ML4GE

The rollout of smart meters is expected to produce significant energy consumption data at an unprecedented rate. Applying Machine Learning (ML) techniques to this massive data will enable the development of intelligent systems that convert energy metering data into easily understandable insights. This is likely to create value-added services to both consumers and utility companies and contribute to higher energy efficiency levels. We successfully developed deep learning-based approach for analysing streams of smart-meter data and create clusters with similar profiles. We further provide per-cluster demand forecasting model based on state-space (Gaussian hidden Markov model) and auto-regressive model for hourly, daily or weekly forecast.