Wind Power Uncertainties Forecasting based on Long Short Term Memory Model for Short-Term Power Market

Renewable sources are volatile and intermittent in nature which produces uncertainties which are non-linear in nature. Generally, quantitative and qualitative forecasting are done based on data available. In this article, a probabilistic forecasting methodology centered on Long Short Term Memory (LSTM) network. This novel forecasting method for wind power generation has been done considering wind uncertainties in the power market. A genuine case study espousing the proposed methodology. It has been perceived that LSTM model forecasts the uncertain WPF more accurately while compared with other states of the art machine learning method and also other forms of a model like Autoregressive Moving Average (ARMA), Autoregressive Integrated Moving Average (ARIMA) etc. The propositioned LSTM model is found to be a suitable method for wind power uncertainties forecasting.