IA2MAIN

The aim is to develop a digital support based on artificial intelligence to predict the evolution of a dynamic system taking into account human factors on three levels:

1. taking into account expert feedback using knowledge capitalization methods,
and
2. transformation of this knowledge through the production of rules concerning product lifespan and maintenance,
and
3. cooperation between systems and human operators to optimize on-site interventions.

Faced with economic, health and environmental issues. the thesis will look at various performance criteria such as plant downtime, safety, health or workload during maintenance operations requiring unavoidable human contributions for assembly and disassembly actions, decontamination. The preventive maintenance support system will draw on control theory to model the operation of the systems to be diagnosed, and on learning and knowledge capitalization to produce expert rules for the evolution of these systems in an uncertain environment. The human factors-centered digital support will also implement the principles of human-machine cooperation to determine optimal maintenance operations on nuclear reactor cooling systems (i.e. primary pump sets). Primary pumps in nuclear power plants are subject to maintenance cycles of between 12 and 18 months. During this period of operation, various sensors monitor pump constants (pressures, temperatures, flow rates, etc.) for preventive purposes. In this context, the thesis proposes an analysis of these parameters in order to optimize maintenance operations or detect failure of Primary Motor Pump Groups (PMPG).