Structural simplification of modular bond-graph models based on junction inactivity

Abstract

The modular modeling paradigm facilitates the efficient building, verification and handling of complex system models by assembling them from general-purpose component models. A drawback of this paradigm, however, is that the assembled system models may have excessively complex structures for certain purposes due to the amount of detail in the component models that has been introduced to promote modularity. For example, a multibody system can be modeled using generic rigid-body models with 6 degrees-of-freedom (DoF) to represent the components of the system, but then constraints have to be added to the model to match the DoF of the system. This work presents a structural simplification technique that can detect such unnecessary complexities in a modular bond-graph system model and eliminate them without compromising accuracy. To this end, the activity concept in the literature is extended to define “inactivity” for junction elements, and simplification is obtained by detecting and eliminating inactive junction elements and by propagating the implications. It is shown by example that this simple idea can result in models that are conceptually and computationally more efficient than the original modular models. The realization-preserving and input-dependent characteristics of this approach are highlighted.