No abstract

No abstract

This study presents a comparison of two developed intelligent systems that carries out, in an integrated manner, failure diagnosis on electric power distribution feeders. These procedures aim to identify and classify critical situations, as highimpedance faults, which can potentially damage the system components and cause power supply interruptions to consumers. The intelligent systems combine the wavelet transform, Dempster-Shafer evidence theory, voting scheme, fuzzy inference system and artificial neural networks. Results show the efficiency, reliability, and robustness of the proposed methodology, allowing its real-time application.

This article presents a method to detect and classify voltage disturbances in electric power distribution systems using a modified Euclidean ARTMAP neural network with continuous training. This decision-making tool accelerates the procedures to restore the normal operation conditions providing security, reliability, and profits to utilities. Furthermore, it allows the diagnosis system to adapt to changes from the constant evolution of the electric system. The voltage signals features or signatures are extracted using discrete wavelet transform, multiresolution analysis, and the energy concept. Results show that the proposed methodology is robust and efficient, providing a fast diagnosis process. The data set used to validate the proposal is obtained by simulations in a real distribution system using ATP software.