| Summary: | In this thesis, computer methods for the transient analysis
of networks are investigated. Numerical transform techniques are
developed to solve the differential equations a rising in network simulation. Extensions to permit inclusion of some nonlinear elements are
considered. Efficient methods for implementing the techniques are
developed.
For the transform techniques, error estimates are derived.
Using these estimates, algorithms for the automatic determination of
solution parameters are developed. Advantages over other numerical
transform and numerical integration techniques are revealed.
For nonlinear networks, it is shown that use of a Newton-Raphson scheme for solving nonlinear algebraic equations is difficult
when coupled with transform methods for solving differential equations.
Instead, an alternative technique is developed. Steps which are easily
generated, but which only approximate Newton-Raphson steps, are used.
The implementation of the transform techniques and the
nonlinear solution is considered. A program using a sparse tableau
form of network equations is discussed. The program is in two sections.
The first reads in the network descriptions, and writes a series of
Fortran subroutines for performing the analysis efficiently. The subroutines
must be compiled, and are used by the second part of the program
to perform the actual analysis.
Examples which illustrate the performance of the various
techniques are presented. === Applied Science, Faculty of === Electrical and Computer Engineering, Department of === Graduate
|
|---|
