Detection of power quality transient disturbances classification based on complex event processing

Effective and efficient classification to transient disturbance is crucial to provide reasonable decision making for solving power quality problems. Existing approaches for detecting transient disturbance classification are often centralized and procedure-oriented, which are developed in a hard code manner, and therefore possibly will face scalability and extensibility problems. In this paper, a novel approach to detecting transient disturbance classification is proposed by utilizing Complex Event Processing (CEP) technology in order to excavate features of transient disturbance effectively and efficiently from real-time, dynamic, and massive data. Each of occurrences of transient disturbance is modeled as an atomic event. A category of transient disturbance is expressed as a complex event that consists of multiple atomic events with logical operations represented by event pattern languages (EPL). The task of detecting classification of transient disturbance can be reduced to filtering and matching events from a real time event stream of transient disturbance in terms of given event patterns. The related experiments show that the CEP technology has significant advantages in the system throughput and latency event processing, and thus has strong scalability and extensibility. Our approach can provide a feasible solution for detecting classification of transient disturbance for massive real-time data with multiple sources.