Method for optimizing deamination reactor based on CFD and BP neural network

Abstract

The invention relates to the technical field of ammonia-nitrogen wastewater degradation, in particular to a deamination reactor optimization method based on CFD and BP neural networks. The method comprises the following steps: performing three-dimensional modeling according to geometric parameters of a deamination reactor, and obtaining a sample set; training a neural network through the sample set; and predicting the cavitation volume fraction and length in the deamination reactor by using a neural network. According to the deamination reactor optimization method based on the CFD and the BP neural network, the cavitation intensity of the deamination reactor is predicted based on a CFD numerical simulation result in combination with the neural network. CFD numerical simulation has the defects of low simulation speed, tedious operation steps and large workload, the cavitation intensity in the device with different parameters is calculated by constructing a neural network model, the amount of information to be processed is large, the calculation speed is high, complex problems can be rapidly processed, and the calculation efficiency is improved.

Description

Technical Field

[0001] The invention relates to the technical field of ammonia nitrogen wastewater degradation, and in particular to a deammonification reactor optimization method based on CFD and BP neural network. Background Art

[0002] The deamination effect of the deamination reactor is affected by many factors such as the structure and operating parameters of the device. When using the deamination reactor to remove ammonia nitrogen, improper device design cannot achieve better deamination effect. The design of the deamination reactor is usually based on past design experience and traditional qualitative analysis methods to determine its structure and operating parameters.

[0003] The traditional analysis method is to analyze the impact of different operating parameters and structural parameters on the deamination effect of the device. Consider all possible results within the parameter value range, and compare different results to obtain the impact of different process parameters on the deamination effect of the deamination reactor, thereby completing the design of the deamination reactor. Since the traditional optimization method needs to consider all feasibility and verify them one by one, the optimization efficiency is low. When encountering complex structures, it is necessary to build multiple geometric shapes and test them in different configurations, resulting in high experimental costs. Traditional analysis methods are gradually no longer applicable.

[0004] The CFD method has a large number of mathematical calculation models that can reflect the fluid flow, heat transfer and mass transfer process in the equipment. However, if only the CFD numerical simulation calculation method is used, in addition to the calculation cost, there will be problems such as slow simulation, large workload and long time consumption.

[0005] Artificial neural network (ANN) is a mathematical model that imitates the information processing work of the human brain. Based on its powerful research capabilities and accurate prediction effects, it has been widely used in many important fields such as medicine, engineering, biology, and agriculture. Among them, the back propagation (BP) neural network is one of the most widely used artificial neural networks. By establishing a BP neural network model coupled with CFD, the various process parameters of the deammoniation reactor can be quickly obtained, thereby improving the optimization efficiency of the device.

[0006] In summary, the existing optimization design of deamination reactors has the following defects: With the complexity of the device structure, the traditional qualitative analysis method is no longer applicable. Although CFD can improve the optimization efficiency while saving experimental costs, it requires a lot of device model design and numerical simulation calculations. The high calculation time cost limits the application of deamination reactor optimization design methods in engineering.

[0007] To this end, a deamination reactor optimization method based on CFD and BP neural network was designed to provide another technical solution to the above technical problems. Summary of the invention

[0008] Based on this, it is necessary to provide a deamination reactor optimization method based on CFD and BP neural network to solve the technical problems raised in the above background technology.

[0009] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0010] A deamination reactor optimization method based on CFD and BP neural network, the steps are as follows:

[0011] According to the geometric parameters of the deamination reactor, three-dimensional modeling is carried out and a sample set is obtained;

[0012] Train the neural network through the sample set;

[0013] Prediction of cavitation volume fraction and length in deamination reactor using neural network.

[0014] As a preferred implementation of the deamination reactor optimization method based on CFD and BP neural network provided by the present invention, three-dimensional modeling is performed according to the geometric parameters of the deamination reactor, and the steps are as follows:

[0015] According to the geometric parameters of the deamination reactor, three-dimensional modeling is carried out, and the steps are as follows:

[0016] Determine the device process parameters that need to be optimized, including water jet hole diameter, inlet flow rate and device top outlet pressure;

[0017] Determine the structure and size of the deamination reactor;

[0018] Divide the geometric model into grids and import it into Fluent fluid calculation software. Set the model boundary conditions, set the inlet as velocity inlet and the outlet as pressure outlet.

