Vehicle compact heat exchanger optimization method based on RF random forest algorithm
Through the optimization method based on RF random forest algorithm and combined with CFD simulation, the overfitting problem in the performance optimization of automotive compact heat exchangers is solved, the prediction accuracy and generalization ability are improved, and more efficient heat exchange is achieved Effect.
Patent Information
- Application Number
- CN202410947657.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-16
- Publication Date
- 2024-11-29
AI Technical Summary
There are bottlenecks in performance optimization of existing automotive compact heat exchangers, especially traditional methods are difficult to effectively reduce the risk of overfitting and improve the generalization ability of model. At the same time, traditional radiators are complex and prone to blockage, making it difficult to meet high efficiency and light quality. , economic demand.
Using the optimization method based on RF random forest algorithm, collect the impact parameter data through numerical simulation and CFD simulation simulation, establish an integrated model, use the prediction results of multiple trees to reduce the risk of overfitting, and evaluate the regression quality and determine the coefficients through decision tree. Model accuracy, optimized heat exchange performance.
It improves the accuracy of predicting heat exchange performance of automotive compact heat exchangers, reduces the risk of overfitting, enhances the generalization ability of model, avoids the influence of irrelevant parameters, and achieves a more efficient heat exchange effect.
Smart Images

Figure CN119047292A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of vehicle-mounted heat exchangers, and in particular to an optimization method for a vehicle-mounted compact heat exchanger based on an RF random forest algorithm. Background Art
[0002] Compact automotive heat exchangers are key devices for heat exchange in automotive cooling systems. With the increasing demand for heat dissipation and increasingly limited layout space, the demand for efficient, lightweight and economical automotive heat exchangers has become more urgent.
[0003] In conventional vehicles, more than 33% of heat is lost during engine combustion, and heat exchangers and other devices enable the power system to dissipate the heat generated during power generation to avoid overheating. Automotive heat exchangers mainly use coolant as the working fluid, such as water and additives, in most cases propylene glycol, to overcome the shortcomings of water as a heat transfer fluid. Most modern cars use aluminum radiators, which are made by welding thin fins to flat tubes in parallel. Modern research is to develop current heat exchanger systems, such as through parametric studies to evaluate the layout of different heat exchanger fins and establish best practices for characterizing these devices. This can be achieved by evaluating the performance of flat tube heat exchangers with louvered fins and optimizing them, or by describing the aerodynamic effects of fin shape on heat exchanger performance. Fin heat exchanger equipment has a wide range of applications and is suitable for heat exchange of a variety of different fluids in the equipment. However, the fin structure is complex and generally has a high cost. The pipes are also more prone to clogging and difficult to clean. Among compact automotive heat exchangers, plate-fin heat exchangers have the advantages of high heat transfer efficiency, low pollution rate, and strong pressure bearing capacity. They can meet the requirements of heat exchanger size and performance in existing automotive application scenarios. Therefore, developing a set of high-performance automotive heat exchanger analysis and optimization methods is the key to breaking through the performance bottleneck of compact automotive heat exchangers. Summary of the invention
[0004] In view of the above-mentioned shortcomings, the present invention proposes an optimization method for a compact vehicle heat exchanger based on the RF random forest algorithm, and adopts the RF random forest algorithm to predict and optimize the heat exchange effect of the compact vehicle heat exchanger, giving full play to the advantage of the RF random forest algorithm that it can integrate the prediction results of multiple trees, thereby reducing the risk of overfitting and improving the generalization ability of the model.
[0005] To achieve the above object, the present invention provides the following technical solution: a method for optimizing a compact heat exchanger for a vehicle based on an RF random forest algorithm, comprising the following steps:
[0006] S1: Collect several parameters affecting the performance of compact heat exchangers for vehicles and perform numerical simulations;
[0007] S2: Based on the numerical simulation results, multiple sets of parameter data that affect the heat transfer performance of the automotive compact heat exchanger are collected;
[0008] S3: Establish an RF random forest algorithm integration model, input various influential parameter data collected by numerical simulation into the RF random forest algorithm integration model, and perform training and learning;
[0009] S4: Predict the numerical simulation data in S2 through the RF random forest algorithm integrated model trained in step S3, compare the prediction results with the collected influential parameter data, and verify the prediction accuracy of the training model;
[0010] S5: Utilize the RF random forest algorithm model to optimize and predict the heat transfer performance of automotive compact heat exchangers.
