Primary supervisorHao Wang
Electricity is an essential part of modern life and the economy. Driven by a combination of policy support and rapidly falling costs of low-carbon technologies, Australia is experiencing a sharp rise in the deployment of distributed energy sources (DERs). Typical DERs include wind, solar photovoltaics (PV), battery storage, and electric vehicles (EVs) on the consumer side.
DERs on the consumer side are becoming smart and able to respond to the system variables, such as prices and availability of sustainable energy generation. Such interactions between the grid and consumers are not captured in the current energy system planning and operation. Effective modelling of DER behaviour in the energy system is a key step for capturing its impact on the energy system operation and planning. An effective approximation is needed to capture such aggregate DER behaviours, and deep learning is a promising solution.
This project aims to leverage deep learning to develop an approximate model capturing the aggregate behaviours of DERs against dynamical energy environments. Using data generated from bottom-up models and/or simulators, this project will
- develop a deep learning model to learn the DERs’ response under various settings, including different system parameters, penetration of solar PV, and other DERs;
- evaluate the gap between the learned response and the optimised response.
 Neel Guha, Zhecheng Wang, Arun Majumdar, Machine Learning for AC Optimal Power Flow, ICML 2019 Climate Change Workshop.
 Ling Zhang, Baosen Zhang, Learning to Solve the AC Optimal Power Flow via a Lagrangian Approach, 2021.
Programming skills in Python (required); Machine Learning and deep learning in particular (preferred); time series data (preferred).