Condition Monitoring (CM) has become an essential tool in complex engineering systems like wind turbines. They can prevent unexpected failures and contribute to a more reliable system. Information attained from monitoring can be employed for maintenance scheduling, hence, minimizing maintenance costs. Remaining Useful Life (RUL) is a critical aspect of CM. This paper introduces a new RUL prediction method for wind turbine blades using a novel fuzzy-based failure dynamic modeling via a Supervisory Control and Data Acquisition (SCADA) system. For this goal, a recursive Principal Component Analysis (PCA) is employed to compress the SCADA data and extract real-time Principal Components (PCs). Next, a wavelet-based Probability Density Function (PDF) is applied to obtain the probability of staying healthy from the extracted PCs. It is anticipated that blade degradation will lead to a subsequent decline in the PDF curve. A failure trajectory is then captured by transforming the PDF into the PC's surface. Subsequently, the T-S fuzzy system is utilized to form the mathematical model of degradation from this failure trajectory. Next, a Bayesian algorithm is adaptively administered to predict the RUL. Experimental test results on Canadian wind farms explain a high performance of the proposed failure prognosis method in comparison with a Bayesian algorithm.