Constrained model predictive control for effective tonal noise attenuation in acoustic ducts

Abstract Active noise control (ANC) effectively reduces tonal noise in industrial ducts, particularly at low frequencies, which is crucial for maintaining a safe and compliant working environment. However, traditional algorithms, such as filtered-x least mean squares (FXLMS), struggle to handle well-known operational constraints, including actuator saturations. Conversely, model predictive control (MPC) effectively manages these constraints but encounters computational challenges, particularly in fast dynamic systems like ANC. This study presents an innovative MPC framework that explicitly addresses operational constraints while ensuring effective noise attenuation. Unlike conventional MPC applications, which treat primary noise as a disturbance in ANC, our framework utilizes primary path modeling to estimate future reference values, thereby enhancing control performance in this fast dynamic system. The methodology is validated through numerical experiments, utilizing identified primary and secondary path models in an active noise control test setup. Results demonstrate the proposed MPC framework’s practical feasibility and significant performance advantages over traditional FXLMS methods, including improved noise attenuation and robust handling of operational constraints.