datasets for Exact First-Passage Time Distributions for Discrete Degradation Processes via Wald's Martingale: A Unified Framework for RUL Prediction

This dataset contains the experimental data used in the study "Exact First-Passage Time Distributions for Discrete Degradation Processes via Wald's Martingale: A Unified Framework for RUL Prediction".

The dataset includes three parts: (1) simulated degradation trajectories generated from multi-step random walk, compound Poisson process, and double exponential jump process models; (2) real optical fiber port power degradation trajectories collected from cloud-computing data-center scenarios; and (3) bearing run-to-failure degradation trajectories used for RUL prediction evaluation. Feature files used for LightGBM baseline experiments are also included.

The data support the simulation experiments, optical fiber port case study, bearing degradation case study, visualization results, and ablation analysis reported in the article. This Zenodo record contains datasets only. The source code for reproducing the experiments will be released separately via GitHub.

Please refer to README_Datasets.txt for the directory structure, file descriptions, and instructions for placing the data into the corresponding code directories.