The Smart Mobility and Low-Altitude Economy Research Team has made significant research progress in the field of data-driven sustainable design for new energy vehicles

This research was jointly accomplished by Nanchang University, Northwestern Polytechnical University, and Nanchang Hangkong University. The relevant findings, entitled Advancing Data-Driven Sustainable Design: A Novel NEV Styling Design Approach for China’s Market, were published in the Journal of Cleaner Production, an authoritative journal in the field of environment and sustainability.

Against the urgent backdrop of global sustainable development and transportation decarbonization, this study responds to the demand for the green transformation of the new energy vehicle industry. It aims to address the core pain points of the traditional automotive form design process: the subjective and inefficient extraction of user needs, and the lengthy design transformation cycle. Meanwhile, to make up for the shortcomings of existing data-driven design research—mostly limited to single-modal and single-perspective generation, and lacking systematic integration with engineering performance verification—an innovative intelligent design framework integrating cross-modal generation and multi-criteria evaluation is proposed. This framework is intended to promote the development of automotive design toward a data-driven, intelligent, efficient, and sustainable direction.

This study innovatively proposes a data-driven intelligent design framework named Cross-Modal Generation with Multi-Criteria Evaluation (CMG-MCE). Methodologically, it comprehensively employs a cross-modal generative adversarial network to convert users’ textual requirements into multi-perspective 2D image schemes, utilizes deep convolutional neural networks (DCNNs) for preliminary screening of Kansei images, adopts the VIKOR multi-criteria decision-making method for quantitative evaluation and ranking of schemes, and finally uses Computational Fluid Dynamics (CFD) simulation to verify the aerodynamic performance of the 3D model of the optimal scheme. Following a systematic experimental approach, the study takes a new energy vehicle (NEV) design for the Chinese market as a case study, fully demonstrating the entire process—from mining user keywords from massive online reviews, generating design schemes, screening and integrating multi-perspective images, to ultimately outputting a 3D model with a low drag coefficient. The research conclusions indicate that this framework can effectively achieve the rapid and objective transformation from vague user language to engineerable design schemes. The generated designs not only meet users’ Kansei expectations but also possess excellent aerodynamic performance, verifying the feasibility and significant value of data-driven methods in improving design efficiency and promoting the sustainable design and clean production of new energy vehicles.

Paper Link：https://doi.org/10.1016/j.jclepro.2024.142626