Optimización de un sistema difuso para la detección automática de tránsitos planetarios en curvas de luz de estrellas individuales

Abstract

The transit method is an effective way to find extrasolar planets. This method is based on the shallow decrease that a planet causes in its host star’s apparent brightness: when the planet passes through our line of sight, it affects the brightness we receive from the star with our telescopes. However, these transit events are very close to the telescopes’ detection sensitivity limit. To confirm a planet observation, it takes at least three (3) transit events, making long-time observations of a star necessary to detect extrasolar planets that may orbit it, which results in large amounts of data to be analized. In this work we did a new software pipeline for autonomous detection of transit traces by analyzing extracted features from stellar light curves using a fuzzy logic classifier, avoiding the task of searching for transit events in each section of the light curves. During the development process of the software pipeline, the Knowledge Discovery in Databases (KDD) methodology is implemented, which presents a way to extract knowledge from large datasets.