Interesting, this happened to me a few times as well and would be nice to understand why. BM said at the time it could be due to the delegate behind or in front of me having a high latency and thus ignoring my block, so guess that could be why in your case.
I looked up the blocks where your fork happened on my site (545580+1 and 549231+1), and I've got them down as signed by "dpos.crazybit" and "cny.bts500" with a latency of 0s. It looks like you tried to produce that first block 6 minutes after and the second 7 minutes after it was produced, which I guess is why you created forks.
Maybe the latency spike was actually on your end, or your system clock was unsynchronized?