- Determine optimal downsampling rate for dipole measurement data [priority:: low] [completion:: 2024-07-29]
Static downsampling rate
THe downsampling rate determines the
Adaptive downsampling rate
For Adaptive downsampling rate, we could base amount of downsampling based on the second order derivative of the signal, which would significantly simplify determining optimal sequence length, since we would be able to include model and predict data only when features are changing.
- Investigate how to implement adaptive sampling rate based on rate of change [priority:: medium] [completion:: 2024-06-28]
Using the RDP algorithm on the Jiles-Atherton simulated data and , for we can reduce the effective number of points by 95 - 98 %, which is effectively downsampling by 20-50, while still retaining important large changes. Here 2.83 % of all points are kept.
For a 10 minute dataset at 1kHz, equivalent to 600 000 points, this is takes 110 ms. For 4 hours at 1kHz, equivalent to 14 000 000 points, this takes 7.6 s.
Applying RDP to vs does not seem to work well, however making a 3D fit with time, current and field and , which leaves 2.27 % of the points.