Neural Network Adaptive Control of Dissolved Oxygen for an Activated Sludge process

Activated sludge wastewater treatment processes are difficult to be controlled because of their complex and nonlinear behavior, however, The removal of nutrients and pollutants is carried out by microorganisms, these require oxygen to break down the waste in the water; oxygen has been delivered by pumping air through diffusers to create bubbles it's a very energy intensive process this can be used to increase biological capacity and creates an ideal environment to support the substrate which absorbs and consumes the polluants and incrased consumption of the polluants found in the wastewater. For this reason, a new approach is explored in this paper using an adaptive control algorithm based on neural networks Cerebellar Model Arithmetic Computer (CMAC) compared with PI control at the desired reference for dissolved oxygen control to maintain a destination point in aerated bioreactors. The controller is tested on a simplified version of the simulation reference model number 1, and provides a high performance and efficiency to disturbances.