[0019] Reasonable turbulence, multiphase flow and cavitation mechanism models are selected to carry out numerical simulation of the deammoniation reactor. According to the set boundary conditions, the cavitation volume fraction and cavitation length distribution data at any position inside the device are obtained. At the same time, random data samples, i.e., sample sets, are generated. The input is the process parameters, and the output is the corresponding response parameters.

[0020] As a preferred implementation of the deamination reactor optimization method based on CFD and BP neural network provided by the present invention, the steps of setting the model boundary conditions are as follows:

[0021] In the boundary condition options, set the reactor inlet flow rate and the reactor top outlet pressure.

[0022] As a preferred implementation of the deamination reactor optimization method based on CFD and BP neural network provided by the present invention, the neural network training steps are as follows:

[0023] A. After preprocessing, the sample set is divided into training set, test set and validation set;

[0024] B. Take the process parameters as input and the response parameters as output, establish a neural network model, and train the neural network model through the training set;

[0025] C. Repeat step B until the error meets the accuracy requirement or the number of iterations reaches the maximum number of iterations. After the training is completed, the cavitation volume fraction and length distribution prediction data are analyzed and evaluated through the test set;

[0026] D. Verify the neural network model through the validation set.

[0027] As a preferred implementation of the deammoniation reactor optimization method based on CFD and BP neural network provided by the present invention, the sample set is divided into a training set, a test set and a validation set in a ratio of 7:1.5:1.5 after preprocessing.

[0028] As a preferred implementation of the deamination reactor optimization method based on CFD and BP neural network provided by the present invention, the response parameters include cavitation volume fraction and cavitation length.

[0029] As a preferred embodiment of the deamination reactor optimization method based on CFD and BP neural network provided by the present invention, the neural network is a three-layer network structure, including an input layer, an output layer and a hidden layer;

[0030] The number of neurons in the input layer is the number of process parameters;

[0031] The number of neurons in the output layer is the number of response parameters;

[0032] The number of hidden layer neurons is used to optimize the neural network fitting and generalization capabilities.

[0033] It can be seen without a doubt that the above-mentioned technical solution of the present application can definitely solve the technical problem to be solved by the present application.

[0034] At the same time, through the above technical solutions, the present invention has at least the following beneficial effects:

[0035] The present invention provides a deamination reactor optimization method based on CFD and BP neural network, which predicts the cavitation intensity of the deamination reactor based on the CFD numerical simulation results combined with the neural network. CFD numerical simulation has the defects of slow simulation speed, cumbersome operation steps and large workload. By constructing a neural network model to calculate the cavitation intensity in the device with different parameters, it can process a large amount of information, has a fast calculation speed, can quickly handle complex problems, and improves the calculation efficiency;

[0036] CFD numerical simulation is used to calculate the cavitation volume fraction and length distribution data at any position inside the deamination reactor, and the complex flow field distribution inside the deamination reactor is visualized through the data;

[0037] The neural network is trained with the data obtained from numerical simulation, and the cavitation effect generated in the device is simulated and predicted to improve the deammoniation efficiency of the device.

[0038] The method of predicting the cavitation intensity of the deamination reactor based on CFD numerical simulation results combined with neural networks greatly saves the calculation time for obtaining results through numerical simulation; and according to the results obtained by neural network prediction, the process parameters of the deamination reactor can be further adjusted and optimized to obtain a better operating state of the device and further improve the deamination efficiency of the deamination reactor. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 is a flow chart of the present invention;

[0041] Figure 2 It is a model diagram of the deamination reactor of the present invention;

[0042] Figure 3 It is the BP neural network structure diagram of the present invention;

[0043] Figure 4 This is a diagram of the cavitation volume fraction prediction results of the present invention;

[0044] Figure 5 This is a diagram of the cavitation length prediction results of the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0046] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings.

[0047] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features and technical solutions in the embodiments may be combined with each other.

[0048] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0049] Reference Figure 1-Figure 5 , a deamination reactor optimization method based on CFD and BP neural network.

[0050] According to the geometric parameters of each part of the deamination reactor, 3D modeling is performed using Spaceclaim 3D modeling software. The steps are as follows:

[0051] 1. Determine the process parameters of the device that need to be optimized. The process parameters here include the diameter of the water jet hole of the device, the inlet flow rate and the outlet pressure at the top of the device.

[0052] 2. Determine the structure and size of the deamination reactor and use 3D modeling software to establish Figure 2 The geometric model of .

