Development of Soft Sensors using Image Processing for Process Monitoring and Advanced Control

Soft sensing (or state estimation) involves online estimation of unmeasured state variables and parameters by combining the measurements available from a system with prediction obtained from a mechanistic dynamic model. Such model-based soft sensors play a vital role in process health monitoring, fault diagnosis, and advanced control. This work aims to estimate tank levels online by combining image processing with nonlinear Bayesian observers, such as extended Kalman filters or unscented Kalman filters. Edge detection algorithms will be used to reconstruct the tank levels in real-time and implement advanced control schemes that use state information for feedback. The project involves experimental studies using the quadruple tank system in the Automation lab.

Co-supervisor: Prof. Mani Bhushan