Pengembangan Model Prediksi Tingkat Risiko Keselamatan Pengemudi Berbasis Data Telematika Kendaraan Menggunakan Metode Hybrid Fuzzy Logic

Wahyuni, Army Agustina (2026) Pengembangan Model Prediksi Tingkat Risiko Keselamatan Pengemudi Berbasis Data Telematika Kendaraan Menggunakan Metode Hybrid Fuzzy Logic. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032241107-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Keselamatan transportasi darat merupakan aspek krusial dalam industri dengan tingkat mobilitas tinggi, khususnya pada operasi yang melibatkan aktivitas transportasi intensif. Meskipun sistem telematika kendaraan telah banyak diterapkan untuk memantau perilaku pengemudi, penilaian risiko yang dihasilkan umumnya masih bersifat deskriptif berbasis jumlah pelanggaran. Pendekatan ini belum mampu menangkap interaksi multidimensi antar indikator perilaku serta ketidakpastian kondisi operasional, sehingga berpotensi menghasilkan penilaian risiko yang kurang komprehensif dan kurang akurat. Permasalahan utama dalam penelitian ini adalah keterbatasan metode yang ada dalam merepresentasikan tingkat risiko pengemudi secara menyeluruh, terutama dalam mengintegrasikan faktor perilaku dan paparan risiko secara simultan. Oleh karena itu, diperlukan suatu model prediktif yang mampu mengakomodasi karakteristik data telematika yang bersifat tidak pasti, non-linear, dan berbasis konteks. Penelitian ini bertujuan mengembangkan model prediksi tingkat risiko keselamatan pengemudi berbasis data telematika kendaraan dengan pendekatan Hybrid Fuzzy Logic. Model dirancang untuk menggabungkan keunggulan sistem inferensi fuzzy Mamdani dalam menangkap interpretasi linguistik perilaku pengemudi dan metode Sugeno dalam menghasilkan output numerik yang presisi. Metodologi penelitian meliputi tahapan preprocessing, normalisasi, pemilihan variabel perilaku utama, serta pengembangan model Fuzzy Inference System dua tahap. Tahap pertama menghasilkan Behavior Risk Score (BRS) berbasis perilaku, sedangkan tahap kedua menghasilkan Contextual Risk Score (CRS) berbasis paparan risiko. Kedua skor tersebut kemudian diintegrasikan menjadi Driver Overall Risk Score (DORS). Validasi dilakukan melalui pendekatan pakar dan analisis kesesuaian dengan indikator keselamatan berbasis kejadian. Hasil penelitian menunjukkan bahwa model yang dikembangkan mampu menghasilkan skor risiko yang kontinu, sensitif terhadap variasi perilaku, serta memiliki karakteristik stabil, monotonik, dan robust. Selain itu, model mampu memberikan klasifikasi risiko yang lebih adaptif dan memiliki daya pembeda yang lebih baik dibandingkan pendekatan konvensional. Kontribusi penelitian ini adalah pengembangan model prediksi risiko keselamatan pengemudi yang fleksibel, interpretable, dan berbasis data, yang mampu mengintegrasikan faktor perilaku dan konteks secara simultan. Model ini memberikan nilai tambah dalam mendukung transformasi manajemen keselamatan dari pendekatan reaktif menuju predictive safety management yang lebih proaktif serta dapat dimanfaatkan sebagai alat bantu pengambilan keputusan dalam pengelolaan keselamatan transportasi.
================================================================================================================================
Land transportation safety is a critical concern in industries with high mobility, particularly in operations involving intensive transportation activities. Although vehicle telematics systems have been widely implemented to monitor driver behavior, existing risk assessment approaches remain largely descriptive, relying on the number of violations. Such approaches are limited in capturing the multidimensional interactions among behavioral indicators as well as the uncertainty of operational conditions, leading to risk evaluations that are often less comprehensive and less accurate. The main issue addressed in this study is the limitation of existing methods in representing driver risk holistically, particularly in integrating behavioral factors with exposure-related risks. Therefore, there is a need for a predictive model capable of handling the uncertain, non-linear, and context-dependent nature of telematics data. This study aims to develop a driver safety risk prediction model based on vehicle telematics data using a Hybrid Fuzzy Logic approach. The model integrates the strengths of the Mamdani fuzzy inference system in capturing the linguistic interpretation of driver behavior and the Sugeno method in producing precise numerical outputs. The research methodology includes data preprocessing, normalization, selection of key behavioral variables, and the development of a two-stage Fuzzy Inference System. The first stage generates a Behavior Risk Score (BRS) representing driver behavior, while the second stage produces a Contextual Risk Score (CRS) reflecting exposure-related factors. These two components are then integrated into a Driver Overall Risk Score (DORS). Model validation is conducted through expert judgment and by assessing consistency with event-based safety indicators. The results indicate that the proposed model is capable of producing a continuous risk score that is sensitive to behavioral variations, while maintaining stability, monotonicity, and robustness. Furthermore, the model demonstrates a more adaptive risk classification and stronger discriminatory power compared to conventional approaches. This research contributes by proposing a flexible, interpretable, and data-driven driver risk prediction model that effectively integrates behavioral and contextual factors. The model provides added value in supporting the transformation of safety management from a reactive approach toward a more proactive predictive safety management, and can serve as a decision-support tool in transportation safety management.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Fuzzy Logic, Journey Management System, Keselamatan Transportasi Darat, Perilaku Pengemudi, Telematika Kendaraan, Driver Behavior, In-vehicle Telematics, Journey Management System, Road Transportation Safety
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) T Technology > TA Engineering (General). Civil engineering (General) > TA169 Reliability (Engineering) T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL152.5 Motor vehicles Driving
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Army Agustina Wahyuni
Date Deposited:	18 Jun 2026 07:45
Last Modified:	18 Jun 2026 07:45
URI:	http://repository.its.ac.id/id/eprint/133895

Actions (login required)

View Item