عنوان مقاله [English]
Short-term prediction of traffic parameters such as traffic flow, speed and occupancy, have an important role in research fields of modern intelligent transportation systems. In this paper, first of all, traffic flow predictability was studied, based on chaos theory, and chaotic properties of traffic volume time-series were evaluated on the basis of Lyapunov exponent. In the prediction module, since the major problem is corrupted and noisy data due to various reasons, effect of corrupted data was reduced by preprocessing techniques. In the next step, suitable classifications were considered, based on the effects of social factors on the traffic volume. In order to predict traffic flow, according to adaptability features, self learning algorithms of neural networks and also comprehensibility of fuzzy rules which are all combined in ANFIS structure were used. The model proposed in this paper is applied to predict the real traffic flow in Ferdowsi Boulevard, Mashhad city, Iran. Comparing the predicted traffic flow value with the flow measured in reality, the results show that the proposed model can predict traffic flow satisfactorily.