A novel traffic optimization method using GRU based deep neural network for the IoV system.

Area of Science:

• Intelligent Transportation Systems
• Deep Learning Applications
• Urban Mobility

Background:

• China's "Industry 4.0" initiative drives intelligent automobile development.
• Increasing urbanization leads to significant traffic congestion and safety concerns.
• The Internet of Vehicles (IoV) offers a potential solution for urban traffic management.

Purpose of the Study:

• To optimize road network traffic conditions within the IoV environment.
• To enhance traffic efficiency and safety through advanced prediction algorithms.
• To address limitations in existing traffic flow prediction methods.

Main Methods:

• Developed a deep neural network utilizing the GRU model for short-time traffic flow prediction.
• Implemented a fine-grained traffic flow statistics approach tailored for IoV.
• Integrated GRU-trained vehicle data into the statistics algorithm for multi-lane analysis.

Main Results:

• The GRU-based deep learning model significantly improved short-time traffic flow prediction accuracy.
• The fine-grained statistics algorithm effectively counted traffic flow across multiple lanes.
• The IoV environment validation confirmed the algorithm's robust performance in diverse scenarios.

Conclusions:

• The proposed deep learning approach enhances IoV traffic flow prediction accuracy and efficiency.
• Fine-grained traffic statistics enable better real-time traffic status assessment.
• The study provides a valuable tool for optimizing urban traffic management and safety.