Title: Research on aerodynamic attachments parameter optimisation by integrating BP neural network and genetic algorithm

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

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

Abstract: In this study, aerodynamic attachments at the rear of a vehicle are optimised using CFD simulation and Latin hypercube design to reduce drag and improve high speed performance. Models are constructed based on four design variables and simulated using ANSYS fluent and realisable k-ε models. Neural networks and genetic algorithms were combined to find the optimal solution, resulting in a drag reduction of more than 11.7%. The study also analyses the effect of each variable on drag using random forest and verifies the reliability of the results. Ultimately, it provides a new approach to optimising the aerodynamic performance of vehicles, helping to reduce energy consumption.

Keywords: computational fluid dynamics; CFD; aerodynamic attachments; BP neural network; optimal Latin hypercube design; genetic algorithm.

DOI: 10.1504/IJVP.2025.144284

International Journal of Vehicle Performance, 2025 Vol.11 No.1, pp.53 - 78

Received: 23 Sep 2024
Accepted: 14 Nov 2024
Published online: 04 Feb 2025 *