AI Empowered Short-Term Wind Forecasting for Participation in Electricity Market

Wind energy is an integral component of a solution to the challenges of global energy crises and environmental degradation, since it is a renewable and environmentally friendly source of electricity. However, the unpredictability of wind speed makes the efficient utilization of wind power challenging. To overcome this challenge, a hybrid wavelet decomposition-ensemble empirical mode decomposition-convolutional neural network-long short-term memory (WD-EEMD-CNN-LSTM) wind speed forecasting model based on WD, EEMD, CNN, and LSTM algorithms is proposed. First, WD is deployed to minimize the fluctuation in the wind speed, and then the wind signal is decomposed into eight IMFs and one residual. The top three IMFs are applied to the input of a proposed model and compared with other state-of-the-art single and simple hybrid models. Furthermore, the proposed model's outputs are incorporated into the traditional approach to formulating the dispatch problem, aiming to reduce the generation cost. The expected cost of power dispatch is calculated for 2-bus and IEEE 24 RTS systems. The results showed that the proposed model has the lowest cost for day-ahead and balance markets. Overall, the WD-EEMD-CNN-LSTM model demonstrates superior performance in wind speed forecasting accuracy, making it a promising solution for increasing wind power utilization.