Improving WFA k-means technique for demand response programs applications

An annual framework for clustering-based pricing for an electricity retailer

“…An application based comparison between weighted fuzzy average (WFA) K-means and modified follow the leader is discussed in [23]. Also, authors have improved WFA K-means in [24].…”

“…Medians can be used in place of averages, although they may throw away good points as well as outliers. For solving this problem, a type of fuzzy averaging is used that puts the center prototype among the more densely situated points [23,24].…”

“…To find its weighted fuzzy average, this algorithm initially takes the sample mean (0) and variance 2 to start the process. We have extended the formulations which are used in [24] by considering L as a vector with H elements that represents the computation of cluster center as following:…”

“…To overcome the drawback in computation of the distance between each vector and cluster center, a new method has been proposed by the authors in [24]. A brief explanation of the method is provided as follows.…”

“…In [24], a new distance computation method for increasing the accuracy of the distance computation in order to put more similar load curves in each cluster has been proposed. The new formulation of the distance is represented by the following equation:…”