Application of Load Forecasting i Thermal Unit Commitment Problems: A Pattern Similarity Approach

This study investigates the application of short-term load forecasting (STLF), which consists of estimating a future demand within a period of time up to one week, to thermal unit commitment problems (TUCP), providing schedule for power plant operations. Both problems have fundamental importance for power system operations and good results on STLF may also influence TUCP performance. The pattern similarity approach is chosen for STLF, which allows the use of regression algorithms based on machine learning applied to time series analysis and forecasting results are used as information for generators scheduling. This study proposes a framework containing these tasks with a deep review of them and provides some statistical information regarding the performance and validation of the framework.