Coordinated control of wastewater oxidation processes under constrained incremental control

This article presents a practical solution to effective control of oxidation processes by coordinating multiple control agents. The proposed methodology of agent coordination is based on Model-Free Learning Control (MFLC), based on Reinforcement Learning (RL) framework. The dynamical system is learnt from the online interaction between agent and the process. From the interaction, the agent provides possible actions, which are satisfying incremental, input-output constraints. The possible actions are defined as a function of symbolic states, which in turn varies with the distance to a goal state. The application on an oxidation process at the laboratory level shows that the proposed MFLC learns to control adequately the process.