Development of a cyber-physical experimental platform for real-time dynamic model updating

Model updating procedures are traditionally performed off-line. With the significant recent advances in embedded systems and the related real-time computing capabilities, online or real-time, model updating can be performed to inform decision making and controller actions. The applications for real-time model updating are mainly in the areas of (i) condition diagnosis and prognosis of engineering systems; and (ii) control systems that benefit from accurate modeling of the system plant. Herein, the development of a cyber-physical real-time model updating experimental platform, including real-time computing environment, model updating algorithm, hardware architecture and testbed, is described. Results from two challenging experimental implementations are presented to illustrate the performance of this cyber-physical platform in achieving the goal of updating nonlinear systems in real-time. The experiments consider typical nonlinear engineering systems that exhibit hysteresis. Among the available algorithms capable of identification of such complex nonlinearities, the unscented Kalman filter (UKF) is selected for these experiments as an effective method to update nonlinear dynamic system models under realistic conditions. The implementation of the platform is discussed for successful completion of these experiments, including required timing constraints and overall evaluation of the system.

► A cyber-physical real-time model updating experimental platform is developed. ► The proposed platform enables updating nonlinear hysteretic systems in real-time. ► Unscented Kalman filter (UKF) is selected for updating nonlinear hysteretic systems. ► Implementation aspects related to “hard” real-time computing are discussed. ► Results from two challenging experiments using the proposed platform are presented.