Monthly sunspot numbers forecast with artificial neural network combined with dynamo model: Comparison with modern methods

Nikolai Safiullin, Sergey Porshnev, Nathan Kleeorin

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

3 Scopus citations

Abstract

In this paper we propose a novel method for a monthly forecast of the total sunspot number time series, based on the combination of a dynamo model with an artificial neural network. The nonlinear autoregressive scheme is used with exo-genous input, consisting of two parts: The prior real observations and the corresponding model estimations at the same time-point. The results of the monthly forecast have been compared to all the modern sunspot forecasting methods, including data assimilation techniques, showing the higher accuracy of the proposed method when using one-step prediction and monthly corrections.

Original languageEnglish
Title of host publicationProceedings - 2018 Ural Symposium on Biomedical Engineering, Radioelectronics and Information Technology, USBEREIT 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages199-202
Number of pages4
ISBN (Electronic)9781538649466
DOIs
StatePublished - 13 Jun 2018
Event2018 Ural Symposium on Biomedical Engineering, Radioelectronics and Information Technology, USBEREIT 2018 - Yekaterinburg, Russian Federation
Duration: 7 May 20188 May 2018

Publication series

NameProceedings - 2018 Ural Symposium on Biomedical Engineering, Radioelectronics and Information Technology, USBEREIT 2018

Conference

Conference2018 Ural Symposium on Biomedical Engineering, Radioelectronics and Information Technology, USBEREIT 2018
Country/TerritoryRussian Federation
CityYekaterinburg
Period7/05/188/05/18

Keywords

  • artificial neural network
  • data analysis
  • solar activity
  • sunspot numbers
  • time series forecast

Fingerprint

Dive into the research topics of 'Monthly sunspot numbers forecast with artificial neural network combined with dynamo model: Comparison with modern methods'. Together they form a unique fingerprint.

Cite this