[0053] 3. Mesh the geometric model and define the properties of the structure;

[0054] 4. Import the meshed model into the Fluent fluid calculation software, set the model boundary conditions, set the reactor inlet flow rate in the boundary condition options, set the reactor top outlet pressure, set the inlet as the velocity inlet, and the outlet as the pressure outlet.

[0055] 5. Select reasonable models of turbulence, multiphase flow, cavitation mechanism, etc., perform numerical simulation on the deammoniation reactor, and calculate according to the set boundary conditions to obtain the cavitation volume fraction and cavitation length distribution data at any position inside the device. In this process, random data samples, i.e., sample sets, are generated. The input is the process parameters, and the output is the corresponding response parameters. At the same time, the influence of the jet hole diameter, inlet flow rate, and outlet pressure on the cavitation volume fraction and cavitation length is adjusted.

[0056] In this embodiment, reasonable models such as turbulence, multiphase flow, and cavitation mechanism are selected. Specifically, in the Fluent software, the Mixture model is selected in the multiphase flow model option, the RNG k-ε model is selected in the turbulence model option, and the Schnerr-Sauer model is selected in the cavitation model; then the boundary conditions are set to start the simulation calculation.

[0057] In this embodiment, the deamination reactor is a prior art, and its working mechanism is as follows: when the ammonia nitrogen wastewater passes through the multiple micro-water jet holes from the jacket, the liquid velocity increases due to the Bernoulli principle, and the porous jets collide after being ejected. At this stage, the liquid flow exists in the form of dispersed droplets and atomization, and the free ammonia in the atomized droplets is extracted together with the water vapor through the negative pressure at the top of the reactor to achieve the purpose of deamination.

[0058] The purpose of adjusting the jet hole diameter parameters and water flow parameters is to strengthen the jet impact effect, so that the liquid phase exists more in the form of dispersed droplets and atomization, and better transfer the free ammonia in the liquid phase to the gas phase; at the same time, the top outlet pressure parameter is set to negative pressure to extract the atomized water vapor and free ammonia produced after the jet impact from the top. In the process of adjusting the jet hole diameter and water flow, the liquid speed inside the jet hole is accelerated, the static pressure is reduced, and local cavitation occurs in the jet hole. These conditions provide a favorable physical and chemical environment for chemical reactions and increase the chemical reaction rate. It is also conducive to the degradation of ammonia nitrogen wastewater.

[0059] Neural Network Prediction:

[0060] A. After preprocessing the sample set through data normalization, it is divided into training set, test set and validation set in a ratio of 7:1.5:1.5;

[0061] B. Take the process parameters as input, namely the water jet hole diameter, inlet flow rate and the outlet pressure at the top of the device, and the response parameters as output (cavitation volume fraction and cavitation length), and establish a neural network model. The neural network is a three-layer network structure. The number of neurons in the input layer is the number of process parameters, the number of neurons in the output layer is the number of response parameters, and the number of neurons in the hidden layer is used to optimize the fitting and generalization ability of the neural network. The number of neurons in the hidden layer is obtained using an empirical formula, and the weights from the input layer to the hidden layer and from the hidden layer to the output layer are defined. The learning rate and error are set. After the parameters are determined, the neural network model is trained using the Levenberg-Marquardt algorithm through the training set;

[0062] The empirical formula is as follows:

[0063]

[0064] Among them, N h is the number of neurons in the hidden layer, N in and Nout are the number of input and output neurons respectively, α∈[1,10].

[0065] C. Repeat step B until the error meets the accuracy requirement or the number of iterations reaches the maximum number of iterations. Specifically, the Tansig function can be used as the activation function, the number of iterations is 1000, and the learning rate is 0.1. After the training is completed, the cavitation volume fraction and length distribution prediction data are analyzed and evaluated through the test set. Specifically, after the neural network training is completed, the process parameters in the test set are input into the neural network, and the results obtained by the neural network prediction are compared with the results obtained by the Fluent numerical calculation. The error within ±5% is considered to meet the neural network prediction accuracy requirements;

[0066] D. Verify the neural network model through the validation set. Specifically, after the neural network model is trained, the required process parameters, i.e., the corresponding water jet hole diameter, inlet flow rate and outlet pressure at the top of the device, can be input to quickly obtain the predicted cavitation volume fraction and length. Compare with the CFD simulation data to verify whether the error is within the allowable range of engineering application. If the error is large, repeat the above steps to train the neural network. After inputting the corresponding process parameters, the trained neural network model can quickly predict the cavitation volume fraction and length under the parameters, which greatly saves the calculation time of the results obtained through numerical simulation. According to the results obtained by the neural network prediction, the process parameters of the deamination reactor can be further adjusted and optimized to obtain a better device operation state and further improve the deamination efficiency of the deamination reactor.

