DISTRIBUTED PROCESS PLANT CONTROLLER USING NETWORK CONTROL THEORY

Network Control Theory is a concept of control which investigates control performance of the system by involving multiple nodes of the system. This method is widely used with help of Artificial Neural Network alongside Reinforcement Learning which is an approach in Machine Learning. In this Project, Network Control is used for a nonlinear process plant (Reactor - Flash - Recycler) to see the interaction of DQN agent with the process plant. The interaction of DQN agent with SISO and MIMO linear process plants control has proven to be successful. The simulation for the process and the DQN Agent is developed and using the Relative Gain Array method, the reward is designed for this system. The concept helped to create set of policies which would help to modify the DQN agent for the process plant.