Robust Identification of Multiple Input Single Output System Response for Efficient Pickup Noise Removal from Tokamak Diagnostics

Abstract

Electromagnetic pickup noise in the tokamak environment imposes an imminent challenge for measuring weak diagnostic photocurrents nA range. The diagnostic signal can be contaminated by an unknown mixture of crosstalk signals from coils powered by currents in kA range. To address this issue, an algorithm for robust identification of linear multi-input single-output (MISO) systems has been developed. MISO model describes the dynamic relationship between measured signals from power sources and observed signals in the diagnostics and allows for a precise subtraction of the noise component. The proposed method was tested on experimental diagnostic data from the DIII-D tokamak, and it has reduced noise by up to 20 dB in 1–20 kHz range.