Modeling the deformations of a lock by means of neuro-fuzzy techniques
Boehm, Stephanie and Hansjorg Kutterer
The survey and modeling of the deformations of large structures is a major task in engineering geodesy. In this paper, a new procedure to describe and predict the deformations is presented and discussed which is based on Neuro-Fuzzy modeling. Neuro-Fuzzy methods are data driven; they deduce the model directly from the data. Hence, they are mostly convenient if there are no physical models available. They allow the automatic derivation of interpretable rules and the data based simulation of complex processes. In particular, the Adaptive Network based Fuzzy Inference System (ANFIS) technique is used here. It represents a fuzzy inference system which is implemented in the framework of adaptive networks. It is based on a supervised learning algorithm to optimize the parameters of a fuzzy inference system. In this study, the procedure of ANFIS is outlined. The corresponding way of modeling is studied based on geodetic data collected at the lock Uelzen I. This lock is located in Northern Germany at the Elbe-Seiten-Kanal which connects the river Elbe with the Mittellandkanal. The surmounted height difference is about 23 m. The lock has a length of 190 m and a width of 12 m. The size of the flood gate is 12 m x 11 m. Since 1978 the Geodetic Institute of the University of Hanover (GIH) has carried out numerous measurement campaigns at the lock Uelzen. The data used for this study were collected during the last campaign in 2004. They comprise time series of several types of geodetic observations. In this study, data of plummet records and of the water level are used to derive different ANFIS models which are discussed in order to exemplarily show the handling and the benefit of Neuro-Fuzzy modeling.
Event: XXIII International FIG Congress : Shaping the change
Only personal, non-commercial use of this document is allowed.