Semi-empirical Thermal Modelling of LAPAN-A3 Satellite Using Machine Learning Method

Understanding LAPAN-A3 thermal characteristics is the key to improve the next generation Indonesian micro-satellite thermal design. LAPAN-A1 until LAPAN-A3 use passive thermal control system by means of heat distribution by their structure, and heat radiation coating. Data-driven approaches in satellite thermal analysis are not rare phenomenon and are increasingly used to reduce the complexity in first principle thermal modelling that require high accuracy in the satellite component's thermal properties. This paper presents a simple semi-empirical thermal model to predict satellite temperature using machine learning method trained on real satellite telemetry data. The model can deduce the variables needed to predict satellite temperature changes. The proposed algorithm is implemented with a 7-node model of LAPAN-A3 (6 nodes for each side and 1 node for middle plate) and trained with data from 19 to 20 May 2018 to create satellite node temperature predictions. From initial performance evaluation, the model shows promising results and has potential for real-life usage in future satellite developments.