Modified Auto Regressive Technique for Univariate Time Series Prediction\n of Solar Irradiance

2020/12/06 by Umar Marikkar, Ahmed S. Hassan, Marikkar, Umar +9
Computer Science · Engineering · Energy · #Solar Radiation and Photovoltaics #Energy Load and Power Forecasting #Photovoltaic System Optimization Techniques

paper · pdf · doi:10.48550/arxiv.2012.03215

Abstract

The integration of renewable resources has increased in power generation as a\nmeans to reduce the fossil fuel usage and mitigate its adverse effects on the\nenvironment. However, renewables like solar energy are stochastic in nature due\nto its high dependency on weather patterns. This uncertainty vastly diminishes\nthe benefit of solar panel integration and increases the operating costs due to\nlarger energy reserve requirement. To address this issue, a Modified Auto\nRegressive model, a Convolutional Neural Network and a Long Short Term Memory\nneural network that can accurately predict the solar irradiance are proposed.\nThe proposed techniques are compared against each other by means of multiple\nerror metrics of validation. The Modified Auto Regressive model has a mean\nabsolute percentage error of 14.2%, 19.9% and 22.4% for 10 minute, 30 minute\nand 1 hour prediction horizons. Therefore, the Modified Auto Regressive model\nis proposed as the most robust method, assimilating the state of the art neural\nnetworks for the solar forecasting problem.\n

Related

• Evaluation of Time-Series, Regression and Neural Network Models for Solar Forecasting: Part I: One-Hour Horizon
• Solar radiation forecasting using ad-hoc time series preprocessing and neural networks
• An Integrated Multi-Time-Scale Modeling for Solar Irradiance Forecasting\n Using Deep Learning
• Data-driven forecasting of solar irradiance
• Short term solar energy prediction by machine learning algorithms