Data-Driven Calibration Scheduling for Measuring Instruments—A Smart Management Architecture

Reliable calibration of measuring instruments is a foundational yet under-reported layer of the smart-manufacturing stack: without traceable calibrations, no downstream process can guarantee the metrological quality of the data it consumes. This research asks whether an administratively managed calibration record base is sufficient to support data-driven recalibration scheduling, and diagnoses the data-governance conditions such scheduling requires. A production multi-faculty university calibration database (3552 registered instruments, 371 calibration events, 2009–2025) is analyzed to characterize fleet inventory and compliance, quantify the historical recalibration late-rate with bootstrap and analytical 95% confidence intervals stratified by faculty and device type, benchmark supervised models against explicit baselines, and propose a transparent rule-based scheduler with an empirical faculty-level safety buffer. The historical late-rate is 47.8% (95% CI 34.1–61.9%), with exploratory inter-faculty heterogeneity, while supervised learning yields only a weak, uncertain signal and no useful deviation regression. Because 86.7% of the fleet lacks the baseline records needed to schedule it, the practical message for instrument managers is direct: completing the missing last-calibration and calibration-period fields is the highest-value first step, after which a transparent, data-governance-informed scheduler—not a black-box predictor—is the defensible deliverable, transferable to comparable research and manufacturing infrastructures.