PREDICTION OF TREE GROWTH UNDER 500 KVS UNGARAN-PEDAN TRANSMISSION CONDUCTORS USING LONG SHORT-TERM MEMORY METHOD

• Main Author: Mustika Aji, Aziz
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/86939
• Institution: Institut Teknologi Bandung

The reliability of electricity supply is a critical aspect in supporting economic activities and daily life across Indonesia. One of the main challenges is maintaining transmission lines free from physical disturbances, including the growth of tree stands under transmission networks. This study aims to predict the growth of Sengon trees (Albizia chinensis), which grow beneath 500-kilovolt high-voltage transmission networks, using a machine learning approach, specifically Long Short-Term Memory (LSTM), focusing on the Ungaran-Pedan network. Data were collected through the Srintami application database, a transmission network inspection platform that records environmental parameters, including tree height, distance to the transmission line, and soil and weather conditions. The LSTM method was chosen for its ability to effectively process time-series data, enabling the model to learn tree growth patterns from historical data. The research results indicate that the prediction outcomes of the LSTM model are highly dependent on the available dataset. The evaluation of the LSTM algorithm showed an accuracy of 99.82%, while the RF algorithm achieved an accuracy of 98.98%. The MSE value of LSTM was also smaller than that of the RF algorithm. If the available historical data is very limited, the RF model is a more suitable choice as it requires minimal interpolation. These predictions can assist transmission network operators in planning maintenance more proactively, reducing the risk of disturbances caused by tree growth, and improving operational efficiency. This research is the first to apply a machine learning approach to transmission network management, particularly at PT. PLN (Persero). It provides a significant contribution to supporting more reliable and sustainable operational efficiency of transmission networks in Indonesia, while also paving the way for broader applications of artificial intelligence in managing other critical infrastructure.