DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Yim, Man-Sung | - |
dc.contributor.advisor | 임만성 | - |
dc.contributor.author | Kim, Tae Ryoun | - |
dc.date.accessioned | 2023-06-26T19:33:20Z | - |
dc.date.available | 2023-06-26T19:33:20Z | - |
dc.date.issued | 2023 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=1032830&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/309779 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 원자력및양자공학과, 2023.2,[iv, 40 p. :] | - |
dc.description.abstract | This research thesis developed an EEG smart helmet to assess the feasibility of intention-based unsafe act monitoring in nuclear power plant. To identify implicit intention of operator behaviors, two deception experiment paradigms were established: face recognition test and deceptive error intention task. Faces were categorized based on familiarity whilst the error interactions are differentiated based on intention. Participants were instructed to conceal certain known faces (probe), and also informed to falsely report about self-aware errors. Then, the unintended and intended errors, and probe and unknown faces were classified using machine learning pipelines. For practical implementation for nuclear operators, EEG smart helmet was constructed in consideration of usability and utility. Flat or spiked, and dry electrodes were adopted to collect psychophysiological data | - |
dc.description.abstract | electrodes were located in frontal and temporal lobes to analyze necessary neural signatures | - |
dc.description.abstract | and wireless connection via Bluetooth was employed for mobility. Event-related potential (ERP) waveforms depicted in face recognition test for both instruments was significant (p < 0.05) in N250 and N400. Classification and prediction results were average 70~80%. Similarly, the ERP components in deceptive error intention task for both devices had statistical significance between unintended and intended errors for ERN, N400, and P600. Despite the high accuracy in validation and prediction steps, the mean AUC scores were less than 0.5 ~ 0.6. Overall, the EEG smart helmet’s performance in comparison to the MUSE 2 was successful as it differentiated concealed known and unknown individual, and falsely reported recognized errors | - |
dc.description.abstract | however, test predictions had underperformed precision and recall scores. To be applied in realistic environment, the experiment needs to be proven with actual operators in NPP and fabricated EEG smart helmet should be adjusted for better signal quality. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | human error▼aunsafe act▼aintention▼aelectroencephalography▼abiosignal▼amachine learning▼anuclear safety | - |
dc.subject | 인적오류▼a불안전 행동▼a의도▼a뇌파▼a생체신호▼a머신러닝▼a원자력 안전 | - |
dc.title | Development of EEG smart helmet for intention-based unsafe act intention monitoring in nuclear power plants | - |
dc.title.alternative | 원전 불안전 행동 모니터링을 위한 의도기반 뇌파 스마트 헬멧 개발 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :원자력및양자공학과, | - |
dc.contributor.alternativeauthor | 김태련 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.