Cephalotus said:
justin_le said:
....
It'll take a lot of additional rework of this trip analyzer web app to present multi-day log files in a way that is easier to visualize since the graph data quickly gets quite noisy, but that's something we'll be working on over the summer so that by the time the race is done we'll be able to show and compare all the different solar ebike performances with this tool.
This would be fantastic!
I'm very interested in the performance of the vehicles and the solar arrays and sadly so very little data is available to the public from the first suntrips.
Well guys, long time since I've piped in here, but the 100 day time limit for the Suntrip race was reached a short while ago and the last of the suntrip riders have just crossed their way to the finish line in Guangzhou, certainly a little battered and tattered but amazing to see!
https://www.thesuntrip.com/francoise-denel-arrivee-suntrip2018/
At last we've started getting the log files from the other suntrip treams and as promised we made a huge number of changes to our trip analyzer software in order to deal with these super large files. For the data junkies, here is the complete data for our trip:
http://www.ebikes.ca/downloads/SuntripLogs/Justin_AnSo.zip
If you unzip this, you'll see there is a folder that has the logged data for each trip in DailyLogs. This is the raw 1Hz log data that's been parsed and split into 24 hr sections (as set by UTC+3) so that it's neatly organized into one file for each day. There's another folder for downsampled data which is the same thing but where we've decimated the data to either 0.1 or 0.2 Hz for smaller file size. And finally, there is a daily summary .csv file that has the statistics (total distance, average speed, wh/km, solar watt-hours etc) from each day.
View attachment 5
Each log file starts with a header that shows the columns. The first column,
UTC, is now a universal unix date/time stamp, the 12 next columns are all the standard things you'd have from a CA3 device (amphours, volts, amps, speed, distance, temperature, rpm etc). Since these databoxes were stand alone devices and not the primary CA, there weren't any PAS, torque, temperature, or throttle sensors attached so much of this data is zero. And the final 7 columns are all new, containing the solar amps and Ah data, the GPS data (heading, elevation, latitude, and longitude) and finally flags that indicate the state of the GPS sensor.
Here's where things get fun. If you have the trip data downloaded on your computer, you can now navigate to the
trip analyzer page here, click to upload CA data, and then control select ANY NUMBER of files that you want to look at and it will concatenate the data and show it in a single plot. The trip analyzer can handle upwards of 100 MB of logged data this way before causing the browser to crash. With the x5 downsampling, it takes a while to load but you can view our entire trip from Lyon to Tehran in a single graph.
Right now after uploading and processing it switches by default to showing a time axis view rather than distance axis, which makes a bit more sense in the context of solar watts. The default viewing shows the solar data in the extended data set, with the watts in pink and the accumulated solar watt-hours in green.
You can now see each daily sun cycle that we received and how much accumulated watt-hours we got from the rise of the green line. If you look closely at the graph, it contains both a dark line as well as a transparent shaded range. Since each pixel represents hundreds or thousands of data points, we have it so that the solid pixel is the average value and then the range of data that is averaged to produce that point is visible in the transparency.
This both helps the graph display much faster, and it makes it a lot less cluttered to see meaningful trends in noisy/fluctuating data like the solar output. Now when you are zoomed in and scroll left and right, it initially previews in the downsampled data to be responsive and render quickly, and then once you release the mouse button the data for that section reloads at a high resolution.
If anyone has tried using the trip analyzer previously with large data files, you'll know how bogged down and unresponsive it would get. These recent updates are a huge improvement that make it properly usable.
So now with these tools are in place, I'll be devoting my next posts to looking at interesting insights that this data has shown us about the performance of our own system, much of which wasn't at all apparent while we were actually riding on the trip!