Green vehicle routing in urban zones – A neuro-fuzzy approach

Abstract

Local city authorities are making a serious effort to expand the number of low-greenhouse gas vehicles (green vehicles) at home. There is no reliable methodology, however, to support the implementation of this passenger transportation concept. In order to optimize the green capacity, a system has been developed to support decision making in urban green vehicle routing. The objective of this paper is to propose a green vehicle distribution model in a public transportation network. The problem has been defined as a problem of non-linear optimization with dispersed input parameters, requiring neuro-fuzzy logic. An adaptive neural network was developed, taking into account the costs to be borne by operators and users, and the environmental parameters along the observed vehicle route. Each input parameter of the neuro-fuzzy model has been placed in a complex context. They were divided into the elements describing in more detail the environmental status, the operator and passenger costs. The advantage of the model is that several factors shaping the input parameters have been taken into consideration. On the other hand, the complexity of urban systems management makes it a considerable challenge, and the surrounding circumstances are difficult to predict accurately. Accordingly, the inputs of the green vehicle model were fuzzified. The Index of Performance (IP) is the output, associated with each branch of the passenger transportation network. The model has been tested on a part of the public transport system in central Belgrade. The results have proven a practical application possible, and a calibration of input parameters allows for full implementation in public transport vehicle routing.