I thought the intermittent splashing of the cooling water/steam in the close-up portion of this J2-X engine test video looked interesting.
So, I used mplayer to dump frames from the video (30 fps) to jpg files (about 1800 images), and the Python Image Library to crop to a rectangle focused on the splashes.
When a splash occurs the pixels in this region become much whiter, so the whiteness of the region should give an indication of the "splashiness". I then converted the images to black and white, and averaged the pixel values to get a scalar time-series. The whole time series is shown in the plot below.
Also, here's a text file if you want to play with the data.
Here's a PSD and autocorrelation for the section of the data excluding the start-up and shut-down transients.
Here's a recurrence plot of that section of the data.
This is a pretty short data set, but you can see that there are little "bursts" of periodic response in the recurrence plot (compare to some of the recurrence plots for Lorenz63 trajectories). I'm pretty sure this is not significant to engine development in any way, but I thought it was a neat source of time series data.
500 second-long test
ReplyDelete