Paper Title
Using K-Means Clustering for Custom Adjustment of Reactive Power in Industrial PFC
Abstract
This paper uses K-means clustering as a strategy to determine the minimum, average and maximum reactive
power ranges required by an industrial plant for the design of its power factor (PF) solution. The strategy calculates three
cluster points in the sensed power data set. The analysis is carried out by comparing the consumption of reactive power
against active power during one week of plant operation. This data is used to analyze the behavior of the power factor, and to
specify the reactive power for a 6-step mixed compensation bank. The strategy was successfully used to design a power
factor corrector in which the specified capacitor bank demonstrated high performance.
Keywords - Compensation Bank, Industrial, K-Means Clustering, Power Factor, Reactive Power, Unsupervised Learning