Finding eclipsing binary star systems and measuring their physical properties using machine learning

食双星系统的研究是恒星物理学中最成熟和最有价值的领域之一，它提供了仅使用观测数据和几何学来确定遥远恒星的质量和半径的独特机会。这一研究领域目前正在经历复兴，这是因为开普勒、K2和TESS卫星等太阳系外行星的空间搜索产生了惊人的质量和数量的数据。这些空间飞行任务产生了如此庞大的数据集，因此使寻找和分析日食双星的过程自动化是至关重要的。这个博士项目的目的是开发这样做所需的计算工具。然后，这些工具可以应用于来自TESS卫星的数据，并在未来将用于来自柏拉图卫星的数据。使用的方法将包括机器学习算法，如深度学习、图像分类和神经网络。该项目的基本目标是确定非常适合验证和改进恒星理论模型的双星，这些模型构成了观测和理论天体物理学的大多数领域的基础。

The study of eclipsing binary star systems is one of the most mature and rewarding areas of stellar physics, offering the unique opportunity to determine the masses and radii of distant stars using only observational data and geometry. This area of research is currently experiencing a renaissance, due to the remarkable quality and quantity of data coming from space-based searches for extrasolar planets such as the Kepler, K2 and TESS satellites. These space missions produce such large datasets that it is vital to automate the processes of finding and analysing eclipsing binaries. The aim of this PhD project is to develop the computational tools needed to do so. The tools can then be applied to data from the TESS satellite, and in future will be used for data from the PLATO satellite. The methods used will include machine learning algorithms such as deep learning, image classification, and neural networks. The fundamental aim of this project is to identify binary stars that are well suited to verifying and improving theoretical models of stars, which form the foundation of most areas of observational and theoretical astrophysics.