Measurement driven models of nonlinear electronic components

A modeling method that allows one to rapidly build data driven models for nonlinear components is discussed. The models are constructed from input/output time domain data and their "embeddings". The notion of models built from embeded data is described in the Taken's Embedding Theorem and has been extensively explored for modeling autonomous systems in the physics liteature. The authors extend these results to nonautonomous systems by creating tools that allow engineers to rapidly build models for driven nonlinear components. These models can be used in simulation, process control, diagonostics, and sensor calibration. We present the results of applying nonlinear modelling techniques to the modelling of measurements taken from a Nonlinear Network Measurement Systems (NNMS). A comparison is made between reconstructed polynomial models and radial basis models.