The thesis deals with the task of monitoring the hydraulic drives conditions in order to detect faults and hence, aims to enhance their robustness and availability.The focus of the work lay on the fluid related problems, and the analysis is performed from the mechatronic point of view. Methodologies were investigated with emphasis on the economic and practical aspects for in-field applications. A novel approach is suggested based on the unsupervised paradigm of machine learning algorithms. Finally, the applicability of the approach in early, intermediate and advanced fault situations, is examined.weiterlesen