Lots of investments are being made to improve the energy efficiencies of buildings. Our client had a pay-for-performance agreement with an efficiency solution provider. In order for this system to work, it was very important to predict the energy usage of buildings without the retrofits given the ambient conditions.
As part of this engagement, we developed the solution by building GBM Light models for energy usage prediction.
We implemented predictive models which learned from the building’s energy usage history to predict energy consumptions (without energy efficiency solutions).
We used 1000 buildings’ energy usage data for 3 years, daily weather data for each building location, and building metadata to develop the solution.
This resulted in providing an improved and scalable solution to forecast energy demand (without retrofits) that resulted in better management of performance-linked payout for energy usage solutions. It also provided a means to measure the savings coming from the optimized energy efficiency of the building.