SolarGAN: Multivariate Solar Data Imputation Using Generative Adversarial Network

Photovoltaic (PV) is receiving increasing attention due to its sustainability and low carbon footprint. However, the penetration level of PV is still relatively low because of its intermittency. This uncertainty can be handled by accurate PV forecasting, which requires high-quality solar data. Nevertheless, up to 40% of solar data can be found missing, which significantly worsens the quality of solar data. This letter proposes a novel solarGAN method for multivariate solar data imputation, in which necessary modifications are made on the input of generative adversarial network (GAN) to effectively tackle the relatively independent solar time series data. Case studies on a public dataset show that the proposed solarGAN outperforms several commonly-used machine learning and GAN based data imputation methods with at least 23.9% reduction of mean squared error.

[1]  P. Lucio,et al.  Data imputation analysis for Cosmic Rays time series , 2017 .

[2]  Rob J Hyndman,et al.  Probabilistic energy forecasting: Global Energy Forecasting Competition 2014 and beyond , 2016 .

[3]  Lei Li,et al.  Multi-Matrices Factorization with Application to Missing Sensor Data Imputation , 2013, Sensors.

[4]  Yajie Zou,et al.  Flexible and Robust Method for Missing Loop Detector Data Imputation , 2015 .

[5]  et al.,et al.  Missing Data Imputation in the Electronic Health Record Using Deeply Learned Autoencoders , 2017, PSB.

[6]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[7]  Trevor Hastie,et al.  Imputing Missing Data for Gene Expression Arrays , 2001 .

[8]  Xiang Wei,et al.  Improving the Improved Training of Wasserstein GANs: A Consistency Term and Its Dual Effect , 2018, ICLR.

[9]  Guido Carpinelli,et al.  A Probabilistic Competitive Ensemble Method for Short-Term Photovoltaic Power Forecasting , 2017, IEEE Transactions on Sustainable Energy.

[10]  Yehuda Koren,et al.  Matrix Factorization Techniques for Recommender Systems , 2009, Computer.

[11]  José Luís Calvo-Rolle,et al.  Missing Data Imputation of Solar Radiation Data under Different Atmospheric Conditions , 2014, Sensors.

[12]  Mihaela van der Schaar,et al.  GAIN: Missing Data Imputation using Generative Adversarial Nets , 2018, ICML.

[13]  Léon Bottou,et al.  Wasserstein Generative Adversarial Networks , 2017, ICML.

[14]  Mehran Amiri,et al.  Missing data imputation using fuzzy-rough methods , 2016, Neurocomputing.

[15]  Ying Zhang,et al.  Multivariate Time Series Imputation with Generative Adversarial Networks , 2018, NeurIPS.

[16]  Stef van Buuren,et al.  MICE: Multivariate Imputation by Chained Equations in R , 2011 .

[17]  Jie Zhang,et al.  A Methodology for Quantifying Reliability Benefits From Improved Solar Power Forecasting in Multi-Timescale Power System Operations , 2018, IEEE Transactions on Smart Grid.

[18]  T. Dudok de Wit,et al.  A method for filling gaps in solar irradiance and solar proxy data , 2011, 1107.4253.

[19]  Yang Wang,et al.  Exploring Key Weather Factors From Analytical Modeling Toward Improved Solar Power Forecasting , 2019, IEEE Transactions on Smart Grid.

[20]  Bri-Mathias Hodge,et al.  Unsupervised Clustering-Based Short-Term Solar Forecasting , 2019, IEEE Transactions on Sustainable Energy.