PREDICTION OF ELECTRICAL METER MATERIAL DEMAND AT PT PLN (PERSERO) USING LONG SHORTTERM MEMORY (LSTM): A CASE STUDY OF WEST JAVA DISTRIBUTION

• Main Author: Rahman Aziz, Toro
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/86681
• Institution: Institut Teknologi Bandung

Supply Chain Management (SCM) is a crucial aspect of PT PLN (Persero)'s business operations to ensure smooth energy distribution across Indonesia. The rapid and diverse market demands make it challenging for PT PLN (Persero) to accurately predict the need for electricity meters, leading to inefficiencies in material planning. The material requirement planning has so far been done by estimating based on the historical usage of electricity meters from the previous year, which is manually filled in by each unit in every city to estimate the material needs for the next three months. This process is carried out manually, and the planned quantity of materials is always overestimated by around 20-30% to account for demand uncertainty. As a result, many unused materials accumulate in the warehouse, increasing storage costs. This study aims to improve inventory management efficiency and reduce costs through more accurate electricity meter demand forecasting using a deep learning-based model, the Long Short-Term Memory (LSTM). LSTM is a deep learning model that has been widely used for time-series data. This prediction is based on the material demand data for electricity meters from 2021 to 2024. The results show that using an LSTM model to forecast material demand led to a 71,32% reduction in storage costs, as the quantities ordered more accurately matched actual demand.