A Digital Twin-Driven Method for Product Performance Evaluation Based on Intelligent Psycho-Physiological Analysis

Digital twin, a new emerging and fast-growing technology which is one of the most promising technologies for smart design and manufacturing, has attracted much attention worldwide recently. With the application of digital twin, product performance evaluation has entered the data-driven era. However, traditional methods for evaluation mainly place emphasis on structure analysis in the stage of manufacturing and service in digital twin. They cannot synthesize multi-source information and take the high-level emotional response into consideration in the design stage. To overcome these disadvantages, a digital twin-driven method is proposed evaluating product design schemes in this study. It enables the acquisition of electroencephalogram (EEG) data, physical data, and emotional feedback. Human factors are systematically considered in the evaluation process to establish the information association between EEG and performance levels. Moreover, intelligent psycho-physiological analysis that incorporates EEG into the fuzzy comprehensive evaluation (FCE) and machine learning methods is adopted within the proposed method. It synthesizes human factors such as psychological requirements, subjective and objective assessment indicators to realize a novel machine learning-based EEG analysis. Taking advantage of the binary particle swarm optimization (BPSO) improved Riemannian manifold mapping, Riemann geometry (RG) features are extracted and selected from EEG signals. Differences of implicit psychological states while using the product produced by different design schemes can be more easily detected and classified. A case study of high-speed elevator is conducted to verify the feasibility and effectiveness of the proposed method. The accuracy of EEG classification for performance evaluation reaches 92%.

Automated summary

Digital twin, as an emerging and fast-growing technology for smart design and manufacturing, has attracted widespread attention globally. A digital twin-driven evaluation method proposed in this study systematically considers human factors and incorporates EEG data, physical data, and emotional feedback to achieve machine learning-based EEG analysis.