Artificial Neural Network-Based Protection Algorithm for Transmission Lines Connecting DFIG-Based Wind Farms

This paper discusses the challenges faced by conventional distance relay in protecting lines emanating from doubly-fed induction generator (DFIG)-based wind farms. During severe faults, the fundamental component of the short circuit current of DFIG is no longer at the system nominal frequency. Besides, the short circuit current rapidly decays to low values shortly after the fault. This behavior may result in incorrect operation of the distance relay, which is configured assuming that the electrical grid is powered only by synchronous generators. Furthermore, an artificial neural network (ANN)-based protection algorithm is proposed for comprehensive fault detection and isolation. The proposed algorithm employs maximal overlap discrete Wavelet Transform to analyze the fault signals. Only local current and voltage measurements are used in the tripping decision; therefore, eliminating the need of a communication link with the other end of the line. To train the ANNs, simulation was carried out including a wide set of system operating conditions. The results were promising, where the proposed algorithm was able to detect and locate all types of faults when the line is connected to a DFIG-based wind farm, regardless of the prevailing system conditions.