Prediction of Vehicle Form Feature for Cross-model Evolution Based on Improved Nonlinear Grey Bernoulli Model

This paper presents a novel approach dealing with quantitative cross-model evolution of vehicle form features. The proposed model relies on a combination of data-driven and algorithm methods. The Morphological-Homogenization Method is used to determine the relationship between the affected and influenced models along different evolutionary axes. Due to the poor samples size, irregularity, and high oscillatory of form feature changes across generations, a combination of nonlinear grey Bernoulli model and Genetic Algorithm is used. The nonlinear grey Bernoulli model is improved with system latency and time-varying parameters, and Genetic Algorithm is used to optimize the model parameters. The results show that the improved model is more accurate than the Nash nonlinear grey Bernoulli model and is efficient in handling nonlinear sequences. This study uses the contour line of a specific brand's headlamp as a design example. Multiple transformations are applied to the three-dimensional styling while ensuring that the predicted headlamp front view profile remains unchanged. The goal is to provide designers with various styling references and valuable recommendations for both mature and developing manufacturers. © 2023 CAD Solutions, LLC.