial basis function (RBF) neural network to predict the nonlinear change process of dissolved oxygen concentration; S3: Use genetic algorithm to optimize the prediction model constructed in S2, improve the global optimization ability and generalization ability of the constructed model, and then improve the accuracy of dissolved oxygen concentration prediction, and obtain the final dissolved oxygen concentration prediction model; S4: Use DSP to collect the prediction data output by the optimization model, and control the opening of the aeration valve in the aeration link through DSP to control the dissolved oxygen concentration; S5: Deploy the optimized dissolved oxygen concentration prediction model and adjust the aeration valve opening of the aeration link in real time, ultimately achieving efficient and accurate control of the dissolved oxygen concentration in the sewage treatment process.
2. A method for controlling dissolved oxygen concentration in a sewage treatment process based on RBF neural network according to claim 1, characterized in that: The collection and processing process of the initial dissolved oxygen concentration control data set includes: S11 initial data collection, using a combined high-precision sensor to collect and save dissolved oxygen concentration Relevant data, including aeration volume ( Indicates no aeration. Indicates the maximum opening of the aeration valve), temperature , sewage treatment volume , Chemical Oxygen Demand , Biological Oxygen Demand ; S12 time series data division, collects sewage treatment information data from sewage treatment plants within a year, and analyzes the dissolved oxygen concentration within each hour. Information, 10 uniform time samplings are performed to obtain the dissolved oxygen concentration time series data corresponding to the time period for: (1); in, Used to represent the dissolved oxygen concentration time series data collected in the first hour of a year; similarly, Represents the dissolved oxygen concentration time series data collected every second hour in a year; For the The dissolved oxygen concentration at each sampling time point, and ; Similarly, the aeration volume collected in the first hour of a year can be obtained ; Similarly, the aeration volume collected in the first hour of a year can be obtained ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand The time series data are: (2); (3); (4); (5); (6); S13 is a data processing method based on normalization. In order to eliminate the influence of differences in units and scales of sewage treatment data indicators, it is necessary to normalize the sampled data and use the maximum and minimum normalization methods to transform all sampled data into intervals. In the formula (7), the normalization formula is shown as: (7); among which, Used to represent the key information collected during the wastewater treatment process, namely dissolved oxygen concentration , Aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand ; Therefore, formula (1) can be rewritten as: (8); in, After normalized data processing The dissolved oxygen concentration at each sampling time point, and ; Similarly, the same data processing is performed on equations (2)-(6) to obtain the corresponding time series data: (9); (10); (11); (12); (13); among which , , , and They are the first aeration volume, temperature, wastewater treatment volume, chemical oxygen demand and biological oxygen demand at each sampling time point, and ; S14 is based on the time series data division method of S12, and the sewage treatment information data (dissolved oxygen concentration , Aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand ) to obtain a total Group sewage treatment data: (14); (15); (16); (17); (18); (19); the aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand and biological oxygen demand As input reference characteristic data, dissolved oxygen concentration As the actual data of dissolved oxygen concentration, it is used to compare with the subsequent predicted values of dissolved oxygen concentration.
3. The method for controlling dissolved oxygen concentration in a sewage treatment process based on RBF neural network according to claim 1 is characterized in that: The process of constructing a dissolved oxygen concentration prediction model based on a radial basis function (RBF) neural network includes: designing a network structure, selecting an activation function, initializing network parameters, and optimizing the network: The input layer S21 is used to connect the external input data with the RBF neural network and the sewage treatment data described in S14 As input information for the dissolved oxygen concentration prediction model, As the output information of the prediction model; The S22 hidden layer is used to process the nonlinear mapping process from the input layer to the hidden layer. In order to improve the local response capability, computational efficiency and multi-dimensional problem processing capability of the dissolved oxygen concentration prediction model, the Gaussian function is selected as the basis function of the hidden layer, as shown in formula (20): (20); among them is the number of hidden layer nodes; Used to represent sewage treatment input data ; is the center of the Gaussian function, and have the same dimensions; is the Gaussian function variance; Euclidean norm To express arrive The distance between exist has a unique maximum value at The S23 output layer is used to process the linear mapping process from the hidden layer to the output to obtain the output prediction information of the dissolved oxygen concentration. It is expressed by formula (16): (21); among them For the The weights of the hidden layer nodes to the output layer.
