Enhanced Estimation of Autoregressive Wind Power Prediction Model Using Constriction Factor Particle Swarm Optimization

TL;DR

This paper improves wind power forecasting accuracy by optimizing autoregressive model parameters using Constriction Factor Particle Swarm Optimization, validated on real-world data and outperforming traditional methods.

Contribution

It introduces a novel application of CF-PSO to optimize AR model parameters for wind power prediction, enhancing accuracy over existing approaches.

Findings

Significant improvement in forecasting accuracy over benchmark methods.

Effective selection of lag order using Akaike information criterion.

Validated results on real wind farm data from Australia.

Abstract

Accurate forecasting is important for cost-effective and efficient monitoring and control of the renewable energy based power generation. Wind based power is one of the most difficult energy to predict accurately, due to the widely varying and unpredictable nature of wind energy. Although Autoregressive (AR) techniques have been widely used to create wind power models, they have shown limited accuracy in forecasting, as well as difficulty in determining the correct parameters for an optimized AR model. In this paper, Constriction Factor Particle Swarm Optimization (CF-PSO) is employed to optimally determine the parameters of an Autoregressive (AR) model for accurate prediction of the wind power output behaviour. Appropriate lag order of the proposed model is selected based on Akaike information criterion. The performance of the proposed PSO based AR model is compared with four well-established approaches; Forward-backward approach, Geometric lattice approach, Least-squares approach and Yule-Walker approach, that are widely used for error minimization of the AR model. To validate the proposed approach, real-life wind power data of \textit{Capital Wind Farm} was obtained from Australian Energy Market Operator. Experimental evaluation based on a number of different datasets demonstrate that the performance of the AR model is significantly improved compared with benchmark methods.