Davidson Frame
Abstract
The Davidson Frame (Structured Information Deterministic Manifolds, SIDM) is a seven-stage procedural architecture for reconstructing latent system states from noisy and information-constrained observations. This study presents an initial empirical validation of the framework using the NASA Commercial Modular Aero-Propulsion System Simulation (CMAPSS) turbofan degradation benchmark, comprising datasets FD001–FD004. The study is intended as a component-level validation of the Davidson Frame's inference and admissibility logic rather than as a competitive Remaining Useful Life (RUL) prediction exercise. A deliberately simple linear Health-Index model was used for Stage 5 reconstruction, while Stage 6 applied an entropy-based admissibility gate that attenuates or rejects candidate estimates when observational irregularity exceeds a defined threshold. The first ten test engines from each CMAPSS subset were evaluated, giving forty test engines in total. The entropy gate activated consistently across all four datasets, with cut-offs occurring in 9/10 FD001 engines, 7/10 FD002, 5/10 FD003, and 7/10 FD004. In each subset, increased observational entropy produced systematically more conservative latent-state estimates. Absolute RUL errors remained substantial, with mean absolute errors ranging from 53.7 to 107.2 cycles. These results are interpreted primarily as a limitation of the deliberately minimal Stage-5 reconstruction, rather than as evidence against the underlying gating architecture. The experiment therefore tests a narrower and more fundamental question: whether the Davidson Frame's entropy-gated admissibility mechanism behaves consistently when exposed to real multivariate degradation data. The observed results support that component-level hypothesis while identifying reconstruction quality, threshold calibration, larger-sample evaluation, and comparison of gated versus ungated estimates as the next validation stages. The longer-term purpose of this validation is architectural. The Davidson Frame is intended to provide a general state-estimation and information-admissibility layer within the wider Pathfinder architecture, including autonomous relay and beacon systems associated with the Phase-Managed Handover System (PMHS) and Poloidal-Shielded Analog Relay Buffer (PSARB). Within that architecture, the present study should therefore be understood as one independent validation step toward a larger system in which sensing, state reconstruction, uncertainty assessment, data handling, and autonomous operational decisions are separately testable before integration.
Related
- Mid-infrared AGN selection criteria comparison for LQAC-6 dense field correction: Towards celestial reference frame densification
- Inextensible Curve Flows in Dual Space D3: A Quasi-Frame Approach
- The EMA Justitia Probe: Terminator Frame-Based Deployment Orbit Selection and Surface Coverage Analysis
- Reflex instabilities I: Boundaries makes grid-based disc simulations frame-dependent
- Ultra-long simulations of collisionless relativistic shocks in front-comoving frame: evidence for a steady state and its properties
BibTeX
@misc{davidson2026davidson,
title = {{Davidson Frame}},
author = {Craig Kyrle Strachan Davidson and Lucie Mary Elise Davidson and Alfie Christian Strachan Davidson and Nolwen Marie Violette Bourlier},
howpublished = {HAL},
year = {2026},
month = sep,
doi = {10.5281/zenodo.22115041},
url = {https://hal.science/hal-05728151v1},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Davidson Frame
AU - Craig Kyrle Strachan Davidson
AU - Lucie Mary Elise Davidson
AU - Alfie Christian Strachan Davidson
AU - Nolwen Marie Violette Bourlier
PY - 2026
DA - 2026/09/01
PB - HAL
DO - 10.5281/zenodo.22115041
UR - https://hal.science/hal-05728151v1
AB - The Davidson Frame (Structured Information Deterministic Manifolds, SIDM) is a seven-stage procedural architecture for reconstructing latent system states from noisy and information-constrained observations. This study presents an initial empirical validation of the framework using the NASA Commercial Modular Aero-Propulsion System Simulation (CMAPSS) turbofan degradation benchmark, comprising datasets FD001–FD004. The study is intended as a component-level validation of the Davidson Frame's inference and admissibility logic rather than as a competitive Remaining Useful Life (RUL) prediction exercise. A deliberately simple linear Health-Index model was used for Stage 5 reconstruction, while Stage 6 applied an entropy-based admissibility gate that attenuates or rejects candidate estimates when observational irregularity exceeds a defined threshold. The first ten test engines from each CMAPSS subset were evaluated, giving forty test engines in total. The entropy gate activated consistently across all four datasets, with cut-offs occurring in 9/10 FD001 engines, 7/10 FD002, 5/10 FD003, and 7/10 FD004. In each subset, increased observational entropy produced systematically more conservative latent-state estimates. Absolute RUL errors remained substantial, with mean absolute errors ranging from 53.7 to 107.2 cycles. These results are interpreted primarily as a limitation of the deliberately minimal Stage-5 reconstruction, rather than as evidence against the underlying gating architecture. The experiment therefore tests a narrower and more fundamental question: whether the Davidson Frame's entropy-gated admissibility mechanism behaves consistently when exposed to real multivariate degradation data. The observed results support that component-level hypothesis while identifying reconstruction quality, threshold calibration, larger-sample evaluation, and comparison of gated versus ungated estimates as the next validation stages. The longer-term purpose of this validation is architectural. The Davidson Frame is intended to provide a general state-estimation and information-admissibility layer within the wider Pathfinder architecture, including autonomous relay and beacon systems associated with the Phase-Managed Handover System (PMHS) and Poloidal-Shielded Analog Relay Buffer (PSARB). Within that architecture, the present study should therefore be understood as one independent validation step toward a larger system in which sensing, state reconstruction, uncertainty assessment, data handling, and autonomous operational decisions are separately testable before integration.
ER -