TIME SERIES WIND SPED PREDICTION WITH ENERGY MAPPING USING HYBRID INTRINSIC MODE FUNCTION (IMF) AND EXTENDED INPUT NEURAL NETWORK (EINN)

Authors

  • S. M. Lawan Department of Electrical Engineering, Kano University of Sci.& Tech. Kano
  • W. A. W. Z. Abidin Department of Electrical and Electronics Engineering, Universiti Malaysia Sarawak
  • T. Masri Department of Electrical and Electronics Engineering, Universiti Malaysia Sarawak
  • F. A. Umari Department of Electrical Engineering, Kano University of Sci.& Tech. Kano
  • A.Y. Abdullaahi Department of Electrical Engineering, Kano University of Sci.& Tech. Kano
  • S. J. Kawu Department of Mechanical Engineering, Baze University Abuja

Keywords:

Artificial Neural Network (ANN), Wind Speed, Intrinsic Mode Function (IMF) and Extended Input Neural Network (EINN), Sarawak

Abstract

Accurate and precise wind speed predictions are a prerequisite requirement that is
necessary before siting of wind turbines. The output power of wind energy system is
completely depends on the behavior of wind speed; a small deviation of wind speed will
lead to large energy losses. This paper presents a new technique for predicting the wind
speed based on hybrid model Intrinsic Mode Function (IMF) and Extended Input Neural
Network (EINN) in the regions where there are limited wind stations. In the first instant,
the important parameters for training the artificial neural network (ANN) are acquired
using the principal component correlation analysis and wind speed signal decomposition,
these parameters used as inputs to the ENN. To illustrate the trend and seasonal factor in
the wind speed time series, the data are decomposed into six empirical time series IMF, the
nonlinear and non- stationary characteristic of wind speed is handled by empirical mode
decomposition (EMD) and EINN respectively. The final predicted values are obtained by
summing all the individual prediction sub models. Wind speed data observed in the
existing wind stations in Sarawak for a period of 1 year from 2015 to 2016 were used for
the simulation. The model implementation confirmed that the proposed model is robust and
capable compared to auto-regression integrated moving average (ARIMA) method.

Downloads

Published

2019-09-25

Issue

Section

Articles