To address the problem of significant prediction accuracy degradation caused by severe second-level and minute-level fluctuations and nonlinear mutations in offshore wind power output during typhoon weather
an Improved Particle Swarm Optimization (IPSO)-Temporal Convolutional Network (TCN)-Transformer wind power forecasting model is proposed
incorporating multiple typhoon meteorological factors. First
the improved particle swarm optimization algorithm is used to efficiently calibrate the key model parameters
enhancing the model's ability to adapt and analyze complex typhoon meteorological factors. Second
the Transformer deep learning model is optimized to process wind power operational data under turbulent interference
outputting high-accuracy point prediction results
while dynamic interval prediction is achieved through a gated mechanism quantile regression approach. Finally
the model is validated using measured data from an offshore wind farm. The case study results show that
compared with existing conventional optimization prediction models
the proposed model exhibits stronger adaptability to extreme typhoon conditions
effectively mitigating prediction deviations caused by sudden output changes. It achieves both higher forecasting accuracy and environmental robustness
providing reliable technical support for accurate offshore wind power prediction under typhoon scenarios.
GUANG. Z Y,LIU L,TANG B,et al ."Ultra-short-term wind power subsection forecasting method based on extreme weather[J ] .IEEE Transactions on Power Systems,2023,38(6):5045-5056.