Setpoint control for reacting to wastewater influent in BSM1

This paper proposes a method for designing and tracking an optimal set point for a biological wastewater treatment, where the set point changes in real time in order to respond to changing influent disturbance. The objectives are to minimize energy consumption even while meeting or exceeding effluent quality standards, even during extreme weather events. The proposed method trains neural networks to estimate NARX models of the system. A nonlinear optimization then predicts an optimal set point, which is used as a search direction for finding the true optimal set point. The BSM1 simulation model provides a benchmark for testing the design. For tracking the set point, the standard PI controls found with BSM1 are replaced by adaptive controls, with an additional feedback loop not found in the original BSM1. Simulation results show that the proposed method improves both effluent quality and reduces energy consumption.