Most of the electric utilities companies are investing in deploying smart meters to replace their aging infrastructure. Recently, I finished a smart grid analytics project with one of the largest electric utilities companies in the US, and found that smart meter data can be use to accurately predict (RSq .86 to .97) the following:

1. A meter which is going to go bad within a 30 days period;
2. Power outage at the meter level within a -1 to -3 hours; and
3. Spike in consumption at the meter level within a -1 to -3 hours

The general smart grid analytics methodology that we developed can be used across the board and is technology agnostic. The value to the electric utility companies is allows them to integrate predictive capabilities to their Meter Data Management Systems (MDMS) in near-real-time. These predictive capabilities can be used in load forecasting, settlement, distribution planning, and efficient deployment of field staff for problems that require manual intervention.