Model-Free Learning Control for Phenol Wastewater Treatment Plant using Fenton's Reagent

S. SYAFIIE; F. TADEO; E. C. MARTÍNEZ

Bali, Indonesia

Conferencia; Asian Control Conference; 2006

This article discusses an alternative approach for controlling of a wastewater treatment plant using Model-Free Learning Control (MFLC) which is based on reinforcement learning algorithms. The proposed MFLC strategy is based on learning directly from the process response, which is adequate for time varying or nonlinear plants where the development of a model is too costly, time consuming or even no feasible. The experimental testing of the proposed algorithm in a semi-batch process where a synthetic phenol wastewater is treated shows that MFLC gives good performance to achieve and maintain maximum degradation rate using readily available measurements such as ORP and pH.