Artificial intelligence unfolding for INTEGRAL IREM data
Solar Energetic Protons (SEPs) and the Radiation Belts (RBs) are
two of the three main sources of particle radiation in the near-Earth
space, the third being Galactic Cosmic Rays. Such radiation can be
detrimental to satellites and the function of their components either
through single event effects (e.g. data corruption) or cumulative
effects, which can eventually render satellite systems
inoperable. Furthermore, particle radiation is ionizing and able to both
damage DNA and cause prompt radiation sickness. It is therefore a very
important factor for manned missions to the Moon and Mars as it can have
severe effects on astronauts' health.
A large volume of measurements during SEPs and from trapped energetic
particles within the RBs are collected from a series of radiation
monitors on-board a multitude of satellites. The INTEGRAL Radiation
Environment Monitor (IREM) on-board INTEGRAL provides important
measurements on SEPs and trapped particles since 2002 (see, e.g.,
INTEGRAL POM's
July 2018,
May 2012,
December 2003).
However, the raw
monitor data, i.e. count-rates, must be rendered into useful information
in the form of particle fluxes in physical units. This flux data can
then be used by satellite operators and radiation environment physicists
and engineers. Nevertheless, the derivation of fluxes from count-rates,
or "unfolding", is not trivial as it is an ill-posed inverse problem and
it can yield grave errors or even completely unphysical results, e.g.
negative particle fluxes.
GenCORUM is an Artificial Intelligence method internally employing a
Genetic Algorithm. With it, it is possible to unfold the count data into
solar proton fluxes and electron fluxes within the outer RB using IREM
raw count measurements to a high level of accuracy. The IREM flux data
were validated with proton fluxes from the ESA SEPEM Reference Dataset
(RDS) and electron flux measurements from the Van Allen Probes (VAP)
mission.