Duke and other utilities are using an Echelon Data Concentrator (or
similar devices made by others but still using LonTalk). I think there
is one per transformer and they act as a router with the ability to
query the meters, shed loads, etc. and send usage data to the central
office over the powerlines. They may not have implemented all of the
features initially. In the Cincinnati area numerous X10 users have
seen random turn-ons which appear to be caused by very high amplitude
86kHz signals from the data concentrators. Duke has told one person
that Echelon is working on a filter that they will try at his
residence when received.
http://www.echelon.com/metering/nes_dc.htm
Jeff Volp has found a combination of clamp-on ferrites plus a filter
across the two hot legs that also seems to filter the signals.
I suspect it's the amplitude (which rises in response to "noise") of
the signals that is causing the problem, acting like spikes which X10
has acknowledged causes similar issues. In two cases, people
experiencing the problem found that Smarthome made X-10 modules set to
the same address appeared to be immune. This was, however, a very
small sample size so YMMV.
This may or may not be the cause of your issue but I think it's
worthwhile for all X10 users to be aware of it with the nationwide
rollout of smart meters.
I live across the river from Cincinnati and my area was an early test
area for Duke about 5 years ago. I saw no X-10 issues (most of my gear
is not X10-made) but did have some inexpensive appliances with
electronic controls go berserk.