본 논문에서는 다중 시그마포인트 세트(MSP)를 사용하는 분산점 칼만필터(UKF)인 UKF-MSP를 소개한다. 비선형 동적시스템을 표현하기 위해 널리 알려진 Bouc-Wen 모델을 사용하였고, 비선형성 고려가 가능한 칼만필터 중 UKF를 선정하였다. 그런데 UKF는 두 가지 인공오차와 시그마포인트의 분포를 결정하는 스케일링 파라미터의 값을 튜닝(Tuning)하는 과정을 통해 적절히 설정해야만 대상 동적시스템의 추정하고자 하는 상태(State)를 정확히 추정할 수가 있다. 본 논문에서는 후자의 스케일링 파라미터 설정 문제를 완화하고자 하였으며, MSP를 사용함으로써 기존 UKF에 비해 칼만필터 튜닝 과정에 덜 민감한 UKF-MSP를 제안하였다. 지진으로 인한 급격한 구조손상 시나리오에 대해 UKF-MSP의 안정성을 검증하였다. 제안된 방법은 튜닝과정을 완화함과 동시에 다른 칼만필 터 파라미터인 인공오차에 대해서도 덜 민감한 거동을 보임을 확인하였다.

In this paper, four damage detection techniques based on dynamic characteristics were utilized and compared to identify damage in a Steel Catenary Riser(SCR). Twelve damage scenarios were simulated by using a numerical model of SCR. The performance on damage detection for each technique was showed and the applicability was discussed.

Since cable members are the major structural components in cable bridges, they should be properly inspected for surface damage as well as inside defects such as corrosion and/or breakage of wires. Starting from August 2010, a new research project supported by Korea Ministry of Land, Transportation Maritime Affairs (MLTM) was initiated to develop the cable inspection robot. In this study, only the vision-based surface damage detection system based on image processing techniques is addressed. The damage detection algorithm combines some image enhancement techniques with principal component analysis (PCA) to detect damages on cable surface. The images from three cameras attached to the cable climbing robot are wirelessly transmitted to the server computer at the cable support. They are processed with image enhancement method together with noise removal technique to improve overall image quality. Then they are projected into PCA sub-space. Finally, the Mahalanobis square distances of the projected images to all sample patterns are calculated. The smallest distance is found to be the match for the input image. The proposed damage detection algorithm was verified through laboratory tests on three types of cables.