[0011] As an improvement, the numerical simulation in step S1 adopts a CFD simulation method, and obtains the discrete distribution of the flow field of the fluid flow in a continuous area by numerically solving the differential equations controlling the fluid flow, thereby approximately simulating the fluid flow. The specific situation of the numerical simulation method based on CFD simulation is as follows:
[0012] In the numerical simulation based on the CFD simulation method, it is assumed that the heat transfer medium is a uniform and stable fluid, the viscous dissipation is ignored, and the heat capacity, density, viscosity and thermal conductivity of the physical parameters remain unchanged. It follows three basic control equations, namely the mass conservation equation, momentum conservation equation and energy conservation equation.
[0013] The mass conservation equation is:
[0014]
[0015] Among them, t represents time, u, v, w represent the components of the velocity vector in the x, y, z directions,
[0016] The momentum conservation equation is:
[0017]
[0018] Where p represents the pressure on the fluid microelement, F x 、F y 、F z represents the components of body force in the x, y, and z directions,
[0019] The energy conservation equation is:
[0020]
[0021] Among them, C p represents the specific heat capacity,
[0022] When calculating through the CFD simulation method, the simulation process is based on the three-dimensional steady-state flow assumption, and the laminar viscosity model is selected. At the same time, the boundary conditions are a velocity inlet and a pressure outlet, and the Coupled algorithm is used to solve the coupling relationship between pressure and velocity. When the residual reaches 10 -6 , the calculation result converges and the simulation calculation is completed.
[0023] As an improvement, the influential parameters in step S2 include environmental parameters and structural parameters. The environmental parameters include inlet air temperature, humidity, inlet air volume, windward speed, and vehicle speed. The structural parameters include the internal surface roughness of the heat exchanger, vehicle body weight, fin spacing and thickness, type and flow rate of the cooling medium.
[0024] As an improvement, the influential parameter data input into the RF random forest algorithm integration model in step S3 is divided into a training set and a test set.
[0025] As an improvement, the RF random forest algorithm model in step S3 consists of several decision tree classifiers. The size of the training set is N. For each decision tree, N training samples are randomly and with replacement drawn from the training set as the training set of this decision tree. The feature dimension of each sample is M. A constant m << M is specified, and m feature subsets are randomly selected from the M features. Each time the decision tree is split, the one with the best regression quality evaluation for the decision tree is selected from these m features.
[0026] As an improvement, the standard for evaluating the regression quality of the decision tree is the mean squared error, and the calculation formula is:
[0027]
[0028] In the above formula, m is the number of nodes on the decision tree, is the value of the parent node, is the value of the child node. In the selection of each node of the decision tree, the node with a smaller MSE will be regarded as the node with high regression quality.
[0029] As an improvement, the error in the comparison in step S4 does not exceed 5%.
[0030] As an improvement, the coefficient of determination R 2 is selected to evaluate the prediction accuracy of the verification training model, and the calculation formula is:
[0031]
[0032] In the above formula, n is the data volume of the data set, y i is the original data, is the forecast data, is the original data mean, R 2 The value range is [0,1]. If the result is 0, it means that the model fit is very poor. If the result is 1, it means that the model is correct.
