With increase in the number of sensors installed on
sub-assemblies of industrial components, the amount of data collected is
rapidly increasing. These data hold information in the areas of operation of
the system and evolution of health condition of the components. Therefore,
extracting the knowledge from the data can bring about significant improvements
in the aforementioned areas.
This dissertation provides a path for achieving such an
objective. It starts by analyzing the data at the sub-assembly level of the
components and creates four frameworks for analysis of operation and
maintenance (O&M) for past, present and future horizons at the component
level. These frameworks allow improvement in operation, maintenance planning,
cost reduction, efficiency and performance of the industrial components. Next,
the dissertation evaluates whether such models can be linked with system level
analysis and how providing such a link could provide additional improvements
for system operators. Finally, preventive maintenance (PM) in generation
maintenance scheduling (GMS) in electric power systems is reviewed and updated
with recent advancements such as connection to the electricity market and
detailed implementation of health condition indicators into the maintenance
models. In particular, maintenance scheduling through game theory in
deregulated power system, for of fshore
w ind farm (OWF) and an islanded
microgrid (MG) are investigated.
The results demonstrate improvements in reducing cost and
increasing profit for the market agents and system operators as well as asset
owners. Moreover, the models also deliver an insight on how direct integration
of the collected operation data through the developed component level models
can assist in improving the operation and management of maintenance.
Keywords: Anomaly Detection, Condition Monitoring,
Maintenance Management, Performance Evaluation, Data Analytics, Mathematical
Modeling, Optimization