Generative adversarial networks for simulating emotional resonance in industrial product design

This paper addresses the lack of emotion-oriented simulation in industrial product design by proposing a novel generative adversarial network framework integrated with a quantifiable emotional model.The core of this approach is an emotion-attention mechanism that dynamically guides the form evolution process.Emotional features are first extracted from e-commerce reviews and modelled via an improved support vector regression algorithm to establish a quantifiable mapping between design elements and user emotions.This emotional model is then integrated into a GAN through a multi-head component attention module, which simulates product form evolution by explicitly weighting the contribution of each component to the target emotional resonance.Experimental results demonstrate the effectiveness of this simulation, with the Fréchet inception distance reduced by at least 38.06%, enabling the generation of industrial products that accurately align with user emotional needs.