Titulo Estágio
Forecasting the Impact of Space Weather in Satellite Orbits
Local do Estágio
DEI-FCTUC
Enquadramento
Space Weather (SW) impacts several aspects of our lives, from communications to power grids. One of the most affected devices are satellites [1]. In middle Earth orbits, such as the Global Navigation Satellites Systems used around the world for navigation,
satellites communicate with ground antennas for monitorisation and other purposes. When a SW event occurs, it is possible to force the satellite out of orbit. In those situations, the antennas need to do extra movements for tracking purposes and the precision decreases [2]. Thus, it is useful to know beforehand when the orbit will be affected by a SW event. By forecasting SW events, it is possible to improve the quality of communication and navigation that the satellites provide.
Machine Learning (ML) is commonly used to model SW events [3]. Despite the low success of ML to forecast SW in the past years, nowadays there are large datasets being made freely available by different organisations, such as ESA [4], with standardised formats that allow the development and training of better ML algorithms. At the same time, in recent years several developments and novel algorithms have also been proposed within the ML community for a variety of different problems, from classification to regression and unsupervised problems.
This work will focus on researching state-of-the-art ML approaches to create models that can predict when the orbit of a satellite will be affected by a SW event. This task will require thoroughly processing existing datasets, using exploratory and descriptive analyses, to identify and characterize the data available from the satellites. This information will then be analyzed to determine which ML techniques are suitable (from dimensionality reduction to algorithm selection), which will afterward be explored to create predictive models for satellite orbit prediction.
References:
[1] J. C. Green, J. Likar, and Y. Shprits, ‘Impact of space weather on the satellite industry’, Space Weather, vol. 15, no. 6, pp. 804–818, 2017, doi: 10.1002/2017SW001646.
[2] E. Doornbos and H. Klinkrad, ‘Modelling of space weather effects on satellite drag’, Advances in Space Research, vol. 37, no. 6, pp. 1229–1239, Jan. 2006, doi: 10.1016/j.asr.2005.04.097.
[3] E. Camporeale, ‘The Challenge of Machine Learning in Space Weather: Nowcasting and Forecasting’, Space Weather, vol. 17, no. 8, pp. 1166–1207, 2019, doi: 10.1029/2018SW002061.
[4] ESA Space Weather Portal, https://swe.ssa.esa.int/current-space-weather
Objetivo
The goal of this thesis is to develop a ML approach to forecast SW impacts in satellites orbits. Thus, making possible the detection of an orbit change and the need to adapt the antenna position.
The student will do a literature review of ML, study different SW events, model existing satellite orbits using a satellite tracking package, and analyse existing datasets to perform the experimental evaluation.
Plano de Trabalhos - Semestre 1
- Familiarisation with different SW events
- Literature review of ML
- Analysis of existing datasets
- Writing of the intermediate report
Plano de Trabalhos - Semestre 2
- Development of a ML approach to detect orbit changes when there is an impact by SW events
- Simulation of satellite orbits
- Experimental analysis
- Writing of the final report
Condições
The work is to be executed at the laboratories of the CISUC’s NCS group. A workplace will be provided as well as the required computational resources.
Observações
Advisors: Noé Godinho (noe@dei.uc.pt) and João R. Campos (jrcampos@dei.uc.pt)
Orientador
Noé Godinho / João R. Campos
noe@dei.uc.pt 📩