Estimating Spinning Reserve Capacity With Extreme Learning Machine Method in Advanced Power Systems Under Ancillary Services Instructions

Abstract

In advanced power systems, to balance load generation mismatches in the face of unexpected events and to meet unforeseen possibilities, load taking and shedding instructions are required under Ancillary Services (AS), which are performed by providing spinning reserve (SR) in the units. SR is an important factor, which is rather uncertain to predict due to the unpredictability of customers’ consumption, excessive or under-energy generation, and unpredictability in the integration of renewable energy sources. In this study, integration of renewable energy is addressed as a factor for predicting SR. Increasing the integration of solar and wind into the power system requires larger amounts of SR, which creates a significant increase in generation and emission costs. Optimum estimation of SR capacity helps system operators (SO) plan generators in advance and in a better bidding environment. In the past, estimation tools such as feed forward networks and time series models have been used to estimate load and electricity generation. In this article, Extreme Learning Machine (ELM) method is used to estimate SR capacity in the day-ahead and intraday market. The results obtained with ELM were also compared with the results observed with ANN, LR and SVM methods.