4. The method for controlling dissolved oxygen concentration in a sewage treatment process based on RBF neural network according to claim 1 is characterized in that: The process of optimizing the prediction model constructed by S2 using the genetic algorithm includes: using the genetic algorithm to optimize the Gaussian function center of the RBF neural network , Gaussian function variance , weight To optimize: S31 sets the parameter set of the RBF neural network (the center of the Gaussian function , Gaussian function variance , weight ) is encoded as a chromosome , used to represent the solution of the required parameter set, and its encoding method adopts binary encoding, that is, the solution of each chromosome is represented by a binary string composed of 0 and 1; The chromosome encoding is shown in formula (22): twenty two); The set of chromosomes is expressed by formula (23): (twenty three); S32 randomly generates the initial population ,in is the population size; and calculate the fitness of each chromosome in S31 , , fitness is used to indicate the quality of the parameter set solution. The higher the fitness, the closer the solution represented by the chromosome is to the optimal solution. is the mean square error of the chromosome set in S3-1; S33 performs crossover and mutation operations on the initial population in S32. In order to improve the robustness and global search capability of the optimization model, the population crossover rate is set according to the empirical value. , mutation rate , and generate a new generation of population ; If the number of crossover and mutation reaches the maximum number of iterations of 1000 in S34, the prediction model optimization is stopped, otherwise, it returns to S32 to continue iteration; and the optimal solution of the RBF neural network parameter set is selected from the crossover and compiled population, that is, a set of parameters with the smallest error between the actual value of dissolved oxygen concentration and the predicted value of dissolved oxygen concentration. At this time, the optimal Gaussian center is , the variance of the optimal Gaussian function is , the optimal weight is , and substitute the optimal solution into the prediction model built by S21 to obtain the output prediction optimization information of dissolved oxygen concentration .
5. The method for controlling dissolved oxygen concentration in sewage treatment process based on RBF neural network according to claim 1 is characterized in that: The DSP is used to collect the prediction data output by the optimization model and control the opening information of the aeration tank to adjust the aeration volume. The process includes: S41 uses the DSP's analog-to-digital converter to collect the predicted dissolved oxygen concentration data output by the optimization model and the actual dissolved oxygen concentration ; Secondly, calculate and Finally, the aeration volume of the aeration tank is controlled in real time according to the difference. , and then the dissolved oxygen concentration in the sewage pool, the high-speed signal processing capability of DSP can ensure that the dissolved oxygen concentration in the sewage treatment process is kept in the most ideal state in real time.
6. The method for controlling dissolved oxygen concentration in sewage treatment process based on RBF neural network according to claim 1, characterized in that: The process of deploying the optimization prediction model to the actual dissolved oxygen concentration control system includes: S51 is based on aeration volume sensor, temperature sensor, sewage treatment volume sensor, chemical oxygen demand sensor, biological oxygen demand sensor and dissolved oxygen concentration sensor, and collects key information of sewage treatment process at 6-minute intervals, including aeration volume ,temperature , sewage treatment volume , Chemical Oxygen Demand , Biological Oxygen Demand and dissolved oxygen concentration , a total of 100 sets of data were collected, and the data were processed using the normalization method to obtain the input reference feature data of the prediction model; S52 uses the genetic algorithm described in S3 to optimize the prediction model built by S2 to obtain the optimized prediction output data of dissolved oxygen concentration in the sewage treatment process ; S53 outputs data based on DSP optimization prediction of dissolved oxygen concentration With actual data Collect and calculate the difference between the two, and use the difference as the control signal of the aeration tank valve opening to achieve precise control of dissolved oxygen concentration; S54 changes the sampling time of the sewage treatment data so that the time series data sample obtained by collecting the sewage treatment data is large enough to meet the real-time control requirements of the system, and gradually replaces the historical time data with the newly collected time series data information, and repeats steps S52 to S53 to achieve automatic adjustment of the aeration tank opening and efficient and precise control of the dissolved oxygen concentration.
Citation Information
Cited By
Method and system for predicting centralized collection rate of urban domestic sewage and storage medium thereof
CN119558547A
Oxygen concentration regulation and control method based on hybrid intelligent model
CN120949841A
Dissolved oxygen closed-loop regulation and control method used in sewage treatment process
CN121832653A
