CHAPTER 14
Nonlinear System Analysis
This chapter reviews some recommended techniques to identify the frequency domain properties of nonlinear systems from measured input/output random data. Procedures are discussed for the following five types of nonlinear systems and models.
1. Zero-memory and finite-memory nonlinear systems
2. Square-law and cubic nonlinear models
3. Volterra nonlinear models
4. Single-Input/Single-Output (SI/SO) models with parallel linear and nonlinear systems
5. SI/SO models with nonlinear feedback
Where appropriate, square-law and cubic nonlinear models are basic systems to apply. Formulas for Volterrra models involve multidimensional functions for Gaussian random data that are difficult to compute and interpret. Formulas for the SI/SO nonlinear models are valid for arbitrary random data and apply to broad classes of nonlinear operations. Direct multiple-input/single-output (MI/SO) linear techniques from Chapters 7 and 9 can be used to solve the SI/SO nonlinear models with parallel linear and nonlinear systems. Reverse MI/SO linear techniques, where input data and output data are interchanged, can be used to solve the SI/SO models with nonlinear feedback. Many examples and applications of these SI/SO models and techniques to solve nonlinear system problems are in Ref. 1 and Chapter 13 of Ref. 2.
14.1 ZERO-MEMORY AND FINITE-MEMORY NONLINEAR SYSTEMS
Two main properties distinguish non-linear systems from linear systems. First, nonlinear systems do not satisfy ...
