We present two contributions in this work: i)Novel electric network frequency (ENF) classification algorithm, and ii)Circuit for measuring power signals from the power grid.We first propose a novel ENF signal estimation algorithm.This algorithm explicitly makes use of the harmonic information present in the signal and estimates the nominal frequency based on the most reliable harmonic. The ENF signal is estimated from the most reliable harmonic by employing a Gaussian weighting window to mitigate the effects of noise. We
then extract features from the ENF signal estimate and train a Neural Network (NN) classifier using the provided training dataset. In addition to the previously proven features for ENF signals, we also use Auto Regressive Moving Average (ARMA) model parameters as features in this work. The proposed classification algorithm performs at 95.06% accuracy on the provided power signal and an accuracy of 85.5% for the audio signal. We also obtained an accuracy of 94% for the practice dataset provided for validation . The circuit to acquire the power signal from the grid is designed on the opensource Arduino board. The accuracy of the proposed circuit
is demonstrated by a comparison with the grid data obtained from the national network frequency monitoring agency in India.