Clustering power demand for feasibility study on transition to plugin hybrid electric vehicles

In order to make comprehensive investigation on feasibility of replacing conventional vehicle with plugin electric vehicle, the electricity grid limitation should be considered. Pdemand is one of the essential limitations. Moreover, historical data of demand profile could be used in simulation. Instead of considering whole data, some representatives could be utilized. Hard (K-means and K-medoid) and fuzzy (Fuzzy C-means and Gustafson Kessel) methods are used so as to find a few representatives for Pdemand, which describe their own clusters perfectly. Using validity measurement indices, optimal number of clusters and the best clustering technique are found out and finally relaying on Separation Invariant index, a comparison with a conventional technique is made.