Solar PV forecasting spells gains for the grid

Solar energy is expected to become the best low-cost solution for generating much of Australia’s future electricity. But as sunlight is available only intermittently, electricity grid operations are vulnerable to variable power-supply issues. Being able to forecast solar energy production over time is important if we’re to stabilise the energy grid, decrease operational costs and achieve high penetration of distributed solar energy production.

Aiming to solve the problem of intermittency, a team of Australian and US researchers is using inexpensive image and radiation sensors to develop low-cost forecasting tools.

The team, which includes researchers from NICTA, ARENA and ANU as well as US energy experts, is collecting and analysing real-time data from rooftop solar PVs within a suburb-sized region of Canberra that spans a range of environmental conditions.

NICTA’s Machine Learning Research Group is developing tools to aid in the collection of this data: rooftop data loggers that measure the real energy production of inverters on the solar panel systems and feed this information into a central analysis platform at five-minute intervals; and a network of low-cost 360° ‘skycameras’ to record cloudburst activity – real-time visual data that enables more accurate estimations of cloud characteristics such as location, motion and opacity when matched to the PV systems.

Data samples from the loggers and cameras are entered into powerful machine-learning algorithms that can predict the energy output from solar panels across Canberra – or, potentially, any region – over time.

The new tools will enable more accurate forecasting of intermittency in solar generation, enabling grid operators to better manage grid fluctuations and implement alternatives in times of lower generation.

In turn, this will simplify the widespread uptake of solar photovoltaics, lower the costs of finance for investment in solar power and boost the market value of solar energy generation capacity.

For more information, go to and  


Error | AgInnovators


The website encountered an unexpected error. Please try again later.