A regime-switching recurrent neural network model applied to wind time series

Nikolay Y. Nikolaev*, Evgueni Smirnov, Daniel Stamate, Robert Zimmer

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

This paper proposes a regime-switching recurrent network model (RS-RNN) for non-stationary time series. The RS-RNN model emits a mixture density with dynamic nonlinear regimes that fit flexibly data distributions with non-Gaussian shapes. The key novelties are: development of an original representation of the means of the component distributions by dynamic nonlinear recurrent networks, and derivation of a corresponding expectation maximization (EM) training algorithm for finding the model parameters. The results show that the RS-RNN applied to a real-world wind speed time series achieves standardized residuals similar to popular previous models, but it is more accurate distribution forecasting than other linear switching (MS-AR) and nonlinear neural network (MLP and RNN) models.
Original languageEnglish
Pages (from-to)723-734
Number of pages12
JournalApplied Soft Computing
Volume80
DOIs
Publication statusPublished - Jul 2019

Keywords

  • FORECASTING WIND
  • DENSITY FORECASTS
  • SPEED
  • POWER
  • ENSEMBLE

Cite this

Nikolaev, Nikolay Y. ; Smirnov, Evgueni ; Stamate, Daniel ; Zimmer, Robert. / A regime-switching recurrent neural network model applied to wind time series. In: Applied Soft Computing. 2019 ; Vol. 80. pp. 723-734.
@article{00d9afcc06c84a278926674abcfecc7d,
title = "A regime-switching recurrent neural network model applied to wind time series",
abstract = "This paper proposes a regime-switching recurrent network model (RS-RNN) for non-stationary time series. The RS-RNN model emits a mixture density with dynamic nonlinear regimes that fit flexibly data distributions with non-Gaussian shapes. The key novelties are: development of an original representation of the means of the component distributions by dynamic nonlinear recurrent networks, and derivation of a corresponding expectation maximization (EM) training algorithm for finding the model parameters. The results show that the RS-RNN applied to a real-world wind speed time series achieves standardized residuals similar to popular previous models, but it is more accurate distribution forecasting than other linear switching (MS-AR) and nonlinear neural network (MLP and RNN) models.",
keywords = "FORECASTING WIND, DENSITY FORECASTS, SPEED, POWER, ENSEMBLE",
author = "Nikolaev, {Nikolay Y.} and Evgueni Smirnov and Daniel Stamate and Robert Zimmer",
year = "2019",
month = "7",
doi = "10.1016/j.asoc.2019.04.009",
language = "English",
volume = "80",
pages = "723--734",
journal = "Applied Soft Computing",
issn = "1568-4946",
publisher = "Elsevier Science",

}

A regime-switching recurrent neural network model applied to wind time series. / Nikolaev, Nikolay Y.; Smirnov, Evgueni; Stamate, Daniel; Zimmer, Robert.

In: Applied Soft Computing, Vol. 80, 07.2019, p. 723-734.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - A regime-switching recurrent neural network model applied to wind time series

AU - Nikolaev, Nikolay Y.

AU - Smirnov, Evgueni

AU - Stamate, Daniel

AU - Zimmer, Robert

PY - 2019/7

Y1 - 2019/7

N2 - This paper proposes a regime-switching recurrent network model (RS-RNN) for non-stationary time series. The RS-RNN model emits a mixture density with dynamic nonlinear regimes that fit flexibly data distributions with non-Gaussian shapes. The key novelties are: development of an original representation of the means of the component distributions by dynamic nonlinear recurrent networks, and derivation of a corresponding expectation maximization (EM) training algorithm for finding the model parameters. The results show that the RS-RNN applied to a real-world wind speed time series achieves standardized residuals similar to popular previous models, but it is more accurate distribution forecasting than other linear switching (MS-AR) and nonlinear neural network (MLP and RNN) models.

AB - This paper proposes a regime-switching recurrent network model (RS-RNN) for non-stationary time series. The RS-RNN model emits a mixture density with dynamic nonlinear regimes that fit flexibly data distributions with non-Gaussian shapes. The key novelties are: development of an original representation of the means of the component distributions by dynamic nonlinear recurrent networks, and derivation of a corresponding expectation maximization (EM) training algorithm for finding the model parameters. The results show that the RS-RNN applied to a real-world wind speed time series achieves standardized residuals similar to popular previous models, but it is more accurate distribution forecasting than other linear switching (MS-AR) and nonlinear neural network (MLP and RNN) models.

KW - FORECASTING WIND

KW - DENSITY FORECASTS

KW - SPEED

KW - POWER

KW - ENSEMBLE

U2 - 10.1016/j.asoc.2019.04.009

DO - 10.1016/j.asoc.2019.04.009

M3 - Article

VL - 80

SP - 723

EP - 734

JO - Applied Soft Computing

JF - Applied Soft Computing

SN - 1568-4946

ER -