Behavior Learning and Prediction
AGT’s analytics use high-resolution energy data gathered from building sensors to learn models for behavioral analysis and consumption prediction. Our solution learns building-specific models, enables application in real time and scales to large data volumes. The results support energy balancing in smart grids, and enable insights about behavior patterns in buildings.
Understanding and predicting the energy consumption of buildings is key to efficient utilization of renewable energy and operations in the smart grid. Consumer-specific behavior and consumption patterns require solutions that take the individual into account, and scale to large numbers of consumers.
Our behavioral analytics automatically extract typical behavior patterns in a building from sensor data, and correlates them with external influencing factors (e.g., project deadlines; VIP visits; holiday seasons). Once learned, these patterns are used to continuously monitor sensor data and automatically detect external influencing factors on the facility.
Our prediction analytics use data derived from energy sensors within buildings to learn building-specific prediction models.
Our solution is highly scalable and capable of applying models over live sensor streams, making continuous predictions in real time. Every building is approached with an individual model that is automatically tailored to it.
AGT’s IoT analytics won an innovation award in 2012, and AGT was invited to present our results at CeBIT 2014.