The high-tech industry is facing an increasing demand from customers to deliver performance and availability-based contracts. This pushes the high-tech industry to change its traditional diagnostics tasks as seen in service and repairs towards an efficient strategy that not only prevents failures, but ensures a system’s performance over the full lifecycle.
Facing this change, the high-tech industry is simultaneously challenged by the increasing complexity of its systems. This turns the development of a diagnostic approach into a difficult engineering task in itself.
For these reasons, we pursue automated diagnostics by design:
We develop efficient system-level techniques for the assisted diagnosis of complex systems and new methodologies to automatically predict the reliability of the system in performing its tasks. Furthermore, we ensure that the diagnosis models at the heart of our approach follow the system design, allowing our industrial partners to realize them with minimal efforts.
The purposes of diagnosis are versatile and cover various stages of a system’s lifecycle.
Next to its role in system health management and quality assurance, we focus on a challenge with raising importance:
The efficient integration of a system into its operational context. Commissioning, parametrization, and adaptation towards a system-of-systems environment as well as to the tasks at hand is necessary due to the increasing demand of customized operations. Doing this effectively and efficiently increases performance and lowers total costs of ownership.
ESI’s diagnosis, prognosis, and root-cause analysis techniques typically center around probabilistic reasoning, e.g., Bayesian networks. This AI technique is suitable for real-time inference, handles uncertainty and missing observations very well, and offers explainable results.
We can fuel the efficient realization of such AI-based diagnostic models with machine-generated data and system engineering models or expertise. This is a crucial advantage for our approach, as it allows us to scale up to industrial use as we enable anomaly detection, forecasting, and root-cause analysis with data science, knowledge graphs, and advanced computations.
Embedding Diagnosability of Complex Industrial Systems Into the Design Process Using a Model-Based Methodology
June 28 - July 2nd, 2021