Background. Manufacturing generates abundant production data, yet operational efficiency, environmental sustainability, and operator well-being are typically analyzed in isolation. Static LCA inventories and aggregate KPIs cannot close this gap.

Objective. This work develops an integrated, data-driven framework that couples object-centric process mining, dynamic LCA, multi-objective optimization, and a hybrid decision layer (Fuzzy AHP and Shannon-Entropy weighting with GRA–TOPSIS ranking) to resolve the efficiency–sustainability–ergonomics trade-off.

Methods. The framework is organized in four stages: (i) automated conversion of raw sensor records into event logs; (ii) process discovery that exposes bottlenecks and rework loops; (iii) dynamic impact attribution that separates productive from idle emissions using a time-resolved Hungarian-grid emission factor EF(t), set to 0.275 kg CO₂/kWh during the evening peak window [16:00, 20:00) and 0.200 kg CO₂/kWh off-peak, together with a composite ergonomic risk index; and (iv) multi-objective optimization followed by a hybrid weighting and ranking layer that selects the preferred operating point from the Pareto front.

Results. Applied to a tube manufacturing case, the framework identified bottleneck stations responsible for nearly half of daily emissions, with about one fifth originating from idle states signatures undetectable by static LCA. The selected operating point delivered a +5.64 percentage-point efficiency gain (87.2% → 92.84%) and a −17.01% CO₂ reduction (863.2 → 716.3 kg CO₂/day), while ergonomic risk remained within the nominal band. Sensitivity analysis confirmed the stability of the top-ranked solutions.

Conclusions. The framework provides a replicable Industry 5.0 methodology for jointly optimizing operational, environmental, and social criteria on live event-level data, with extensions toward multi-product scheduling and closed-loop ergonomic feedback. For the TL209 line, the rank-1 configuration P₅ is recommended as the preferred operating point and should be staged into production through the closed-loop MES write-back of the FAHP–Entropy / GRA–TOPSIS selection.