Enhancing Power Communication Network Efficiency: A Hybrid Approach Using DBSCAN Clustering and Sliding Window Techniques for Alarm Data Reduction

Abstract

The original alarm data in power communication networks exhibit issues such as discreteness, redundancy, and temporal asynchrony. To address these challenges, this paper proposes a method that integrates DBSCAN clustering with a sliding window approach. Initially, the DBSCAN clustering algorithm is utilized to effectively tackle the problem of discreteness. Subsequently, the sliding window technique is employed to mitigate alarm redundancy and temporal asynchrony. Simulations demonstrate that the proposed method significantly reduces the number of alarms, achieving an average reduction of up to 47.68%.