Some Initial Experiments in Calibrating Emissions Models Using an Evolutionary Algorithm

Abstract

Vehicle routing has traditionally been considered from an optimisation perspective in relation to minimising costs associated with distance travelled and number of vehicles used. Increasingly however, there is a demand to additionally optimise journeys according to the levels of carbon emissions. Incorporating this criterion into an optimisation algorithm necessitates the use of vehicle emission models. Although a number exist, they are often complex to use and customise due to the number of parameters involved. In this paper, we evaluate the use of an Evolutionary Algorithm to calibrate the parameters of a vehicle emissions model against real-world observed emissions data . The calibrated model can then be used with confidence within the fitness function of an optimisation technique on similar data. Initial results obtained suggest that this approach shows promise. The work forms the initial stages of a wider programme of work to investigate the use of nature inspired methods to construct accurate vehicle emissions models for use in green logistics.