An Estimation of Distribution Algorithm based Approach for the Order Batching Problem

Up to 50 per cent of all warehouse operations costs can be attributed to order picking [1]. It is an aspect critical to each supply chain and the planning and control of warehouse processes. According to experience from logistics environments, a reduction of picking times is an integral part of the delivery lead-time and may result in an improvement of the customer service provided by a warehouse. Order batching consists of combining customer orders into picking orders to minimize the picking time. Different evolutionary algorithms haven been proposed to build better batches for the order picking. These algorithms try to preserve batches or sequences of batches that appear frequently in all members of the population in order to keep track and inherit these characteristics exhibited by the parents to the next generation. However, the traditional evolutionary operators used in current research may lose the characteristics mentioned. To address this situation, an estimation of distribution algorithm based approach is proposed for the order batching problem in an order-picking warehouse. A probability model is built to describe the characteristics exhibited by the parents as a distribution of the solution space. In a series of extensive numerical experiments, an acceptable performance using the algorithm mentioned is shown against different evolutionary algorithms known in the literature.