Initially, I could not see any clear pattern in it, so I called it "irregular". Next, Alain Figer from France pointed out it was not so irregular after all. His own images from August 16th (URL's here) show a very neat pattern of one major flash, then two secondary flashes, then a major flash again. On his images, the major flashes are 0.73 seconds apart.
Going back to my observations, it turns out it is indeed possible to find a similar periodicity of 0.73 +/- 0.03 seconds. The secondary flashes appear to be 0.73/3 = ~0.24 from these main peaks: in other words, main and secondary peaks fit a ~0.24s pattern.
Below diagram shows it. the lowest line gives the observed peaks, with the red triangles representing the main 0.73s cycle. The yellow tringles point out several secondary maxima at ~0.24s after the main cycle peak; the grey/white triangles point at a hint of a similar pattern ~0.24s before he main cycle peaks.
The upper line shows the modelled behaviour: a main peak (black triangle) each 0.73s, with secondary and tertiary peaks at ~0.24s intervals.