A Study of BCG Signal-based Sleep Classification Technology through Ensemble Running Signal Processing and Piezoelectric Sensor Surface Material Change

Because drowsy driving increases the incidence of traffic accidents and leads to fatal accidents, a lot of social attention is needed. The number of accidents by drowsy drivers increases every year. So various studies have been performed to solve the issue all over the world. Among others, we focus...

Full description

Saved in:
Bibliographic Details
Published inJournal of electrical engineering & technology Vol. 18; no. 5; pp. 3881 - 3886
Main Authors Yang, Chulseung, Ku, Giwon, Jung, Jiseong, Choi, Junhyuk, Kim, Kyungho
Format Journal Article
LanguageEnglish
Published Singapore Springer Nature Singapore 01.09.2023
대한전기학회
Subjects
Electrical Engineering
Electrical Machines and Networks
Electronics and Microelectronics
Engineering
Instrumentation
Original Article
Power Electronics
전기공학
CNN
Piezoelectric sensor
LSTM
RNN
BCG
Signal processing
Ensemble learning
Drowsiness detection
Online AccessGet full text

Cover

More Information
Summary:Because drowsy driving increases the incidence of traffic accidents and leads to fatal accidents, a lot of social attention is needed. The number of accidents by drowsy drivers increases every year. So various studies have been performed to solve the issue all over the world. Among others, we focus on the non-contact method. Various vibration signals such as engine, tire, and body vibration are generated in a driving or stationary car. The existing static system is used to measure microscopic vibrations generated from the drivers breathing and heartbeat. It is necessary to design a structure that can cushion vehicle vibration through an advanced signal processing program. In this study, actual vehicles test was conducted to analyze the transmission characteristics of fine vibration through the change of plate structure under sensors. The plate structure is a urethane structure plate that can cushion vehicle vibration while driving. Furthermore, we developed an AI algorithm that classifies whether a subject is in a sleep state or not using a piezoelectric sensor-based BCG signal. This paper shows the AI method is more accurate than the method of classifying sleep states according to analyzing HRV and the ratio of LF/HF with ECG signal. In order to train the system, the subject's biosignals were acquired every 30 s, and 797 data were comparatively analyzed.
ISSN:1975-0102
2093-7423
DOI:10.1007/s42835-023-01468-1