Application of comparative strainer clustering as a novel method of high volume of data clustering to optimal power flow problem

Clustering-Based Penalty Signal Design for Flexibility Utilization

“…Furthermore, as shown in (23), (24), the SOC of a battery has a direct relation with the I dc of it. Therefore, by decreasing the I dc of battery, the SOC of the battery decreases.…”

“…Due to the high volume of these datasets, clustering, as a recent method of reducing the volume of data, is utilized in this paper. Clustering is applied to the data of each hour separately and representatives of clusters are extracted [24]. The discrete levels are defined based on the representatives of clusters.…”

“…To decrease the computation complexity and volume of data, clustering is utilized to segment the load and generation datasets. Generally speaking, clustering methods aim to group given data points into the optimal number of classes based on their similarity and define a centroid for each cluster [24]. These centroids are representatives of the whole data and researchers can analyze these to extract information about the whole dataset.…”

“…point just belongs to one cluster, and overlapping ones, that a data point belongs to different clusters. The main algorithms of exclusive clustering are partitional and hierarchical clustering [24]. K-means is a famous method of partitional clustering and defines the similarity of data points based on the distance of data points.…”

“…In divisive algorithms, at first all data points belong to a single cluster, then based on the diversity this cluster is divided into different numbers of clusters. Agglomerative algorithms treat each data point as a single cluster and then these clusters merge based on their similarity until all data points belong to one cluster [24]. In this paper, the Linkage-Ward (LW) clustering algorithm, the most suitable method for quantitative variables, has been used for segmenting given data into the predefined number of clusters.…”