Machine Learning and Knowledge Acquisition in a Computational Architecture for Fault Diagnosis in Engineering Systems

Abstract

We present a computational architecture in a domain of extraordinary economical importance: fault diagnosis in engineering systems. We describe the underlying domain requirements leading to this architecture with a special focus on the included learning tasks. In the sense of KADS, the presented architecture represents an operational design model for technical diagnosis within which machine learning techniques are used to fill the model with concrete knowledge. We show how the underlying computational architecture constrains the involved learning processes.