Description
Brain-computer interface(BCI) is a means of communication that can provide a direct connection from the brain to communication and control devices. Many researchers have focused on the application of BCI for the control of assistive tools, with the use of steady-state visually evoked potential-based brain-machine interfaces (SSVEP-BCI) for assistive control being popular. SSVEP-BCI has the advantage of high accuracy, high transmission rate, relatively high signal-to-noise ratio and no training required. Unfortunately, in SSVEP-BCI, the user needs a lot of effort to focus on the visual stimuli to produce a strong enough SSVEP. Due to high luminance, excessive low-frequency stimuli and single tasks, the user may easily become fatigued.To address this problem, this dissertation presents a real-time fatigue detection system for SSVEP-BCI. The overall work and results are divided into the following sections. (1) This project completes the design and implementation of a wearable EEG device with 4 signal channels, which uses a physiological signal acquisition chip BMF21A1 and GD32 as MCU. it is characterized by high convenience, high accuracy and low power consumption. (2) This project investigated three entropies and found that fuzzy entropy, sample entropy and approximate entropy have consistency in fatigue detection. Fuzzy entropy has better sensitivity. (3) Support vector machine was used to classify the non-linear dynamics fatigue features, and the highest classification accuracy of 98.56% was achieved. (4) The project compared signals from the forehead and occipital regions, and the classification accuracy of the frontal was generally higher than that of the occipital signals . In contrast, the accuracy of the two signals combined was higher than either one of them. (5) This paper provides a systematic BCI-based human state detection process, which includes:point selection, feature selection, feature analysis, feature recognition, and result display.
| Period | Jan 2022 → Dec 2022 |
|---|