[OpenTRV-dev] Adaptive...

[Damon Hart-Davis]

Hi,

The collection (and dumping) of stats for temperature and ambient light is now running.

I have added some companding to the temperature to get decent resolution (0.125C) in the area of main interest [16C,24C[ and particularly to get meaningful smoothed values in 8 bits with roughly a weekly time-constant, while recording temperatures across the full gamut [0C,100C] for everything from frost monitoring to Bo's DHW!

All stats are stored in EEPROM and wear should not be an issue.

I shall leave it running to look for gotchas.

I was very pleased to have my 'unit test' mode to do test-driven-development for some of the tricker parts!

Rgds

Damon


On 26 May 2013, at 11:03, Damon Hart-Davis wrote:

> Hi,
> 
> Though I should be doing other things, over the next week or so I may start building into my Arduino TRV implementation the vestiges of an adaptive/learning system, by at least, for example, recording samples temperatures and ambient lighting levels for the last 24h, with a set smoothed over (more than) a weekly cycle, which may be interesting to the MSc student(s) and other researchers I have spoken to.  (Eg to gather research data and to help deduce how much of the heating effort may be wasted on unoccupied rooms.)
> 
> I may also try to set it up (though it will be difficult to test in anger before winter) to note how many times warm/bake or frost modes were manually engaged below / above certain temperature thresholds or in certain time buckets as a way of adjusting the thresholds or pre-empting the need for manual intervention.
> 
> I have yet to find a suitable PIR occupancy sensor, which would help a lot, or to do any work on using a microphone to detect voices, etc.
> 
> Rgds
> 
> Damon