인공지능(Artificial Intelligence, AI)은 1950년대 초기개념과 이론을 앨런 튜링이 튜링 테스트를 제안하여 기계가 인간과 같은 수준의 지능을 가질 수 있는지 대한 질문을 던지면서 시작되었다. 1980년대부터 특정 분야의 전문 지식을 모방하여 지원하는 AI 시스템인 전 문가 시스템이 부상하기 시작하면서 Machine Learning이 중요성을 얻기 시작하였다. 특히, Decision Tree, Clustering 그리고 Neural Network Algorithm 등이 연구되기 시작하였다. Clustering 기법은 다양한 분야에서 통계분석에 사용되는 자료를 정제하기 위한 비지도 학습 중 하나로, 군집화 알고리즘을 사용하여 자료의 값(Pointer)들을 특정 그룹으로 분류하는 방법이다. 이러한 Clustering을 활용하여 기존 데이터에서 숨겨진 데이터들의 특성을 파악할 수 있으며, 일정 패턴이나 특징을 가진 데이터들끼리의 군집화를 할 수 있게 된다. 이러한 클러스터링은 다양한 산업 분야에서 적용 및 활용하고 있다. 산업화 이후 미국, 벨기에 등 많은 나라에서 효율적인 도로 관 리를 위해 자국의 특성에 맞는 Pavement Management System (PMS)를 운영하고 있지만 현재 많은 분야에서 적용하고 있는 AI를 활용한 사례가 매우 드물다. 한국에서도 수십년 동안 국토교통부와 한국도로공사에서 PMS를 이용하여 도로를 관리해 왔으며, 최근에 는 몇 개 지자체에서 PMS를 도입하였다. 하지만 한국에서는 오랜 PMS 운영 경험에도 불구하고 AI를 활용하지 않고 전통적 방법인 회귀모형을 활용하여 개발한 공용성 예측모형을 사용하고 있기 때문에 그 성능이 떨어지고 있다. 따라서 본 연구에서는 Machine Learning Clustering 기법 을 PMS 자료에 적용이 가능한지 확인하였다. 공용성 예측모형의 종속변수인 Performance Factors와 독립 변수인 Influencing Factors 간의 상관성을 확인할 수 없는 경우 클러스터링을 적용하여 종속변수와 독립변수 간의 상관성을 분명히 나타내고 회귀분석이 가능하도록 하였다. Delaunay Triangulation을 적용하여 인천광역시 기상관측소의 삼각망을 형성하였다. 삼각망 의 각 꼭짓점과 도로 각 지점 간의 거리에 대하여 Inverse Distance Weighted 방법을 적용하여 도로 각 구간의 PMS 자료와 영향인 자를 매칭하였다. 클러스터링 기법을 원자료에 적용한 결과 공용성인자와 영향인자 간의 상관성이 분명해졌다. 또한, 클러스터링 이전 과 이후 자료의 확률밀도함수의 분포를 비교하여 클러스터링 이후의 자료가 이전의 대해서 대표성을 갖고 있는지 확인하였다.

PURPOSES : This study was conducted to evaluate the performance of pressure relief joints (PRJs) installed on highways for preventing blow-ups. METHODS : To evaluate the performance of a PRJ, pavement surface images were acquired by conducting a follow-up survey through on-site visual inspection and an automated pavement condition survey. The PRJ widths in the acquired pavement images were measured using image analysis software, and major distress in the PRJ was identified. The relationship between the performance period and the joint width was bi-linearized, and the causes of distress and repair methods for PRJ were suggested. RESULTS : As a result of the first survey, it became known that the width of a PRJ rapidly decreases after installation owing to an expansion of the concrete slab. The width of a PRJ continued to decrease, with 94% of the joint below 30 mm and 64% of the joint being below 10 mm, based on the fourth survey. The PRJ width of an Alkali-Aggregate Reaction (AAR) section decreased more than that of a normal section, but the difference in the average width between both sections decreased from 9.7 mm to 2.7 mm over the measured period. A bilinear estimation equation was developed based on the survey data. Through the estimation equation, it could be confirmed that joint contraction accelerated after installation owing to the effects of AAR, and that the joint widths of both sections converged to a similar level over the measurement period. As the result of the major distress analysis indicates, joint seal failure occurred in 70% of the joints, and the distress rates in terms of cracks, edge breakouts and spalling, and patching were 73%, 57%, and 28%, respectively. CONCLUSIONS : With the follow-up survey it could be confirmed that the concrete slab continued to expand despite seasonal changes, maintenance of the PRJ, and additional installation of new PRJs. The expansion of the concrete slab due to AAR is considered to be closely related to PRJ behavior. In addition, it is judged that major distress of the PRJ occurs at an early age owing to the instantaneous release of excessive compressive stress inherent in the slab during joint cutting.

PURPOSES : This study deals with a pressure relief joint, which is one of primary preventive methods of blow-up in concrete pavement. The purpose of the study is to estimate the joint sealant protrusion of pressure relief joint filler types according to horizontal displacement of concrete pavement by applying a variety of joint sealants and joint fillers. And test method for resistance of concrete to chloride ion penetration and test method for resistance of concrete to rapid freezing and thawing were conducted to analyze the improvement of concrete durability according to the primer types on concrete surface of stress relief joint. METHODS : Joint fillers of pressure relief joint were categorized into four different types, which are was styrofoam+backer+sealant(type 1), styrofoam+sealant(type 2), foaming styrofoam+sealant(type 3), and preformed joint+sealant(type 4). By varying the depth (10, 20, 30, 50 mm) from the top of the test specimens to the sealant’s surface, the test factors were evaluated for a total of 16 variables. When the specimen’s joint spacing decreased from 70mm to 10mm, the load was stopped. And the displacement of the center of the joint protrusion was measured. The test was terminated when the specimen joint spacing was reduced to 60 mm. The horizontal displacement at the time when the joint protrudes over the specimen surface is recorded and analyzed as the critical threshold displacement. RESULTS : According to the test results according to the type of joint filling material, it was found that there was a difference in the protrusion of the horizontal compression displacement according to the joint filling type. Under the current installation standard of 20mm, the preformed seal joint member showed the best crimping characteristics by securing the safety against protrusion until the horizontal displacement of 50mm occurred. CONCLUSIONS : The most common failures in pressure relief joints are those related to joint sealants, which can be minimized by changing the current joint type, installation depth, etc. to suppress them.