CN116595845A - Structure optimization method of spiral fin tube type heat exchanger - Google Patents
Structure optimization method of spiral fin tube type heat exchanger Download PDFInfo
- Publication number
- CN116595845A CN116595845A CN202310649248.XA CN202310649248A CN116595845A CN 116595845 A CN116595845 A CN 116595845A CN 202310649248 A CN202310649248 A CN 202310649248A CN 116595845 A CN116595845 A CN 116595845A
- Authority
- CN
- China
- Prior art keywords
- fin tube
- spiral
- spiral fin
- factor
- structural parameters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 52
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000013528 artificial neural network Methods 0.000 claims abstract description 43
- 238000004088 simulation Methods 0.000 claims abstract description 40
- 230000017525 heat dissipation Effects 0.000 claims abstract description 35
- 238000011156 evaluation Methods 0.000 claims abstract description 30
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 29
- 238000002474 experimental method Methods 0.000 claims abstract description 17
- 230000000694 effects Effects 0.000 claims abstract description 12
- 238000012360 testing method Methods 0.000 claims description 18
- 238000012549 training Methods 0.000 claims description 18
- 238000004364 calculation method Methods 0.000 claims description 10
- 239000007787 solid Substances 0.000 claims description 10
- 230000001419 dependent effect Effects 0.000 claims description 8
- 108010074506 Transfer Factor Proteins 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 239000012530 fluid Substances 0.000 claims description 5
- 230000005855 radiation Effects 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 3
- 230000008878 coupling Effects 0.000 claims description 3
- 238000010168 coupling process Methods 0.000 claims description 3
- 238000005859 coupling reaction Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 abstract description 9
- 238000013461 design Methods 0.000 abstract description 6
- 230000007547 defect Effects 0.000 abstract description 3
- 210000002569 neuron Anatomy 0.000 description 16
- 238000010586 diagram Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 239000013598 vector Substances 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 6
- 239000000463 material Substances 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000035772 mutation Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Geometry (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computer Hardware Design (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Genetics & Genomics (AREA)
- Physiology (AREA)
- Medical Informatics (AREA)
- Fluid Mechanics (AREA)
- Algebra (AREA)
- Computational Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a structural optimization method of a spiral fin tube heat exchanger, which comprises the steps of firstly using three-dimensional modeling software to establish a spiral fin tube heat exchanger heat dissipation simulation model, carrying out finite element mesh division, carrying out heat dissipation simulation experiments in combination with orthogonal experiment methods, calculating heat exchange performance evaluation index j factors and resistance performance evaluation index f factors, then establishing a BP neural network regression model between spiral fin tube structural parameters and the j factors and f factors, finally adopting NSGA-II algorithm to optimize the spiral fin tube structural parameters, modeling the spiral fin tube according to the structural parameters obtained by optimization, carrying out heat dissipation simulation experiments, and comparing the optimization effects. The method applies the BP neural network regression model and NSGA-II algorithm to the multi-objective optimization problem of the spiral fin tube structural parameters, so that the optimized spiral fin tube heat exchanger has better heat transfer performance and lower resistance performance, and the defects of high time cost and economic cost of the conventional optimization design based on physical experiments are overcome.
Description
Technical Field
The invention belongs to the technical field of structural optimization of fin tube heat exchangers, and particularly relates to a structural optimization method of a spiral fin tube heat exchanger.
Background
Spiral fin tube heat exchangers play a very important role in air energy systems. By optimizing the structural parameters of the spiral finned tube, the efficiency and performance of the air energy system can be improved, and the quality of services such as heating, refrigerating, hot water and the like can be further improved. Meanwhile, the energy consumption and the running cost of the air energy system can be reduced, and the application of the air energy system in sustainable development is promoted. The research on the fin tube type heat exchanger is very rich, the structural optimization method of the spiral fin tube type heat exchanger is generally a traditional physical experiment method, a method of combining simulation analysis and an optimization algorithm is adopted in part of the research, and the optimization algorithm is mainly a genetic algorithm.
However, the current structure optimization method of the spiral fin tube type heat exchanger has the following problems:
1. the spiral fin tube heat exchanger structure is optimally designed by a physical experiment development method, the time cost and the economic cost are high, and the heat transfer and flow proceeding process is difficult to directly observe.
2. The structural parameters of the spiral fin tube are optimized by adopting a traditional genetic algorithm, the optimized variable is single, and two variables of a heat exchange performance evaluation index j factor and a resistance performance evaluation index f factor are difficult to optimize simultaneously, so that the heat transfer performance and the resistance performance of the spiral fin tube heat exchanger can not reach a better level at the same time.
