System identification and modal tracking on ship structures

[摘要] ENGLISH SUMMARY: are currently based mainly on dynamic response feedback. Navigators decideon how to operate the vessel based on how they feel it pitching, heaving, rollingand vibrating. The aim of this thesis is to investigate the idea of using systemidentification and modal tracking on polar vessels towards the development ofa decision aiding system. System identification provides a powerful tool forbuilding mathematical models of dynamic systems. An open source toolbox(openSID) for system identification using Stochastic Subspace Identification(SSI) was developed as a research and learning tool. Full scale measurementswere performed on the research vessel Polarstern during an expedition to theArctic. This is the first comprehensive data set including vibration responsesand environmental parameters to span the entire operational profile of a researchvoyage to the Arctic. System identification successfully identified sevenglobal modes in the bandwidth 2 - 10 Hz. Comparisons between differentmethods were used to cross validate results. A modal tracking algorithm wasdeveloped and relationships between identified modes and system inputs wereobserved. A novel method is developed to improve the uncertainty and sensitivityof system identification and tracking, based on a data driven statisticalmodel and a Kalman filter. A key objective is to make experimental data maximallyinformative by using additional system inputs. The model was foundto accurately re-create the training data set and was used to make predictionsbased on future system inputs. The Kalman filter estimates were observedto produce balanced and consistent results. These results demonstrate thepotential of an ice force estimation and structural health monitoring system.

[发布机构] Stellenbosch University