Neural network interpolation of the magnetic field for the LISA Pathfinder Diagnostics Subsystem

TL;DR

This paper evaluates neural network interpolation for accurately estimating magnetic fields at test masses in the LISA Pathfinder mission, demonstrating robustness under various conditions and improving over classical methods.

Contribution

It introduces and assesses a neural network approach for magnetic field interpolation, enhancing measurement reliability in space-based gravitational wave detection.

Findings

Neural networks provide accurate magnetic field estimates at test masses.

The method is robust against sensor position variations and environmental changes.

Classical interpolation methods are less reliable in this context.

Abstract

LISA Pathfinder is a science and technology demonstrator of the European Space Agency within the framework of its LISA mission, which aims to be the first space-borne gravitational wave observatory. The payload of LISA Pathfinder is the so-called LISA Technology Package, which is designed to measure relative accelerations between two test masses in nominal free fall. Its disturbances are monitored and dealt by the diagnostics subsystem. This subsystem consists of several modules, and one of these is the magnetic diagnostics system, which includes a set of four tri-axial fluxgate magnetometers, intended to measure with high precision the magnetic field at the positions of the test masses. However, since the magnetometers are located far from the positions of the test masses, the magnetic field at their positions must be interpolated. It has been recently shown that because there are not enough magnetic channels, classical interpolation methods fail to derive reliable measurements at the positions of the test masses, while neural network interpolation can provide the required measurements at the desired accuracy. In this paper we expand these studies and we assess the reliability and robustness of the neural network interpolation scheme for variations of the locations and possible offsets of the magnetometers, as well as for changes in environmental conditions. We find that neural networks are robust enough to derive accurate measurements of the magnetic field at the positions of the test masses in most circumstances.