Disclosure of Invention
In order to solve the technical problems, the invention provides a structural optimization method of a spiral fin tube heat exchanger, which is based on a BP (back propagation) neural network regression model and an NSGA-II algorithm, and solves the problems of high time and economic cost and single optimization variable of the structural optimization of the traditional spiral fin tube heat exchanger.
The invention adopts the technical scheme that: the structure optimizing method of the spiral fin tube heat exchanger comprises the following specific steps:
s1, using three-dimensional modeling software to establish a heat dissipation simulation model of a spiral fin tube heat exchanger;
s2, carrying out finite element mesh division on the heat dissipation simulation model, carrying out a heat dissipation simulation experiment by combining an orthogonal experiment method, and calculating a heat exchange performance evaluation index factor j and a resistance performance evaluation index factor f;
s3, establishing a BP neural network regression model between the structural parameters of the spiral finned tube and the factors j and f by utilizing a BP neural network algorithm;
s4, combining a BP neural network regression model, adopting an NSGA-II algorithm to optimize the structural parameters of the spiral finned tube, modeling the spiral finned tube according to the structural parameters obtained by optimization, carrying out a heat dissipation simulation experiment, and comparing the optimization effects.
Further, the step S1 specifically includes the following steps:
s11, determining structural parameters of the spiral finned tube according to structural characteristics of the spiral finned tube, determining a value range of the structural parameters according to actual requirements, and establishing a geometric model of the spiral finned tube;
the structural parameters of the spiral fin tube include: the diameter of the fin tube, the thickness of the fin tube, the screw pitch, the thickness of the fin and the height of the fin.
S12, establishing an air domain in a geometric model of the spiral fin tube, wherein a fluid-solid coupling surface exists between the air and the spiral fin tube;
s13, recording the volume and the surface area of the air domain, and calculating the characteristic length of the air domain;
the feature length is calculated as follows:
wherein V is f Representing the volume of the air field, A f Representing the surface area of the air-domain.
Further, the step S2 specifically includes the following steps:
s21, carrying out finite element mesh division on an initial heat dissipation simulation model, carrying out mesh independence verification, and determining an optimal mesh number range;
s22, designing an orthogonal experiment table according to structural parameters of the spiral fin tube by combining an orthogonal experiment method;
s23, carrying out a heat dissipation simulation experiment by using simulation software under the condition of different wind speeds;
s24, recording a fluid-solid interface Knoop coefficient, an inlet-outlet pressure difference and a wind speed at the minimum section under a heat dissipation stable state according to a heat dissipation simulation experiment result;
s25, according to the Knoop coefficient Nu of the fluid-solid interface and the wind speed v at the smallest section of the spiral finned tube max Calculating heat exchange performance evaluation index j factor according to the characteristic length De of the air domain, the inlet and outlet pressure difference delta P and the wind speed v at the minimum section of the spiral finned tube max Calculating a resistance performance evaluation index f factor;
the calculation formulas of the factor j and the factor f are respectively as follows:
wherein Re represents the Reynolds number, pr represents the Plandth number of air, ρ represents the air density, L represents the length in the flow direction, μ 0 Aerodynamic viscosity (kg/mS) is shown.
S26, establishing a key section cloud picture, and observing the temperature change condition in the experimental process.
Further, the step S3 specifically includes the following steps:
s31, randomly dividing a sample set into a training set and a test set according to the test set proportion of 30%;
s32, normalizing the training set and the test set, carrying out model training after normalization treatment, and establishing a BP neural network regression model;
the independent variables of the regression model are the structural parameters and wind speed of the spiral finned tube, and the dependent variables are the heat exchange performance evaluation index j factor and the resistance performance evaluation index f factor; the wind speed is the air speed at the inlet.
S33, using the trained regression model for the test set, verifying the accuracy of the regression model, judging whether the hyper-parameters of the BP neural network regression model need to be continuously adjusted according to the accuracy, and optimizing the regression model.
Further, the step S4 specifically includes the following steps:
s41, determining an objective function and constraint conditions of the multi-objective optimization problem;
s42, combining a BP neural network regression model, and optimizing structural parameters of the spiral finned tube by using an NSGA-II algorithm;
the input variable of the NSGA-II algorithm is a spiral finned tube structural parameter, the output variable is an optimized spiral finned tube structural parameter, and the optimization criterion is as follows: the optimized spiral fin tube has higher heat transfer factor j and lower resistance factor f.
S43, a heat radiation simulation model is built by combining the optimized spiral fin tube structural parameters, a heat radiation simulation experiment is carried out, a knowler coefficient cloud chart, a fluid domain pressure and a speed distribution cloud chart of the surface of the optimized spiral fin tube are observed, an optimized heat transfer factor j and a resistance factor f are calculated, and the optimization effect is compared.
