Load reliability analysis using ARMA wind speed modeling

Abstract

Due to the continuous increase in power demand and fast technological improvements and developments, the future of power systems is expected to be a challenging issue. The integration of new sources and technologies are needed to satisfy the utility requests and to decrease the use of fossil fuel sources. One of the main challenging issues in integrating wind power sources is how to forecast the long-term wind speed and study its impact on the reliability of the load. In this paper, the AutoRegression and Moving Average (ARMA) model is used to forecast the wind speed for Dhahran area in Saudi Arabia. Then, the wind power is simulated using the speed-power curve. Finally, the impact of the forecasted wind power on the reliability of residential, commercial, and industrial loads is assessed using Monte Carlo simulation.Index Terms--wind speed, forecasting, ARMA, wind power, reliability.

Impact of wind speed modelling on the predictive reliability assessment of wind‐based microgrids

“…With these two methods, the order of the ARMA series is specified, and the Akaike information criterion (AIC) is used to measure the GOF of the model. When modelling and forecasting the wind speed and power using an ARMA time series, the following steps are performed [42]: i. Hourly wind speed data are collected.…”

“…When modelling and forecasting the wind speed and power using an ARMA time series, the following steps are performed [42]: Hourly wind speed data are collected. The best order for the ARMA model is found using the autocorrelation function (ACF) and partial ACF. The two methods will determine the MA parameter ( q ) and AR parameter ( p ) values. The ARMA model is used to find the best time‐series model for forecasting the wind speed depending on the p and q values. The forecasted wind speed and speed–power curve for a WTG are used to simulate the forecasted wind power. …”

Modelling and simulation of wind flow: A gradient method of identifying windy region

“…The diagnostic wind model is more accurate than the predictive wind model because it is based on the existing wind data (Cox et al, 2003), but cannot extrapolate average wind speed as CALMET (Wang et al, 2008), WindNinja (Quill et al, 2019), elliptic differential equation (Moussiopoulos et al, 1988), the mass consistent model, MCSCIPUF (Cox et al, 2003) and the stationary wind field and turbulence, SWIFT (Cox et al, 2003). On the other hand, prognostic wind models are inaccurate as the prediction is based on probability density functions like Markov function (Bizrah and AlMuhaini, 2015; Hocaoğlu et al, 2010; Tang et al, 2015), Monte Carlo function (Koivisto et al, 2016), Bayesian function (Galanis et al, 2017), normal Weibull and mixed Weibull functions (Gómez-Lázaro et al, 2016; Kollu et al, 2012; Onoruoiza et al, 2022), 2-parameter Gamma function (Pobočíková et al, 2017; Touré, 2019), 2-parameter lognormal (Pobočíková et al, 2017), the operational multiscale environment model with grid adaptivity, OMEGA (Cox et al, 2003) and RAMS (Cox et al, 2003; Thomas et al, 2014).…”