Solar Generation Forecasting

January 31st, 2015

More and more Australians are relying on the benefits of solar power, however the impact to electricity grid stability needs to be determined.

Further uptake of solar panels is at risk as electricity grid operators grapple with the instability caused by overflow solar energy entering the grid.

Data61’s Solar Generation Forecasting project is using sensor data from solar panels on residential houses across the city of Canberra to predict future energy production. The ability to predict energy production is the first step in enabling power grid operators to be able to ensure the quality and control of electricity supplies in an environment of greater solar panel usage.

Project Background

Seeking to solve the problem of the unpredictability of solar generation, Data61 is installing data loggers on residential solar systems across Canberra. ‘Skycameras’ are also being set up on the top of buildings to capture information about cloudbursts.

Energy production by solar panels is intermittent when it is cloudy. Weather satellite images are not sufficient to predict solar energy output to the grid at a suburb or even finer locale level.

The data loggers will measure the real energy production of the inverter on the solar panel systems and send this information back to a central analysis platform every five minutes.

Data61 developed skycameras to record images of cloudbursts which is key to determining impacts of clouds on generation.

What is the technology?

Data samples collected from the loggers and skycameras are entered into powerful machine learning algorithms that will predict the energy output from solar panels across the entire city of Canberra over time.

By receiving information about the amount of energy distributed solar systems are giving back to the grid, grid operators will have the necessary data to put measures in place to control fluctuations on the grid as solar panel uptake increases. Ideally data from the system will be made available in real-time to all interested energy market participants.

Data61’s unique approach

Data61 is attempting to create a highly granular prediction capability for solar energy generation in Australia which will allow grid operators to better forecast the intermittency of solar
generation and plan for alternatives in times of lower generation.

An example of Data61 Cloud imagery

An example of Data61 Cloud imagery



Australian National University

Download the Solar Generation Forecasting brochure.

Our first project is launching in Canberra – and we’d like you to help! If you have a home solar PV system, you can sign up via the Solar Monitoring Portal if you’re interested.