The beneficial effects of the invention are as follows: according to the method, three-dimensional modeling software is used first, a heat dissipation simulation model of the spiral fin tube heat exchanger is built, finite element mesh division is carried out, a heat dissipation simulation experiment is carried out in combination with an orthogonal experiment method, a heat exchange performance evaluation index j factor and a resistance performance evaluation index f factor are calculated, a BP neural network regression model between the structural parameters of the spiral fin tube and the j factor and the f factor is built again, NSGA-II algorithm is adopted to optimize the structural parameters of the spiral fin tube finally, the spiral fin tube is modeled according to the structural parameters obtained through optimization, the heat dissipation simulation experiment is carried out, and the optimization effect is compared. The method applies the BP neural network regression model and NSGA-II algorithm to the multi-objective optimization problem of the spiral fin tube structural parameters, so that the optimized spiral fin tube heat exchanger has better heat transfer performance and lower resistance performance, and the defects of high time cost and economic cost of the conventional optimization design based on physical experiments are overcome.
Drawings
FIG. 1 is a flow chart of a method of optimizing the structure of a spiral fin tube heat exchanger of the present invention.
FIG. 2 is a two-dimensional view of a helical finned tube geometry model in an embodiment of the invention.
FIG. 3 is a three-dimensional simulation model diagram of an air-domain containing embodiment of the present invention.
Fig. 4 is a finite element mesh diagram in an embodiment of the present invention.
Fig. 5 is a flowchart of a BP neural network regression model establishment in an embodiment of the present invention.
FIG. 6 is a schematic diagram of artificial neural network neuron weights and biases in an embodiment of the present invention.
Fig. 7 is a diagram of a BP neural network model in an embodiment of the present invention.
Fig. 8 is a regression model diagram of a BP neural network according to an embodiment of the present invention.
FIG. 9 is a graph of heat transfer j factor prediction in an embodiment of the invention.
FIG. 10 is a graph showing the predicted condition of the resistance factor f in the embodiment of the present invention.
FIG. 11 is a flowchart of the NSGA-II algorithm in accordance with an embodiment of the present invention.
FIG. 12 is a comparison of geometric models of the helical finned tubes before and after optimization in an embodiment of the invention.
FIG. 13 is a graph showing the comparison of the surfaces Nu of the pre-and post-optimized turn-fin tubes in the example of the present invention.
Detailed Description
The method of the present invention will be further described with reference to the accompanying drawings and examples.
As shown in fig. 1, the structural optimization method flow chart of the spiral fin tube heat exchanger comprises the following specific steps:
s1, using three-dimensional modeling software to establish a heat dissipation simulation model of a spiral fin tube heat exchanger;
s2, carrying out finite element mesh division on the heat dissipation simulation model, carrying out a heat dissipation simulation experiment by combining an orthogonal experiment method, and calculating a heat exchange performance evaluation index factor j and a resistance performance evaluation index factor f;
s3, establishing a BP neural network regression model between the structural parameters of the spiral finned tube and the factors j and f by utilizing a BP neural network algorithm;
s4, combining a BP neural network regression model, adopting an NSGA-II algorithm to optimize the structural parameters of the spiral finned tube, modeling the spiral finned tube according to the structural parameters obtained by optimization, carrying out a heat dissipation simulation experiment, and comparing the optimization effects.
In this embodiment, the step S1 is specifically as follows:
s11, determining structural parameters and a value range of the spiral finned tube, and establishing a geometric model of the spiral finned tube;
the structural parameters of the spiral fin tube include: the diameter of the fin tube, the thickness of the fin tube, the screw pitch, the thickness of the fin and the height of the fin.
In this example, the structural parameters and the values of the spiral finned tube are shown in Table 1, and the geometric model two-dimensional diagram of the spiral finned tube is shown in FIG. 2.
TABLE 1
S12, establishing an air domain in a geometric model of the spiral fin tube, wherein a fluid-solid coupling surface exists between the air and the spiral fin tube;
wherein, the three-dimensional simulation model diagram containing air domain is shown in fig. 3.
S13, recording the volume and the surface area of the air domain, and calculating the characteristic length of the air domain;
the feature length is calculated as follows:
wherein V is f Representing the volume of the air field, A f Representing the surface area of the air-domain.
