A tool-box approach to hybrid symbolic-numerical modeling

Modeling of the non-linear dynamic behavior of technical systems is not yet sufficiently supported by computer programs. Encouraging approaches to behavioral modeling are symbolic analysis and approximation, numerical order reduction for systems of linear differential equations, and simulation-based parameter identification. Non-linear order reduction is a topic of research but these algorithms are not yet used in industrial simulation environments. An approach is presented which combines some of these powerful methods. It is based on partitioning the technical system into linear and non-linear parts, application of symbolic approximation to the non-linear parts, numerical order reduction to the linear parts, and combining the various partial models.