Enhancing Reliability Predictions by Considering Learning Effects Based on One-Parameter Lindley Distribution

The development of non-homogenous Poisson process (NHPP) models has attracted many researchers during recent decades. Incorporating learning effects with NHPP models may improve the predictive capability of these models and leads to more accurate predictions. In this paper, a NHPP model [1] based on one-parameter Lindley distribution is integrating with learning effects. More specifically, the new model is built by considering two influential factors: the autonomous errors-detected factor and the learning factor. Then, its performance is validated and compared to the classical NHPP model both objectively and subjectively based on five real reliability datasets and using three different criteria. The application results show that: in term of influential factors when the autonomous errors-detected factor is lowest and the learning factor is highest, the improved NHPP L model is efficient. Also, the optimized model by incorporating learning effects is more accurate and better predicted than the classical NHPP model.