[TowerTalk] Particle Swarm Optimization of LPDA'S

[Jim Jarvis]

Just when you thought you understood antenna modelling....

The August edition of the IEEE Antenna & Propagation Magazine arrived
yesterday, having obviously travelled long-path to get here.  It contains an
article by a group at the University of Grenada, Spain, which uses a design
based on genetic algorithms, called Particle Swarm Optimization.

Their statistical approach, coupled with NEC code, produced a 10 element
LPDA which has the f/b, fwd gain, and vswr bandwidth of a 20 element lpda
using existing optimization.

Over a range of 1.5 octaves, the antenna produced average gain of 8.2dB,
20dB f/b, and a mean vswr of 1.5:1.  It took 300 iterations of the
optimization algorithm to get there.

Their model operates from 450MHz to 1350MHz.  Scaled to HF, that would be
14.0 to 35MHz.  Or 10 to 25MHz.  One might come up with something useful for
30-10m, sacrificing some parameter at the edges.

I assume the gain figures are dBi, not dBd, but the article doesn't specify.
This still puts a 10 element lpda in the same category as a 3 element yagi,
rather than the 2 element yagi that it had approximated before.

They got the performance out of half the materials that it would have
otherwise taken.  Any math-jocks out there feel like exploring multiband
yagis?  

n2ea