In this embodiment, the step S2 is specifically as follows:
s21, as shown in FIG. 4, performing finite element mesh division on an initial heat dissipation simulation model, performing mesh independence verification, and determining an optimal mesh number range;
s22, designing an orthogonal experiment table according to structural parameters of the spiral fin tube by combining an orthogonal experiment method;
wherein the spiral fin tube contains 5 structural parameters, and the orthogonal experiment table is shown in Table 2.
TABLE 2
S23, carrying out a heat dissipation simulation experiment by using simulation software under the condition of different wind speeds; the heat dissipation simulation experiment is set as follows:
1. boundary condition setting:
inlet: setting the air inlet face as a velocity-inlet boundary, and setting the velocity to be 0-4m/s, the temperature to be 300K and the pressure to be 0Pa (the static pressure to be 0Pa is the same as the ambient pressure);
and (3) an outlet: setting an air outlet face as a pressure-outlet boundary, wherein the boundary temperature is 300K and the pressure is 0Pa;
wall surface: the "wall-hot" plane was set as the wall boundary at 373.15K.
2. Algorithm setting: coupled algorithm.
3. The material setting: the spiral fin tube material is set as a burner and the fluid domain material is set as air.
S24, recording a fluid-solid interface Knoop coefficient, an inlet-outlet pressure difference and a wind speed at the minimum section under a heat dissipation stable state according to a heat dissipation simulation experiment result;
s25, according to the Knoop coefficient Nu of the fluid-solid interface and the wind speed v at the smallest section of the spiral finned tube max Calculating heat exchange performance evaluation index j factor according to the characteristic length De of the air domain, the inlet and outlet pressure difference delta P and the wind speed v at the minimum section of the spiral finned tube max Calculating a resistance performance evaluation index f factor;
the calculation formulas of the factor j and the factor f are respectively as follows:
wherein Re represents the Reynolds number, pr represents the Plandth number of air, ρ represents the air density, L represents the length in the flow direction, μ 0 Aerodynamic viscosity (kg/mS) is shown.
S26, establishing a key section cloud picture, and observing the temperature change condition in the experimental process.
In this embodiment, the step S3 is specifically as follows:
s31, randomly dividing a sample set into a training set and a test set according to the test set proportion of 30%;
s32, normalizing the training set and the test set, carrying out model training after normalization treatment, and establishing a BP neural network regression model;
normalization may allow the model to better learn the relationships between the data, thereby improving its generalization ability.
The normalization formula is as follows:
wherein x is i Representing the actual value of the argument, y i Representing the actual value of the dependent variable. X is x i ' and y i ' represents the normalized value of the corresponding variable.
In the prediction process, after a regression model is adopted to calculate a predicted value, the predicted value is subjected to normalization reduction, and a reduction formula is as follows:
wherein,,representing the predicted value of the dependent variable,/->Representing the predicted value of the normalized dependent variable.
The independent variables of the regression model are the structural parameters and wind speed of the spiral finned tube, and the dependent variables are the heat exchange performance evaluation index j factor and the resistance performance evaluation index f factor.
Wherein the wind speed is the air speed at the inlet.
As shown in fig. 5, the steps for establishing the BP neural network regression model are specifically as follows:
1) Collecting data: collecting a data set containing independent variables and dependent variables, wherein the data are structural parameters of the spiral finned tube, wind speed, a fluid-solid interface Knoop coefficient obtained according to a numerical simulation experiment, inlet-outlet pressure difference and wind speed at the minimum section of the spiral finned tube;
2) Data preprocessing: preprocessing data, including: filling missing values, processing abnormal values and normalizing variables;
3) Splitting the data set: splitting the data set into a training set and a testing set, and generally adopting a random sampling method;
4) Selecting a model: selecting an appropriate model based on the characteristics of the data and the requirements of the problem, including: linear regression, ridge regression, multiple linear regression;
5) Fitting a model: fitting the selected model by using training data to obtain model parameters;
6) Model evaluation: evaluating the fitted model using the test set, comprising: calculating a Mean Square Error (MSE), a Mean Absolute Error (MAE), a determinable coefficient (R 2 );
7) The model was used: predicting new data by using the trained model to obtain a prediction result;
8) Model optimization: model optimization is carried out according to the evaluation result, and the method comprises the following steps: selecting new characteristics and adjusting model super parameters; and 5) bringing the adjusted parameters into the step 5) to restart model training, and finally comparing model effects under different parameters to select an optimal model and parameters.
The concept of the artificial neural network algorithm is as follows: an artificial neural network (Artificial Neural Network, ANN) is a class of computational models consisting of a plurality of nodes or neurons. Neurons in an ANN are typically organized in multiple layers, where each layer of neurons is connected only to neurons of a previous layer, and there is no connection between neurons in the same layer. The input layer accepts the input of raw data, and the output layer outputs the final result. Each layer is composed of a plurality of neurons, each neuron has a weight and a bias, and the output is obtained by weighting and biasing the input layer and then performing nonlinear conversion through an excitation function. The neuron weights and offsets are shown in fig. 6.
