Question Answering for MonitoRed Fuel Cell systEms

Modern advanced and high value fuel cell systems are monitored by multiple embedded sensors which transmit a large amount of data every few seconds.

Unfortunately, service engineers are still faced with the challenging task of identifying the causes of a failure by manually investigating not only the streaming sensor data but also a wide range of structured, semi-structured and unstructured monitoring data.

At the same time, they are required to have a thorough knowledge of the full operating mechanism.Our overarching aim is to utilise next generation deep learning and knowledge technology paradigms (i.e. ontology-based systems, knowledge-graph based systems) to represent this monitoring knowledge in a human and machine processible form such that decision-making processes can be automated and deeper engineering insights can be obtained.

To achieve this, we will implement a radically cross-disciplinary methodological approach, by developing new spatio-temporal knowledge representations and reasoning and instilling them with natural language processing techniques. This will result in a novel paradigm for truly intelligent cyber physical systems.

The QuAre paradigm will be put to test and fine tuned on the diagnosis and prognosis of polymer electrolyte fuel cell systems.On the training side, this project is designed to instill the applicant with a niche set of core skills on question answering over knowledge graph embeddings, knowledge management retrieval, and natural language generation; these will position the researcher at the fore-front of intelligent knowledge representation and establish her as a leading researcher in the field of question answering.

The project is further designed to provide the researcher with cutting edge teaching, leadership, and communication skills so that by the end of this project she will be ready to pursue her first permanent academic position.