The Application of Multicriteria Genetic Algorithms for Signal Setting Design at a Single Junction

Abstract

The purpose of this paper is to solve the Signal Setting Design (SSD) at a single junction. Two methods can be applied in SSD: the monocriteria optimisation in which one objective function is considered and the multicriteria optimisation in which two or more objective functions can be involved in the optimisation. This paper aims at the implementation of the multicriteria Genetic Algorithms (GAs). Two Pareto-based methods are applied to a single "T" junction: the Goldberg's Pareto ranking and the Non dominated Sorting Genetic Algorithm II (NSGA II). The combinations of functions considered for multicriteria optimisation are: (i) the total delay minimisation and the queue length minimisation, (ii) the total delay minimisation and the total number of stops minimisation. Some concluding remarks are made with respect to the effect of population size, crossover rate and mutation rate, with respect to the effectiveness of criteria, with respect to the effectiveness of algorithms.