Rana Adhikari - Research – Intelligent Controls

Intelligent Controls

The LIGO interferometers (and indeed all of the ground based GW detectors in the world), are limited by technical noise from the control systems at some frequency bands below 100 Hz. This noise is not fundamental. The feedback loops need to control the motion of the auxilliary mirrors at low frequencies (< 10 Hz). There is some leakage of the control forces into the GW detection band (> 10 Hz). [cite Denis noise paper]

There is seismic noise. We can cancel it using adaptive feed-forward (distinct from feedback). We measure the noise on the ground. La machina applies filters to all of the ground sensors and feeds it forward into the interferometer so as to cancel the motion. We have been able to reduce the seismically induced motion by 10-20x in the past. [cite JenneFF, RyanFF]

One of the key techniques that has emerged under the applications of machine learning in control systems is Reinforcement Learning. To learn more about what Reinforcement Learning is and how it is being used in LIGO control systems click here

Other papers from our group on controls for GW detectors: