Insider Brief
- WISER and E.ON completed a research collaboration exploring hybrid quantum machine learning approaches for electricity demand forecasting.
- The project evaluated quantum models using an anonymized dataset of 103 residential customers on both simulators and real quantum hardware.
- Researchers found that projected quantum kernel models improved forecasting accuracy against selected classical baselines under near-term quantum computing conditions.
Press release – The Washington Institute for STEM, Entrepreneurship and Research (WISER) announces the successful completion of a research collaboration with E.ON on energy demand forecasting using quantum machine learning. The joint project, published on arXiv, explores two hybrid quantum-classical approaches for forecasting correlated electricity consumption time series, Kernelized Quantum Reservoir Computing with Repeated Measurement (KQRC-RM), and a Projected Quantum Kernel Gaussian Process (QGP).
Industrial use-case problem statement
Imagine you are trying to predict how much electricity 100 different families/ households will use in the coming hours. Some families use more power when it’s cold, others when they watch TV, and their habits often mirror each other.
This research addresses a central challenge in energy systems planning, which is how to forecast demand accurately across multiple correlated customers while operating under realistic near-term quantum hardware constraints. Using an anonymized Smart Meter dataset of 103 residential customers, the team evaluated both models on simulator and real quantum hardware, showing that quantum-enhanced approaches can already be studied experimentally on practical forecasting tasks at meaningful scale.
This is critical for the reliable operation of modern energy systems, including load balancing and renewable integration.
“It’s possible!.. We can now run these complex, multi-output time-series forecasts on real quantum computers with over 100 qubits. While the ‘perfect quantum advantage’ is still waiting for the hardware to get a bit quieter and more reliable, we come very close to it. And in the process establish that hybrid quantum models can outperform specific classical baselines in structured energy forecasting tasks under NISQ constraints. This represents a significant step toward practical quantum-assisted forecasting.” said Vardaan Sahgal, WISER
From Quantum Methods to Energy Forecasts
The project combined two complementary approaches. KQRC-RM used recurrent quantum dynamics and repeated measurements to model temporal structure and cross-stream correlations in smaller customer subsets, while QGP used a more hardware-efficient projected quantum kernel to support multi-output forecasting on larger groups of customers.
In the smaller-scale benchmarking study, the QGP model reduced average MAE relative to a classical multi-output Gaussian Process baseline by 62.01% on simulator and 40.37% on hardware. In the same study, KQRC-RM reduced average MAE relative to an Echo State Network with kernel ridge regression by 36.92% on simulator, while the hardware implementation remained more sensitive to noise.
“Quantum machine learning models that can forecast multiple time-series values has been somewhat elusive in the field, yet classically exists everywhere in industry. We were happy to push the boundaries of hybrid quantum algorithm development to make that happen for a real world use-case and run benchmarks using 100+ qubits on IBM Quantum Computers. ” said Dr. Corey O’Meara, Chief Quantum Scientist, E.ON Digital Technology GmbH
Practical Relevance for Energy Systems
Forecasting electricity demand is challenging due to nonlinear dynamics, multi-scale seasonality, and strong dependencies across correlated series. Classical statistical models often struggle with these nonlinearities, while flexible machine learning approaches typically require substantial data and computational resources.
The study shows that quantum machine learning may become useful for structured multi-output forecasting where relationships between time series matter, especially as quantum hardware improves.
The paper also demonstrates the importance of hardware-aware design. The QGP model scaled to a 100-qubit utility-scale experiment, where 80% of customers fell into low or medium error categories, underscoring both the promise of the approach and the influence of device noise on performance.
The findings are especially relevant for utilities and energy operators dealing with volatile load patterns, distributed consumption, and increasing demand for predictive intelligence.
