A genetic algorithm for vehicle routing problems with stochastic demand and soft time windows

Abstract

We study the stochastic vehicle routing problem with soft time windows (SVRPSTW). Vehicles with limited capacity are routed from the central depot to a set of geographically dispersed customers with unknown demands, predefined presence probability and time windows. The late arrival at the customer is allowed by adding a penalty to the objective value. A mathematical model is developed to describe the behavior of this kind of delivery system. A novel age based genetic scheduling algorithm is proposed as an optimization tool to solve this intractable vehicle routing problem in order to minimize the total cost. The effectiveness of the proposed scheduling algorithm is illustrated by using a set of randomly generated numerical examples. The results indicate that the proposed genetic approach is a simple but effective means for solving these problems