Automatic Control System for Wastewater Collection and Treatment in MATLAB Using Classical Controllers and a Neural Network

Abstract

Urban wastewater systems are increasingly exposed to strong hydraulic and organic disturbances caused by rainfall events and daily load variations, while treatment plants must satisfy strict effluent quality and energy-efficiency requirements. This paper presents a MATLAB/Simulink-based automatic control framework that integrates wastewater collection and activated-sludge treatment into a unified control architecture. Local PI/PID controllers are applied at pump stations to regulate wet-well levels and prevent overflow, whereas a constraint-aware model predictive controller is implemented for dissolved oxygen regulation in the biological reactors. To address limited availability and delay of online quality measurements, a neural-network module is introduced as a soft sensor and short-horizon predictor for influent load and key process variables. The plant dynamics are modeled using ASM1 combined with a BSM1-style layout to ensure reproducibility and benchmark-oriented evaluation. Simulation results under dry-weather and storm conditions demonstrate improved disturbance handling, stable dissolved oxygen control, and enhanced operational flexibility. The proposed modular framework supports rapid controller tuning and coordinated supervision between collection and treatment subsystems, making it suitable for research, education, and practical control-system prototyping.