Fine, that solves the problem for exactly one ephemeral correlate of power consumption surges. What about the next one?
If we're talking armchair first-stabs at automation, I'd much rather have 1) a large set of historical data from distributed consumption data feeds across the service area and 2) analysis of that same data live and with near-realtime latency. Given that, it seems plausible that a well-chosen machine learning based system could provide useful automated assistance.
That said, there are a bunch of practical reasons why it probably never makes sense to take human judgement out of the equation, esp. given the potential infrastructure impact here.
If we're talking armchair first-stabs at automation, I'd much rather have 1) a large set of historical data from distributed consumption data feeds across the service area and 2) analysis of that same data live and with near-realtime latency. Given that, it seems plausible that a well-chosen machine learning based system could provide useful automated assistance.
That said, there are a bunch of practical reasons why it probably never makes sense to take human judgement out of the equation, esp. given the potential infrastructure impact here.