It doesn’t matter if it’s a Plug-In; it will be detected based on the charger level and frequency. A customer with a Plug-In Hybrid will likely be detected if they use a Level 2 charger and frequently charge at home.
Bidgely EV Questions
As long as the customer has net metering, Bidgely is able to detect and estimate the solar generation and still have full disaggregation for their EV. There is an effect of accuracy stacking, but it is minimal given our solar estimation has an accuracy above 90%.
Another problem can occur which is a “false negative” where the customer does indeed have an EV, but the algorithm does not detect it. This can be due to a number of reasons including:
- The customer does not charge at home (or does so infrequently). The effectiveness of the algorithm relies upon charging actually taking place in order to detect the pattern and presence of an EV. A customer may have an EV but mostly charges at work, in this case it’s likely we won’t detect the EV since there is no home charging.
- The customer leverages level 1 charging (plugging into a standard outlet). The algorithm currently deployed detects Level 2 charging where the customer has installed a level 2 charger on a 240V outlet and it has greater than 3 kW of demand. Charging below this level is considered Level 1 charging. Bidgely does have an algorithm for this, however, it’s not currently deployed in the standard package as it has lower accuracy and generally requires ground truth from the territory of the utility.
While disaggregation is very accurate, it’s not perfect and “false positives” can occur. A signature of another appliance could resemble an EV depending on how the customer uses it. For example, a customer may have an unusual appliance in their home that draws significant power and is used in the same manner as an EV. If this occurs, the customer can leverage the home profile survey to correct this and the algorithm will update along with the EV disaggregation.
Bidgely leverages disaggregation (described above) to detect electric vehicle loads. Electric vehicle charging has unique patterns of electricity consumption, essentially a fingerprint of usage, that artificial intelligence (AI) can detect. A few of the patterns or indicators the AI looks for includes instantaneous demand (level 2 chargers have an average demand of 7kW), duration of high demand (average charging session duration is 4 hours), and time of demand (charging often occurs in the night and does not correlate with temperature). By detecting the pattern, the algorithm is able to detect the presence of an EV and also estimate the amount of charging.