RANDOM FOREST-BASED PREDICTIVE MODEL FOR OPTIMIZING ASSET HEALTH MANAGEMENT IN POWER PLANTS

• Main Author: Ghafur Abdulah W.S., Muhammad
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/86666
• Institution: Institut Teknologi Bandung

Effective asset management is crucial in the power plant industry to maintain operational reliability and minimize maintenance costs. The ability to accurately predict asset conditions is key to making informed decisions regarding maintenance and asset replacement. This research aims to develop a predictive model using the Random Forest method for enhancing asset health management and maintenance in power plants operated by a major power generation company in Indonesia. The dataset used includes information such as Asset Criticality Ranking (ACR), ACRRANK, System Criticality Ranking (SCR), Equipment Criticality Ranking (ECR), and Maintenance Priority Index (MPI). The results demonstrate that the Random Forest model achieves a predictive accuracy of 95% in predicting asset conditions. Feature importance analysis reveals that MPI has the most significant influence in predicting asset conditions, followed by ACR, ACRRANK and SCR. This information can be utilized as a basis for decision-making regarding maintenance and asset replacement, as well as identifying areas that require improvement to maintain operational reliability in power plants.