Time delay neural networks designed using genetic algorithms for short term inter-city traffic forecasting

Abstract

Estimation of short-term traffic volumes is an important issue in the development of intelligent transportation systems (ITS). This paper uses genetic algorithms to maximize statistical correlation for selecting connections between input and hidden layers of a time delay neural network for inter-city traffic volume estimations. The predictions for high traffic volume hours using proposed approach reflect a high degree of accuracy.