Nonlinear System Analysis and Identification from Random Data

Describes procedures to identify and analyze the properties of many types of nonlinear systems from random data measured at the input and output points of physical systems. Improvements are offered in applying older techniques, and problems that traditionally have been difficult to analyze are solved by new, simpler procedures. Formulas are stated for optimum nonlinear system identification in both general models consisting of parallel, linear bilinear and trilinear systems, and special models consisting of parallel linear, finite-memory square-law systems and finite-memory cubic systems. New results, obtained here, show when and how to replace complicated single input/output nonlinear models with simpler alternative multiple input/single output linear models. New error analysis formulas are presented to design experiments and to evaluate estimates obtained from measured data. Includes many illustrative examples.