Microgrid protection under wind speed intermittency using extreme learning machine

Operation of wind turbine-based microgrids is characterized by increased intermittency due to the uncertainty in wind speed, which significantly affects the voltage-current profile. Conventional overcurrent relays based on prespecified threshold setting quite often are not able to detect faults under sporadic behavior of wind generators. With the aim of developing a protection scheme which is immune to the stochastic variation in wind speed under both grid-connected and islanding mode, a technique based on Discrete wavelet transform (DWT) and Extreme learning machine (ELM) has been proposed for mode detection, fault detection/classification and section identification. Uncertainty in wind speed has been modeled using Weibull distribution function and further incorporated in the protection modules. The proposed scheme has been validated using different statistical indices and compared with reported techniques for varying fault scenarios. Further, the effectiveness of the proposed scheme has also been validated for practical field applications by performing real-time simulations.