Gao, Yuan

Energy Analysis

Deep Learning, Reinforcement Learning, Time series prediction

Associate Professor

Research Thrust : Multiscale Science and Engineering for Energy and the Environment Thrust
E-mail :
Website : https://hyoka.ofc.kyushu-u.ac.jp/search/details/K008675/english.html

Research Outline

Reinforcement learning-based optimization of renewable energy operation

Develop reinforcement learning control methods suitable for renewable energy systems, especially the design of reward functions and the processing of mixed action spaces to get the optimal operational policy.

Interpretable time series prediction using deep learning models in renewable energy

Use historical data to predict future wind power and solar power, and improve the interpretability of the model through the model structure.

Time series fault diagnosis based on convolutional neural network

By turning time series into pictures, convolutional neural networks are used for classification, and at the same time, the model is lightweighted through channel pruning.

Research Methods and Facilities

Times series processing

Through the deep learning model, I can make full use of historical data to predict the future, and perform interpolation completion and fault diagnosis.

Decision making: Sequence control

As long as it involves controlling temperature in the time dimension, I can use reinforcement learning and model predictive control to make decisions, especially for control problems in renewable energy equipment and systems.

I²CNER Research Seeds