You are currently viewing a new version of our website. To view the old version click .
Applied Sciences
  • This is an early access version, the complete PDF, HTML, and XML versions will be available soon.
  • Article
  • Open Access

26 December 2025

Fuzzy Model-Based Predictive Control Applied to Wastewater Treatment Plants Represented by the BSM1 Benchmark

and
Supervision and Process Control Research Group, Computing and Automatic Department, University of Salamanca, 37008 Salamanca, Spain
*
Authors to whom correspondence should be addressed.
Appl. Sci.2026, 16(1), 272;https://doi.org/10.3390/app16010272 
(registering DOI)
This article belongs to the Special Issue AI in Wastewater Treatment

Abstract

The control of wastewater treatment plants (WWTPs) is an ecologically, economically, and socially important objective. In the case of plants using activated sludge (ASP) processes, their control presents a significant challenge due to the complexity of the dynamics of these processes (a consequence of their biological nature). To objectively evaluate control strategies, standardized benchmark simulation models (BSMs) are used. This article tests the feasibility and evaluates the performance, in a simulation environment, of a specific fuzzy model-based predictive control strategy, called FMBPC/CLP, applied to the BSM1 reference model. In each iteration, this strategy first uses an FMBPC-type algorithm, which determines the basic control action (based on a fuzzy model and applying functional predictive control) that guarantees the local stability of the closed-loop system. Then, a second predictive control algorithm, called closed-loop predictive control (CLP-MPC), calculates a compensating term that is added to the basic control law and ensures compliance with constraints in the control action. In the simulation experiments carried out, the plant structure described in the BSM1 benchmark (reactor divided into five tanks, followed by a settling tank) was maintained, but the default control configuration was modified. The alternative control configuration designed for the BSM1 test bench includes two control loops: one to regulate the oxygen concentration in compartment 5 of the reactor (maintaining the PI algorithm of the default control configuration) and another loop to regulate the nitrate concentration (nitrate and nitrite) in tank 2 and, simultaneously, the ammonia concentration in tank 5, using the alternative FMBPC/CLP strategy. This control hybrid configuration was tested and evaluated considering values of the influent (dry, rainy, and stormy weather), and performance measurement criteria, both standardized in the BSM1 platform. The base model of the plant to be controlled, necessary for the FMBPC strategy, is obtained by prior fuzzy identification, from open-loop input and output data. The identification is achieved with the help of a software tool that uses mathematical clustering methods (based on the Gustafson–Kessel algorithm) that allow for the extraction of fuzzy models of the Takagi–Sugeno type from the numerical input–output data of a given plant. The FMBPC strategy is potentially appropriate for the control of complex, changing or unknown systems and this article demonstrates that this strategy is viable, with satisfactory performance, and that it can even be competitive when compared with more traditional control strategies.

Article Metrics

Citations

Article Access Statistics

Article metric data becomes available approximately 24 hours after publication online.