Ensemble Nonlinear Autoregressive Exogenous Artificial Neural Networks for Short-Term Wind Speed and Power Forecasting.

Zhongxian Men,Yongqian Liu,Fue-Sang Lien,Zhiling Yang,Eugene Yee

doi:10.1155/2014/972580

Zhongxian Men, Yongqian Liu + Show 3 more

Open Access

https://doi.org/10.1155/2014/972580

Copy DOI

Abstract

Short-term wind speed and wind power forecasts (for a 72 h period) are obtained using a nonlinear autoregressive exogenous artificial neural network (ANN) methodology which incorporates either numerical weather prediction or high-resolution computational fluid dynamics wind field information as an exogenous input. An ensemble approach is used to combine the predictions from many candidate ANNs in order to provide improved forecasts for wind speed and power, along with the associated uncertainties in these forecasts. More specifically, the ensemble ANN is used to quantify the uncertainties arising from the network weight initialization and from the unknown structure of the ANN. All members forming the ensemble of neural networks were trained using an efficient particle swarm optimization algorithm. The results of the proposed methodology are validated using wind speed and wind power data obtained from an operational wind farm located in Northern China. The assessment demonstrates that this methodology for wind speed and power forecasting generally provides an improvement in predictive skills when compared to the practice of using an “optimal” weight vector from a single ANN while providing additional information in the form of prediction uncertainty bounds.

Highlights

With recent rapid advances in the development of clean energy, wind power has increasingly become an important component of a renewable and sustainable energy system
We proposed an ensemble artificial neural network (ANN) model for wind speed and wind power forecasting based on a nonlinear autoregressive exogenous (NARX)(p, q) model structure
The examples considered suggest that combining forecasts for wind speed and power from the ensemble ANN members using a weighted averaging scheme generally provided predictions with improved forecast accuracy compared to those obtained from a single best set of ANN parameters

Summary

Introduction

With recent rapid advances in the development of clean energy, wind power has increasingly become an important component of a renewable and sustainable energy system. The third methodology for wind speed forecasting utilizes machine learning approaches Within this class of methods, artificial neural network (ANN), fuzzy systems theory, and other related paradigms such as grey predictors and support vector machine (SVM) have been applied. Unlike the more conventional statistical and physicsbased approaches for wind speed forecasting, machine learning approaches do not explicitly formulate a specific parametric model for the process but instead base the analysis on the use of a “black box” or “grey box” formulation of the problem. An attractive alternative approach to using simple or weighted averaging is to apply a Bayesian methodology and average the outputs of many ANN realizations having weights sampled from a posterior distribution (giving essentially a probability weighting) In this context, Blonbou [17] proposed applying an ANN trained using an adaptive Bayesian learning method for very short-term wind power forecasting.

Architecture of ANNs

Empirical Analysis of Wind Speed and

Empirical Analysis of the Censored Wind

Findings

Conclusions