Page 2 of 2
Re: Cumulus behaviou when sensor contact lost
Posted: Fri 10 Jan 2014 8:09 pm
by BCJKiwi
OK, I'll just keep testing and see how it goes.
When I'm happy that it does not generate false 'sensor lost' flags, then I'll consider making it available.
Re: Cumulus behaviour when sensor contact lost
Posted: Thu 30 Jan 2014 1:50 am
by BCJKiwi
Update.
After a lot of fiddling and changes I have been running an algorithm which stores the last 5 <#temp> <#hum> <#dew> <#rmidnight> <#wlatest> <#bearing> values at every interval which is 10 mins in the test setup.
The routine has been comparing the current sum of these variables with the sum from 30 minutes before. If the sum is the same then the sensorlost var is set to 1.
The logs show that in the last 19 days with data logged every 10 minutes, there have been 14 sensor lost indications.
The primary issue is the obvious one - when it is calm there is not enough variation in the sensor returned data to always return a difference in the sum.
There are a few ways the tests could be changed;
1. compare each of the values individually rather than the sum
2. compare more than one of the current and stored sets.
I feel it is unlikely to be a reliable technique in the long term.
However, since I have all this working, I will try adding some different comparison techniques and see what the outcome is.