Abstract:

Objectives Offshore wind power is of great significance for achieving carbon peaking and carbon neutrality goals, yet its power prediction remains challenging. To address this issue, this study proposes a novel offshore wind power prediction method integrating temporal convolutional network (TCN) with the Transformer framework. Methods Firstly, discrete cosine transform (DCT) technology is incorporated to effectively extract frequency-domain features from time-series data, optimizing feature weight allocation by fully exploiting frequency-domain characteristics. Based on this, a hybrid TCN-Transformer model is constructed, which is capable of comprehensively integrating both time-domain and DCT-processed frequency-domain information, enabling the deep learning and accurate prediction of the data. Results Comparative experiments are conducted using two sets of actual operational data from an offshore wind farm. The results show that the proposed method not only performs well in feature extraction but also significantly improves the accuracy of power prediction, fully verifying its effectiveness and application value. Conclusions The TCN-Transformer model integrated with DCT technology demonstrates exceptional prediction performance in offshore wind power prediction. Compared to other benchmark models, it achieves a significant improvement in prediction accuracy. This approach provides robust support for the efficient operation and optimal management of offshore wind farms.

Key words: offshore wind power, power prediction, discrete cosine transform, temporal convolutional network, machine learning, deep learning, renewable energy