The occurrence of cyber and physical disturbances in power systems is increasing, resulting in public attention on cyber-physical architectures. It is observed that disturbances can propagate between cyber and physical systems, emphasizing the need to study interdependencies. In this work, we present an approach toward improving cyber-physical interdependency characterization through modeling techniques. The improved assessment of these dependencies can then aid system design optimization to improve functional resilience. To achieve this, we transform the cyber-physical architecture to a graph and apply a bio-inspired network analysis using bipartite network methods to characterize the system during the disturbances. Moreover, a DeepWalk method is applied to cluster the component based on their interdependencies. In this paper, a WSCC-9 bus system is used for numerical study and quantification.