AKBAR, SYUKRI (2026) DETEKSI DAN PREDIKSI PANJANG ANTRIAN KENDARAAN BERBASIS DATA REAL-TIME CCTV DAN CUACA UNTUK PENGENDALIAN LAMPU LALU LINTAS ADAPTIF. Masters thesis, Institut Teknologi Nasional Malang.
![[img]](http://eprints.itn.ac.id/style/images/fileicons/text.png)
Abstract
i ABSTRAK Pertumbuhan jumlah kendaraan bermotor di wilayah perkotaan menyebabkan kemacetan lalu lintas yang semakin serius, khususnya pada jam-jam sibuk. Sistem pengendalian lampu lalu lintas di Indonesia umumnya masih menggunakan sistem fixed time, yaitu pengaturan sinyal dengan durasi waktu yang statis dan tidak mempertimbangkan kondisi lalu lintas aktual secara real-time. Pendekatan ini kurang adaptif terhadap fluktuasi arus kendaraan maupun faktor eksternal seperti cuaca, sehingga sering menimbulkan penumpukan antrian dan peningkatan waktu tunggu kendaraan di simpang. Penelitian ini mengusulkan rancangan sistem pengendalian lampu lalu lintas adaptif berbasis integrasi data real-time dari kamera CCTV dan data cuaca daring. Data kendaraan diperoleh melalui deteksi objek menggunakan metode You Only Look Once version 8 (YOLOv8) untuk mengukur panjang antrian kendaraan pada setiap fase sinyal. Data hasil deteksi kemudian digabungkan dengan data cuaca dari layanan Application Programming Interface (API) OpenWeather dan digunakan sebagai dataset untuk pelatihan model prediksi berbasis Long Short-Term Memory (LSTM). Hasil prediksi panjang antrian dari model LSTM digunakan sebagai input pada simulasi lalu lintas menggunakan perangkat lunak SUMO (Simulation of Urban Mobility), di mana sistem pengendalian lampu lalu lintas adaptif dirancang dengan metode fuzzy logic untuk menyesuaikan durasi lampu hijau secara dinamis sesuai kondisi lalu lintas yang diprediksi. Kinerja sistem adaptif dibandingkan dengan sistem fixed time berdasarkan data observasi lapangan di pertigaan Belimbing–Borobudur, Kota Malang. Evaluasi dilakukan dengan menggunakan parameter waiting time (waktu tunggu rata-rata kendaraan) sebagai indikator efisiensi sistem. Hasil simulasi menunjukkan bahwa sistem pengendalian lampu lalu lintas berbasis fuzzy logic yang memanfaatkan prediksi panjang antrian dari model LSTM mampu menurunkan waiting time kendaraan secara signifikan dibandingkan sistem fixed time. Keunggulan rancangan ini terletak pada integrasi multi-sumber data real-time (CCTV dan cuaca), penerapan model prediksi berbasis deep learning, serta penerapan kontrol cerdas berbasis fuzzy logic yang mendukung pengembangan Smart Traffic Management dalam ekosistem Smart City. Kata Kunci : YOLOv8, LSTM, fuzzy logic, SUMO, fixed time, panjang antrian, waiting time, CCTV, data cuaca, Smart Traffic Management, Smart City ABSTRACT The rapid growth of motor vehicles in urban areas has led to increasingly severe traffic congestion, especially during peak hours. Traffic light control systems in Indonesia generally still employ fixed-time mechanisms, where signal durations remain static and do not account for real-time traffic conditions. This approach is less adaptive to fluctuations in traffic flow and external factors such as weather, often resulting in long vehicle queues and increased waiting times at intersections. This study proposes the design of an adaptive traffic light control system based on the integration of real-time data from CCTV cameras and online weather services. Vehicle data were obtained through object detection using the You Only Look Once version 8 (YOLOv8) algorithm to measure the queue length of vehicles at each signal phase. The detection results were combined with weather data retrieved from the OpenWeather Application Programming Interface (API) and used as a dataset for training a prediction model based on Long Short-Term Memory (LSTM). The predicted queue length from the LSTM model was then used as input in a traffic simulation conducted using SUMO (Simulation of Urban Mobility). Within the simulation, an adaptive traffic light control system was developed using the fuzzy logic method to dynamically adjust green light durations based on predicted queue conditions. The performance of the adaptive system was compared with the fixed-time system using field observation data collected at the Belimbing–Borobudur intersection in Malang City. The evaluation was conducted using average vehicle waiting time as the main performance indicator. The simulation results indicate that the fuzzy logic-based adaptive traffic control system, powered by LSTM queue length prediction, significantly reduces vehicle waiting time compared to the fixed-time system. The advantages of this design lie in the integration of multisource real-time data (CCTV and weather), the application of deep learning-based prediction models, and the use of intelligent fuzzy logic control, which collectively contribute to the development of Smart Traffic Management within the Smart City ecosystem. Keywords : YOLOv8, LSTM, fuzzy logic, SUMO, fixed time, queue length, waiting time, CCTV, weather data, Smart Traffic Management, Smart City
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Keywords : YOLOv8, LSTM, fuzzy logic, SUMO, fixed time, queue length, waiting time, CCTV, weather data, Smart Traffic Management, Smart City |
| Subjects: | Engineering > Electrical Engineering |
| Divisions: | Program Pasca Sarjana > Teknik Elektro S2 > Teknik Elektro S2 (Tesis) |
| Depositing User: | SYUKRI AKBAR |
| Date Deposited: | 26 Feb 2026 01:58 |
| Last Modified: | 26 Feb 2026 01:58 |
| URI: | http://eprints.itn.ac.id/id/eprint/16116 |
Actions (login required)
 |
View Item |
