Incorporating Spatial and Temporal Correlations to Improve Aggregation of Decentralized Day-Ahead Wind Power Forecasts

Abstract

In some electricity markets, individual wind farms are obliged to provide point forecasts to the power purchaser or system operator. These decentralized forecasts are usually based on on-site meteorological forecasts and measurements, and thus optimized for local conditions. Simply adding decentralized forecasts may not capture some of the spatial and temporal correlations of wind power, thereby lowering the potential accuracy of the aggregated forecast. This paper proposes the explanatory variables that are used to train the kernel density estimator and conditional kernel density estimator models to derive dayahead aggregated point and probabilistic wind power forecasts from decentralized point forecasts of geographically distributed wind farms. The proposed explanatory variables include (a) decentralized point forecasts clustered using the clustering large applications algorithm to reduce the high-dimensional matrices, (b) hour of day and month of year to account for diurnal and seasonal cycles, respectively, and (c) atmospheric states derived from self-organizing maps to represent large-scale synoptic circulation climatology for a study area. The proposed methodology is tested using the day-ahead point forecast data obtained from 29 wind farms in South Africa. The results from the proposed methodology show a significant improvement as compared to simply adding the decentralized point forecasts. The derived predictive densities are shown to be non-Gaussian and time-varying, as expected given the time-varying nature of wind uncertainty. The proposed methodology provides system operators with a method of not only producing more accurate aggregated forecasts from decentralized forecasts, but also improving operational decisions such as dynamic operating reserve allocation and stochastic unit commitment.INDEX TERMS Aggregated wind power forecasting; diurnality; large-scale atmospheric circulations; probabilistic ; seasonality

K-Means and Alternative Clustering Methods in Modern Power Systems

“…As power usage data often contain noise and outliers due to varying factors such as weather conditions, time of day, or unexpected appliance use, PAM's robustness to such irregularities makes it an ideal choice for this application [303], [304]. efficiently identify unique consumption patterns by sampling and processing smaller subsets of the large dataset [306], [307], [308].…”

A Review of Modern Wind Power Generation Forecasting Technologies

“…In order to further improve the accuracy of short-term wind power forecasting, kernel density estimation is used to estimate the probability density function of the random variables required for predictive models to avoid the density leakage problem estimated for probabilistic wind power forecasting (WPF) of a region at both the wind farm and regional levels [9][10][11]. Quantile regression (QR) approximates the conditional probability distribution of a random variable by quantiles.…”

“…Due to their high computation intelligence and accuracy, such methods have been widely used in the past few years to improve the accuracy and performance of traditional WPPF models. The machine learning-based wind speed predictions for k-NN and conditional KDE, Adaboost-PSO-ELM, and enhanced bee swarm optimization (EBSO), to perform parameter optimization for least squares support vector machine (LSSVM) [11,[26][27][28][30][31] models, were proposed to identify meaningful training data to reduce the volume of modeling data and improve the computing efficiency. They have good generalization ability and robustness and can provide more accurate wind power forecasting.…”