Neural Network based calibration of atmospheric density models

Pérez, David; Bevilacqua, Riccardo

doi:10.1016/j.actaastro.2014.12.018

Cited by 36 publications

(20 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Numerous studies on calibrating the density of atmospheric models have used neural network techniques. Perez and Bevilacqua [22] used the density from DTM-2013, NRLMSISE-00, and JB2008 as the neural network targets, with CHAMP and GRACE satellite data used for training, verification, and testing. The resulting density error was better than that before correction.…”

Section: Introductionmentioning

confidence: 99%

Density Correction of NRLMSISE-00 in the Middle Atmosphere (20–100 km) Based on TIMED/SABER Density Data

et al. 2020

View full text Add to dashboard Cite

This paper describes the density correction of the NRLMSISE-00 using more than 15 years (2002–2016) of TIMED/SABER satellite atmospheric density data from the middle atmosphere (20–100 km). A bias correction factor dataset is established based on the density differences between the TIMED/SABER data and NRLMSISE-00. Seven height nodes are set in the range between 20 and 100 km. The different scale oscillations of the correction factor are separated at each height node, and the spherical harmonic function is used to fit the coefficients of the different timescale oscillations to obtain a spatiotemporal function at each height node. Cubic spline interpolation is used to obtain the correction factor at other non-node heights. The spatiotemporal correction function depends on six key parameters, including height, latitude, longitude, local time, day, and year. The evaluation results show that the spatiotemporal correction function proposed in this paper achieves a good correction effect on the atmospheric density of NRLMSISE-00. The correction effect becomes more pronounced as the height increases. After correction, the relative error of the model decreased by 40%–50% in July, especially at ±40° N in the 80–100 km region. The correction effect of the spatiotemporal correction function under different geomagnetic activity may have some potential relationships with geomagnetic activities. During geomagnetic storms, the relative errors in atmospheric density at 100, 70, and 32 km decrease from 41.21%, 22.09%, and 3.03% to −9.65%, 2.60%, and 1.44%, respectively, after correction. The relative errors in atmospheric density at 100, 70, and 32 km decrease from 68.95%, 21.02%, and 3.56% to 3.49%, 2.20%, and 1.77%, respectively, during the geomagnetic quiet period. The correction effect during the geomagnetic quiet period is better than that during geomagnetic storms at a height of 100 km. The subsequent effects of geomagnetic activity will be considered, and the atmospheric density during magnetic storms and quiet periods will be corrected separately near 100 km. The ability of the model to characterize the mid-atmosphere (20–100 km) is significantly improved compared with the pre-correction performance. As a result, the corrected NRLMSISE-00 can provide more reliable atmospheric density data for scientific researches and engineering fields, such as data analysis, instrument design, and aerospace vehicles.

show abstract

Section: Introductionmentioning

confidence: 99%

Density Correction of NRLMSISE-00 in the Middle Atmosphere (20–100 km) Based on TIMED/SABER Density Data

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Uncertainties in the atmospheric density forecast result in errors in the guidance trajectories, and vice versa, any improvement in the atmospheric density forecast will allow to calculate more realistic guidance trajectories. Frequently used global atmospheric models are often designed to calculate more than just a specific parameter (such as the density) leading to higher computation time and less accurate results for the specific quantity [114,115]. A critical assessment of atmospheric modelling can be found in [62].…”

Section: Coping With Uncertaintiesmentioning

confidence: 99%

On the exploitation of differential aerodynamic lift and drag as a means to control satellite formation flight

et al. 2019

View full text Add to dashboard Cite

In order for a satellite formation to maintain its intended design despite present perturbations (formation keeping), to change the formation design (reconfiguration) or to perform a rendezvous maneuver, control forces need to be generated. To do so, chemical and/or electric thrusters are currently the methods of choice. However, their utilization has detrimental effects on small satellites' limited mass, volume and power budgets. Since the mid-eighties, the potential of using differential drag as a means of propellant-less source of control for satellite formation flight is actively researched. This method consists of varying the aerodynamic drag experienced by different spacecraft, thus generating differential accelerations between them. Its main disadvantage, that its controllability is mainly limited to the in-plain relative motion, can be overcome by using differential lift as a means to control the out-of-plane motion. Due to its promising benefits, a variety of studies from researchers around the world have enhanced the state-of-the-art over the past decades which results in a multitude of available literature. In this paper, an extensive literature review of the efforts which led to the current state-of-the-art of different lift and drag based satellite formation control is presented. Based on the insights gained during the review process, key knowledge gaps that need to be addressed in the field of differential lift in order to enhance the current state-of-the-art are revealed and discussed. In closer detail, the interdependence between the feasibility domain / the maneuver time and increased differential lift forces achieved using advanced satellite surface materials promoting quasi-specular or specular reflection, as currently being developed in the course of the DISCOVERER project, is discussed. Keywords Satellite aerodynamics • differential lift • differential drag • formation flight control • propellant less control Acknowledgements This project has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No. 737183. This reflects only the author's view and the European Commission is not responsible for any use that may be made of the information it contains. The author would like to thank the reviewers, the DISCOVERER team as well as several colleagues from IRS for their valuable feedback and suggestions.

show abstract

“…Автономность особенно важна для аппаратов, движущихся на низких орбитах, так как для них плотность атмосферы является вторым по значимости фактором после гравитационного притяжения. С использованием методов обучения с учителем нейронные сети могут помочь уточнить модель атмосферы во время движения аппарата [1,2].…”

Section: Introductionunclassified

Correction and prediction of the spacecraft’s orbital motion by artificial neural networks

Sorokin¹,

Shirobokov

2018

KIAM Prepr.

View full text Add to dashboard Cite

Neural Network based calibration of atmospheric density models

Cited by 36 publications

References 37 publications

Density Correction of NRLMSISE-00 in the Middle Atmosphere (20–100 km) Based on TIMED/SABER Density Data

Density Correction of NRLMSISE-00 in the Middle Atmosphere (20–100 km) Based on TIMED/SABER Density Data

On the exploitation of differential aerodynamic lift and drag as a means to control satellite formation flight

Correction and prediction of the spacecraft’s orbital motion by artificial neural networks

Contact Info

Product

Resources

About