[0067] The predicted cavitation volume fraction and length are specifically calculated as follows: the process parameters are used as input quantities, and the input quantities are called through the net call command in the neural network, so as to obtain the cavitation volume fraction and cavitation length of the corresponding process parameters calculated according to the neural network.

[0068] Neural network is used to predict the cavitation volume fraction and length in the deamination reactor, so as to achieve the purpose of optimizing the deamination reactor.

[0069] In summary, in order to solve the problems of high experimental and computational costs and long time consumption in the existing optimization design methods, a new deamination reactor optimization design method is proposed by combining CFD numerical simulation and artificial neural network. The traditional optimization design method of engineering equipment is mainly based on qualitative analysis and exhaustive method, which requires experiments and comparisons for each parameter combination within the feasible range, and the optimization efficiency is low. When only using CFD method to optimize engineering problems, it is usually necessary to model, mesh, and perform numerical simulation calculations for various structural devices separately, which takes a long time to calculate and affects the optimization efficiency of CFD method. The proposed deamination reactor optimization design method is based on artificial neural network, and CFD method assists in generating sample data. The trained neural network can realize the rapid analysis of deamination efficiency of deamination reactor. Upgrading the physical drive and mathematical drive mode in the neural network model to data drive mode can effectively reduce the cost of calculation and analysis, and achieve efficient and low-cost deamination reactor optimization design effect.

[0070] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A deamination reactor optimization method based on CFD and BP neural network, characterized in that: Here are the steps: According to the geometric parameters of the deamination reactor, three-dimensional modeling is carried out and a sample set is obtained; Train the neural network through the sample set; Prediction of cavitation volume fraction and length in deamination reactor using neural network.

2. The deamination reactor optimization method based on CFD and BP neural network according to claim 1, characterized in that: According to the geometric parameters of the deamination reactor, three-dimensional modeling is performed, and the steps are as follows: According to the geometric parameters of the deamination reactor, three-dimensional modeling is carried out, and the steps are as follows: Determine the device process parameters that need to be optimized, including water jet hole diameter, inlet flow rate and device top outlet pressure; Determine the structure and size of the deamination reactor; Divide the geometric model into grids and import it into Fluent fluid calculation software. Set the model boundary conditions, set the inlet as velocity inlet and the outlet as pressure outlet. Reasonable turbulence, multiphase flow and cavitation mechanism models are selected to carry out numerical simulation of the deammoniation reactor. According to the set boundary conditions, the cavitation volume fraction and cavitation length distribution data at any position inside the device are obtained. At the same time, random data samples, i.e., sample sets, are generated. The input is the process parameters, and the output is the corresponding response parameters.

3. The deamination reactor optimization method based on CFD and BP neural network according to claim 2, characterized in that: The steps for setting the model boundary conditions are as follows: In the boundary condition options, set the reactor inlet flow rate and the reactor top outlet pressure.

4. The deamination reactor optimization method based on CFD and BP neural network according to claim 1, characterized in that: The neural network training steps are as follows: A. After processing, the sample set is divided into training set, test set and validation set; B. Take the process parameters as input and the response parameters as output, establish a neural network model, and train the neural network model through the training set; C. Repeat step B until the error meets the accuracy requirement or the number of iterations reaches the maximum number of iterations. After the training is completed, the cavitation volume fraction and length distribution prediction data are analyzed and evaluated through the test set; D. Verify the neural network model through the validation set.

5. The deamination reactor optimization method based on CFD and BP neural network according to claim 4, characterized in that: After preprocessing, the sample set is divided into training set, test set and validation set in the ratio of 7:1.5:1.

5.

6. The deamination reactor optimization method based on CFD and BP neural network according to claim 4, characterized in that: The response parameters include cavitation volume fraction and cavitation length.

7. The deamination reactor optimization method based on CFD and BP neural network according to claim 4, characterized in that: The neural network is a three-layer network structure, including an input layer, an output layer and a hidden layer; The number of neurons in the input layer is the number of process parameters; The number of neurons in the output layer is the number of response parameters; The number of hidden layer neurons is used to optimize the neural network fitting and generalization capabilities.

Images