Supervised Non-Intrusive Load Monitoring Algorithm for Electric Vehicle Identification

Abstract

Transport sector electrification represents an increase in the number of electric vehicles (EV), producing significant variations in the distribution network dynamics. As a result, bidirectional power flow, overload and load unbalances are caused at the low voltage level due to unexpected increased load peaks. Non-intrusive load monitoring (NILM) methods have been developed as a strategy for energy management systems, applied to the customer side producing energy savings. This research presents a NILM methodology based on a low complexity conventional supervised machine learning pipeline. Our approach uses Principal Component Analysis (PCA) and Random Forest (RF) to detect the presence of a charging electric vehicle on the electricity network. By processing low sampling rate active power data, this approach provides a simple but feasible method that can be applied to smart meters. This provides useful data analysis for distribution network operators (DNO) to effectively deal with variability caused by these low carbon loads in the distribution grid. Achieving an overall efficacy of 92.68%, the proposed method can be compared with other state of the art methods developed under higher complexity techniques.

Load modelling and non-intrusive load monitoring to integrate distributed energy resources in low and medium voltage networks

“…Using an online irradiance measurement as a reference, aggregated active and reactive power were modelled to analyse solar PV farms and distributed generation scenarios. In a previous research study, an NILM method was proposed to identify EV from aggregated measurements of household load profiles as explained in [96]. Using principal components analysis as feature projection method and support vector machines as classification algorithm,…”

Effective identification of distributed energy resources using smart meter net‐demand data

“…Gini impurity is a metric based on erroneous classifications, which provides 1.0 for a misclassification and 0.0 for a perfect prediction. Entropy measures how well organised is a dataset [28].…”

Recent trends of smart nonintrusive load monitoring in buildings: A review, open challenges, and future directions

“…In this regard, NILM has been presented as a promising solution to split the individual load and identify charging EVs. 163…”

“…To that end, detecting the presence of a charging EV is essential to issue the appropriate actions and void any problems. In this regard, NILM has been presented as a promising solution to split the individual load and identify charging EVs 163 …”