The backward propagation neural network (Back Propagation Neural Network, BPNN) is a branch of the ANN, and is characterized by adding a backward propagation link of the error based on the forward transmission of the neural network signal, as shown in fig. 7.
The neuron model shown in fig. 6 is described mathematically as follows:
Y=XW+B
wherein Y= [ Y ] 1 ,y 2 ,y 3 ],y j A value representing an output layer neuron node, Y representing an output vector; x= [ X ] 1 ,x 2 ,x 3 ,x 4 ],x i Values representing input layer neuron nodes, X representing input vectors; w represents the weight matrix of the layer of neurons, W ij Representing the weighting coefficient of the ith neuron node of the input layer for the jth neuron in the output layer; b= [ B ] 1 ,b 2 ,b 3 ],b j Represents the bias value of the j-th neuron node of the output layer, and B represents the bias vector of the output layer.
In the back propagation link of the error, the weight matrix and the bias vector are updated by the following formula:
wherein W is (k) Representing the value of the kth iteration W, B (k) The value of the kth iteration B is represented, eta represents the learning rate of the BPNN and can be adjusted manually according to the iteration effect.
S33, using a trained regression model for the test set, verifying the accuracy of the regression model, judging whether the hyper-parameters of the BP neural network regression model need to be continuously adjusted according to the accuracy, and optimizing the regression model;
the input layer data of the training set is a two-dimensional matrix of 67×6, representing 67 sets of sample data, each set of samples containing 6 features as arguments. The output layer data is a 67 x 2 two-dimensional matrix representing a total of 67 sets of samples, each set containing 2 features as dependent variables. In this embodiment, the data of the regression model training set of the BP neural network is shown in Table 3.
TABLE 3 Table 3
After training is completed, parameter values and training error conditions of the BP neural network regression model under the initial learning rate are obtained, and training set evaluation indexes are shown in table 4. At this time, the learning rate of the BP neural network regression model is 0.005, the hidden layer number is 1, and the hidden layer node number is 10.
TABLE 4 Table 4
Wherein RMSE (root mean square error) represents the square root of MSE (mean square error), the calculation formula of MSE and MAE (mean absolute error) is as follows:
wherein n represents the sample size of the data, Y i Representing the true value of the data,representing predicted values of the data.
From Table 4, it can be seen that the coefficient R 2 The regression model can also continue to optimize as seen by the values of less than 0.8.
By adjusting the learning rate of the BP neural network, a better model is obtained, the learning rate of the model is 0.2, and the determination coefficients R of j and f are determined as shown in the table 5 2 All are larger than 0.85, the model has good regression effect, and the structure of the regression model is shown in figure 8. The evaluation indexes of the optimized regression model on the training set are shown in table 5.
TABLE 5
The relation between the structural parameter of the spiral fin tube and the heat transfer j factor and the resistance f factor is expressed by a function:
Y=f(X)
wherein X represents a five-dimensional vector of structural parameters of the spiral finned tube, namely X= [ X ] d ,x t1 ,x s ,x t2 ,x h ],x d ,x t1 ,x s ,x t2 ,x h Respectively representing the diameter of the finned tube, the thickness of the finned tube, the screw pitch, the thickness of the fins and the height of the fins; y represents a two-dimensional vector composed of the j factor and the f factor corresponding to the group of spiral finned tubes, namely Y= [ j, f]。
The formula of the BP neural network regression model is:
Y m1 =W 1 X+B 1
Y=W 2 Y m2 +B 2
wherein X represents an input variable, Y m1 Representing weighted and biased intermediate variables in hidden layer, Y m2 Representing the hidden layer output variable of the intermediate variable after the activation function processing, and Y represents the output variable. W (W) 1 B (B) 1 Respectively represent the weighting coefficient and bias of the input layer, W 2 B (B) 2 Respectively represent the weighting coefficients and offsets of the hidden layers. ReLU represents the activation function corresponding thereto.
Using a trained regression model for the test set, and verifying the accuracy of the regression model; in the prediction set, the input layer data is a two-dimensional matrix of 33 x 6, the output layer data is a two-dimensional matrix of 33 x 2, and the BP neural network regression model test set data is shown in table 6.
TABLE 6
And predicting the test set by using the trained regression model, and verifying the prediction accuracy of the regression model, wherein the prediction accuracy is known from the table 7, the determinable coefficients of j and f are both larger than 0.8, and the prediction effect is good. The evaluation indexes used on the test set by the BP neural network regression model are shown in Table 7.
