The New Algorithm for Symbolic Network Analysis.
Abstract
A new and highly efficient tree identification algorithm is derived here for obtaining the determinant and the cofactors of a circuit's node admittance matrix, and hence, for obtaining various symbolic network functions for oneport and twoport reciprocal and nonreciprocal networks, with the network's topological description as its input. The algorithm is so devised that it is practically memorystorage free, and it is simple enough that even a microcomputer can obtain symbolic network functions for a fairly large circuit in a reasonably short time. It is worth noting that the algorithm can handle topological branches with infinite admittance values. Making use of this special feature, we have derived a simple topological model for the ideal operational amplifier, hence providing the ability to obtain various topological formulas of operational amplifier circuits in a reasonable time. By choosing appropriate symbolic network functions, along with some measured transfer function data, the circuit's nominal element values, and a nonlinearequation solving subroutine, we have constructed a computer program to perform analog circuit fault diagnosis. This program can identify which of a circuit's elements are faulty or out of design tolerances. In the course of this research we have also identified an application to a biological problem, one in which the resistor values of an electrical model of the guineapig cochlea can easily be deduced even when some nodes are inaccessible for measurements. All these features have been implemented on a very modest microcomputer, the Apple II. Obviously, a larger computer will not only accomplish the same result faster but also it will be capable of analyzing much larger circuits.
 Publication:

Ph.D. Thesis
 Pub Date:
 1983
 Bibcode:
 1983PhDT.......135C
 Keywords:

 Physics: Electricity and Magnetism