MrSkinny wrote:
Nick B wrote:
PowerCal just takes your HR to estimate your power output ...
AFAICT: HRV, not "just HR".
out of idle curiosity, just three weeks ago I applied a two-part first order model to hr as a function of power in some garmin files and it was shocking how good the relationship was.
It was pretty much straight out of engineering controls class.
Say HRtotal = HRbaseline + HRoxy + HRheat
Total heart rate = some baseline number for heart rate, whatever your heart rate is sitting on your bike and not pedaling.
+ the heart rate required to flow oxygen to the working muscle cells.
+ the heart rate required to move blood into the surface cells to be cooled - I say due to heat but can be thought of as all the heart rate drift rolled into one.
Then do a differential mass or energy balance on the sections as needed
for example for HRoxy
tau1 * dHRoxy(t) / dt = [k1* power(t)] - HRoxy(t)
to make it discrete then
tau1 * [HRoxy(t) - HRoxy(t-1)] = k1 * power(t-1)] - HRoxy(t-1)
repeat with the HRheat and solve the constants and you get pretty good results. The one I have the r^2 goes from 16% with a simple linear fit to 68% with the first order kinetics.
so in this way you can eliminate different baseline heart rates from the system, the idea that some days heart rate is higher than others even before you start and also eliminate monotonic HR drift either up from heat or dehydration or down from having a previous swim effort.
And that doesn't take into account pure time delays in the system. I need to rework it with pure time delays. Though looking at things I don't think there is much ore of a completely pure time delay, just looking at other systems.
All that to say, with a first order plus dead time model tuned over a few dozen users, or a few hundred users, you can probably get relatively decent power estimates from hr alone.
Not sure how one would calibrate the power though.
In the end though, not sure why one would want this. To me if HR is good enough, it's good enough. Just use HR to train with rather than translating it back to power.
Running power meter, isn't that what the rpm insoles do?