Abstract：In order to improve the prediction accuracy of day-ahead electricity prices in the purchase and sales electricity market, this paper combines variational mode decomposition (VMD), improved coyote algorithm (ICOA) and bi-directional long and short-term memory (BiLSTM) network to propose a novel method for day-ahead electricity price forecasting. Firstly, to address the non-smoothness of the electricity price sequence, VMD is used to decompose the original sequence into several subsequences. Secondly, to address the problems of slow convergence and insufficient optimization performance of the coyote algorithm, the Sobol sequence is introduced into the coyote initialization, and the global optimum and local optimum is introduced into the group culture trend. Then, ICOA is used to optimize parameters of the BiLSTM and build an ICOA-BiLSTM prediction model for each subsequence. Finally, the prediction results of all sequences are superimposed to obtain the final prediction result of electricity price. Experiments are conducted on Denmark electricity market data, and the results show that the proposed method has good forecasting accuracy and generalization ability.
2023, Vol. 24 