[0033] Compared with the prior art, the advantages of the present invention are:
[0034] (1) The present invention uses the RF random forest algorithm to predict and optimize the heat exchange effect of the automotive compact heat exchanger, giving full play to the advantage that the RF random forest algorithm can integrate the prediction results of multiple trees, thereby reducing the risk of overfitting and improving the generalization ability of the model;
[0035] (2) The numerical simulation method of the present invention adopts a CFD simulation method. The principle of this method is to numerically solve the differential equations controlling the fluid flow, and obtain the discrete distribution of the flow field of the fluid flow in a continuous area, thereby approximately simulating the fluid flow situation, making the numerical simulation results more accurate, thereby improving the accuracy of the RF random forest algorithm in optimizing the prediction of the heat transfer performance of the automotive compact heat exchanger;
[0036] (3) The present invention collects several parameters that affect the performance of the compact vehicle heat exchanger, and determines the important influencing parameters that affect the heat transfer effect of the vehicle heat exchanger through numerical simulation using a CFD simulation method, thereby effectively avoiding the influence of irrelevant parameters on the training of the RF random forest algorithm, thereby affecting the prediction accuracy of the training model. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:
[0038] Figure 1 It is a simplified model schematic diagram in the numerical simulation part;
[0039] Figure 2 This is the model diagram of the RF random forest algorithm;
[0040] Figure 3 The figure is a flow chart of an optimization method for compact heat exchangers for vehicles based on the RF random forest algorithm. DETAILED DESCRIPTION
[0041] In the present invention, it is necessary to understand that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "planar direction", "circumferential" 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, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.
[0042] like Figure 1 As shown in the figure, it is a simplified model of a compact heat exchanger. There are several factors that affect the efficiency of automotive compact heat exchangers. Environmental parameters such as inlet temperature, humidity, air volume, windward speed, vehicle speed, etc., and structural parameters such as the roughness of the internal surface of the heat exchanger, vehicle weight, fin spacing and thickness, cooling medium type and flow rate will affect the heat transfer performance of the heat exchanger. These influencing parameters can be verified in numerical simulations.
[0043] For the influencing parameters such as inlet air temperature, humidity, cooling medium flow rate and inlet air volume, adjustments are made to the inlet settings in the boundary conditions. The idea of controlling variables is applied. When discussing the influence of inlet air temperature, the inlet air temperature is changed, and the other variables remain unchanged. The inlet air volume can be calculated by selecting the windward area:
[0044]
[0045] Where V m is the mass flow rate on the windward surface, m is the gas mass flow rate, and F is the windward area.
[0046] For the influencing parameters such as headwind speed and vehicle speed, adjustments are made to the wall settings in the boundary conditions. The idea of controlling variables is applied. When discussing the influence of headwind speed, the headwind speed is changed, and the other variables remain unchanged.
[0047] For parameters influencing the heat exchanger's internal surface roughness, vehicle weight, and type of cooling medium, adjustments are made to the material settings, and the effects of these parameters on the heat exchange performance of the heat exchanger are studied by adjusting the different physical properties of the materials.
[0048] The influencing parameters, fin spacing and thickness, were adjusted when the model was established. The heat exchanger model was changed by changing the fin spacing and thickness, and finally imported into the numerical simulation software to explore the influence of these parameters on the heat transfer performance of the heat exchanger.
[0049] In the numerical simulation based on the CFD simulation method, the influence of influencing parameters on the performance of the heat exchanger is explored. Under the premise of controlling the variables, the influence of a single variable is judged by observing the Nu number of the heat exchanger to determine the quality of the heat exchange effect. Under the relevant heat exchange effect, the more compact the heat exchanger and the smaller the volume, the higher the performance of the heat exchanger and the more it meets the requirements.
[0050] like Figure 2 As shown in the figure, the random forest algorithm is an algorithm based on ensemble learning. It performs regression tasks by constructing multiple decision trees and integrating their prediction results. In a random forest, each decision tree is trained independently on a randomly selected subsample, which can effectively reduce the risk of overfitting. The random forest obtains the final regression result by averaging or weighted averaging the prediction results of multiple decision trees.
