Short-term load forecasting plays a major role in energy planning, its accuracy having a direct impact on the way the power system is operated and managed specifically for an operational planning timeframe. We propose a new Clustering-based Similar Pattern Forecasting algorithm (CSPF) for short-term load forecasting. It resorts to a clustering algorithm to identify load patterns using a combination of distance measures to capture differences between time series trends, values, cyclical, and autocorrelation behaviours. The day-ahead load forecasting is then obtained based on a selection of sequences of days that exhibit similar load patterns and similar temperature patterns. We apply the algorithm to provide the aggregated national load forecast of Portugal for the day ahead with a discretization of 15-minutes. The forecasting results, referring to one year, exhibit a better performance - evaluated by RMSE, MAE and MAPE - when compared to the alternative Pattern Sequence-based Forecasting (PSF).

--