** 3e prix du jury Fastercom - Meilleure présentation orale **
This research project introduces a machine learning framework to detect and diagnose problems or “faults” in industrial processes from sensor data. By doing so, it aims to prevent accidents that can jeopardize safety, the environment, and cause high production costs. The method uses neural networks which are considered black-boxes, but modifications to the model architecture to reflect the physical system allow interpretations as to why the model detected a given fault. This inherent interpretability allows for greater trust in the model, especially given the high consequences of an industrial accident.
Негізгі бет Ғылым және технология Benjamin Nguyen - AI for interpretable fault detection and diagnosis in industrial systems
Пікірлер