Approach to Sensitivity Analysis of Stochastic Freeway Capacity Model Based on Applying Analysis of Finite Fluctuations

Abstract

The paper introduces approach to Sensitivity Analysis based on applying Analysis of Finite Fluctuations to neural network model showing the dynamics of freeway section capacity depending on several external factors. The presented numerical examples, conducted on data from loop and radar detectors describing the capacity within long-term work zones on sections of German freeways, contain calculated scores on factors significance. There is also given the comparison between the proposed approach and Garson algorithm which is common in Sensitivity Analysis of neural network models; the similarity of both results proves the relevance of applying Analysis of Finite Fluctuations in this field.