Neural Network Model Based on Fuzzy ARTMAPfor Forecasting of Highway Traffic Data

摘要

In this chapter, a neural network model is presented for forecastingthe average speed values at highway traffic detectors locations using the FuzzyARTMAP theory. The performance of the model is measured by the deviationbetween the speed values provided by the loop detectors and the predictedspeed values. Different Fuzzy ARTMAP configuration cases are analysed intheir training and testing phases. Some ad-hoc mechanisms added to the basicFuzzy ARTMAP structure are also described to improve the entire modelperformance. The achieved results make this model suitable for beingimplemented on advanced traffic management systems (ATMS) and advancedtraveller information system (ATIS).