Autocorrelated Kriging-Based Predictions for System Reliability Evaluation

EasyChair Preprint 14790

Abstract

System reliability analysis is crucial for ensuring the safe and efficient operation of engineering systems. Traditional methods often struggle with the complexities and dependencies inherent in real-world systems. This article explores the integration of autocorrelation into Kriging models to enhance system reliability evaluation. Autocorrelated Kriging models account for temporal and spatial dependencies in data, offering improved prediction accuracy and uncertainty quantification. We provide a detailed framework for applying autocorrelated Kriging, including data collection, model development, and validation techniques. A case study involving a wind turbine gearbox demonstrates the practical application and benefits of this approach. Results indicate that autocorrelated Kriging outperforms traditional methods in predicting system failures, providing more accurate reliability insights and optimizing maintenance strategies. Despite some challenges, such as computational complexity and data quality, ongoing advancements in statistical modeling and computational tools hold promise for further enhancing predictive reliability models.

Keyphrases: Autocorrelated Kriging, Data Autocorrelation, Variogram Analysis, engineering systems, failure prediction, maintenance optimization, predictive modeling, reliability assessment, system reliability, time series analysis

@booklet{EasyChair:14790,
  author    = {Alakitan Samad},
  title     = {Autocorrelated Kriging-Based Predictions for System Reliability Evaluation},
  howpublished = {EasyChair Preprint 14790},
  year      = {EasyChair, 2024}}