Ultra-Short-Term Offshore Wind Power Prediction Based on VMD-TCN-BiLSTM with Self-Attention Mechanism

Wind power exhibits significant volatility and randomness, which adversely impacts power quality and grid security. To enhance the forecasting accuracy of wind power, this study proposes an ultra-short-term power forecasting model The model employs a Temporal Convolutional Network (TCN) to capture local temporal patterns, a Bi-directional Long Short-Term Memory (BiLSTM) network to capture temporal dependencies, and a multi-head self-attention mechanism to dynamically allocate weights to critical features. Additionally, Variational Mode Decomposition (VMD) is utilized to decompose the original time series into intrinsic mode functions (IMFs), effectively isolating trend, seasonal, and residual components. Experimental results demonstrate that the proposed model outperforms benchmark models across several error metrics, , and the MAE is reduced by 30.4% compared to the traditional LSTM-based approach. This framework provides a robust solution for improving ultra-short-term wind power forecasting, thereby supporting grid stability and renewable energy integration.