hysteresis code. for example for thermostat

[hater 43]

subj

(c) 

Please login or register to see this link.

 

Please login or register to see this code.

 

Edited October 23, 2021 by hater

[hater 43]

Please login or register to see this code.

 

Please login or register to see this attachment.

Edited October 23, 2021 by hater

[hater 43]

Please login or register to see this code.

 

 

Please login or register to see this attachment.

Edited October 23, 2021 by hater

[hater 43]

        -- don't use the current reading but average it out over
        -- the past <SMOOTH_FACTOR> readings (data smoothing) to get rid of noise, wrong readings etc
 

add some useful functions

Please login or register to see this code.

 

and add code improvements

 

Please login or register to see this code.

 

[jgab 2,105]

Thanks, this is very useful code and nice to have in the library.

 

Another similar piece is to calculate a sliding average, i.e . give me the average of the last 10 values. It gives a smoothed value instead of a jumpy value , like power measurements usually are.

Normally we do this by saving the n last values and return the average of this - the problem is that we need a buffer and recalculate the sum.

There is a neat trick without a buffer of values that is not exactly a sliding average but a good approximation.

What we do is that we deduce the average from the current sum before adding the new value. (we only need to keep track of n and sum)

 

Please login or register to see this code.

 

Edited October 23, 2021 by jgab

[hater 43]

10 hours ago, jgab said:

Thanks, this is very useful code and nice to have in the library.

Jan, and thank you, as well, it's your school  

[hater 43]

Jan, btw. another minor problem.

smoothFilter like your or my monster ? smoothFilter do not guaranty with satiation below

 

10:20:00 15*C

[no data from sensor]

11:00:00 11*C

 

in this case we should ignore to use smoothFilter cuz accumulating data with different time is tricky, IMHO 

[hater 43]

Please login or register to see this code.

able to reset, @jgab

[hater 43]

Sample simple QA for heating

Please login or register to see this attachment.

[jgab 2,105]

2 hours ago, hater said:

Please login or register to see this code.

able to reset, @jgab

 

Alternatively

Please login or register to see this code.

 

[petrkl12 214]

2 hours ago, jgab said:

 

Alternatively

Please login or register to see this code.

 

Nice function

It could be perfect to have similiar function for moving average (where is also time how long values inluence final number)

[hater 43]

2 hours ago, petrkl12 said:

moving average

Please login or register to see this code.

Please login or register to see this attachment.

[jgab 2,105]

On 10/24/2021 at 2:22 PM, petrkl12 said:

Nice function

It could be perfect to have similiar function for moving average (where is also time how long values inluence final number)

Yes, but to use the same trick (approximation without buffer) it would be easiest if the value was sampled at regular intervals.

Like this:

Please login or register to see this code.

Otherwise a buffer is needed and re-summing for every average calculation...

Edited October 25, 2021 by jgab

[hater 43]

10 hours ago, jgab said:

Please login or register to see this code.

 

10 hours ago, jgab said:

Please login or register to see this code.

warning.

setInterval in ms NOT min.

so pay attention mul on 1000

[hater 43]

and then some.

IMHO

1) do not need to push the same value to function. only if value a new one according to stamp from device

2) we should use stamp from function fibaro.get