Online Detection of Electrical Vehicle Charging Activity

Abstract

Energy analytics is gaining attention with the wide adoption of smart meters. Such meters can provide updates about energy consumption at least at the rate of every 15 minutes. This amounts for 96 reads a day . The analysis of such consumption data can give insight and help predict energy demand.

Demand-response is one scenario that can be automated based on such analysis. Demand-response is about detecting or predicting potential upcoming high demands and reacting by suggesting moving of unnecessary consumption to off-peak intervals. To manage a demand-response scenario ideally, sub-meter level consumption data need to be reported. That is, reporting consumption at the device level. This is also known as intrusive load monitoring. Yet, this requires a level of infrastructure and maturity of households not widely achieved. Still, another way to disaggregate non-intrusive consumption data is by detecting load profiles within consumption data. It is very hard to enumerate all-devices in all-categories consumption profiles. For instance, a hair dryer and an iron can have very similar load profiles. Yet, it is possible to identify unique load profiles.

Electrical vehicles have a unique consumption profiles when connected to a plug at a household. However, the accuracy of the profile detection depends on several factors among which is the granularity of the consumption sampling rate.

In [1], the authors propose an approach to identify electric vehicle charging profile by means of profile-consumption cross correlation. The approach proposed was mainly applied offline. In this work, we aim at extending the work to happen online, based on streams, so that detection can happen as soon as possible. This will give better value for demand-response management. Moreover, other contextual data can be used to enhance the accuracy of the detection and make it more resilient to the granularity of the sampling rate. For instance, if we can use the location, e.g. GPS data of a vehicle and the location data of the meter, we can better decide if the energy consumption includes vehicle charging.

 

References

[1] Ingebrigtsen, Karoline, Arne-Jørgen Berre, and Volker Hoffmann. "Energy Analytics-Opportunities for Energy Monitoring and Prediction with smart Meters." (2017).