Offshore wind turbine units are susceptible to typhoon attacks

and extreme typhoon weather can trigger a disaster chain of strong winds

heavy rain

storm surges

etc. Traditional operation and maintenance strategies only consider typhoon wind speeds

making it difficult to systematically characterize the impact of typhoons on turbine reliability and formulate corresponding maintenance strategies. To address this

the predictive maintenance strategy for offshore wind power

considering the typhoon disaster chain

is proposed in this article. Firstly

to address the difficulty in modeling reliability stemming from the varying impact mechanisms and load characteristics of typhoons and their disaster chains on different components of offshore wind turbines

a joint multi-disaster model based on 3-D Vine Copula is constructed to differentially model the incremental failure rates of components. Secondly

to tackle complex implicit spatio-temporal relationships in supervisory control and data acquisition (SCADA) correlation data under complex sea conditions such as typhoons

which hamper accurate situation description

Information Entropy and Recurrent Neural Network are introduced into Temporal Convolutional Network

and a unit situational awareness model is proposed to correct the reliability under the typhoon disaster chain. Finally

aiming at the coordination of maintenance strategies across different time scales

considering the decline in turbine reliability under the coupled influence of situational changes and typhoon disaster chains

a multi-time-granularity hierarchical predictive maintenance strategy for offshore wind turbine units is established. The superiority and effectiveness of the proposed model are verified through case simulations.