Technology Outline (Process Description)

Rapidly varying load demand is one of the greatest problems that distribution system operators are now experiencing. At the same time, the need to face new intermittent sources in the power system, has given more importance to probabilistic load forecasting in recent years. The uncertainties associated with the renewable sources force the load demand to alter as per the availability of the power generation. This becomes more prominent in the regions, where the utility grid does not reach and the entire demand is depending on the intermittent sources through microgrid. Hence, this causes uncertain demand governed by weather conditions. So, in order to improve demand certainty, load forecasting is necessary. An advance machine learning technique (LSTM) is applied for performing forecasting. LSTM is a complex version of RNN and a type of RNN with more layers inside of the cell. To overcome the drawback of vanishing or exploding gradient, RNN was introduced and has shown impressive performance, while dealing with sequential data. LSTM, a modified version of RNN, was introduced for capturing long-term dependencies. The operation of LSTM is handled by the three main layers, i.e., input gate layer, output gate layer and forget gate layer, which provide them with the power to selectively learn, unlearn or retain information.