TABLE 7
The heat transfer j factor prediction case is shown in fig. 9, and the resistance f factor prediction case is shown in fig. 10.
In this embodiment, the step S4 is specifically as follows:
s41, determining an objective function and constraint conditions of the multi-objective optimization problem;
s42, combining a BP neural network regression model, and optimizing structural parameters of the spiral finned tube by using an NSGA-II algorithm;
the input variable of NSGA-II algorithm is the structural parameter of the spiral finned tube, the output variable is the structural parameter of the spiral finned tube after optimizing, the optimization criterion is: the optimized spiral fin tube has higher heat transfer factor j and lower resistance factor f.
As shown in fig. 11, the NSGA-II algorithm includes the following steps:
(1) Initializing a population: randomly generating a group of initial individuals to form an initial population;
(2) Evaluating the fitness of the population: calculating the fitness value of the individual to represent the performance of the individual in the target space;
(3) Non-dominant ranking and congestion level calculation: non-dominant ranking is carried out on each individual, the population is divided into a plurality of levels, and the crowding degree value of each individual is calculated;
(4) Selecting a new population: selecting a group of optimal individuals as parents of the next generation population according to the non-dominant ranking and the crowding degree value;
(5) Performing crossover operation: performing cross operation on parent individuals to generate a group of new child individuals;
(6) Performing mutation operation: performing mutation operation on the offspring individuals to generate a group of new mutated offspring individuals;
(7) Evaluating fitness of the newly generated individual: calculating the fitness value of each new individual according to the fitness function;
(8) Retaining elite individuals: combining the parent individuals and the new child individuals to form a group of new population, and performing non-dominant sorting and congestion degree calculation on the new population;
(9) Selecting a new population: taking a group of optimal individuals as parents of the next generation population according to the non-dominant ranking and the crowding degree value;
(10) Repeating steps (5) to (9) until a termination condition is satisfied.
In this example, the heat transfer factor j and the resistance factor f are used as the heat radiation performance and the resistance evaluation index of the spiral fin tube heat exchanger, so that 2 optimization targets are included, and the fin tube diameter d and the fin tube thickness t of the spiral fin tube are calculated 1 Pitch s, fin thickness t 2 The fin height h is used as an optimal design variable, and meanwhile, 5 optimal design variables also need to meet a certain value range, and the value range of the design variable is shown in table 8.
TABLE 8
The mathematical model of the multi-objective optimization problem in this embodiment is described as follows:
wherein x= [ X ] d ,x t1 ,x s ,x t2 ,x h ]X represents a five-dimensional vector composed of helical finned tube structural parameters, f 1 (X)=1/j,f 2 (X)=f。
The calculation of j and f is performed through a BP neural network regression model, and the calculation formula is as follows:
[j,f]=Y=f(X)=W 2 ·ReLU(W 1 X+B 1 )+B 2
s43, establishing a heat dissipation simulation model by combining the optimized spiral fin tube structural parameters, developing a heat dissipation simulation experiment, observing a Knoop coefficient cloud image, a fluid domain pressure and a speed distribution cloud image of the surface of the optimized spiral fin tube, calculating an optimized heat transfer j factor and a resistance f factor, and comparing the optimization effect;
in the multi-objective optimization process, j and f are calculated through a BP neural network regression model, so that the optimization speed is increased.
The structural parameter changes of the spiral finned tube before and after optimization are shown in table 9, and a decimal value is reserved after the optimized variable is rounded, so that the data meets the requirement of processing precision.
TABLE 9
And (3) establishing a corresponding geometric model according to the optimized structural parameters of the spiral fin tube, adopting Fluent to carry out heat dissipation simulation calculation on the geometric model, and carrying out aftertreatment by Tecplot to obtain a knoop-Seer coefficient cloud image, a fluid domain pressure and a velocity distribution cloud image of the surfaces of the fin before and after optimization.
The geometric models before and after the optimization of the spiral fin tube are shown in fig. 12, and the pipe diameter of the spiral fin tube is reduced, the height of the fin is reduced and the screw pitch is increased after the optimization.
The pair of surfaces Nu of the helical fin tubes before and after optimization, such as fig. 13, shows that under the same operation condition, the surface Nu of the helical fin tube after optimization is larger, which indicates that the surface heat dissipation capability of the helical fin tube after optimization is stronger.
