|Ph.D.,||Texas A&M University,||(2006)|
|B.S.,||Indian Institute of Technology, Roorkee, India,||(2000)|
In the process industries, there are rewards to producing higher quality products, to reduce rejection rates, limiting downtime, and to satisfying safety and environmental regulations. Standard process controllers are designed for steady state operating conditions by compensating for small deviations in process variables. In today's chemical processes there are process deviations that go undetected by these steady state controllers.
My research focuses on developing model bases fault diagnosis strategies, which apply process modeling to monitor measurable (temperature, pressure) and immeasurable (concentration, viscosity , density) variables simultaneously and diagnose faults. The idea revolves around generating residuals (the mismatch between actual and ideal values of variables) against model uncertainty and nonlinearity of the process. Specifically, I am considering various nonlinear observers for residual generation which can directly cope with the process nonlinearity rather than linearizing the process around a favorable operating point and testing its efficacy against modeling uncertainty.
Currently, I am setting up a lab scale continuous stirred tank reactor (CSTR) with a nonlinear chemical process which will act as a test bed for various fault diagnosis algorithms.