An intelligent system approach for industrial data in prescriptive decision-making: wastewater treatment plant as a case study

Abstract

The exponential growth of today's industry demands innovation in the control and data analytics systems in charge of monitoring these operations to improve efficiency. The research conducting in this thesis presents a hybrid intelligent system to prescribe pH and dissolved oxygen setpoints to reduce the chemical oxygen demand (COD) in an industrial wastewater treatment plant. The intelligent system comprises prediction models, estimators, desirability functions alongside composite desirability, and a genetic algorithm (GA) for optimization. As a novelty, this research focuses on including more than one variable to manipulate for optimization and mixing machine learning and GA techniques. The hybrid intelligent system optimizes the COD in two tanks (equalizer tank and discharge pit) by 21.71% and 4.79% based on an optimization rate defined in this work, respectively.