AERF: Adaptive Evidence-Assured Predictive-Maintenance Framework
AERF is a modular research framework for governed evaluation of predictive-maintenance models across classification, regression, remaining-useful-life, and physical-edge execution routes. The capsule provides the base framework, campaign executor, model adapters, dataset registry, physical-edge validation utilities, result-production code, and a consolidated notebook with code-linked equations. Raw datasets and experimental results are intentionally excluded. Dataset directories contain placeholders and require separately licensed source data. The default run validates the capsule structure and Python source files. Full experiments require the corresponding datasets and compute environment. Federated learning, drift adaptation, cross-domain extensions, extended uncertainty estimation, and advanced explainability modules are outside the capsule’s validated core and are retained as future extensions.