Design Optimization of Product Forms using Neural Networks: a case study of cellular phones

The purpose is to examine the design optimization on product forms. The optimum combinations of form elements of a product are examined using the grey relational analysis model. To predict and infer the design optimization in the future, Neural Networks models are used. This study focuses on investigating and categorizing of various cellular phones in the consuming market. Fifty-four cellular phones are used as experimental samples and a form elements table, inclusive of nine items and twenty-seven categories, is generalized from the experimental samples. The results of the case study show that the grey relational analysis can find the most important form element and can rise up Neural Networks prediction ability. The results are summarized in three points. First, the “top shape” element of product forms primarily affects the “Simple-Complex” kansei words from the grey relational analysis. Second, we discard the less influential elements of product forms to simplify the model structure, and particularly to have a better prediction ability. Third, the errors of the root mean square of Neural Networks show that the simple model has the higher prediction consistency on the “Simple-Complex” kansei words. This study can develop a decision support system to product designers as an important reference for their design work. It can help the designers do the best choice as they design new products’ form. Although the cellular phones are chosen as the object of the case study, this methodology can be used to develop other products.