Ecology, Environment and Conservation Paper

Vol 27, Issue 4, 2021; Page No.(1719-1733)

EFFLUENT CONTROL AND IMPROVING OF THE PERFORMANCE OF BIOLOGICAL WASTEWATER TREATMENT PLANT USING NEURAL NETWORKS

Drioui Nissrine, El Mazoudi El Houssine and El Alami Jamila

Abstract

In this paper we have developed two models to control the dissolved oxygen concentration which is an important key in the operation of wastewater treatment processes (WWTP) and the nitrate concentration of the sewage treatment plant in the activated sludge process: the first model is the Adaptive Control Neural Networks (ACNN) and the second model is the Adaptive Control Neural Networks PI (ACNN-PI). In order to use the BMS1 (Benchmark Simulation Model Nº 1) for the evaluation of the technical performance of these controls applied to wastewater treatment plants (WWTP). The control of wastewater treatment plants is not trivial because of the large disturbances of the influent, non-linearities, delays and interactions between variables and operating conditions. The predictive controllers of nonlinear models based on neural networks (NN) are designed to model unknown nonlinearities of wastewater treatment plants with high predictive performance. The control results obtained under disturbances in dry weather are satisfactory when our models are well applied and provide a very useful tool which can be used by WWTP operators in their daily management. Especially when you mix the ACNN nonlinear predictive control and the classic PI control. As well as the optimal control of oxygen and nitrates, it is possible to benefit from important properties in terms of process performance and energy costs such as energy consumption (EC) and effluent quality (EQ). The system adopted in this work offers significant reliable and economic and environmental benefits, depending on the performance criteria sought.