Robust Controller Development Via Iterative Model Updating for Active Magnetic Bearing Rotor Systems

Abstract Modern turbomachinery equipped with active magnetic bearings (AMBs) requires increasingly high-performance controllers to ensure safe and efficient operations. These controllers are typically tuned based on frequency responses, or in more advanced cases based on the plant models. However, effects such as interference fits and shrink fits can significantly impact rotor dynamics, reducing the accuracy and reliability of these models. This study proposes a novel model updating method to deal with problems related to practical applications of model-based controllers. First, this paper presents an iterative model updating method based on an optimization algorithm to adjust the finite element (FE) model of complex rotors used in oil and gas applications, exploiting frequency measurements from the actual rotor. Subsequently, the proposed iterative updating method is utilized to develop a set of rotor models that differ only for some designed rotor modal shapes that are not well identified by the rotor frequency measurements. This set of rotor models is then exploited by a nonsmooth optimization algorithm to synthesize a robust controller—an augmented PID—optimized to ensure a specified level of performance across all scenarios described by the developed set of rotor models, thereby guaranteeing safe operations over time.