CHOSUN

뇌파 데이터를 이용한 뉴럴 네트워크 기반의 운전자 이상상태 분석

Metadata Downloads
Author(s)
김지수
Issued Date
2017
Keyword
neural network brain wave
Abstract
As the amount of automobile usage increases, carelessness of drivers incur more big traffic accidents and casualties. According to the causal analysis index of big traffic accidents, drowsiness and drinking accounts for a high percentage of causes of traffic accidents, and it is thought that the incidence of big traffic accidents is directly related to the driver's abnormal condition.
In this paper, the abnormal condition of the driver is analyzied by collecting brainwave data, and Power Spectral Analysis to quantitatively determine each vibration component in the brainwave data. Accordingly, the analyzed data are collected for learning and test data, and classify drivers' abnormal condition with high accuracy through learning modeling for drowsy driving and drunken driving by using DNN(Deep Neural Network), which is a kind of machine learning technique. In conclusion, the performance evaluation of learning modeling about the comparable subjects' data is discussed.
Alternative Title
Analyzing driver's abnormalities based on neural network using brain wave data
Alternative Author(s)
Kim, Ji Su
Department
산업기술융합대학원 소프트웨어융합공학과
Advisor
최준호
Awarded Date
2018. 2
Table Of Contents
ABSTRACT

Ⅰ. 서론 1
A. 연구 배경 및 목적 1
Ⅱ. 관련 연구 3
A. 생체신호를 이용한 운전자 이상상태 분석 연구 3
B. 뇌파 분석 방법 7
1. 뇌파의 종류와 특징 7
a. 뇌파 특성 및 주파수 대역과 파형의 범위 7
b. 주파수별 뇌파의 특징 8
2. 뇌파의 분석 9
C. Machine Learning 분석 기법 11
1. Deep Learning 12
2. Tensorflow Library 14
Ⅲ. 뇌파 데이터를 이용한 운전자 이상상태 분석 17
A. 데이터 셋 18
B. 학습 알고리즘 20
Ⅳ. 실험 및 평가 26
A. 실험 환경 26
B. 실험 방법 27
1. 뇌파 측정 및 데이터 취득 27
2. 뇌파 분석을 통한 학습 모델링 구현 33
3. 실험 평가 및 결과 분석 40
Ⅳ. 결론 및 제언 43

참고문헌 44
Degree
Master
Publisher
조선대학교 산업기술융합대학원
Citation
김지수. (2017). 뇌파 데이터를 이용한 뉴럴 네트워크 기반의 운전자 이상상태 분석
Type
Dissertation
URI
https://oak.chosun.ac.kr/handle/2020.oak/1983
http://chosun.dcollection.net/common/orgView/200000266591
Appears in Collections:
Engineering > Theses(Master)(산업기술창업대학원)
Authorize & License
  • AuthorizeOpen
Files in This Item:

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.