The Forecasting Approach of Short-term Traffic Volume based on Hierarchical Genetic Algorithm and ANFIS

Adaptive Neural-Fuzzy Inference System(ANFIS) is a perfect method for nonlinear mapping problem which combines neural network and fuzzy inference system. But it only optimizes the internal parameters of the network, not the global structure parameters. In order to solve the problem, an approach to optimize the global structure parameters of ANFIS used by Hierarchical Genetic Algorithm (HGA) is presented. In HGA, every chromosome is coded by hierarchical binary. The genes in chromosome include FIS population, input variables and bit strings of membership function. The demonstration of short-term traffic volume forecasting proves that it can optimize global structure parameters and the results are satisfactory.