Dynamic Demand Response Through Decentralized Intelligent Control of Resources

With the advent of machine learning and IoT capabilities built into resources, it is becoming feasible to implement data analytics-based strategies where the loads and renewable resources dynamically predict and publish their ability and extent of reduction or generation. This paper explores how the traditional demand response programs can be made automated, independent and dynamic—in the sense of allocation of load reduction dynamically to various resources, based on the data published by them regarding how much each of them is able to contribute. This will facilitate extending the demand response paradigm itself—by making it a tool to achieve load-curve tweaking at any point of time, rather than using it just when demands peak. With the increasing penetration of renewable power generators and EV loads which are highly variable in nature, we believe our dynamic demand response paradigm would be of high utility for future smart grids.