Mean Time to Failure Prediction for Complex Systems With Adaptive Surrogate Modeling

Abstract

Abstract The Mean Time to Failure (MTTF) is a critical metric for assessing the reliability of non-repairable systems, and it plays a significant role in incident management. However, accurately estimating MTTF can be challenging due to the limited availability and quality of historical experimental data on a group of identical parts or systems. To address this challenge, we propose an adaptive surrogate modeling method, which approximate the system’s performance functions with a more computationally efficient model to predict the MTTF during the design stage. In this paper, we develop an adaptive Gaussian process (GP) approach that combines with Monte Carlo simulation (MCS) to predict MTTF. The proposed method initially trains a GP model for the system’s performance function, and then an MTTF surrogate model can be obtained. With a learning function, these models are updated dynamically based on new information as it becomes available. We demonstrate the effectiveness of our method on series system, parallel system, and mixed system with two examples, and the results show that the adaptive surrogate modeling method can accurately predict the MTTF of the system. Our method has the potential to improve the reliability and safety of complex system and reduces the need for costly performance testing.