In conclusion, based on the embodiment, simulation experiments are carried out, and the heat distribution of the spiral fin tubes before and after optimization and the improvement of the heat exchange performance of each part can be visually observed. The method of the invention overcomes the defects of high time cost and high economic cost of the conventional optimization design based on physical experiments, simultaneously adopts the NSGA-II algorithm to simultaneously optimize the heat transfer performance and the resistance performance of the spiral fin tube, combines the BP neural network regression model with the NSGA-II algorithm, and ensures that the calculation speed of heat transfer factor j and resistance factor f in the NSGA-II algorithm is faster.
Those of ordinary skill in the art will recognize that the embodiments described herein are for the purpose of aiding the reader in understanding the principles of the present invention and should be understood that the scope of the invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific modifications and combinations from the teachings of the present disclosure without departing from the spirit thereof, and such modifications and combinations remain within the scope of the present disclosure.
Claims (5)
1. The structure optimizing method of the spiral fin tube heat exchanger comprises the following specific steps:
s1, using three-dimensional modeling software to establish a heat dissipation simulation model of a spiral fin tube heat exchanger;
s2, carrying out finite element mesh division on the heat dissipation simulation model, carrying out a heat dissipation simulation experiment by combining an orthogonal experiment method, and calculating a heat exchange performance evaluation index factor j and a resistance performance evaluation index factor f;
s3, establishing a BP neural network regression model between the structural parameters of the spiral finned tube and the factors j and f by utilizing a BP neural network algorithm;
s4, combining a BP neural network regression model, adopting an NSGA-II algorithm to optimize the structural parameters of the spiral finned tube, modeling the spiral finned tube according to the structural parameters obtained by optimization, carrying out a heat dissipation simulation experiment, and comparing the optimization effects.
2. The method for optimizing the structure of a spiral fin tube heat exchanger according to claim 1, wherein the step S1 is specifically as follows:
s11, determining structural parameters of the spiral finned tube according to structural characteristics of the spiral finned tube, determining a value range of the structural parameters according to actual requirements, and establishing a geometric model of the spiral finned tube;
the structural parameters of the spiral fin tube include: the diameter of the fin tube, the thickness of the fin tube, the screw pitch, the thickness of the fin and the height of the fin;
s12, establishing an air domain in a geometric model of the spiral fin tube, wherein a fluid-solid coupling surface exists between the air and the spiral fin tube;
s13, recording the volume and the surface area of the air domain, and calculating the characteristic length of the air domain;
the feature length is calculated as follows:
wherein V is f Representing the volume of the air field, A f Representing the surface area of the air-domain.
3. The method for optimizing the structure of a spiral fin tube heat exchanger according to claim 1, wherein the step S2 is specifically as follows:
s21, carrying out finite element mesh division on an initial heat dissipation simulation model, carrying out mesh independence verification, and determining an optimal mesh number range;
s22, designing an orthogonal experiment table according to structural parameters of the spiral fin tube by combining an orthogonal experiment method;
s23, carrying out a heat dissipation simulation experiment by using simulation software under the condition of different wind speeds;
s24, recording a fluid-solid interface Knoop coefficient, an inlet-outlet pressure difference and a wind speed at the minimum section under a heat dissipation stable state according to a heat dissipation simulation experiment result;
s25, according to the Knoop coefficient Nu of the fluid-solid interface and the wind speed v at the smallest section of the spiral finned tube max Calculating heat exchange performance evaluation index j factor according to the characteristic length De of the air domain, the inlet and outlet pressure difference delta P and the wind speed v at the minimum section of the spiral finned tube max Calculating a resistance performance evaluation index f factor;
the calculation formulas of the factor j and the factor f are respectively as follows:
wherein Re represents the Reynolds number, pr represents the Plandth number of air, ρ represents the air density, L represents the length in the flow direction, μ 0 Represents aerodynamic viscosity (kg/mS);
s26, establishing a key section cloud picture, and observing the temperature change condition in the experimental process.
4. The method for optimizing the structure of a spiral fin tube heat exchanger according to claim 1, wherein the step S3 is specifically as follows:
s31, randomly dividing a sample set into a training set and a test set according to the test set proportion of 30%;
s32, normalizing the training set and the test set, carrying out model training after normalization treatment, and establishing a BP neural network regression model;
the independent variables of the regression model are the structural parameters and wind speed of the spiral finned tube, and the dependent variables are the heat exchange performance evaluation index j factor and the resistance performance evaluation index f factor; the wind speed is the air speed at the inlet;
s33, using the trained regression model for the test set, verifying the accuracy of the regression model, judging whether the hyper-parameters of the BP neural network regression model need to be continuously adjusted according to the accuracy, and optimizing the regression model.
