Application of evolutionary computation techniques to nonlinear microwave circuit analysis

Abstract

The application of genetic algorithms to the simulation of nonlinear microwave one-port circuits using hybrid techniques is reviewed. To date, harmonic balance and sample balance have been the two major types of hybrid technique, and genetic algorithms have been successfully used to implement both methods. However, genetic algorithms are "transparent" to the solution domain, which allows new classes of hybrid technique to be developed: these are termed frequency-domain sample balance (FDSB) and time-domain harmonic balance (TDHB). FDSB, in particular, has useful properties for circuits with noncommensurate multitone inputs. By formulating the error function in the time domain whilst representing the solution in the frequency domain, it is possible to circumvent the use of Fourier transforms altogether. This allows, in principle, any arbitrary steady-state input to be applied.

A genetic algorithm for generating RF circuit models from calibrated broadband measurements

“…Genetic algorithms are a biologically inspired, iterative approach to complicated optimization problems [8]. Genetic algorithms have been widely used to design and model RF circuits [9], [10]. Two-port error boxes have been determined from a broadband measurement of a thru by use of a genetic algorithm [11].…”