Structure-preserving model reduction
High-precision machines are often modelled as assemblies of flexible substructures, which preserves physical insight but leads to models that are too large for efficient dynamic analysis and control design.
This article compares two model order reduction methods on a two-axis gantry: state-of-the-art substructure-based modal truncation with residualisation, and interconnected second-order balanced
reduction (iSOBR), which accounts for the selected assembly input-output behaviour. Results presented in this Mikroniek article show that iSOBR achieves similar input-output accuracy with 3.4 times fewer states and helps identify which parts of the machine are most relevant to the performance of interest. (Image courtesy of Luuk Poort)
