Under the "dual carbon" target

the high proportion of new energy access poses severe challenges to the real-time balance of the power system and the operation of the spot market. Existing wind power forecasting research often focuses on a single accuracy indicator

making it difficult to couple market deviation assessment rules with scheduling requirements. To this end

an IBKA-VMD-CNN-GRU Attention wind power intelligent prediction model was constructed that integrates deviation assessment economic constraints. Firstly

by improving the Black Winged Kite Algorithm（IBKA） and optimizing the Variational Mode Decomposition（VMD） parameters

signal denoising and feature enhancement are achieved; Furthermore

the Convolutional Neural Network Gated Recurrent Unit Attention Mechanism（CNN-GRU Attention） network is utilized for spatial feature extraction

long and short-term temporal dependency modeling

and key period weight allocation

resulting in high-precision prediction results. At the same time

the deviation assessment of electricity consumption is directly integrated into the model fitness function

meeting the requirements of 95% accuracy and differentiated time periods

which solves the limitation of traditional research ignoring economic constraints. Verified by data from a wind farm in Xinjiang

the prediction accuracy of this model is significantly better than traditional models

which can effectively reduce the cost of electricity market imbalance and provide technical support for market declaration

pre control deviation assessment

and grid optimization scheduling.