IMPLEMENTASI SISTEM VEHICLE ROUTING PROBLEM (VRP) MENGGUNAKAN DEEP Q-NETWORK PADA PENDISTRIBUSIAN

DWIHANGGORO, MUAMAR and Putra, Rakhmadi Irfansyah and Prayitno, Budi (2025) IMPLEMENTASI SISTEM VEHICLE ROUTING PROBLEM (VRP) MENGGUNAKAN DEEP Q-NETWORK PADA PENDISTRIBUSIAN. Diploma thesis, ITPLN.

[thumbnail of 202131036_Muamar Dwihanggoro_Revisi_Skripsi_Muamar Dwihanggoro.pdf] Text
202131036_Muamar Dwihanggoro_Revisi_Skripsi_Muamar Dwihanggoro.pdf
Restricted to Registered users only

Download (8MB)

Abstract

Penelitian ini bertujuan untuk mengimplementasikan dan mengevaluasi algoritma Deep Q-Network (DQN) dalam menyelesaikan Vehicle Routing Problem (VRP) dinamis untuk menentukan rute terbaik dengan mempertimbangkan faktor cuaca dan lalu lintas secara real-time. Data distribusi aktual pada empat rute utama di wilayah Jabodetabek diperkaya dengan integrasi OpenWeatherMap API dan simulasi lalu lintas untuk membentuk state vector berdimensi sembilan. Hasil pelatihan menunjukkan average reward sebesar 149,01, completion rate 100%, dan penurunan Mean Squared Error (MSE) sebesar 80,8% dari awal pelatihan. Pengujian membuktikan model mampu menyesuaikan rute dan estimasi waktu tempuh secara adaptif pada berbagai skenario operasional. Sistem yang di-deploy dalam bentuk dashboard interaktif memberikan visualisasi rute dan rekomendasi optimal, sehingga cocok digunakan sebagai decision support system untuk meningkatkan efektifitas distribusi di wilayah padat lalu lintas.

This study aims to implement and evaluate the Deep Q-Network (DQN) algorithm in solving the dynamic Vehicle Routing Problem (VRP) to determine the best distribution routes by incorporating real-time weather and traffic factors. Actual distribution data from four main routes in the Greater Jakarta area was enriched with OpenWeatherMap API integration and traffic simulation to construct a nine-dimensional state vector. The training results achieved an average reward of 149.01, a completion rate of 100%, and an 80.8% reduction in Mean Squared Error (MSE) from the initial value. Testing demonstrated that the model can adapt routes and travel time estimates dynamically under various operational scenarios. The deployed system, presented as an interactive dashboard, provides route visualization and optimal recommendations, making it suitable as a decision support system to enhance distribution effectiveness in high-traffic areas.

Item Type: Thesis (Diploma)
Uncontrolled Keywords: Vehicle Routing Problem, Deep Q-Network, Reinforcement Learning, Distribusi, Cuaca, Lalu Lintas. Vehicle Routing Problem, Deep Q-Network, Reinforcement Learning, Distribution, Weather, Traffic.
Subjects: Skripsi
Bidang Keilmuan > Teknik Informatika
Divisions: Fakultas Telematika Energi > S1 Teknik Informatika
Depositing User: Sudarman
Date Deposited: 13 Oct 2025 04:34
Last Modified: 13 Oct 2025 04:34
URI: https://repository.itpln.ac.id/id/eprint/2123

Actions (login required)

View Item
View Item