[0051] The prediction process of the RF random forest algorithm mainly includes three parts: data splitting, training and testing. In the data splitting stage, the data set is divided into two subsets, namely the training set and the test set. The data part obtained by numerical simulation is used as the training set to train the model, and the remaining part is used as the test set to compare the model obtained by prediction fitting to observe whether the error result is within 5%. The RF random forest algorithm is a type of machine learning algorithm based on decision trees. In the process of using the training set for model training, most of the data will be directly packaged and fitted through different decision trees.
[0052] The standard used by decision trees to evaluate regression quality is mean square error, and the calculation formula is:
[0053]
[0054] In the above formula, m is the number of nodes in the decision tree. is the value of the parent node, is the value of the child node. In the selection of each node in the decision tree, the node with a smaller MSE will be regarded as a node with high regression quality.
[0055] The main function of the test set is to verify the effect of the model. The decision tree used in the test phase does not need to be trained again, and the tree determined in the training phase can be directly used. Similarly, the parameters used for packaging are also saved during the training process. After obtaining the final prediction results of the test data set, the accuracy of the model can be calculated. The present invention uses the determination coefficient R 2 To evaluate the model, the calculation formula is:
[0056]
[0057] In the above formula, n is the amount of data in the dataset, and y is i is the original data, is the forecast data, is the original data mean. R 2 The value range is [0,1]. If the result is 0, it means that the model fit is very poor. If the result is 1, it means that the model is correct.
[0058] like Figure 3 As shown, the present invention proposes a method for optimizing a compact heat exchanger for a vehicle based on an RF random forest algorithm, comprising the following steps:
[0059] S1: Collect parameters that affect the performance of several types of compact heat exchangers for vehicles
[0060] S2: Through numerical simulation, the performance of compact heat exchangers for vehicles of various models and under different conditions is evaluated, and the volume fraction is used as the evaluation index. Under the same heat transfer effect, the more compact the heat exchanger, the smaller the volume, and the higher the performance;
[0061] S3: Based on the numerical simulation results, the influence of two types of influencing parameters, environmental parameters and structural parameters, on the heat transfer performance of automotive compact heat exchangers are analyzed. Environmental parameters include inlet air temperature, humidity, inlet air volume, windward speed, and vehicle speed. Structural parameters include the roughness of the internal surface of the heat exchanger, vehicle weight, fin spacing and thickness, cooling medium type, and flow rate.
[0062] S4: collecting multiple groups of parameter data that affect the heat transfer performance of the vehicle compact heat exchanger, and dividing the data into a training set and a test set. Preferably, 75% of the data is used as the training set, and 25% of the data is used as the test set.
[0063] S5: Establish an RF random forest algorithm integration model, input various influential parameter data collected by numerical simulation into the RF random forest algorithm integration model, and perform training and learning;
[0064] S5: Predict the numerical simulation data in S2 through the RF random forest algorithm integrated model trained in step S3, compare the prediction results with the collected influential parameter data, ensure that the error does not exceed 5%, and verify the prediction accuracy of the training model;
[0065] S6: Use the RF random forest algorithm model to optimize and predict the heat transfer performance of automotive compact heat exchangers.
[0066] The present invention is described above by way of example in conjunction with the accompanying drawings. It is obvious that the specific implementation of the present invention is not limited to the above-mentioned method. As long as various non-substantial improvements are made using the technical solution of the present invention, or the concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.
Claims
1. A method for optimizing a compact heat exchanger for a vehicle based on RF random forest algorithm, characterized in that: It includes the following steps: S1: Collect several parameters affecting the performance of vehicle compact heat exchangers and conduct numerical simulations; S2: According to the numerical simulation results, collect multiple sets of parameter data of the parameters affecting the heat transfer performance of vehicle compact heat exchangers; S3: Establish an RF random forest algorithm integration model, input the various influential parameter data collected by numerical simulation into the RF random forest algorithm integration model, and conduct training and learning; S4: Use the RF random forest algorithm integration model trained in step S3 to predict the numerical simulation data in S2, compare the prediction results with the collected influential parameter data, and verify the prediction accuracy of the training model; S5: Use the RF random forest algorithm model to optimize and predict the heat transfer performance of vehicle compact heat exchangers.
