Reproducibility archive for Towards digital twins through maintenance inference with uncertain intervention times and alarm-derived records

Computational reproducibility release accompanying “Towards digital twins through maintenance inference with uncertain intervention times and alarm-derived records” by Shoniwa Mosekari.

The release contains the scalar maintenance-record estimator, a 64-case held-out simulation with uncertain intervention times and alarm-derived category evidence, a separate measured-motor chronological-replay feasibility study, and the corrected mechanical reference. It includes the original-calibration error audit and preserves the original faulty source solely for transparent reproduction of that correction.

Contents include source code, fixed configurations and seeds, pinned Python dependencies, input data and provenance, figure-generation scripts, expected result tables and figures, mathematical-property checks, and recorded clean-run verification reports. The README documents reproduction and comparison tolerances. The checksum manifest covers 190 files. The current release retains the previously verified numerical results; the final update adds this Zenodo DOI to the README and regenerates the checksums.

Third-party data rights and licence notices are retained and documented in the README and provenance records.