Random Forest Ensemble of Support Vector Regression Models for Solar Power Forecasting

📝 Abstract

To mitigate the uncertainty of variable renewable resources, two off-the-shelf machine learning tools are deployed to forecast the solar power output of a solar photovoltaic system. The support vector machines generate the forecasts and the random forest acts as an ensemble learning method to combine the forecasts. The common ensemble technique in wind and solar power forecasting is the blending of meteorological data from several sources. In this study though, the present and the past solar power forecasts from several models, as well as the associated meteorological data, are incorporated into the random forest to combine and improve the accuracy of the day-ahead solar power forecasts. The performance of the combined model is evaluated over the entire year and compared with other combining techniques.

💡 Analysis

To mitigate the uncertainty of variable renewable resources, two off-the-shelf machine learning tools are deployed to forecast the solar power output of a solar photovoltaic system. The support vector machines generate the forecasts and the random forest acts as an ensemble learning method to combine the forecasts. The common ensemble technique in wind and solar power forecasting is the blending of meteorological data from several sources. In this study though, the present and the past solar power forecasts from several models, as well as the associated meteorological data, are incorporated into the random forest to combine and improve the accuracy of the day-ahead solar power forecasts. The performance of the combined model is evaluated over the entire year and compared with other combining techniques.

📄 Content

Note: This is a pre-print of the full paper that published in Innovative Smart Grid Technologies, North America Conference, 2017, which can be referenced as below: M. Abuella and B. Chowdhury, “Random Forest Ensemble of Support Vector Regression for Solar Power Forecasting,” in Proceedings of Innovative Smart Grid Technologies, North America Conference, 2017. Random Forest Ensemble of Support Vector Regression Models for Solar Power Forecasting

Mohamed Abuella, Student Member, IEEE Energy Production and Infrastructure Center Department of Electrical and Computer Engineering University of North Carolina at Charlotte Charlotte, USA Email: mabuella@uncc.edu Badrul Chowdhury, Senior Member, IEEE Energy Production and Infrastructure Center
Department of Electrical and Computer Engineering University of North Carolina at Charlotte Charlotte, USA Email: b.chowdhury@uncc.edu

Abstract— To mitigate the uncertainty of variable renewable resources, two off-the-shelf machine learning tools are deployed to forecast the solar power output of a solar photovoltaic system. The support vector machines generate the forecasts and the random forest acts as an ensemble learning method to combine the forecasts. The common ensemble technique in wind and solar power forecasting is the blending of meteorological data from several sources. In this study though, the present and the past solar power forecasts from several models, as well as the associated meteorological data, are incorporated into the random forest to combine and improve the accuracy of the day-ahead solar power forecasts. The performance of the combined model is evaluated over the entire year and compared with other combining techniques.
Keywords—Ensemble learning, post-processing, random forest, solar power, support vector regression. I. INTRODUCTION
The wind and solar energy resources have created operational challenges for the electric power grid due to the uncertainty involved in their output in the short term. The intermittency of these resources may adversely affect the operation of the electric grid when the penetration levels of these variable generations are high. Thus, wherever the variable generation resources are used, it becomes highly desirable to maintain higher than normal operating reserves and efficient energy storage systems to manage the power balance in the system. The operating reserves that use fossil fuel generating units should be kept to a minimum in order to get the maximum benefit from the deployment of the renewable resources. Therefore, the forecast of these variable generations becomes a vital tool in the operation of the power systems and electricity markets [1]. As in wind power forecasting, the solar power also consists of a variety of methods based on the time horizon being forecasted, the data available to the forecaster and the particular application of the forecast. The methods are broadly categorized according to the time horizon in which they generally show value. Methods that are common in solar power forecasting include Numerical Weather Prediction (NWP) and Model Output Statistics (MOS) to produce forecasts, as well as hybrid techniques that combine ensemble forecasts and Statistical Learning Methods [2]. Applying machine learning techniques directly to historical time series of solar photovoltaic (PV) production associated with NWP outcomes have placed among the top models in the recent global competition of energy forecasting, GEFCom2014 [3]. Just to name a few of the machine learning tools, the artificial neural networks (ANN) and support vector regression (SVR), gradient boosting (GB), random forest (RF), etc. are believed to be the most common. Hybrid models of two or more statistical and physical techniques are often combined to capture complex interactions and provide useful insights and better forecasts. In ref. [4], the authors implement a hybrid model that consists of ARMA and ANN to forecast the solar irradiance by NWP data for 5 locations of a Mediterranean climate. They found the proposed model outperforms the naïve persistence model and improvement with respect to its core techniques as well. The study reported in ref. [5] presents the benefits of combining the data of solar irradiance that is derived from a satellite with ground-measured data to improve the intraday forecasts in the range up to six hours ahead. In ref [6], the authors combine satellite images with ANN outcomes to forecast the solar irradiance of leading time up to two hours for two sites in California.
In ref. [7], several statistical combining methods are used to combine multiple linear regression models for load forecasting, and the authors conclude that the regression combining technique is the best. While ref. [8] uses several statistical models to forecast the hourly PV electricity production for the next day at some power plants in France, t

This content is AI-processed based on ArXiv data.