ABSTRACT
The paper addresses the challenges of daily, weekly, and yearly traffic fluctuations, as well as gradual but uncertain increases in traffic loads on the road network impacting tunnel user’s safety. Since the problem is multidimensional, a test of the machine learning for tunnel operation was conducted for tunnel operation enhancements. The main focus is to evaluate tunnel traffic conditions as they change over time for dynamic traffic control, safety measures efficiency and traffic flow. Since machine learning technologies are based on data, the Cenkova tunnel (Slovenia) data was used for the test. The risk-based data was engineered by implementing the modified tunnel user’s risk model TuRisMo (normally used in Slovenia to confirm tunnel safety according to EU Directive 2004/54/EC). The tested approach offers data integration on a completely new level, enabling other sources and data integration e.g. automatic event detection, C-ITS, IoT-based technologies, weather data, etc.
