Harmonics distortion is a major arising issue in power networks. In almost every study done, the source of harmonics often referred to non-linear loads. In this study, the author concerntrates only on harmonics induced by synchronous generators at generation-end. The objective of paper is to provide a complete process review on implementing a practical and efficient technique to forecast triplen harmonics voltages and corresponding phase angles at every network busses. Real-time data from the 11-kV generation system at Gas District Cooling (GDC) distribution network in Universiti Teknologi PETRONAS (UTP) would be fed into the proposed algorithm for harmonic estimation and forecasting. Based on a thorough review on various methods of harmonics state estimation (HSE), neural networks (NN) method was suggested and studied to estimate and forecast the harmonic voltage vectors. © 2011 IEEE.