Networked assembly of nonlinear physical system models

Engineering design is evolving into a global strategy that distributes model information through computer networks. This strategy requires companies to provide dynamic models of supplied physical components. Component models are transmitted through the Internet to a common location and then assembled to obtain a product dynamic model. Internet connection permitting, real-time, automated assembly of models requires four characteristics. Specifically, physical models must have a unique standard format, the exchange of model information must be executed in a single-query transmission, the models must describe only external behavior, and the assembly process must be recursive. The Modular Modeling Method (MMM) is an energy based model distribution and assembly algorithm that satisfies these four requirements. The MMM distributes and assembles linear and affine physical systems models using dynamic matrices. Though the MMM procedure can be used for a large class of systems, the dynamic matrices cannot be used to represent nonlinear behavior. A more general nonlinear model representation is required. This work is an extension of the MMM algorithm to assemble physical systems models characterized by analytic nonlinearities. This is a more general procedure that uses Volterra transfer functions to represent nonlinear behavior. Any analytic nonlinear system can be represented through a Volterra model. The reason why we use Volterra models instead ODEs is because Volterra models are only in function of input and output variables. This characteristic facilitates their use in an energy based model assembly method such as the MMM procedure. A procedure to assemble standard Volterra models using conservation energy principle is described. Even though there are extensive literature about gluing models, these techniques do not have all the characteristics needed by the MMM procedure. Using the approach proposed here, complex model assemblies can be executed recursively while hiding the topology and characteristics of their structural model subassemblies.