Unusual events in road tunnels (fire, accidents, landslide, etc) entail high risk situations that put people's lives in danger and cause significant material damage. These events are so diverse and infrequent that it is very difficult to take good decisions in situations of limited and contradictory initial information, with quick changes and uncertain development.
The main aim of this project was to develop an automated decision support system in road tunnels in order to improve the management of emergency situations and minimise damage/risk in case of unusual events.
Initially, the project analyses, classifies, typifies and statistically processes data on unusual events in different road tunnels. This will allow the development of mathematical models that can predict impact variables (number of deaths, people who need to be rescued, people who can be evacuated by themselves, etc). This is used as input to computational simulation models for evacuation. These models predict human movement and behaviour during evacuation.
Simulation of evacuation phenomena, with the use of optimisation algorithms, then allows the project to obtain recommendations for decision-making.
Results will provide an interactive system with feedback, allowing the proposals of decisions to be tuned while the event is still ongoing, through the introduction of complementary information about the development and efficacy of the measures adopted before, as an evaluation for future research. Therefore, the system could have the possibility to include a learning module. This learning module will study neural networks.