Large-Scale Gradual Change Detection [Elektronische Ressource] / Heiko Hofer. Gerhard Staude. Konrad Pilzweger. Universität der Bundeswehr München, Fakultät für Elektrotechnik und Informationstechnik

¨ ¨UNIVERSITAT DER BUNDESWEHR MUNCHENFakultat fur Elektrotechnik und Informationstechnik¨ ¨Large-ScaleGradual Change DetectionHeiko HoferVorsitzender des Promotionsausschusses: Prof. Dr.-Ing. G. Bauch1. Berichterstatter Priv.-Doz. Dr.-Ing. G. Staude2. Berichterstatter Prof. Dr.rer.nat. K. PilzwegerTag der Prufung 8.4.2010¨Mit der Promotion erlangter akademischer Grad:Doktor-Ingenieur(Dr.-Ing.)Neubiberg, den 12. April 2010AbstractObservation is one of the import tasks of scientific work. A quantified observationover time is termed a signal. A signal is a documentation of changes in the systemand can be used for analysing the past states of the system. This empirical methodis a central concept in science since the European Renaissance.Inthisthesis,amethodisdevelopedforthecomputerisedanalysisofsignals. Theproblemisthesegmentationofasignalinsegmentswithconstantsystembehaviour.The computer aided analysis has the advantages to decide objectively and togive reproducibleresults. In addition to that, it allows to process a large amount ofdatainreasonabletime. Thesetwopointsareespeciallyimportantinintensivecaremonitoring, where a reliable segmentation is the precondition for the description ofthe patients state.Other applications of segmentation can be found in almost all scientific areas.Among them is the study of global climate changes and the detection of trends ineconometrics.