“Statistical signal processing and learning in smart grid”. Yue Zhao, Vincent Poor.

This special session aims to serve as a catalyst for the many promising applications of statistical signal processing and statistical learning in the rapidly developing field of smart grid. SSP plays a central role in addressing a variety of key challenges that emerge in smart grid advancement. In particular,

a) Renewable energy sources such as wind and solar are being widely integrated into power systems. The outputs of such energy sources are however highly uncertain and variable in nature, demanding new stochastic modeling and learning techniques to effectively represent and predict such random signals.

b) With significant potential in customer-side participation for making energy demand more flexible, a paradigm shift in balancing power supply and demand is underway. As customer behavior is very difficult to model or even anticipate, statistical and online learning are essential for exploring and exploiting customers’ demand response potentials.

c) Critical reliability and security issues persist in power system operation, in particular facing potential failures and attacks. Power network monitoring systems must be able to identify highly complex failures and attacks in real time, which demands new advanced SSP techniques beyond classic detection and estimation techniques.

An emerging theme in addressing these problems is to learn information and patterns from extensive real world data, such as wind and solar generation, customers’ responses to pricing signals, etc. As such, SSP will enable the development of insight into the open areas of data-driven research in smart grid. Smart grid applications also bring unique challenges and new perspectives to SSP, e.g., the co-design of SSP schemes with power system operation and markets.