Outage Detection in Power Distribution Networks

Outage detection and management has been a long-standing problem in power distribution networks. Outages are caused by protective devices closing off a part of the network to automatically isolate it. Usually, a short circuit fault will trigger this protective operation. We employ the term outage detection to denote the task of finding the status of the protective devices, and the term fault detection to denote finding the faults that caused the resulting outage situation. Many methods for outage and fault detection based on artificial intelligence have been developed. Outage detection is often performed prior to fault detection and can greatly improve the accuracy of fault diagnosis. For outage detection, fuzzy set approaches have been proposed based on customer calls and human inspection [11], and based on real-time measurement with a single sensor at the substation [1]. In networks where supervisory control and data acquisition (SCADA) systems are available, a subset of the protective devices’ status can be obtained via direct monitoring. When two-way communications from the operator and the smart meters are available, AMI polling has been proposed to enhance outage detection [6]. There have also been knowledge based systems that combine different kinds of information (customer calls, SCADA, AMI polling) [5]. For fault detection, using only a single digital transient recording device at the substation, fault location and diagnosis systems have been developed based on fault distance computation using impedance information in the distribution system [12]. Using only the outage detection results, i.e., the status of the protective relays, expert systems have been applied to locate the underlying faults [3]. Incorporating voltage measurements in the distribution system with the outage detection results, fault detection methods based on knowledge based systems have been proposed [2]. Fault detection that uses during fault voltage-sag measurements and matching has