Deep-based conditional probability density function forecasting of residential loads

Afrasiabi, Mousa and Mohammadi, Mohammad and Rastegar, Mohammad and Stankovic, Lina and Afrasiabi, Shahabodin and Khazaei, Mohammad (2020) Deep-based conditional probability density function forecasting of residential loads. IEEE Transactions on Smart Grid, 11 (4). pp. 3746-3757. 8988175. ISSN 1949-3053 (https://doi.org/10.1109/TSG.2020.2972513)

[thumbnail of Afrasiabi-etal-IEEE-2020-Deep-based-conditional-probabiity-density-function-forecasting]
Preview
Text. Filename: Afrasiabi_etal_IEEE_2020_Deep_based_conditional_probabiity_density_function_forecasting.pdf
Accepted Author Manuscript

Download (6MB)| Preview

Abstract

This paper proposes a direct model for conditional probability density forecasting of residential loads, based on a deep mixture network. Probabilistic residential load forecasting can provide comprehensive information about future uncertain-ties in demand. An end-to-end composite model comprising convolution neural networks (CNNs) and gated recurrent unit (GRU) is designed for probabilistic residential load forecasting. Then, the designed deep model is merged into a mixture density network (MDN) to directly predict probability density functions (PDFs). In addition, several techniques, including adversarial training, are presented to formulate a new loss function in the direct probabilistic residential load forecasting (PRLF) model. Several state-of-the-art deep and shallow forecasting models are also presented in order to compare the results. Furthermore, the effectiveness of the proposed deep mixture model in characterizing predicted PDFs is demonstrated through comparison with kernel density estimation, Monte Carlo dropout, a combined probabilistic load forecasting method and the proposed MDN without adversarial training