Electrical customers classification through the use of self-organizing maps

This paper shows the ability of the Self-Organizing Maps (SOM) as a tool of classification for different kinds of electrical customers. This approach allows to extract the pattern of customer behavior from historic load demand series. Several ways of data analysis from load profiles can be used to get different input data to "feed" the neural network. Here, two methods are proposed for improving customer clustering: the use of frequency-based indices and the use of the hourly load profile. The results obtained clearly show the classification and aggregation capacities through the use of daily load profile parameters. Self-Organizing Maps allow to achieve in a no supervised way, coherent clusters from the original customers data.