Two-Layer Productivity Diagnosis–Prescription Methodology Using Manufacturing Data: A Data-Driven Preliminary Stage Toward SME AI Transformation (AX)

This study proposes a two-layer methodology linking productivity diagnosis and improvement prescriptions using manufacturing platform data from manufacturing SMEs. As a preliminary stage toward AI transformation, it involves no training or deployment of AI models. The macro layer decomposes equipment idle time by cause and converts it into opportunity cost. A screening gate selects microanalysis targets based on macro priorities, while the micro layer analyzes NC blocks using axis loads to estimate potential savings. Application to 17 machines over five business days and one CNC lathe cycle yielded 74.7% weekly utilization and approximately KRW 9.7 million in idle cost. Operator allocation waiting (46.6%) and changeover (33.0%) were the principal improvement targets. The estimated ex ante cycle time reduction was 12.8-15.1%, requiring field validation. This integrated framework supports data-driven improvement prioritization in manufacturing SMEs.