Article: Multi-objective optimisation of automotive drag coefficient and lift coefficient via artificial neural network and genetic algorithm Journal: Progress in Computational Fluid Dynamics, An International Journal (PCFD) 2025 Vol.25 No.2 pp.79 - 94 Abstract: In terms of combining traditional methods with others, group method of data handling and radial basis function neural networks have been applied to the study of automotive aerodynamics. Backpropagation neural networks have relatively few applications. The aerodynamic performance of a vehicle is improved by optimising its shape parameters. In this paper, a shape parameter optimisation method is proposed which employs backpropagation neural network model and genetic algorithm to reduce the air resistance and air lift. Five form parameters of the car are taken as design variables and 50 sets of sample data are designed using optimal Latin hypercube experimental design method. After performing CFD simulation, all the data are used for backpropagation neural network learning. Genetic algorithm is then used to perform multi-objective optimisation on all the data and the Pareto front is developed. The results show that the CD and CL of the optimised car are reduced by about 15.28% and 25.85%, respectively. Inderscience Publishers - linking academia, business and industry through research

Title: Multi-objective optimisation of automotive drag coefficient and lift coefficient via artificial neural network and genetic algorithm

Authors: Zihou Yuan; Hongwei Zhang; Wangyang Xiang; Yanming Du; Xingren Zheng

Addresses: Hubei Key Laboratory of Digital Textile Equipment, Wuhan Textile University, Wuhan, Hubei, 430200, China ' Hubei Key Laboratory of Digital Textile Equipment, Wuhan Textile University, Wuhan, Hubei, 430200, China ' Hubei Key Laboratory of Digital Textile Equipment, Wuhan Textile University, Wuhan, Hubei, 430200, China ' Hubei Key Laboratory of Digital Textile Equipment, Wuhan Textile University, Wuhan, Hubei, 430200, China ' Hubei Key Laboratory of Digital Textile Equipment, Wuhan Textile University, Wuhan, Hubei, 430200, China

Abstract: In terms of combining traditional methods with others, group method of data handling and radial basis function neural networks have been applied to the study of automotive aerodynamics. Backpropagation neural networks have relatively few applications. The aerodynamic performance of a vehicle is improved by optimising its shape parameters. In this paper, a shape parameter optimisation method is proposed which employs backpropagation neural network model and genetic algorithm to reduce the air resistance and air lift. Five form parameters of the car are taken as design variables and 50 sets of sample data are designed using optimal Latin hypercube experimental design method. After performing CFD simulation, all the data are used for backpropagation neural network learning. Genetic algorithm is then used to perform multi-objective optimisation on all the data and the Pareto front is developed. The results show that the C_D and C_L of the optimised car are reduced by about 15.28% and 25.85%, respectively.

Keywords: shape parameters; computational fluid dynamics; CFD; BP neural network; NSGA-II; multi-objective optimisation; Pareto front.

DOI: 10.1504/PCFD.2025.144811

Progress in Computational Fluid Dynamics, An International Journal, 2025 Vol.25 No.2, pp.79 - 94

Received: 22 Apr 2024
Accepted: 20 Jun 2024
Published online: 03 Mar 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Multi-objective optimisation of automotive drag coefficient and lift coefficient via artificial neural network and genetic algorithm

Keep up-to-date