Loading...
Loading...
Contributors:
Dataset URI:
jgh87tpkmnReproducibility package for the manuscript "Preserving Channel Identity in Multivariate Fault Diagnosis: A Controlled Study on a Six-Wheel Independently Driven Chassis". Contents: (1) Simulation-generation code and run manifests for 39-class wheel-level diagnosis on a six-wheel independently driven chassis. (2) Observation windows sampled at 1 kHz, with 1 s duration, 0.5 s stride, and 21 channels, stored as little-endian float32, together with supporting simulation outputs. Simulation files are distributed across two independently extractable ZIP archives according to the run identifiers in their paths: simulation_data_part1.zip contains run01–run09 files, including applicable independent-realization runs; simulation_data_part2.zip contains run10–run18 files, applicable independent-realization runs, and remaining simulation files. (3) Trained checkpoints and saved predictions for the main benchmark and representation ablation, based on 18 original simulation runs, together with frozen-model evaluation on six new stochastic realizations. (4) Healthy measured CAN records and a speed-agreement check script, together with synthetic sensor-corruption experiments on measured healthy backgrounds. (5) An independent measured-data benchmark using the UCI Condition Monitoring of Hydraulic Systems dataset (Helwig et al., 2015; https://doi.org/10.24432/C5CW21), including preprocessing, a fixed split by auxiliary-component condition combinations, 15 model-training runs, checkpoints, and predictions. The original dataset is licensed under CC BY 4.0. Download code_results.zip, measured_support.zip, simulation_data_part1.zip, simulation_data_part2.zip, and external_hydraulic_validation.zip. Extract all five archives into the same empty directory, preserving their internal paths. The two simulation archives are ordinary ZIP files and do not need to be concatenated. See README.md for reproduction instructions and the included manifests for file checksums. This release supports the revised experiments, using a corrected simulation model, regenerated simulation data, and rerun training and evaluation. The earlier dataset at https://doi.org/10.17632/n5rzmpx7zx.2 corresponds to the previous simulation protocol and is not the dataset supporting the revised results reported in this manuscript.
Keyword(s):
Computer Science; Engineering.
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
No Associated Publication Data Available
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)