Online Optimizing Control of Nonlinear Processes using Machine Learning Techniques


Process industry is moving towards use of nonlinear dynamic models in advanced process control solutions such as fault tolerant control, process monitoring and online real time optimization. Development of control relevant models that capture system behavior over a wide range is always a challenge in chemical processes because of tight mass and energy integrations through recycle
streams. Availability of huge amounts of historical data has spurred interest in machine learning based approaches for developing grey-box models that adequately capture system non-linearity over a wide range of operation. This project aims at using AIML tools, such as neural nets, to model the underlying systems for deploying in advanced control and monitoring applications.

Research Area(s)