Under the combined actions of wind, wave, and current, the floating wind turbine is prone to creep in mooring, which can accelerate corrosion, increase the failure probability, and affect the stability of the platform. In order to ensure the safe operation of floating wind turbine and achieve early warning in the early stage of mooring creep, this paper proposes a mooring fault diagnosis method for floating wind turbine based on multifractal analysis. First, the mooring fault nonlinear information was extracted by variational mode decomposition (VMD) method, and the impact of mooring creep and mooring failure at different locations on the stability of floating wind turbine was analyzed. Then, considering the multi-metric characteristics of the nonlinear signal, multiple fractal detrended fluctuation analysis was used to extract the fault signal characteristics, and it was estimated whether mooring creep occurred and the locations of mooring creep. Finally, the platform response data under mooring creep at different locations were analyzed. Results show that when the original signal was processed by VMD and the fractal box dimension was used to filter the feature signal, the noise was effectively filtered and more representative nonlinear features were extracted. The mooring fault signal had multiple fractal features, and the mooring creep and its location were effectively estimated by the singularity index α₀. The nonlinear features extracted by VMD could be estimated according to the data complexity by the multiple fractal detrended fluctuation analysis. The state of the mooring could be determined based on the complexity of the data. The research results can provide theoretical methods for information extraction and fault determination of offshore floating wind turbine moorings.