Putra, Dicky Eka (2026) Pengembangan Algoritma untuk Menyelesaikan Dynamic Vehicle Routing Problem With Time Dependent Pickup and Delivery (DVRP-TDPD) dengan Prediksi Waktu Tempuh Berbasis Machine Learning. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5010221072-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5010221072-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (4MB) | Request a copy |
Abstract
Distribusi logistik perkotaan sering kali dipengaruhi oleh ketidakpastian kemacetan, cuaca, dan masuknya pesanan pelanggan secara mendadak. Asumsi waktu tempuh statis pada model Vehicle Routing Problem (VRP) konvensional kurang mampu menggambarkan kondisi sebenarnya karena sering memicu ketidakakuratan jadwal dan pembengkakan denda keterlambatan. Untuk Mengatasi masalah tersebut, penelitian ini mengembangkan sistem optimasi rute dinamis yang mengintegrasikan Machine Learning dan Constraint Programming. Estimasi waktu tempuh diprediksi menggunakan algoritma regresi eXtreme Gradient Boosting (XGBoost) dengan optimasi menggunakan Optuna. Hasil prediksi ini diintegrasikan ke solver OR-Tools untuk memecahkan model Time-Dependent VRP (TDVRP). Untuk mencegah bias antara asumsi waktu keberangkatan dan pembentukan rute (look-ahead bias), sistem mengimplementasikan evaluasi iteratif Successive Approximation. Sementara itu, pesanan yang masuk secara real-time ditangani menggunakan mekanisme Dynamic Injection dengan teknik stitching, sehingga rute baru dapat disisipkan tanpa mengganggu jalur yang sudah dilewati armada. Kinerja sistem dievaluasi berdasarkan kemampuannya meminimalkan total biaya operasional yang mencakup durasi tempuh dan penalti keterlambatan (Soft Time Windows), sekaligus mempertahankan stabilitas rute pengemudi di lapangan.
=================================================================================================================================
The urban logistics sector faces significant challenges due to travel time uncertainty caused by fluctuating traffic congestion and weather conditions, as well as the high dynamism of sudden customer orders. The use of static travel time assumptions in conventional Vehicle Routing Problems (VRP) often yields inaccurate schedules, triggering time window violations, and increasing operational costs due to late penalties. This study aims to design and build a dynamic logistics route optimization system that integrates Machine Learning algorithms and Constraint Programming. Travel time prediction is modeled using the eXtreme Gradient Boosting (XGBoost) regression algorithm, optimized with the Optuna framework. The dynamic time matrix is then integrated into the OR-Tools solver to address the Time-Dependent Vehicle Routing Problem (TDVRP). To solve the interdependency issue between departure times and route formation (look-ahead bias), the system implements an iterative Successive Approximation mechanism. Furthermore, the system is equipped with a Dynamic Injection mechanism utilizing Stitching techniques to accommodate real-time orders without disrupting the track record of routes already traversed by the fleet. Based on this architecture, the system is evaluated on its ability to minimize total operational costs, comprising travel duration and late penalty costs (Soft Time Windows), while maintaining the stability of the drivers' working rhythm in the field.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Dynamic Vehicle Routing Problem, Pickup and Delivery, Prediksi Kemacetan, Machine Learning, Optimasi Rute, Time-Dependent Travel Time, Dynamic Vehicle Routing Problem, Pickup and Delivery, Traffic Congestion Prediction, Machine Learning, Route Optimization, Time-Dependent Travel Time |
| Subjects: | T Technology > T Technology (General) > T57.6 Operations research--Mathematics. Goal programming |
| Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26201-(S1) Undergraduate Thesis |
| Depositing User: | Dicky Eka Putra |
| Date Deposited: | 30 Jul 2026 06:40 |
| Last Modified: | 30 Jul 2026 06:40 |
| URI: | http://repository.its.ac.id/id/eprint/139569 |
Actions (login required)
 |
View Item |