Application of Support Vector Machine Model based on Particle Swarm Optimization for the Evaluation of Products’ Kansei Image

With the development of science and technology and ever complicating process of design objects, scientific methods are needed for the evaluation of products’ Kansei image. This present quantitative experiment proposed a SVM model based on PSO for the evaluation of Kansei image as reflected in the Computer numerical control (CNC) machine. For the first place, we obtained the average scores of Kansei image evaluation of the CNC machine by the questionnaire survey. Then the form elements of the CNC machines were analyzed. Lastly, comparison was made concerning the accuracies of the three methods, namely, BP neural networks, SVM based on cross-validation method and SVM based on PSO in the evaluation of CNC machine’s Kansei image. The research used 35 sets of samples for training as the experimental group, and five other groups as the control group. The results showed that statistics obtained from SVM based on PSO came closer to the mean value from the questionnaires than the other two methods, thus justified its role in intervening consumer’s evaluation on the Kansei image of products.