AI 학습 기반 키워드분석을 통한 노면 낙하물 사고분석 연구 = A Study on the Analysis of Falling Objects from Vehicle Accidents through AI Learning-Based Keyword Analysis

다국어 초록 (Multilingual Abstract)

Traffic accidents caused by road drops have been steadily increasing in recent years, and have become a social issue. In this study, keyword analysis using big data-based AI was conducted to prevent accidents caused by domestic road drops based on traffic accident data from 2015 to 2019. Keyword analysis performed text mining using Python. According to the analysis, the biggest causes of road drop accidents were poor loading and overloading. Consequently a crackdown on small trucks within the haulage industry is urgently needed. Meanwhile, road managers do not have the authority to crack down on the vehicles that are in violation of domestic load regulations, and the method of fixing the load has not been specifically proposed as a law or a related clause. Therefore, there is a problem in that the criteria for determining violations when cracking down on poor loading can only be changed according to the supervision of the regulator. In order to address the limits of the crackdown on poorly loaded vehicles in Korea, preparing a manual that can be used for driver training or crackdown is urgently required, also new legal regulations that dictate specific standards are needed with haste. In future research, new R&D research projects such as the prevention of road drop by classification of the type of traffic accident and analysis of road drop type data mining using big data should be proposed.

국문 초록 (Abstract)

최근 노면 낙하물로 인한 교통사고는 꾸준히 증가하고 있으며, 사회적으로 이슈화되고 있다. 본 연구에서는 2015년부터 2019년까지의 한국도로공사의 교통사고 발생자료를 바탕으로 노면 낙하물로 인한 사고 예방을 위해 Big Data 기반의 AI를 활용한 키워드 분석을 실시하였다. 키워드 분석은 파이썬을 이용하여 Text Mining을 수행하였다. 분석 결과, 노면 낙하물 사고의 가장 큰 발생원인은 적재불량과 과적에 해당했으며, 화물차 중에서도 소형 화물차의 단속이 절실한 실정이다. 한편, 국내 적재불량 차량의 경우 도로관리자에게 위반차량 단속권한이 없고, 적재물 고정방법이 법이나 관련 조항으로 구체적인 제시가 되어 있지 않다. 따라서 적재불량 단속 시 위반을 판단하는 기준이 단속자의 주관에 따라 달라질 수밖에 없는 문제가 발생하기도 한다. 국내 적재불량 차량의 단속 한계를 해결하기 위해, 구체적인 기준이 명시된 법적 규정뿐만 아니라, 운전자 교육이나 단속에 활용할 수 있는 메뉴얼 마련이 시급하다. 향후 연구에서는 교통사고 유형 분류에 따른 노면 낙하물 발생 예방 연구와 빅데이터를 이용한 노면 낙하물 종류 Data Mining 분석 등 새로운 R&D 연구사업을 제안하고자 한다.

참고문헌 (Reference)

1 National Police Agency (Korea ITS Association, Road Traffic Authority), "Translations of Foreign Road Traffic Act III: Germany"

2 British Department of Transportation (DfT) Homepage, "Translations of Foreign Road Traffic Act II: British"

3 Korea Legislation Research Institute, "France and Japan’s Road Traffic Act"

4 National Police Agency (Korea ITS Association, Road Traffic Authority), "Foreign Road Traffic Act Translation Book IV: United States"

5 National Police Agency(Korea ITS Association, Road Traffic Authority), "Foreign Road Traffic Act Translation Book I: Japan"

6 Road Traffic Authority, "A Study on the Traffic Safety Consciousness Survey of Truck Drivers and the Measures to Prevent Drop Accidents of Loading" 2018

7 Road Traffic Authority, "A Study on the Regulation of ｢Poor Loading｣ Open Cargo Truck on Highway Entry" 2017

8 Korea Expressway Corporation Research Institute, "A Study on the Improvement of Highway Traffic Restriction Vehicle Control System Using Automatic Measuring Equipment" 2013