5. The method for optimizing the structure of a spiral fin tube heat exchanger according to claim 1, wherein the step S4 is specifically as follows:
s41, determining an objective function and constraint conditions of the multi-objective optimization problem;
s42, combining a BP neural network regression model, and optimizing structural parameters of the spiral finned tube by using an NSGA-II algorithm;
the input variable of the NSGA-II algorithm is a spiral finned tube structural parameter, the output variable is an optimized spiral finned tube structural parameter, and the optimization criterion is as follows: the optimized spiral fin tube has higher heat transfer factor j and lower resistance factor f;
s43, a heat radiation simulation model is built by combining the optimized spiral fin tube structural parameters, a heat radiation simulation experiment is carried out, a knowler coefficient cloud chart, a fluid domain pressure and a speed distribution cloud chart of the surface of the optimized spiral fin tube are observed, an optimized heat transfer factor j and a resistance factor f are calculated, and the optimization effect is compared.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310649248.XA CN116595845A (en) | 2023-06-02 | 2023-06-02 | Structure optimization method of spiral fin tube type heat exchanger |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310649248.XA CN116595845A (en) | 2023-06-02 | 2023-06-02 | Structure optimization method of spiral fin tube type heat exchanger |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116595845A true CN116595845A (en) | 2023-08-15 |
Family
ID=87593640
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310649248.XA Pending CN116595845A (en) | 2023-06-02 | 2023-06-02 | Structure optimization method of spiral fin tube type heat exchanger |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116595845A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117648778A (en) * | 2024-01-29 | 2024-03-05 | 地平线(天津)科学技术应用研究有限公司 | Optimal design method of single-tube multi-fin calandria evaporator |
-
2023
- 2023-06-02 CN CN202310649248.XA patent/CN116595845A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117648778A (en) * | 2024-01-29 | 2024-03-05 | 地平线(天津)科学技术应用研究有限公司 | Optimal design method of single-tube multi-fin calandria evaporator |
CN117648778B (en) * | 2024-01-29 | 2024-04-23 | 地平线(天津)科学技术应用研究有限公司 | Optimal design method of single-tube multi-fin calandria evaporator |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112966954B (en) | Flood control scheduling scheme optimization method based on time convolution network | |
Xie et al. | Performance predictions of laminar and turbulent heat transfer and fluid flow of heat exchangers having large tube-diameter and large tube-row by artificial neural networks | |
CN113705877B (en) | Real-time moon runoff forecasting method based on deep learning model | |
Wang et al. | Prediction of heat transfer rates for shell-and-tube heat exchangers by artificial neural networks approach | |
CN112084727A (en) | Transition prediction method based on neural network | |
CN116595845A (en) | Structure optimization method of spiral fin tube type heat exchanger | |
CN109948920B (en) | Electric power market settlement data risk processing method based on evidence theory | |
CN114777192B (en) | Secondary network heat supply autonomous optimization regulation and control method based on data association and deep learning | |
CN114811713B (en) | Two-level network inter-user balanced heat supply regulation and control method based on mixed deep learning | |
CN111461404A (en) | Short-term load and hydropower prediction method based on neural network prediction interval | |
CN114202111A (en) | Electronic expansion valve flow characteristic prediction based on particle swarm optimization BP neural network | |
CN114068051A (en) | Method for calculating temperature and flow of coolant of main pipeline of nuclear reactor based on ultrasonic array | |
CN108376294A (en) | A kind of heat load prediction method of energy supply feedback and meteorologic factor | |
CN114741961A (en) | Method and system for optimizing wing type fin arrangement structure of printed circuit board type heat exchanger | |
CN114818487A (en) | Natural gas and wet gas pipeline liquid holdup prediction model method based on PSO-BP neural network | |
CN115796011A (en) | Hydrogen storage bed heat transfer performance optimization method based on neural network and genetic algorithm | |
CN109840335A (en) | Based on the radial forging pit prediction optimization method for strengthening T-S fuzzy neural network | |
Jafarkazemi et al. | Performance prediction of flat-plate solar collectors using MLP and ANFIS | |
CN117034808A (en) | Natural gas pipe network pressure estimation method based on graph attention network | |
CN114611418B (en) | Natural gas pipeline flow state prediction method | |
CN115034133A (en) | Jet pump heat supply system implementation method based on information physical fusion | |
CN117077508B (en) | Performance prediction and optimization method for thermoelectric generator | |
Díaz et al. | Analysis of data from single-row heat exchanger experiments using an artificial neural network | |
CN116562094B (en) | AUV formation flow field prediction method based on PINN model | |
CN117973216A (en) | Supercritical carbon dioxide boiler heat transfer characteristic prediction method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |