Energy AI

In operating Smart Grid, AI and machine learning become more and more important due to the increasing number of renewable generation, such as solar and wind, and the distributed energy resources (DERs). We apply deep learning and reinforcement learning for prediction and control.

Deep Learning for Short-term Load Forecasting

We are applying deep learning techniques for short-term load forecasting from residential households to a large scale customers. Specifically, we apply deep neural network (DNN), long short-term memory (LSTM), convolutional neural network (CNN), residual network (ResNet) and improve the forecasting accuracy substantially.

Funded by KEPCO

Deep Learning for PV Generation Forecasting

Photovoltaic (PV) generation is essential to reduce the carbon footprint and sustainable energy. However, its generation is fluctuating due to weather conditions including cloud movement. We are applying the state-of-the art CNN technique to collectively predict the multi-site PV generation and significantly improve forecasting accuracy.

Funded by KEPCO and NRF

Deep Learning for Energy Data Analytics

Due to the increasing number of smart meters, energy big data needs be effectively processed. For better analysis of customers' behavior, design of new tariff, missing data imputation, etc., deep learning techniques, such as convolutional autoencoder and variational autoencoder, can be utilized and combined with clustering.

Funded by KEPCO and NRF

Deep Reinforcement Learning

Deep reinforcement learning (DRL) can be used to control various components of Smart Grid including ESS, microgrids, active distribution networks. Furthermore, prediction and control are no longer separate but collectively considered using reinforcement learning framework. Please keep an eye on us innovating DRL.

Funded by NRF