dyno chart smoothing

Hey guys im new to this forum site i recently did a pull and logged it to try yo get some numbers on the virtual dyno now with smoothing zero it just seemed like the numbers didnt add up and after putting in smoothing 1 the hp and tq numbers seemed more legit just really curious as to what soothing really does and what does it mean? Any feed back is appreciated thanks

Virtual dyno calculates power by measuring rpm rate and calculating the power required to accelerate through that gear by taking into account gearing and rolling diameter (to work out speed change) and aerodynamic drag. The thing here is the calculations rely on time signature resolution, tacho accuracy/consistency, log data frequency etc. Those numbers don't all necessarily to be the PRECISE rpm, time offset etc at each reference point - there is pretty much always a little bit of error before you even take into account any wheel slip, road surface noise etc so if you saw the raw data measured by scatter graph it'd all look a little erratic depending on how reliable the data you have been provided with is.

What Virtual Dyno attempts to do is effectively create a trend graph out of all the points made, you use the smoothing to fine tune out all the "noise" resulting from less than perfectly reliable samples from the original data. The better the original data the less smoothing will be required. The more smoothing applied the less precise the final curve will be, though if you can't make any sense of it with no smoothing then that isn't really a problem. Smoothing will pull out dips and peaks in the power curve, so your peak power will tend to be a little lower with more smoothing though also look "nicer". On the flipside, if there is messy reference points in the original data you could end up with an artificially high power reading too.

It's probably worth mentioning that ACTUAL dynos need to do this, too. Actually a LOT of things which electronically collect and represent data as curves tend to do a certain amount of this kind of processing to make things easier to make sense of a trend.

Google for images on "Moving linear regression curve" if you want to see some examples of this kind of thing in effect.

Ok that makes perfect sensethanks alot lith