Pemodelan Stokastik Hibrida (Physics-Data Driven) menggunakan Proportional Hazards Model (PHM) untuk Prediksi Remaining Useful Life (RUL) Komponen Kritis Mesin Stamping A2

Abstract

The low availability of Stamping Machine A2 at PT XYZ, which only reaches 82% due to high accumulated downtime, necessitates a transformation of maintenance strategies from time-based to predictive maintenance. This study aims to develop a hybrid prognostic model (Grey-Box) to precisely predict the Remaining Useful Life (RUL) of the critical Piston Rod component. The methodology integrates Weibull Distribution statistical analysis and Inverse Power Law (IPL) failure mechanics into a Proportional Hazards Model (PHM) mathematical framework, solved using 100 million Monte Carlo simulation iterations. The results indicate that the Piston Rod is in the infant mortality phase (= 0,783643) and exhibits high failure rate sensitivity to operational load fluctuations with a fatigue exponent (p) of 5,60544. Sensitivity tests demonstrate that increasing the load from normal to extreme conditions drastically reduces the safe remaining life limit (P10 RUL) from 396.4 hours to 208.5 hours at the 2,000-hour observation point. This study concludes that the P10 RUL threshold should serve as the predictive maintenance schedule benchmark to ensure machine reliability remains at 90% and precisely mitigate catastrophic failure risks according to daily production load dynamic.