Sensor network based solar forecasting using a local vector autoregressive ridge framework

The significant improvements and falling costs of photovoltaic (PV) technology make solar energy a promising resource, yet the cloud induced variability of surface solar irradiance inhibits its effective use in grid-tied PV generation. Short-term irradiance forecasting, especially on the minute scale, is critically important for grid system stability and auxiliary power source management. Compared to the trending sky imaging devices, irradiance sensors are inexpensive and easy to deploy but related forecasting methods have not been well researched. The prominent challenge of applying classic time series models on a network of irradiance sensors is to address their varying spatio-temporal correlations due to local changes in cloud conditions. We propose a local vector autoregressive framework with ridge regularization to forecast irradiance without explicitly determining the wind field or cloud movement. By using local training data, our learned forecast model is adaptive to local cloud conditions and by using regularization, we overcome the risk of overfitting from the limited training data. Our systematic experimental results showed an average of 19.7% RMSE and 20.2% MAE improvement over the benchmark Persistent Model for 1-5 minute forecasts on a comprehensive 25-day dataset.

[1]  Vahan Gevorgian,et al.  Review of PREPA Technical Requirements for Interconnecting Wind and Solar Generation , 2013 .

[2]  J. Kleissl,et al.  Embedded nowcasting method using cloud speed persistence for a photovoltaic power plant , 2015 .

[3]  Iain MacGill,et al.  The potential role of forecasting for integrating solar generation into the Australian National Electricity Market , 2010 .

[4]  Carlos F.M. Coimbra,et al.  Cloud-tracking methodology for intra-hour DNI forecasting , 2014 .

[5]  T. Soubdhan,et al.  A benchmarking of machine learning techniques for solar radiation forecasting in an insular context , 2015 .

[6]  M. Diagne,et al.  Review of solar irradiance forecasting methods and a proposition for small-scale insular grids , 2013 .

[7]  Gordon Reikard.  Predicting solar radiation at high resolutions: A comparison of time series forecasts , 2009 .

[8]  Dazhi Yang,et al.  Very short term irradiance forecasting using the lasso , 2015 .

[9]  Jin Xu,et al.  Solar irradiance forecasting using multi-layer cloud tracking and numerical weather prediction , 2015, SAC.

[10]  Jan Kleissl,et al.  Deriving Cloud Velocity From an Array of Solar Radiation Measurements , 2012 .

[11]  Mark Z. Jacobson,et al.  100% clean and renewable wind, water, and sunlight (WWS) all-sector energy roadmaps for the 50 United States , 2015 .

[12]  Aoife Foley,et al.  Current methods and advances in forecasting of wind power generation , 2012 .

[13]  Laura M. Hinkelman.  Differences between along-wind and cross-wind solar irradiance variability on small spatial scales , 2013 .

[14]  Jin Xu,et al.  A Stochastic Framework for Solar Irradiance Forecasting Using Condition Random Field , 2015, PAKDD.