2. The optimization method for a compact heat exchanger for a vehicle based on the RF random forest algorithm according to claim 1, characterized in that: The numerical simulation in step S1 adopts the CFD simulation method. By numerically solving the differential equations governing fluid flow, the discrete distribution of the fluid flow field in the continuous region is obtained, thereby approximately simulating the fluid flow situation. The specific situation of the numerical simulation method based on CFD simulation is as follows: In the numerical simulation based on the CFD simulation method, it is assumed that the heat transfer medium is a uniform and stable fluid, viscous dissipation is ignored, and the heat capacity, density, viscosity, and thermal conductivity in the physical property parameters remain unchanged. It follows three basic control equations, namely the mass conservation equation, the momentum conservation equation, and the energy conservation equation. The mass conservation equation is: where t represents time, and u, v, w represent the components of the velocity vector in the x, y, and z directions. The momentum conservation equation is: Where p represents the pressure on the fluid microelement, F x 、F y 、F z represents the components of body force in the x, y, and z directions, The energy conservation equation is: Among them, C p represents the specific heat capacity, When the CFD simulation method is used for calculation, the simulation process is based on the assumption of three-dimensional steady-state flow, and the laminar viscosity model is selected. At the same time, the boundary conditions use velocity inlet and pressure outlet, and the Coupled algorithm is used to solve the coupling relationship between pressure and velocity. When the residual reaches 10 -6 When , the calculation results converge and the simulation calculation is completed.
3. The optimization method for a compact heat exchanger for a vehicle based on the RF random forest algorithm according to claim 1, characterized in that: The influential parameters in step S2 include environmental parameters and structural parameters. The environmental parameters include inlet air temperature, humidity, inlet air volume, windward speed, and vehicle speed. The structural parameters include the internal surface roughness of the heat exchanger, vehicle body weight, fin pitch and thickness, type and flow rate of the cooling medium.
4. The method for optimizing a compact heat exchanger for a vehicle based on the RF random forest algorithm according to claim 1, characterized in that: The influential parameter data input into the RF random forest algorithm integration model in step S3 is divided into a training set and a test set.
5. The method for optimizing a compact heat exchanger for a vehicle based on the RF random forest algorithm according to claim 4, characterized in that: The RF random forest algorithm model in step S3 consists of several decision tree classifiers. The size of the training set is N. For each decision tree, N training samples are randomly and with replacement drawn from the training set as the training set of this decision tree. The feature dimension of each sample is M. Specify a constant m << M, and randomly select m feature subsets from the M features. Each time the decision tree is split, the one with the best regression quality evaluation for the decision tree is selected from these m features.
6. The optimization method for a compact heat exchanger for a vehicle based on the RF random forest algorithm according to claim 5, characterized in that: The standard for evaluating the regression quality of the decision tree is the mean squared error, and the calculation formula is: In the above formula, m is the number of nodes in the decision tree. is the value of the parent node, It is the value of the child node. In the selection of each node in the decision tree, the node with smaller MSE will be regarded as a node with high regression quality.
7. The method for optimizing a compact heat exchanger for a vehicle based on the RF random forest algorithm according to claim 1, characterized in that: The error in the comparison in step S4 does not exceed 5%.
8. The optimization method for a compact heat exchanger for a vehicle based on an RF random forest algorithm according to claim 1, characterized in that:The prediction accuracy of the validation training model in step S4 is determined by the coefficient R 2 To evaluate the model, the calculation formula is: In the above formula, n is the amount of data in the dataset, and y is i is the original data, is the forecast data, is the original data mean, R 2 The value range is [0,1]. If the result is 0, it means that the model fit is very poor. If the result is 1, it means that the model is correct.
Citation Information
Cited By
Heat exchanger optimization method and device based on multi-source data
CN120562312A
Heat exchanger optimization method and device based on multi-source data
CN120562312B
Heat dissipation control system and method based on efficient heat management
CN120568671A
