ENERGY MANAGEMENT
Given the need for improved energy management tools, availability of real-time energy data with sophisticated metering systems (e.g. “smart” meters), multiple energy sources on the grid, and dynamic price structures, an adaptive and intelligent energy management program has never made more sense, or been more attainable. ARCs™ energy management program couples the superior real-time predictive accuracy of artificial neural networks with formal optimization so that optimal purchasing/selling strategies can be identified in real-time in accordance with recent, existing, and forecasted operational, weather, and price data, for both energy consumers and providers.
ARCs™ has the ability to optimize even the most complex energy systems, like those that can both produce and purchase energy from any number of sources and providers. The system of ANN models, which forecasts the short-term energy supply and demand metrics for the entity, is integrated with a multiobjective optimization program. The program can compute in real-time the optimal strategy for reducing energy costs to the extent possible, while providing the required level of reliability for satisfying energy demand, in accordance with the uncertainty of model forecasts (e.g. weather uncertainty). These two objectives are often conflicting, as a larger purchase/allocation of renewable energy sources, while increasing reliability, may reduce energy cost savings, requiring an acceptable trade-off in accordance with dynamic conditions (i.e. on- and off-peak periods) and institutional issues.
Figure 1 is a flow-chart depicting the logical development and inter-relationships between individual components, as well as coupling the ANN-based real-time energy production and demand forecasting models with the energy management optimization program.
As depicted, the renewable energy production and energy demand models are combined with the multiobjective optimization model, which identifies in real-time, based upon recent historical, existing, and forecasted conditions, the optimal energy purchasing/allocation strategies that balance reduced energy cost with reliability for satisfying energy demand. Although the utility would like to minimize energy cost to the extent possible, there will often be inherent uncertainty with regard to energy production and demand forecasts, largely due to weather uncertainty. This uncertainty determines the expected reliability of the selected energy sources (e.g. renewable) for satisfying the forecasted energy demand. The program explicitly accounts for this uncertainty/reliability on the basis of historical model forecast performances under similar weather and demand conditions. ARCs™ automatically generates alternative solutions as a function of different trade-off values because cost and reliability, from which the utility can select their optimal purchasing option in accordance with operational requirements, regulatory policies, structure of the energy industry, and new technologies.