# Algorithm for constant signal determination

• 11-29-2010
anthei
Algorithm for constant signal determination
Hi all,

I am looking for an algorithm which assesses the stability of a noisy signal over time. To be more specific letīs say a program sets a new setpoint in a temperature controller and the algorithm inside the program now should decide if the setpoint has been reached and - more important - if the temperature can be considered as constant.

Couldn't find anything useful on the internet.
Of course I have some ideas how to tackle this, like linear regression in a moving window and slope checking and so on, but I assume since this is a fairly basic and general problem, there should be many tried-and-tested solutions for this.

Some useful keywords would be fine to have a starting point for further search in the internet.

• 11-29-2010
Shakti
What about using some statistical analysis? Something along the lines "if standard deviation is lower than some_constant for some time t then temperature is considered to be constant at mean <mean_of_all_measured_temps>"
• 11-30-2010
anthei
Yes, thatīs what we will try if we have to invent an algorithm ourselves. Linear regression inside a user-defined time window and evaluating the slope and its standard error. But since it's such an elementary problem there have to be plenty of algorithms out there.
We have to implement the algorithm in LabView, which comes with tons of predefined functions but so far we couldn't find anything useful for our purpose.

Best regards
• 11-30-2010
Perspective
If you want to find more literature on the subject, search for "stream query processing" and "stream data mining".
• 12-02-2010
anthei
Great. Thanks