Deep Neural Networks for Traffic Flow Estimation for Spatial–Temporal DOMAIN
International Journal of Software Engineering and Knowledge Engineering, vol.35, no.4, pp.547-567, 2025 (SCI-Expanded, Scopus)
- Publication Type: Article / Article
- Volume: 35 Issue: 4
- Publication Date: 2025
- Doi Number: 10.1142/s0218194025500159
- Journal Name: International Journal of Software Engineering and Knowledge Engineering
- Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Applied Science & Technology Source, Business Source Elite, Business Source Premier, Communication Abstracts, Compendex, Computer & Applied Sciences, INSPEC, Metadex, Civil Engineering Abstracts
- Page Numbers: pp.547-567
- Keywords: deep learning stacked autoencoders, deep neural networks, gated recurrent units, long short-time memory units, Tra±c °ow prediction
- Ankara Yıldırım Beyazıt University Affiliated: Yes
Abstract
Traffic flow estimation is critical for road traffic management. However, the traditional systems measuring procedures demand a significant amount of time and money to continually collect the essential data and keep the associated hardware, like cameras and sensors. On the other hand, deep learning approaches give a scientifically solid framework for modeling the ambiguity and complicated relationships between multiple variables. Accordingly, this study utilized deep learning networks and built a model to estimate traffic flows using anticipated journey times. Stacked Autoencoders (SAEs), Gated Recurrent Units (GRUs) and Long Short-Time Memory units (LSTMs) techniques were specifically used to train the model. A number of experiments were carried out using various time sequence values and compared the estimated traffic flows produced by the suggested model and the actual ones collected from real sensors. The results demonstrate that the suggested model is capable of capturing the general trend of actual traffic flows. Our research proposes building a model to estimate traffic flows using anticipated journey times to capture the general trend of actual traffic loads using a deep learning model. Different experiments are completed using various time sequence values and comparing the estimated traffic flows produced by the suggested models and real data sensors. This work provides a solution using a deep learning model to estimate traffic flow that provides a consummate for e-businesses to use this approach to reduce their transportation cost and increase their customer satisfaction.