An Australian quantum computing project has demonstrated that quantum-enhanced processing could improve the accuracy of household energy forecasts, which could give utilities a new tool for managing increasingly complex power grids.
Silicon Quantum Computing (SQC), a government-funded startup, and French electrical giant Schneider Electric have reported an average 20% improvement in next-day household energy forecasting accuracy using SQC’s quantum processor, with improvements reaching as high as 41% in some cases.
The project has received A$3.6 million (about US$2.5 million) from the Australian government through the second stage of its Critical Technologies Challenge Program. The companies are now expanding the project to hundreds of homes across Australia. Researchers from the University of New South Wales are also participating in the effort.
The technology is not intended to replace conventional CPUs and GPUs. Instead, SQC’s approach uses a quantum processor as a specialized accelerator that generates additional features for conventional machine learning models.
And it is not moving the whole workload to the quantum computer. Instead, selected portions of a problem are processed using quantum hardware, while conventional computing systems continue to handle the rest.
At the heart of the project is SQC’s Watermelon quantum processor. The silicon-based processor uses atomic-scale engineering to generate quantum features that can be combined with conventional data-processing techniques.
For energy forecasting, the approach is particularly relevant because power consumption and generation are becoming increasingly difficult to predict. The growth of rooftop solar, residential batteries and electric vehicles means that households can increasingly act as both consumers and producers of electricity.
A home with solar panels, for example, can switch from drawing power from the grid to supplying electricity to it depending on weather conditions, battery levels and household demand.
That variability creates additional challenges for utilities attempting to determine how much electricity will be required hours or days ahead.
During the first phase of the SQC-Schneider project, researchers used Watermelon to analyze 12 months of next-day energy-forecasting data, where it racked up the 20% improvement.
The next stage will provide a more significant test of the technology as researchers expand the trial to hundreds of Australian homes and integrate the quantum system more directly into Schneider Electric’s existing energy-management workflows.
Improved forecasting could allow energy companies to make better decisions about when electricity should be generated, stored or consumed. In a grid with increasing amounts of distributed renewable generation, even incremental improvements in prediction could potentially reduce the need for expensive reserve capacity and improve utilization of batteries and other energy-storage systems.
SQC says its Watermelon technology has already been applied to areas including telecommunications, banking and high-frequency trading. The company’s quantum systems can be accessed through the cloud or deployed as hardware, including in data-center environments.
That model could allow quantum technology to enter commercial applications before fully fault-tolerant, general-purpose quantum computers become available.
The expanded Australian trial will determine whether the forecasting improvements demonstrated in the initial phase can be reproduced at larger scale. If they can, the project could provide another example of quantum computing moving beyond laboratory demonstrations and into practical workloads where conventional computing has struggled to deliver the desired level of prediction accuracy.
