Short-Term Electricity Load Forecasting Using Modified Hidden Markov Model

This study proposes a modified Hidden Markov Model (HMM) as a method for predicting electricity on an hourly basis in the Delhi region. Typically, load prediction involves utilizing statistical techniques that need significant modifications in the data to adapt to the random nature of the energy demand. Alternatively, data-based methods like artificial neural networks (ANNs) rely heavily on data to deliver reliable findings. An attempt is made to implement HMM taking into account short-term electricity demand as a non-stationary time series. This provides satisfactory prediction results even with limited data. Furthermore, the proposed modified HMM technique outperforms alternative techniques in terms of computational time and complexity.