SamC and ConorML have designed some sensor correction filters using some matlab swarm optimization routines. Their designs are summarized in SEI log 1234 (yes, really) and a technical note in the DCC (https://dcc.ligo.org/DocDB/0147/T1700541/001/T1700541-V1.pdf ). I've installed these to test on ETMY and I'm leaving them running overnight. Microseism is pretty high right now so it's a good time to test. I've already collected some data, but the differences aren't dramatic, but I wasn't comparing very long stretches, I found a long window from last January with high microseism, that I'll try to use for comparison tomorrow.

First attached plot is the St1 Y motion measured by the T240 (thick lines), their RMS (dashed) and the ground motions (thin), red traces are for the swarm filters, blue are for the broadband filter we used during O2. Second plot shows the comparison of the CPS (SCSUM_CPS, thick lines), the cps RMS (dashed) and ground again (thin lines), again red traces are for the swarm filters, blue are for the O2 configuration. Last attached plot shows the two filter designs in displacement, and their complements, stolen from the DCC doc. Blends were not changed during these measurements. 

Sam says that the filters were design when .1-.3 hz gnd blrms was around 450 nm/s. I've put ETMY back to damped to give them some time with higher microseism to design with.