NUST Institutions Library Catalogue

This thesis introduces a novel methodology for optimal interface expansion in dynamic
substructuring, focusing on inaccessible and continuous interfaces. Unlike traditional
approaches relying on modal parameters from Frequency Response Functions (FRFs), it
employs a direct frequency-based method targeting interface Degrees of Freedom (DoFs),
bypassing errors linked to modal identification. The System Equivalent Model Mixing
(SEMM) technique is utilized to expand dynamics at the interface by integrating numerical
and experimental models into a mixed model. Coherence is employed as a robust
correlation metric, evaluating both phase and magnitude of FRFs. This research addresses
optimal sensor placement (OSP) for effective expansion, testing nineteen stochastic
metaheuristics categorized into swarm intelligence, surrogate algorithms, and physicsinspired methods to alleviate computational challenges of exhaustive searches. The
Mountain Gazelle Optimizer (MGO) algorithm proved highly efficient for larger systems,
while exhaustive search was effective for smaller cases. Validation involved cantilevered
beam models and experimental setups, demonstrating strong correlation at interface DoFs.
MGO proved to be 49 times faster than exhaustive search in identifying optimal sensor
locations. The proposed methodology showcases practical applicability in achieving
accurate dynamic expansions.