Classification of operator’s workload based on physiological response

• Autor(a) principal: Mendonça, Beatriz Afonso
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/153929
• topic: Workload; Fatigue; Machine Learning; Operator; Occupational Risks; Industry 4.0; Domínio/Área Científica::Engenharia e Tecnologia::Outras Engenharias e Tecnologias

People spend most of their lives at work, during which time they are exposed to mechanical and environmental conditions that can harm their health. This risk can occur in an hour- long or over long periods, even when performed at a light to moderate intensity due to cumulative fatigue. Several measures have been proposed in order to prevent or reduce fatigue-inducing repetitive work. However, these measures are essentially subjective or only measure fatigue locally. Wearables are an attractive solution to measure work-related fatigue globally and at any time. The purpose of this study is to quantify biosignals information for the determination of fatigue while performing repetitive work. Electrocardiogram (ECG), electromyography (EMG), respiratory inductance plethysmography (RIP) and Accelerometer (ACC) signals were collected from 25 healthy participants. The participants were instructed to perform a repetitive task after induced fatigue. Their biosignals were processed, and different families of features were extracted. These features were used to fit a classifier in order to evaluate fatigue. Self-Similarity Matrix (SSM) was used to select and segment the data in Baseline and Fatigue. Autocorrelation of inertial measures, respiratory synchrony, and the root mean square of the cardiovascular load features achieved 88% of accuracy. It was possible to verify that the ACC’s features lead to the best classification results, followed by the RIP, EMG and finally the ECG’s features. Multimodal data allows global classification of when a person is working after expe- riencing fatigue. Motor information contributes significantly to this classification due to compensations that occur while performing the repetitive task. More studies should be done to develop an index characterising the fatigue state.