Improved orbit predictions using two-line elements through error pattern mining and transferring

Li, Bin; Zhang, Yan; Jian, Huang; Sang, Jizhang

doi:10.1016/j.actaastro.2021.08.002

Cited by 13 publications

(4 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is typically used with the classification and regression tree (CART) (Breiman et al 1984) of a fixed size as base learners, which is known as GBDT. GBDT has been one of the most commonly used techniques in Kaggle competitions and achieved excellent performances in many scientific applications, such as the orbit predictions of space debris (Li et al 2020(Li et al , 2021, real-time GNSS precipitable water vapour sensing (Zheng et al 2022), and GPS signal reception classification (Sun et al 2020). Therefore, we hypothesise that the GBDT could perform well in modelling and prediction of GNSS time series.…”

Section: Gradient Boosting Decision Treementioning

confidence: 96%

“…The errors of the physics-based OP results over the future 7 days achieve at least 50% accuracy improvement with the application of the model. Furthermore, Li et al (2021) applied the GBDT and convolutional neural networks (CNN) to model the underlying orbit error patterns. With the correction by model-predicted errors, the accuracy of OP over the future 14 days are improved by more than 75%, 90%, and 90% in the along-track, cross-track, and radial directions, respectively.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches

Gao

Chen

et al. 2022

J Geod

View full text Add to dashboard Cite

Global navigation satellite system (GNSS) site coordinate time series provides essential data for geodynamic and geophysical studies, realisation of a regional or global geodetic reference frames, and crustal deformation research. The coordinate time series has been conventionally modelled by least squares (LS) fitting with harmonic functions, alongside many other analysis methods. As a key limitation, the traditional modelling approaches simply use the functions of time variable, despite good knowledge of various underlying physical mechanisms responsible for the site displacements. This paper examines the use of machine learning (ML) models to reflect the effects or residential effects of physical variables related to Sun and the Moon ephemerides, polar motion, temperature, atmospheric pressure, and hydrology on the site displacements. To form the ML problem, these variables are constructed as the input vector of each ML training sample, while the vertical displacement of a GNSS site is regarded as the output value. In the evaluation experiments, three ML approaches, namely the gradient boosting decision tree (GBDT) approach, long short-term memory (LSTM) approach, and support vector machine (SVM) approach, are introduced and evaluated with the time series datasets collected from 9 GNSS sites over the period of 13 years. The results indicate that all three approaches achieve similar fitting precision in the range of 3–5 mm in the vertical displacement component, which is an improvement in over 30% with respect to the traditional LS fitting precision in the range of 4–7 mm. The prediction of the vertical time series with the three ML approaches shows the precision in the range of 4–7 mm over the future 24- month period. The results also indicate the relative importance of different physical features causing the displacements of each site. Overall, ML approaches demonstrate better performance and effectiveness in modelling and prediction of GNSS time series, thus impacting maintenance of geodetic reference frames, geodynamics, geophysics, and crustal deformation analysis.

show abstract

Section: Gradient Boosting Decision Treementioning

confidence: 96%

Section: Introductionmentioning

confidence: 99%

Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches

Gao

Chen

et al. 2022

J Geod

View full text Add to dashboard Cite

show abstract

“…Under the same hybrid SGP4 approach, four independent works show that Convolutional Neural Networks (CNN) [65], Feed-Forward Neural Networks (FNN) [66], Recurrent Neural Networks (RNN) [67] and Long-Short Term Memory (LSTM) [68] can be used to improve SGP4 orbit prediction. Using a Gradient Boosting Decision Tree (GBDT) and a CNN, Li et al [69,70] improved by more than 75% the along-track direction prediction for five satellites in LEO and GEO. All of this work has been done…”

Section: Orbit Predictionmentioning

confidence: 99%

Machine Learning in Orbit Estimation: a Survey

Caldas¹,

Soares²

2022

Preprint

View full text Add to dashboard Cite

Since the late '50s, when the first artificial satellite was launched, the number of resident space objects (RSOs) has steadily increased. It is estimated that around 1 Million objects larger than 1 cm are currently orbiting the Earth, with only 30,000, larger than 10 cm, presently being tracked. To avert a chain reaction of collisions, termed Kessler Syndrome [1], it is indispensable to accurately track and predict space debris and satellites' orbit alike. Current physics-based methods have errors in the order of kilometres for 7 days predictions, which is insufficient when considering space debris that have mostly less than 1 meter. Typically, this failure is due to uncertainty around the state of the space object at the beginning of the trajectory, forecasting errors in environmental conditions such as atmospheric drag, as well as specific unknown characteristics such as mass or geometry of the RSO. Leveraging datadriven techniques, namely machine learning, the orbit prediction accuracy can be enhanced: by deriving unmeasured objects' characteristics, improving non-conservative forces' effects, and by the superior abstraction capacity that Deep Learning models have of modelling highly complex non-linear systems. In this survey, we provide an overview of the current work being done in this field.

show abstract

“…Takahashi et al [34] explored Gaussian process regression using GPS (Global Positioning System) data to reconstruct continuous data from sparse positioning data, providing a valuable perspective on data-driven orbit prediction. Li et al [35] implemented two learning methods, gradient boosting decision tree and convolutional neural networks to improve TLE-based orbit prediction accuracy. Salleh et at.…”

Section: Introductionmentioning

confidence: 99%

A PCA-SVM Regression Model for LEO Space Debris Orbit Prediction in an Optical Space-Based Space Surveillance Network

Hamidian,

Kosari

2024

IEEE Access

View full text Add to dashboard Cite

The primary objective of this study is to investigate the efficacy of Support Vector Machine (SVM) regression method to enhance the accuracy of Low Earth Orbit (LEO) space debris orbit prediction using the historical data. Principal Component Analysis (PCA) is employed to efficiently reduce the dimensionality of dataset's feature space and hence optimize the model's performance. This investigation is motivated by the limitations of conventional orbit prediction methods, which often rely on dynamic models with unknown coefficients of perturbation forces and other relevant characteristics of space debris, leading to errors during the prediction process. On the other hand, while the Collision Avoidance Maneuver (CAM) strategy remains crucial for mitigating the threat posed by such debris, precise knowledge of debris coordinates is essential for effective CAM implementation. However, traditional ground-based optical equipment encounters challenges in observing fastmoving debris within the dynamic LEO environment, including atmospheric interference and limited Field of View (FOV). To address these limitations, the secondary objective of this study involves exploring the potential of an in-orbit optical space surveillance network as a promising solution. The system utilizes optical sensors distributed across multiple spacecraft within the Above the Horizon (ATH) constellation, specifically designed to continuously monitor the most densely populated altitude band in LEO. Simulations under different conditions demonstrate that the proposed scheme successfully complements groundbased equipment and dynamic models for debris tracking, thereby improving orbit prediction accuracy. The results of simulations under different conditions demonstrate that proposed scheme successfully complements ground-based equipment and dynamic models for debris tracking, and improving orbit prediction accuracy.

show abstract

Improved orbit predictions using two-line elements through error pattern mining and transferring

Cited by 13 publications

References 20 publications

Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches

Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches

Machine Learning in Orbit Estimation: a Survey

A PCA-SVM Regression Model for LEO Space Debris Orbit Prediction in an Optical Space-Based Space Surveillance Network

Contact Info

Product

Resources

About