HUNG-CHANG LIAO2026-10-162026-10-162010-10-01Pure ID: 15473634Pure UUID: cc5e9c73-7a2c-4a4f-ace1-98488aee8b93https://www.scopus.com/pages/publications/80052525825https://scholars.csmu.edu.tw/handle/123456789/22958The robust design method developed by Taguchi is a popular way of reducing response variations in products and processes through selecting control factors that provide the best performance and the least sensitivity to noise factors. However, the method has some limitations in practical applications, especially in dealing with dynamic problems. This study proposes a novel approach to modeling and optimizing double-dynamic robust design experiments. The proposed approach is based on artificial neural networks (ANN) and an electromagnetism-like mechanism (EM) algorithm. An ANN is used to build a system response function model. The performance measures, signal-to-noise ratio (SNR), and sensitivity used by Taguchi are then integrated by exponential desirability functions to formulate an objective function of the system. An EM algorithm then obtains the optimal factor settings through the response function model. The proposed approach allows the optimal parameter settings for quantitative factors to be any real values within the parameter bounds. A numerical example demonstrates the feasibility and the effectiveness of the proposed approach.15enModeling and optimization of double-dynamic robust design experimentsjournal article