![]() |
AutoCBM Automated Adaption of Condition-Based Maintenance methods for Manufacturing Systems Maintaining the proper functionality of manufacturing machines is a crucial factor in the automotive industry. Highly efficient maintenance systems are needed to stay competitive. In the course of the ongoing digitalization, new possibilities arise, to further improve condition-based maintenance systems (CBM). Conventional condition monitoring systems demand a high level of domain expertise and manual tuning when implemented on individual machines. The exhausting task of manually adapting condition-based maintenance systems for individual machines, which is typically done by multiple specialists from different areas, is set to be mostly automated. For this purpose, a machine learning based methodology will be developed to select suitable diagnostic and prognostic methods automatically. A set of machine learning tools and conventional stochastic methods for time series analysis shall be combined to a learning algorithm in such a way, that the quality of the prognostics and its capability to detect anomalies in a manufacturing process can be improved over time. The core of the approach will be a meta-learning system for automatic selection and optimization of prognostic models via self-collected experience data. This way, the usual manual adaptation workload is set to be reduced. Contact persons: H. Engbers ![]() ![]() S. Leohold ![]() ![]() Funded by: Land Bremen / EFRE Duration: 01.07.2020 - 30.06.2022 See project's publications List all projects |
