Convolutional neural network design for the prediction of the focusing temperature of PLATO telescopes under simulated space operating conditions

The main science goal of the PLAnetary Transits and Oscillations of stars mission is to detect and characterize extrasolar planets around solar-type stars. For this purpose, the telescopes developed for the mission have a focusing system consisting on to tune the thermal surrounding of the opto-mechanics for achieving the optimal focus. We propose an alternative method for determining the best focus temperature. We designed a convolutional neural network that uses images obtained in ground performance evaluation to implement a calibration method to be used during pre-launch tests and orbital tuning. The convolutional neural network effectively predicts focusing temperatures, improving calibration efficiency and accuracy. This approach demonstrates the potential of artificial intelligence methods for optimizing telescope focusing in operational conditions.