Typhoon weather significantly increases the volatility and randomness of wind power. Traditional prediction models usually standardize data using known installed capacity or the maximum and minimum value method

which cannot effectively map the strong local fluctuations. At the same time

the dynamic changes in the available installed capacity of wind farm stations caused by the shutdown of wind turbines also have strong uncertainty. Therefore

this paper reveals the influence mechanism of typhoon weather on the power generation of wind farm stations

and then constructs a TCN-Transformer neural network model to predict the available installed capacity and unit power generation of wind farm stations during typhoon transit. Finally

the predicted available installed capacity is used to invert the unit predicted power. The predicted power of wind farm station under typhoon weather is obtained. A wind farm in Guangdong Province of China was taken as an example to verify the method

and the verification results showed that the accuracy and reliability of the proposed power prediction method under typhoon scenarios were significantly improved.