ESPAI (Enhancing Signal Purity with Artificial Intelligence) is a project that aims to enhance signal purity using Artificial Intelligence in X-band telescopes. The project’s goal is to research and devise an automated system for removing celestial disturbances from astronomical signal observations in the X-band. ESPAI aims to reduce the quantity of data discarded during analyses and to improve the quality of filters used for cleaning observations, consequently improving the quality of analyses.

The ESPAI project has been selected and funded under the ICSC Programme, financed as part of the PNRR, and funded by the European Union – NextGenerationEU.

A significant challenge in astronomy is related to celestial interference, which acts as noise that complicates the identification of astronomical signals.

Various sources of disturbance exist, among the main ones being solar activity (flares, coronal ejections), energetic particles in the Van Allen belts, and cosmic rays. These factors contribute to obscuring X-ray signals, which require careful filtering and modelling to isolate the astrophysical sources of interest.

In addition to the scientific challenges, this noise exacerbates technological challenges by increasing the volume of data collected without adding any value.

To address these challenges, ESPAI plans to:

1. Use Generative AI (Variational Autoencoders, Stable Diffusion, and GANs) to develop a large dataset of synthetic observations free from celestial disturbances, serving as a benchmark to differentiate genuine signals from noise.
2. Develop an anomaly detection system for identifying and removing celestial disturbances from X-band telescope signals, with potential real-time implementation to reduce data storage needs.
3. Design a system adaptable to individual telescope needs, allowing customisation based on geographic location and specific disturbances.