Wenqing Yao
This research aims at enhancing the functionality and reliability of the control and monitoring system to benefit nuclear power plant (NPP) safety and economy. To improve monitoring capabilities and reduce calibration cost, a dynamic data-driven approach is proposed to filter sensor time series data, detect and isolate anomalous/faulty events as well as calibrate both redundant and non-redundant sensing information. From these perspectives, a resilient control scheme is proposed to enhance NPP safety against disturbances and severe accidents in an adaptive and “intelligent” manner.
The project is being performed in collaboration with researchers from mechanical engineering. Topics include: resistance thermometer (RTD) dynamical behavior classification with multiphysics simulation tool COMSOL, cross-calibration of redundant sensors, Support Vector Regression based non-redundant sensor calibration, PWR main steam line break and BWR turbine trip early detection and thermocouple failure detection in a TRIGA reactor.