Neural network-based approach to noise identification of laser interferometric GW antennas

A neural network-based approach is presented for the real time noise identification of gravitational wave (GW) laser interferometric antennas. The 40-meter Caltech laser interferometer output data provides a good testbed of algorithms for noise identification (violin resonances in the suspensions, main power harmonics, ring-down noise from servo control systems, electronics noises, glitches and so on) of the first interferometric long GW antennas like VIRGO, LIGO, GEO and TAMA. We used for noise identification both simulated data for VIRGO and the real data of the 40-meter Caltech laser interferometer. The results obtained are quite good notwithstanding its high initial computational cost. The algorithm we propose is quite general and robust, taking into account that it does not requires a priori information on the data, nor a precise model, and constitutes a powerful tool for quality data analysis