Online Parameter Estimation in Digital Twins for Real-Time Condition Monitoring

This paper introduces innovative online parameter estimation algorithms that employ both deterministic and stochastic methodologies in digital twins for real-time condition monitoring. The deterministic approach utilizes an exponential forgetting factor adaptive observer, while the stochastic approach involves an adaptive Kalman filter. In contrast to conventional methods, these online algorithms demonstrate robustness against variations in initial conditions and measurement noise. Notably, the algorithms exhibit the capability to manage multiple parameters and directly estimate them from sensor measurements. The effectiveness of the proposed algorithms is demonstrated through experiments focused on parameter estimation in DC motors and marine surface vessels. The results highlight the algorithms’ accuracy in estimating parameters under diverse conditions. This research contributes to the advancement of online parameter estimation techniques for condition monitoring, showcasing their applicability and reliability in real-world scenarios involving complex systems.