After decades of rapid development, the wind power industry may be entering a period of technological maturity and maintenance. The components of the wind turbines mostly fail in a high-failure-risk period. Research on the reliability components can offer a basis for decisions on maintenance operations. In this paper, a wind turbine blade is subjected to wind load impacts year-round. The Lévy index is used to describe the instantaneous wind law. The data-driven method is used to describe the variation of blade failure related parameters. Based on this, the degradation failure and Lévy index of Wiener process are constructed. Based on the example analysis of blade failure of wind turbine in a wind farm in Daban city, Xinjiang province, a competitive model of failure probability of wind turbine under non-gaussian wind load is established. The statistical predictive model for competitive failures is closer to the actual characteristics of the wind farm. This offers data to help support site selection for individual turbines, and subsequent blade maintenance.