Kohonen map approach for vehicle routing problem with pick-up and delivering

Generally in the vehicle routing problem with delivery and pick-up (VRPDP), we work with a single central depot, a fleet of homogeneous vehicles and a set of customers. Customers are divided into two types: customers with known quantity of goods to be delivered from a depot and customers with a known quantity of goods to be collected to a depot. The aim of this work is, to present a variant of VRPDP and to minimize the vehicle fleet and the sum of travel time, with the restriction that the vehicle must have enough capacity for transporting the commodities to be delivered and those ones picked-up at customers for returning them to the depot. We propose an approach to update the Kohonen map algorithm using unsupervised competitive neural network concepts. The proposed algorithm extends existing research by reinforcing the term bias in order to encourage the network to explore a greater number of feasible solutions and introducing an enhancement through the well-known 2-opt procedure.