Detection and estimation of mental fatigue in manual assembly process of complex products

Abstract

Purpose This paper aims to investigate an approach for mental fatigue detection and estimation of assembly operators in the manual assembly process of complex products, with the purpose of founding the basis for adaptive transfer and demonstration of assembly process information (API), and eventually making the manual assembly process smarter and more human-friendly. Design/methodology/approach The proposed approach detects and estimates the mental state of assembly operators by electroencephalography (EEG) signal recording and analysis in an engine assembly experiment. When the subjects perform assembly tasks, their EEG signal is recorded by a portable EEG recording system called Emotiv EPOC+ headset. The feature set of the EEG signal is then extracted by calculating its power spectrum density (PSD), followed by data dimension reduction based on principal component analysis (PCA). The dimension-reduced data are classified by using support vector machines (SVMs), and hence, the mental state of assembly operators can be estimated during the assembly process. Findings The experimental result shows that the proposed approach is able to estimate the mental state of assembly operators within an acceptable accuracy range, and the PCA-based dimension reduction method performs very well by representing the high-dimensional EEG feature set with just a few principal components. Originality/value This paper provides theoretical and experimental basis for the API transfer and demonstration based on human cognition. It provides a new idea to seek balance between the improvement of production efficiency and the sustainable utilization of human resources.