Skip to main navigation Skip to search Skip to main content

Outlier Removal of Discontinuous Satellite Telemetry Data Based on Deconvolutional Reconstruction Network

  • School of Astronautics, Harbin Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Satellite telemetry data is important strategic resource. We can monitor and predict the status of the satellite through analyzing the telemetry data. However, due to interference on the satellite and sensor failure, the telemetry data will jump and generate outliers. Therefore, it is necessary to identify and remove the outliers. This paper proposes an outlier removal method based on deconvolutional reconstruction network. The deconvolutional reconstruction network is composed of multiple convolution and deconvolution which is used to learn the internal laws from massive telemetry data. The learned network can make accurate predictions for normal data except for outliers. Our method use this difference to set the threshold and perform outlier removal. The deconvolutional reconstruction network proposed in this paper uses a very few parameters for rapid learning. The network can converge within 20 epochs for multiple sets of telemetry datasets which contains more than 60k discontinuous points. Numerical experiments show that the proposed method can achieve perfect removal effects.

Original languageEnglish
Title of host publicationProceedings of 2022 IEEE 11th Data Driven Control and Learning Systems Conference, DDCLS 2022
EditorsMingxuan Sun, Zengqiang Chen
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages6-10
Number of pages5
ISBN (Electronic)9781665496759
DOIs
StatePublished - 2022
Externally publishedYes
Event11th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2022 - Emeishan, China
Duration: 3 Aug 20225 Aug 2022

Publication series

NameProceedings of 2022 IEEE 11th Data Driven Control and Learning Systems Conference, DDCLS 2022

Conference

Conference11th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2022
Country/TerritoryChina
CityEmeishan
Period3/08/225/08/22

Keywords

  • Deconvolutional Reconstruction Network
  • Outlier Removal
  • Satellite Telemetry Data

Fingerprint

Dive into the research topics of 'Outlier Removal of Discontinuous Satellite Telemetry Data Based on Deconvolutional Reconstruction Network'. Together they form a unique fingerprint.

Cite this