Abstract:

Due to the increased usage of Electric Vehicles (EVs), the burden on the power grid utilities has increased in the recent past. But the time-shiftable nature of the charging process of these EVs allows them to participate in Demand Response Programs (DRP) which offers a whole new dimension of flexibility in a Power system. To identify the availability of EV in a household, the grid operator can employ the concerned household's consumption data that is available through the smart meter installed there. A Hidden Markov Model (HMM) based approach similar to Machine Learning has been proposed in this paper for identifying houses with EVs. For operational purposes, this algorithm helps in the desegregation of EV charging from smart meter's energy consumption data along with determining their charging/discharging state and SOC estimation. This has been done in a non-intrusive way and aims to identify time-flexible or shiftable loads for Demand Response (DR) participation. Data from the smart meters of over 1000 households have been taken into account at different sampling frequencies. Several of these households have also incorporated rooftop PV to make them self-reliant. The practicality of the proposed approach is ascertained by the simulation results that are based on real-world situations. An error analysis has been performed to observe the effectiveness of this method. The scope of participation for the identified EVs in DR programs has also been discussed at the end.