Short-Term Traffic Flow Prediction Based on Interval Type-2 Fuzzy Neural Networks

Abstract

In this paper, a TSK interval type-2 fuzzy neural network is proposed for predicting the short-term traffic flow. The proposed fuzzy neural network is adaptively organized from the collected short-term traffic flow data. The whole process includes structure identification and parameter learning. In structure identification, the hierarchical fuzzy clustering algorithm performs the training traffic flow data set in order to generate the network structure. After the structure identification is finished, the BP algorithm is adopted to perform the parameter learning. Then the trained fuzzy neural network is employed the collected short-term traffic flow test set and the prediction result verifies that the TSK interval type-2 fuzzy neural network has high prediction accuracy.