By 2040 the requirement for global electricity generation is expected to rise by 45%: renewable energies will play a key role to satisfy the energy demand. This project aims to develop deep learning models correcting the bias in NWP forecasts to support smart energy systems by using multivariate time series, sky images and atmospheric parameters to build an xAI model.

Variable Renewable Energy (VRE) e.g., photovoltaics and wind are the fastest growing resources with contributory shares expected to rise to 15% by 2040 and more than half arising from wind. Yet, renewable energy resources must be explored for their feasibility in meeting the constant supply requirement in order to eliminate any imbalance between supply and demand that can disproportionately increase financial risks to energy utilities and the supply risk to consumers.

Numerical Weather Prediction (NWP) models use forecasts and availability of wind and solar energies in the future (e.g. inter-daily scales) or monitor very short-term supply (e.g. hourly scales) ensuring that the energy utilities can maintain their businesses while the consumers remain satisfied. However, NWP models have significant uncertainties in their global climate model parametrisations that can also cause major errors in the meteorological variables fed into solar and energy prediction systems. Incorrectly predicted solar and wind energy causes power outrage, reducing the confidence in meeting energy demand and security.

The project will develop advanced analytics with big data and deep learning to operate the whole energy fleet sustainably, based on demand, supply and price the forecasting of energy resources in a robust informatics platform. Here, Optimal = Reliable + Economical + Sustainable. Integration of VREs represented by frontiers offer the highest return for a defined level of risk or lowest risk for a given return. Deep learning will use its capability to establish patterns, trends and fluctuations in NWP variables in respect to a demand load, producing pareto optimal solution. The project will use machine learning to build xAI models for bias correction in NWP variables, that can increase confidence in solar and wind energy forecasts.

The student will learn industry-standard Python, r or MATLAB programming, extensive data analysis and explainable AI modelling tool. The project suits students with interest in renewable energy, mathematics, atmospheric science, computing and environmental science.

The research project is strongly centred on AI and Renewable Energy Technologies. The project may be undertaken with Queensland-based energy company and an industry mentor.

This project is suitable for PhD, Research or Coursework Masters Thesis. It provides opportunity to publish in high quality Q1 journals.

The research student will be part of the Advanced Data Analytics Research Group under Prof Ravinesh Deo.

For more details see https://eportfolio.usq.edu.au/view/view.php?id=1

Professor Ravinesh Deo

• Assoc Prof Nathan Downs
• Professor Jeffrey Soar

• Institute for Agriculture and the Environment
• School of Agricultural and Environmental Sciences

• Artificial Intelligence and Image Processing
• Atmospheric Sciences

• Doctor of Philosophy (DPHD)
• Doctor of Philosophy (DPHD)
• Master of Research (MRES)
• Master of Research (MRES)