Vol.:(0123456789)
1 3
Journal of Ambient Intelligence and Humanized Computing
https://doi.org/10.1007/s12652-018-0886-0
ORIGINAL RESEARCH
Novel forecasting model based on improved wavelet transform,
informative feature selection, and hybrid support vector machine
on wind power forecasting
Zhenling Liu
1,2
· Mahdi Hajiali
3
· Amirhosein Torabi
4
· Bahman Ahmadi
5
· Rolando Simoes
6
Received: 25 October 2017 / Accepted: 3 May 2018
© Springer-Verlag GmbH Germany, part of Springer Nature 2018
Abstract
Wind speed/power prediction plays an important role in large-scale wind power penetration because of the wind volatility
and uncertainty. In this paper, an accurate forecast model is presented based on improved wavelet transform, informative
feature selection and hybrid forecast engine. The proposed forecasting engine is based on support vector machine which
is an appropriate prediction forecast engine due to its ability to discover natural structures of wind speed/power variation.
The mentioned forecast engine is equipped with an intelligent algorithm and enhances its prediction accuracy. For this
purpose, we applied a new version of enhanced particle swarm optimization in this work as the optimization algorithm.
Effectiveness of the proposed forecast model is extensively evaluated by real-world electricity market through comparison
with well-known forecasting methods. Obtained numerical results and analysis demonstrate the validity and superiority of
the proposed method.
Keywords Improved wavelet transform · Informative feature selection · ISVM · E-PSO · Wind power
1 Introduction
1.1 Problem definition
Increasing awareness of worldwide climate change and its
prospective effects on our environment, economy, as well as
on welfare in general, has caused a re-think in many fields of
energy politics. Regarding the fact that the power generation
is responsible for momentous emissions of greenhouse
gases, decreasing the negative effects of such energy sources
is the priority in current global energy policy (Abedinia et al.
2018; Liu et al. 2017). Accordingly, one of the most impor-
tant strategies to decrease the power generation’s negative
effects is using the renewable energy sources such as wind
power. The electrical power system is one of the biggest and
most complex engineering systems, where its operation has
always been essential and also a very big problem for human
life (Ghadimi et al. 2017a, b; Mirzapour et al. 2017).
Since the generation means of wind power is different
from the conventional power plants in terms of characteris-
tics and the output variability. These impose several draw-
backs and problems to an independent system operator (ISO)
that should deal with balance and security of the network.
Furthermore, an accurate forecast of wind power can play a
serious role in power system operation (Aghajani and Ghad-
imi 2018). For this purpose, an accurate prediction model
for wind power is considered by researchers in recent years.
1.2 Literature review
Several forecasting methods have been proposed by research-
ers in recent years. In Ghadimi et al. (2017a, b), the authors
* Zhenling Liu
liuzhenling1858@126.com
1
School of Management, Henan University of Technology,
Zhengzhou, Henan 450001, China
2
College of Engineering, Winona State University, Winona,
MN 55987, USA
3
Department of Computer Engineering and Information
Technology, Amirkabir University of Technology, Tehran,
Iran
4
Department of Electrical Engineering, Faculty
of Engineering, University of Isfahan, Esfahān, Iran
5
Department of Electrical Engineering, Istanbul Technical
University, Istanbul, Turkey
6
Northeastern University, Boston, MA 082715, USA
@Phil_Robichaud
@deepthiw
@JoseServera