A novel algorithm for the parallel arc fault identification in DC aircraft power plants

Arc faults are phenomena associated to unwanted electric discharge between two or more conductors. Since energy related to an arc may fuse the conducting material, very often an arc is responsible for a fire. This condition is very critical in aerospace environment, where a fire may produce catastrophic effects. For this reason, many efforts have been produced in order to develop a reliable method for arc detection. The main problem in facing this aim is represented by the intermittent and random nature of an arc in an aircraft, due to the effects of the in-flight vibrations. A continuous monitoring of the electric system is mandatory, and a smart measurement system is also required. In particular, different identification methods have been proposed, based on Kalman filter, neural network, fuzzy logic, etc., that usually require a complex learning activity which could be not exhaustive. In this paper a novel approach to the arc identification is presented, that is based on the evaluation of two physical parameters: rate of concurrency of current spikes and their specific energy. By combining these two parameters is possible to detect the arc and estimate its severity. The method has been tested using a standardized emulation system under different conditions and the experimental results clearly show its effectiveness.