퍼지뉴럴네트워크를 이용한 상충데이터 정제기법 연구

Abstract

When the number of reported accidents to evaluate the safety of intersections is insufficient, we can use the traffic conflict technique. However, It raises a problem related to the variation of individual surveyor's recognition. This paper, to reduce the variation of individual surveyor's recognition, uses the Fuzzy Neural Network System(FNN). It is made up of a Fuzzy which is used to reduce the human obscurity, and a Neural Network System which is used to abstract, learn and memorize the certain events like the activity of human brain. At the result, the proposed model of this paper showed that it reduced the variation of surveyor's recognition until about 70 percentiles and could construct a more accurate traffic conflict model.