Supplementary materials for "SQ-CAP: Sensor-Quality-Guided Conflict-Aware Pooling for Auditable Edge-AI Fault Screening"

Supplementary package for the article "SQ-CAP: Sensor-Quality-Guided Conflict-Aware Pooling for Auditable Edge-AI Fault Screening" (Periodica Polytechnica Electrical Engineering and Computer Science). Contents: the synthetic two-modality benchmark generator and recorded per-event scores for twenty seeds (1001–1020); summary and per-seed result tables; twenty-seed refitted ablations; the conflict–fault sensitivity experiment; cached CWRU drive-end vibration features with the feature-perturbation experiment and quality-covariate audit; the analytic resource model; scripts that generate the manuscript figures; method, provenance and reproduction documentation; and SHA-256 file hashes. All recorded analyses run offline with Python 3.12 and the pinned dependencies in requirements.txt. Run scripts/verify_package.py to check data, results, figures and file integrity. See README.md for how each file maps to the manuscript's tables and figures. Scope: the two-modality study uses generated features; the CWRU study uses vibration features with injected feature perturbations, not synchronized camera–microphone data; resource costs are analytic estimates, not hardware measurements.