Distribution fault location using graph neural network with both node and link attributes

A fault detection apparatus for fault detection and localization in distribution feeders having branches and nodes, comprising:

• a computing system having, a transceiver, data storage with instructional modules and circuitry configured for processing, to cause the apparatus to
• receive, via the transceiver, feeder raw data in a distribution feeder of a power distribution system;
• process, via the processor, the feeder raw data with predetermined operational electrical characteristics of the distribution feeder data accessed via the data storage, to generate a branch attribute dataset for each branch separated by a pair of nodes for all branches, and generate a node attribute dataset for each node for all the nodes in the distribution feeder;
• input the branch attribute datasets and the node attribute datasets into a trained fault detection neural network to determine whether a branch has a fault and a fault location within the branch, to output a classification of the fault and the fault location, and generate an alert signal based upon determining the classified fault and the fault location;
• and
• send, via the transceiver, the alert signal to an alert control system, upon the alert signal being received, the alert control system generates an action in response to the alert signal to an outage response system to reroute power and restore service to the disconnected power of the distribution feeder with the fault.

A non-transitory computer readable medium, having a computer program thereon, wherein the computer program, when executed by a processor of a fault detection apparatus, causes the processor to:

• receive feeder raw data including real-time measured pre-fault branch regulations and energizations data, and real-time measured during-fault node voltages and currents raw data, in a distribution feeder of a power distribution system;
• process the feeder raw data with predetermined operational electrical characteristics of the distribution feeder to generate a branch attribute dataset for each branch separated by a pair of nodes for all branches, and generate a node attribute dataset for each node for all the nodes in the distribution feeder;
• input the branch attribute datasets and the node attribute datasets into a trained fault detection neural network to determine whether a branch has a fault and a fault location within the branch, to output a classification of the fault and the fault location, and generate an alert signal based upon determining the classified fault and the fault location;
• and
• reroute power and restore service to the disconnected power of the distribution feeder with the fault, based upon the alert signal being sent to, and received by, an outage response system.