Electrical Fault Detection On Downed DC Trolley Lines

The National Institute for Occupational Safety and Health (NIOSH), Pittsburgh Research Center (PRC), has conducted research to improve electrical fault detection on coal mine direct current (dc) trolley systems. Present circuit breaker protection systems are current-magnitude-based and cannot discern between normal traffic and high resistance electrical faults. Tests were conducted at cooperating mines to assist in developing a neural-network-based detection algorithm for distinguishing between electrical faults and normal operation. The Fast Fourier Transform of the trolley rectifier current signature served as baseline data for the neural network. During field tests, the algorithm performed with greater than 90% accuracy.