Improving hierarchical ammonium control operation in wastewater treatment with non-ideal sensors and actuators using filters, adaptive control and measurement error prediction

This study builds upon a hierarchical control strategy for ammonium that regulates the dissolved oxygen set-point in wastewater treatment plants. First, the study aims to improve the ammonium sensor measurement by considering its noise and delay. Noise in ammonium measurement can lead to incorrect control actions by controllers, which can have negative consequences on the aquatic ecosystem, and greater wear on the actuators. Filters are applied with the main goal of reducing variations in ammonium measurement and, consequently, in the actuator, while improving dissolved oxygen set-point tracking. However, filters cause a delay in measurement and therefore also in the controllers' performance, reducing the effects of control strategies on environmental objectives. Thus, ammonium measurement errors are predicted to correct the delay caused by the filters. Finally, adaptive controls are proposed to vary the proportional gain of the controller when there is a change in dissolved oxygen set-point, aiming for a faster actuator response to set-point changes. For the filters, a weighted average filter and an event-based filter are proposed. For ammonium measurement error prediction, linear regressions and artificial neural networks are suggested. Finally, two adaptive controls are applied based on the filter used. Both vary the proportional gain of the controller based on the set-point variation, with one of them being activated only when an event is detected. Various combinations of the proposed techniques were tested, reducing abrupt actuator variations and achieving the maximum integral of squared error reduction of 74% in ammonium measurement error, and 88% in dissolved oxygen control.