Title: Application of interval type-2 fuzzy neural networks to predict short-term traffic flow

Authors: Liang Zhao

Addresses: College of Electrical Engineering, Henan University of Technology, Henan Zhengzhou 450007, China

Abstract: This paper presents a new prediction model based on interval type-2 fuzzy neural network (IT2FNN) and self-organising learning algorithm. Unlike traditional intelligent prediction models, whose structure and parameters must be predetermined by expert experience or professional knowledge, the IT2FNN model determines its own form by the self-organising structure identification and parameter optimisation algorithm. In the structure identification stage, the hierarchical clustering algorithm which includes lower-layer subtractive clustering and upper-layer FCM clustering is employed to determine the size of the IT2FNN predictor. Then, in the parameters optimisation stage, the steepest gradient descent algorithm is also utilised to optimise the free parameters. Finally, two groups normalised traffic flow data, which came from the 3rd ring freeway, Beijing and I880 urban freeway, California are employed to train and evaluate the IT2FNN predictor. Experiment results have illustrated its effectiveness.

Keywords: traffic flow prediction; interval type-2 fuzzy neural networks; IT2FNN; structure identification; parameter learning; short-term traffic flow; fuzzy logic; self-organisation; clustering.

DOI: 10.1504/IJCAT.2012.045843

International Journal of Computer Applications in Technology, 2012 Vol.43 No.1, pp.67 - 75

Published online: 13 Mar 2012 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article