本研究透過人因實驗，探討不同聲音的情境下研究參與者的專注程度， 使用腦波儀量測人體頭部的表面電訊號，提取在放鬆與專注狀態下的腦波， 經訊號處理與特徵萃取流程，使用分析工具及機器學習進行腦波狀態的辨 識與分類。 特徵萃取的部分使用了碎形維度(fractal dimension)、近似熵(approximate entropy)與使用希爾伯特-黃轉換(Hilbert-Huang transform)後得到的頻帶功率 作為二元分類的指標，後續使用接收者操作特徵曲線(receiver operating characteristic curve)、支持向量機(support vector machine)及長短期記憶(long short-term memory)三種分析工具進行腦波狀態分類。進而討論研究參與者 在各情境間的腦波差異與不同聲音情境對專注力的影響。;In this study, we explores the effects of different background sounds on the brainwave states through ergonomics experiments. EEG device was used to measure electrical signals on the scalp, extracting participants′ brainwaves during relaxed and focused states. After signal processing and feature extraction, the EEG data were analyzed and classified using analytical tools and machine learning techniques to identify brain states. Feature extraction included fractal dimension, approximate entropy, and band power computed from the Hilbert-Huang transform. These features were used as indicators for binary classification. Subsequently, three analytical tools—receiver operating characteristic curve (ROC), support vector machine (SVM), and long short-term memory (LSTM) neural networks—were employed to classify brain states. The analysis further explored differences in EEG signals across conditions and examined the influence of different sound environments on participants’ levels of concentration.
