Prediksi Pelayanan Petikemas dengan Pendekatan Jaringan Syaraf Tiruan Backpropagation

Abstract

Ports are crucial nodes in the flow of trade and goods distribution, with 40% of 90% of the world's trade routes passing through Indonesia. PT Pelabuhan Indonesia (Persero) or Pelindo is a state-owned enterprise that provides port services. International container handling at Belawan port is managed by PT Belawan New Container Terminal (BNCT). Data on target and actual container handling showed significant differences in 2019 and 2021. In 2019, actual handling exceeded the target by 10.09%, while in 2021, it fell short by 11.92% due to the global economic situation caused by Covid-19. Without analyzing the trends and patterns observed in historical data, it is difficult to develop accurate and responsive planning to meet future needs. This underscores the importance of using historical data-based predictions as a decision support tool in determining realistic and responsive operational targets that can adapt to changing conditions, to minimize the gap between actual and target container handling. This research uses Backpropagation Neural Networks to predict container handling. The goal is to determine the optimal network architecture and its prediction results. This Neural Network method includes training, testing, and finally making predictions. Through 30 research schemes, the architecture with 6 nodes in 1 hidden layer showed the smallest MAPE of 31.54%. The prediction for 2024 is 528,578 TEUs, showing an increase of 15.09% compared to the actual 2023 figure of 459,284 TEUs, indicating a positive growth trend for BNCT