A framework for now-casting and forecasting in augmented asset management

Abstract

Asset Management of a complex technical system-of-systems needs cross-organizational operation and maintenance, asset data management and context-aware analytics. Emerging technologies such as AI and digitalisation can facilitate the augmentation of asset management (AAM), by providing data-driven and model-driven approaches to analytics, i.e., now-casting and forecasting. However, implementing context-aware now-casting and forecasting analytics in an operational environment with varying contexts such as for fleets and distributed infrastructure is challenging. The number of algorithms in such an implementation can be vast due to the large number of assets and operational contexts for the fleet. To reduce the complexity of the analytics, it is required to optimize the number of algorithms. This can be done by optimizing the number of operational contexts through a generalization and specialization approach based on both fleet behaviour and individual behaviour for improved analytics. This paper proposes a framework for context-aware now-casting and forecasting analytics for AAM based on a top-down, i.e., Fleet2Individual and bottom-up, i.e., Individual2Fleet approach. The proposed framework has been described and verified by applying it to the context of railway rolling stock in Sweden. The benefits of the proposed framework is to provide industries with a tool that can be used to simplify the implementation of AI and digital technologies in now-casting and forecasting.