Introduction  With the rapid development of offshore wind turbine installed capacity, the owner attaches more and more importance to safe operation of offshore wind turbines, and imposes more stringent requirements for reliability. The traditional after-fault trouble-shooting pattern cannot ensure reliability of offshore wind power equipment. Moreover, as the accessibility of the offshore wind power equipment is unfavorable, the passive trouble-shooting pattern leads to huge loss of outgoing power, which fails to meet the latest requirements of the modern offshore wind farm. The intelligent fault warning system can predict the abnormal conditions of equipment and eliminate the hidden danger at its very beginning stage, preventing it from further deterioration.

  Method  The forecast of critical faults, such as generator temperature abnormal, generator bearing abnormal and gear box heat dissipation abnormal and cog belt rupture, can be achieved in advance, by collecting the data of wind turbine unit critical components, summing up the fault characteristics from historical data and employing the big data algorithm including neural network.

  Result  In accordance with early warning of equipment faults, active and preventive maintenance strategy can be practiced in a planned manner, in combination with the OWF maintenance characteristics of meteorology, typhoon, oceanic and maritime conditions. Thus, large component faults and wind turbine unit failures can be effectively prevented.

  Conclusion  The research results could enhance the wind turbine unit reliability and ensure the overall gains of the offshore wind farm.