Machine Learning Enhanced NARMAX Model for Dst Index Forecasting

As many systems and equipment are sensitive to magnetic disturbances, it is important to understand the magnetosphere system, to reduce the negative effect caused by severe space weather situations. The disturbance storm time (Dst) index is used to measure the magnetic disturbances and it is correlated with solar wind variables. This study presents a new machine learning enhanced NARMAX (MLE- NARMAX) model for 3 hours ahead forecasting of Dst index. An important advantage of the MLE-NARMAX model is that it provides a transparent and explainable model structure. The model performance is tested over three typical strong storm periods, where the prediction skills are 0.9734, 0.9598 and 0.9206 in terms of correlation, and 0.9474, 0.9173, and 0.8333 in terms prediction efficiency (PE). Compared to the conventional NARX model, the MLE-NARMAX produces better model predictions.

[1]  Dong-Hun Lee,et al.  Comparison of Dst Forecast Models for Intense Geomagnetic Storms , 2012 .

[2]  P. Wintoft,et al.  Evaluation of Kp and Dst Predictions Using ACE and DSCOVR Solar Wind Data , 2018, Space Weather.

[3]  S. A. Billings,et al.  A century of variation in the dependence of Greenland iceberg calving on ice sheet surface mass balance and regional climate change , 2014, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences.

[4]  S. A. Billings,et al.  Forecasting the geomagnetic activity of the Dst index using multiscale radial basis function networks , 2007 .

[5]  Sheng Chen,et al.  Orthogonal least squares methods and their application to non-linear system identification , 1989 .

[6]  H. Lundstedt,et al.  Prediction of geomagnetic storms from solar wind data using Elman Recurrent Neural Networks , 1996 .

[7]  Stephen A. Billings,et al.  Data derived NARMAX Dst model , 2011 .

[8]  Julia L. Blanchard,et al.  Quantifying heterogeneous responses of fish community size structure using novel combined statistical techniques , 2016, Global change biology.

[9]  Xinlin Li,et al.  Dst model for 1995–2002 , 2006 .

[10]  Stephen A. Billings,et al.  Identification of nonlinear time-varying systems using an online sliding-window and common model structure selection (CMSS) approach with applications to EEG , 2016, Int. J. Syst. Sci..

[11]  Michael A. Balikhin,et al.  System Identification and Data‐Driven Forecasting of AE Index and Prediction Uncertainty Analysis Using a New Cloud‐NARX Model , 2019, Journal of Geophysical Research: Space Physics.

[12]  Hua-Liang Wei,et al.  Significant Indicators and Determinants of Happiness: Evidence from a UK Survey and Revealed by a Data-Driven Systems Modelling Approach , 2018 .

[13]  Jinfu Liu,et al.  Deep Learning Based Solar Flare Forecasting Model. I. Results for Line-of-sight Magnetograms , 2018 .

[14]  S. A. Billings,et al.  Using the NARMAX OLS-ERR algorithm to obtain the most influential coupling functions that affect the evolution of the magnetosphere , 2011 .

[15]  B. Tsurutani,et al.  Prediction of peak-Dst from halo CME/magnetic cloud-speed observations , 2004 .

[16]  Enrico Camporeale,et al.  The Challenge of Machine Learning in Space Weather: Nowcasting and Forecasting , 2019, Space Weather.

[17]  Hua-Liang Wei,et al.  Nonlinear predictive model selection and model averaging using information criteria , 2018 .

[18]  C G Billings,et al.  The prediction of in-flight hypoxaemia using non-linear equations. , 2013, Respiratory medicine.

[19]  Manuel Grande,et al.  Current understanding of magnetic storms: storm-substorm relationships , 1998 .

[20]  H. W. Kroehl,et al.  What is a geomagnetic storm , 1994 .

[21]  Henrik Lundstedt,et al.  Geomagnetic storm predictions from solar wind data with the use of dynamic neural networks , 1997 .

[22]  Geomagnetic storm forecasting service StormFocus: 5 years online , 2018, 1802.06590.

[23]  Stephen A. Billings,et al.  A novel logistic-NARX model as a classifier for dynamic binary classification , 2017, Neural Computing and Applications.

[24]  Giuseppe Consolini,et al.  Comparison between three algorithms for Dst predictions over the 2003-2005 period , 2008 .

[25]  Sheng Chen,et al.  Representations of non-linear systems: the NARMAX model , 1989 .

[26]  Chen Jie,et al.  Wind speed forecasting using nonlinear-learning ensemble of deep learning time series prediction and extremal optimization , 2018, Energy Conversion and Management.

[27]  Xinlin Li,et al.  A new model for the prediction of Dst on the basis of the solar wind , 2002 .

[28]  S. Billings.  Nonlinear System Identification: NARMAX Methods in the Time, Frequency, and Spatio-Temporal Domains , 2013 .

[29]  S. A. Billings,et al.  Using the NARMAX approach to model the evolution of energetic electrons fluxes at geostationary orbit , 2011 .

[30]  S. Billings,et al.  Prediction of the Dst index using multiresolution wavelet models , 2004 .

[31]  M. X. Zhao,et al.  Support Vector Machine combined with Distance Correlation learning for Dst forecasting during intense geomagnetic storms , 2016 .

[32]  Steve A. Billings,et al.  Term and variable selection for non-linear system identification , 2004 .