Linearization algorithms for computer-aided control engineering

Numerical algorithms and heuristic logic for the accurate and robust linearization of nonlinear dynamic system models are given. The numerical algorithms deal with cases in which the system nonlinearities are differentiable (possess a Taylor series expansion), and the logic handles a variety of anomalous situations. New conventional methods and algorithms for linearization, as well as an expert system to aid the control engineer in determining linearized models for nonlinear systems, have been obtained. Included are an approach that minimizes the effects of truncation and round-off errors incurred through numerical differentiation and techniques for accurately identifying certain discontinuities in the mathematical description of a nonlinear system, as well as other problems that make linearization difficult or meaningless.<>