After offshore wind power enters the parity era

there is a more urgent need for accurate fault prediction to improve the reliable operation of wind turbines and reduce power generation losses; however

relying only on physical or data models for wind turbine early fault warning often faces the problem of model accuracy. Based on the idea of model-data fusion modeling

a fusion-driven offshore doubly-fed wind turbine early-fault-warning method based on equivalent thermal network model and Stacking integration algorithm is proposed. Firstly

the equivalent thermal network method is used to construct the wind turbine thermal balance matrix

the matrix is solved to obtain the steady state temperature values of each node

and the first-order RC thermal network model is adopted to describe the temperature trend over time. Then

the stator winding temperature and other related variables calculated by the thermal model are used as input features of the Stacking integration algorithm to correct the stator winding temperature values. Finally

the K-S (Kolmogorov-Smirnov) test principle is utilized to determine the adaptive threshold value

and early fault warning is carried out according to the trend of the residuals. The SCADA data of a domestic offshore wind farm are analyzed as an example to verify the effectiveness of the fusion model. The early warning method for offshore wind turbine faults based on the temperature-heat model driven with data fusion is generalizable and provides a technical support for the healthy and sustainable development of offshore wind power in the era of parity.