Combining Kansei Engineering and SSA-XGBoost for new energy vehicle front-face form design
Xinhui Kang, Huiqi Wan, Shuyao Li
Abstract
With the worsening global energy crisis and environmental pollution, new energy vehicles (NEV) have emerged as a vital solution for sustainable development. As a vehicle's most visually striking part, the front-face significantly influences user perception and market competitiveness. This paper proposes a novel front-face design method for NEV by integrating Kansei Engineering (KE) with the Sparrow Search Algorithm-Extreme Gradient Boosting (SSA-XGBoost) to enhance emotional appeal. The research involves four stages. First, representative Kansei words are extracted using the ChatGPT language model and prioritised through the Entropy-AHP method. Second, to break free from existing market design limitations, the Stable Diffusion Model (SDM) generates 99 new NEV images, from which a morphological database and deconstruction table are created using morphological analysis. Third, the SSA-XGBoost model builds a mapping between user emotions and design features, identifying optimal parameter combinations. Finally, use Rhino to construct the three-dimensional model and conduct a sensory and rational evaluation of the car's frontal shape design through aerodynamic performance computational fluid dynamics (CFD) simulations and expert evaluations. Experimental results show that this method achieves higher prediction accuracy and customer satisfaction compared to traditional machine learning approaches, providing a creative and efficient framework for NEV front-face design.