Dynamic Network Models

Abstract

The Dynamic Network Model (DNM) approach represents the physical model of dynamic systems in a network framework. The DNM approach has two main roots, namely, bond graph and Artificial Neural Networks (ANNs). This method employs the concepts of effort and flow to create the building blocks of network structure. The proposed DNM approach can be used in different applications, such as direct simulation, system identification, and design control systems. Other applications include analyzing interacting systems and modeling linear and nonlinear systems. The main advantages of DNM method include obtaining approximated difference equations, revealing possible drawbacks of backpropagation algorithm, and respecting Occam's razor principle.