Genetic Approach to Modeling of a Dispatcher in Discrete Transport Systems

Abstract

The essential problem in transport system functional and economic analysis is a technique of modeling the human dispatcher presented in real systems. It is very hard to find an ”intelligent” algorithm of dispatcher – an algorithm which is giving results which significantly differs from pure random algorithms. In a paper we propose an evolutionary approach to this problem. A set of heuristic rules for dispatcher is searched by genetic algorithm. The fitness function is defined by the economic measure. The discrete transport system is modeled using Monte-Carlo simulation. A more accurate model of dispatcher, found by genetic algorithm, allows to obtain more realistic results of functional and economic analysis. The proposed, novelty approach can serve for practical solving of essential decision problems related to an organization and parameters of transport systems.