This paper introduces a method for measuring the Reliability Function of Carbon Fiber Reinforced Polymer (CFRP) under cyclic stress using nonparametric estimators. The study's goal is to gain a better understanding of material deterioration by combining microscopic and macroscopic approaches while accounting for the uncertainty associated with the data and models. Several statistical techniques were used to estimate the reliability function from failure data generated under cyclic stress, such as Nadaraya-Watson, spline estimation, kernel density estimation, Bayesian estimation, and Gaussian process regression.

Real data will be simulated as a result of experiments conducted in Stanford structures and the vehicle laboratory (SACL) in partnership with NASA's AMES Research Center for Predictive Excellence (PCoE). Thus, developing the method of generating data based on failure rate and the number of cycles that allow calculation of possible failure periods, the parametric models will accurately predict physical behavior under periodic pressure, as evidenced by data analysis using performance measures such as Mean Squared Error (MSE) and the Determination Factor (R 2). The study also confirms the need to include uncertainty in forecasts to increase accuracy of results-level estimation stresses the use of statistical models that explain uncertainty in particular Bayesian Kernel Estimator and Gaussian Process Regression to estimate the reliability function of carbon fiber reinforced polymers CFRP maintenance costs that are important in the development of vehicle safety methodologies.