A graph cut strategy for transmission path problems in statistical energy analysis

In this paper, it is shown how a strategy based on algorithms for computing cuts in undirected networks, can be applied to reduce energy transmission in realistic statistical energy analysis (SEA) models, with the sole modification of a limited number of internal and coupling loss factors. The frequency dependent case of SEA systems with multiple sources and targets is considered. A mathematical justification for the strategy is also provided, which relies on an analysis of the series expansion of the energy vector, in terms of the powers of the SEA graph adjacency matrix. A numerical example of vibroacoustic transmission in a simple test building has been included to show the performance of the approach.

► Several energy transmission path problems in SEA can be solved in the framework of graph theory. ► We aim at reducing the energy at a set of target subsystems due to external energy input at sources. ► An algorithm to compute graph cutsets is used to separate sources from targets. ► The loss factors of all subsystems involved in the cutsets become modified to achieve the expected energy reduction. ► A realistic example of vibroacoustic transmission in the SEA model of a building is included.