ARTIFICIAL INTELLIGENCE TECHNIQUES IN THE DEVELOPMENT OF EXPRESSWAY INCIDENT DETECTION SYSTEMS

In this paper, the authors describe how genetic algorithms and probabilistic neural networks can be applied as freeway automatic incident detection (AID) models are developed. AID model development has been hampered by insufficient training data and a lack of model adaptability when site conditions change. The authors propose solutions to these two problems using artificial intelligence techniques. They use genetic algorithms to calibrate system parameters and neural networks to achieve site adaptation potential.