Model order reduction of dynamical systems: An approach to investigate real and complex poles

摘要

The process of model order reduction facilitates analysis and realization of complex systems while marinating the essential properties of the system and input/output behavior. This research proposes a novel method for model order reduction. The novelty of the proposed method lies in addressing the limitations of dimensionless representation and pole-zero dominance. Both real and complex poles are treated for order reduction. The reduction of dimensionless systems, which considers real poles only, in the field of controls has not been considerably investigated. Even MATLAB has not completely exploited it. This work also investigates the augmentation of dimensionless systems and singular values based order reduction technique. For systems with complex poles, pole-to-pole and pole-to-zero ratios are considered for the reduction. The performance of the proposed method is demonstrated with simulated examples of multiple systems. The results indicate that the error between the unit step responses of the reduced and the high order model is less than 5%. This research may find potential in various applications involving model order reduction including but not limited to; circuit simulation, micro-electromechanical systems, avionics and structural dynamics.