Polynomial Regression in Evaluating Reliability for Dynamic Strength Using Jaya Algorithm

Objective: This study aims to develop a robust model for evaluating the reliability of a component subjected to fatigue loading. The focus is on enhancing predictive accuracy and reducing computational time while accounting for the dynamic nature of material strength over time. Method: The methodology assumes that the material strength follows a normal distribution that evolves dynamically over time. A conditional probability density function is derived based on this assumption. Polynomial regression is utilized to model the degradation of the mean strenght of material while the variance is considered to increase linearly. The model parameters are estimated using the Jaya algorithm combined with the maximum likelihood method. To validate the proposed approach, the methodology is applied to SCM440 steel strength data. Results and Discussion: The proposed method demonstrates its effectiveness by presenting reliability assessments with reduced computation time. The application to SCM440 steel strength data validates the implementation of the approach. The analysis highlights the significant impact of degradation type on the variation of reliability estimates. Research Implications: The study provides a framework for assessing the reliability of components under fatigue loading. By reducing computational time and improving predictive accuracy, the methodology offers a valuable tool for industries where reliability evaluation plays a critical role. Additionally, it emphasizes the importance of considering degradation type in reliability assessments. Originality/Value: This study presents a novel approach to reliability assessment by integrating polynomial regression for modeling dynamic mean strength degradation with the Jaya algorithm for parameter estimation. Unlike conventional methods, which often assume static material properties or employ computationally intensive techniques, this framework achieves a balance between accuracy and computational efficiency.