Development of an intelligent system for automated generation of diagnostic and prognostic models for health management

In the era of digital transformation and interconnectivity, the useful life of a technical system can be increased and optimized through condition monitoring. As the name implies, condition monitoring is the continuous measurement of the condition of a technical system through wear- and degradation-related quantities such as vibration and temperature with appropriate sensors. The data is used to infer changes caused by friction and wear to derive the state or the remaining useful life of a system. The predictions can then be used to schedule appropriate maintenance or to adaptively control a system to ensure that its mission is accomplished.

Mechatronic systems are able to provide insights on the current system status from condition monitoring data. To evaluate these data and identify an existing fault or estimate the remaining useful life of a system, intelligent diagnostic and prognostic methods are necessary. A toolbox has been developed at the chair that comprises feature extraction and selection methods, state of the art diagnostic and prognostic algorithms. Such a toolbox that incorporates intelligent prognostic and health management methodologies for mechatronic systems allows the estimation of the current health of a component or system and the prediction of the future condition of the component or system based on condition monitoring data. The further development is subject of a current project for a direct application in industry at shop floor level. Modularization, automation and scalability are focused. The overall goal is to create a holistic platform for small and medium-size companies in which the toolbox is used to generate diagnostic and prognostic models.

Publications of the chair for this research area can be found here.
The PhD thesis of Dr. Kimotho is available here.