A Modeling Design Method for Complex Products Based on LSTM Neural Network and Kansei Engineering

Complex products (CPs) modeling design has a long development cycle and high cost, it is difficult to accurately meet the needs of enterprises users. At present, Kansei Engineering (KE) method based on back-propagated (BP) neural networks applied solve problem that meets users’ affective preferences for simple quickly effectively. However, feature data CPs have wide range dimensions, parameter codes, characteristics time series. As result, BP recognize from an overall visual perception level as humans do. To address problems above assist designers with efficient high-quality design, CP Long Short-Term Memory (LSTM) network KE (CP-KEDL) was proposed. Firstly, improved MA carried out transform product features into codes sequence characteristics. Secondly, mapping model between perceptual images established LSTM predict evaluation value product’s images. Finally, optimal sets were calculated by Genetic Algorithm (GA). The experimental results show MSE only 0.02, whereas traditional Deep Neural Networks (DNN) Convolutional (CNN) models are 0.30 0.23, respectively. verified proposed can effectively grapple timing factor, improve satisfaction shorten R&D industrial design.